Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
49.84% |
157 / 315 |
|
18.75% |
3 / 16 |
CRAP | |
0.00% |
0 / 1 |
| MergeHistory | |
49.84% |
157 / 315 |
|
18.75% |
3 / 16 |
530.57 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
2 | |||
| toProperPageIdentity | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
6 | |||
| getRevisionCount | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
2 | |||
| getMergedRevisionCount | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| authorizeInternal | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
12 | |||
| probablyCanMerge | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
1 | |||
| authorizeMerge | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
1 | |||
| isValidMerge | |
96.00% |
24 / 25 |
|
0.00% |
0 / 1 |
14 | |||
| hasOverlappingTimestamps | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| wouldClobberDestLatest | |
0.00% |
0 / 23 |
|
0.00% |
0 / 1 |
20 | |||
| wouldClobberSourceLatest | |
0.00% |
0 / 10 |
|
0.00% |
0 / 1 |
6 | |||
| merge | |
81.25% |
65 / 80 |
|
0.00% |
0 / 1 |
8.42 | |||
| updateSourcePage | |
0.00% |
0 / 41 |
|
0.00% |
0 / 1 |
72 | |||
| getTimestampLimit | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| getTimeWhere | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| initTimestampLimits | |
71.62% |
53 / 74 |
|
0.00% |
0 / 1 |
12.29 | |||
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * Copyright © 2015 Geoffrey Mon <geofbot@gmail.com> |
| 5 | * |
| 6 | * @license GPL-2.0-or-later |
| 7 | * @file |
| 8 | */ |
| 9 | |
| 10 | namespace MediaWiki\Page; |
| 11 | |
| 12 | use InvalidArgumentException; |
| 13 | use MediaWiki; |
| 14 | use MediaWiki\Content\Content; |
| 15 | use MediaWiki\Content\IContentHandlerFactory; |
| 16 | use MediaWiki\EditPage\SpamChecker; |
| 17 | use MediaWiki\HookContainer\HookContainer; |
| 18 | use MediaWiki\HookContainer\HookRunner; |
| 19 | use MediaWiki\Logging\ManualLogEntry; |
| 20 | use MediaWiki\Message\Message; |
| 21 | use MediaWiki\Permissions\Authority; |
| 22 | use MediaWiki\Permissions\PermissionStatus; |
| 23 | use MediaWiki\Revision\SlotRecord; |
| 24 | use MediaWiki\Status\Status; |
| 25 | use MediaWiki\Storage\PageUpdater; |
| 26 | use MediaWiki\Storage\PageUpdaterFactory; |
| 27 | use MediaWiki\Title\Title; |
| 28 | use MediaWiki\Title\TitleFactory; |
| 29 | use MediaWiki\Title\TitleFormatter; |
| 30 | use MediaWiki\Utils\MWTimestamp; |
| 31 | use MediaWiki\Watchlist\WatchedItemStoreInterface; |
| 32 | use Wikimedia\Assert\Assert; |
| 33 | use Wikimedia\Rdbms\IConnectionProvider; |
| 34 | use Wikimedia\Rdbms\IDatabase; |
| 35 | use Wikimedia\Timestamp\TimestampException; |
| 36 | use Wikimedia\Timestamp\TimestampFormat as TS; |
| 37 | |
| 38 | /** |
| 39 | * Handles the backend logic of merging the histories of two |
| 40 | * pages. |
| 41 | * |
| 42 | * @since 1.27 |
| 43 | */ |
| 44 | class MergeHistory { |
| 45 | |
| 46 | /** Maximum number of revisions that can be merged at once */ |
| 47 | public const REVISION_LIMIT = 5000; |
| 48 | |
| 49 | /** @var ProperPageIdentity Page from which history will be merged */ |
| 50 | protected $source; |
| 51 | |
| 52 | /** @var ProperPageIdentity Page to which history will be merged */ |
| 53 | protected $dest; |
| 54 | |
| 55 | /** @var IDatabase Database that we are using */ |
| 56 | protected $dbw; |
| 57 | |
| 58 | /** @var ?string Timestamp up to which history from the source will be merged */ |
| 59 | private $timestamp; |
| 60 | /** @var ?string Timestamp from which history from the source will be merged */ |
| 61 | private $timestampStart; |
| 62 | |
| 63 | /** |
| 64 | * @var array|false|null SQL WHERE condition that selects source revisions |
| 65 | * to insert into destination. Use ::getTimeWhere to lazy-initialize. |
| 66 | */ |
| 67 | protected $timeWhere = false; |
| 68 | |
| 69 | /** |
| 70 | * @var MWTimestamp|false|null Timestamp upto which history from the source will be merged. |
| 71 | * Use getTimestampLimit to lazily initialize. |
| 72 | */ |
| 73 | protected $timestampLimit = false; |
| 74 | |
| 75 | /** |
| 76 | * @var MWTimestamp|false|null Timestamp upto which history from the source will be merged. |
| 77 | * Use getTimestampLimit to lazily initialize. |
| 78 | */ |
| 79 | protected $timestampStartLimit = false; |
| 80 | |
| 81 | /** |
| 82 | * @var string|null |
| 83 | */ |
| 84 | private $revidLimit = null; |
| 85 | /** |
| 86 | * @var string|null |
| 87 | */ |
| 88 | private $revidStart = null; |
| 89 | |
| 90 | /** @var int Number of revisions merged (for Special:MergeHistory success message) */ |
| 91 | protected $revisionsMerged; |
| 92 | |
| 93 | private IContentHandlerFactory $contentHandlerFactory; |
| 94 | private WatchedItemStoreInterface $watchedItemStore; |
| 95 | private SpamChecker $spamChecker; |
| 96 | private HookRunner $hookRunner; |
| 97 | private PageUpdaterFactory $pageUpdaterFactory; |
| 98 | private TitleFormatter $titleFormatter; |
| 99 | private TitleFactory $titleFactory; |
| 100 | private DeletePageFactory $deletePageFactory; |
| 101 | |
| 102 | /** |
| 103 | * @param PageIdentity $source Page from which history will be merged |
| 104 | * @param PageIdentity $dest Page to which history will be merged |
| 105 | * @param ?string $timestamp Timestamp up to which history from the source will be merged |
| 106 | * @param ?string $timestampStart Timestamp after which history from the source will be merged |
| 107 | * @param IConnectionProvider $dbProvider |
| 108 | * @param IContentHandlerFactory $contentHandlerFactory |
| 109 | * @param WatchedItemStoreInterface $watchedItemStore |
| 110 | * @param SpamChecker $spamChecker |
| 111 | * @param HookContainer $hookContainer |
| 112 | * @param PageUpdaterFactory $pageUpdaterFactory |
| 113 | * @param TitleFormatter $titleFormatter |
| 114 | * @param TitleFactory $titleFactory |
| 115 | * @param DeletePageFactory $deletePageFactory |
| 116 | */ |
| 117 | public function __construct( |
| 118 | PageIdentity $source, |
| 119 | PageIdentity $dest, |
| 120 | ?string $timestamp, |
| 121 | ?string $timestampStart, |
| 122 | IConnectionProvider $dbProvider, |
| 123 | IContentHandlerFactory $contentHandlerFactory, |
| 124 | WatchedItemStoreInterface $watchedItemStore, |
| 125 | SpamChecker $spamChecker, |
| 126 | HookContainer $hookContainer, |
| 127 | PageUpdaterFactory $pageUpdaterFactory, |
| 128 | TitleFormatter $titleFormatter, |
| 129 | TitleFactory $titleFactory, |
| 130 | DeletePageFactory $deletePageFactory |
| 131 | ) { |
| 132 | // Save the parameters |
| 133 | $this->source = self::toProperPageIdentity( $source, '$source' ); |
| 134 | $this->dest = self::toProperPageIdentity( $dest, '$dest' ); |
| 135 | $this->timestamp = $timestamp; |
| 136 | $this->timestampStart = $timestampStart; |
| 137 | |
| 138 | // Get the database |
| 139 | $this->dbw = $dbProvider->getPrimaryDatabase(); |
| 140 | |
| 141 | $this->contentHandlerFactory = $contentHandlerFactory; |
| 142 | $this->watchedItemStore = $watchedItemStore; |
| 143 | $this->spamChecker = $spamChecker; |
| 144 | $this->hookRunner = new HookRunner( $hookContainer ); |
| 145 | $this->pageUpdaterFactory = $pageUpdaterFactory; |
| 146 | $this->titleFormatter = $titleFormatter; |
| 147 | $this->titleFactory = $titleFactory; |
| 148 | $this->deletePageFactory = $deletePageFactory; |
| 149 | } |
| 150 | |
| 151 | private static function toProperPageIdentity( |
| 152 | PageIdentity $page, |
| 153 | string $name |
| 154 | ): ProperPageIdentity { |
| 155 | // Make sure $source and $dest are proper pages |
| 156 | if ( $page instanceof Title ) { |
| 157 | $page = $page->toPageIdentity(); |
| 158 | } |
| 159 | |
| 160 | Assert::parameterType( |
| 161 | ProperPageIdentity::class, |
| 162 | $page, |
| 163 | $name |
| 164 | ); |
| 165 | '@phan-var ProperPageIdentity $page'; |
| 166 | |
| 167 | return $page; |
| 168 | } |
| 169 | |
| 170 | /** |
| 171 | * Get the number of revisions that will be moved |
| 172 | * @return int |
| 173 | */ |
| 174 | public function getRevisionCount() { |
| 175 | $count = $this->dbw->newSelectQueryBuilder() |
| 176 | ->select( '1' ) |
| 177 | ->from( 'revision' ) |
| 178 | ->where( [ 'rev_page' => $this->source->getId(), ...$this->getTimeWhere() ] ) |
| 179 | ->limit( self::REVISION_LIMIT + 1 ) |
| 180 | ->caller( __METHOD__ )->fetchRowCount(); |
| 181 | |
| 182 | return $count; |
| 183 | } |
| 184 | |
| 185 | /** |
| 186 | * Get the number of revisions that were moved |
| 187 | * Used in the SpecialMergeHistory success message |
| 188 | * @return int |
| 189 | */ |
| 190 | public function getMergedRevisionCount() { |
| 191 | return $this->revisionsMerged; |
| 192 | } |
| 193 | |
| 194 | /** |
| 195 | * @param callable $authorizer ( string $action, PageIdentity $target, PermissionStatus $status ) |
| 196 | * @param Authority $performer |
| 197 | * @param string $reason |
| 198 | * @return PermissionStatus |
| 199 | */ |
| 200 | private function authorizeInternal( |
| 201 | callable $authorizer, |
| 202 | Authority $performer, |
| 203 | string $reason |
| 204 | ) { |
| 205 | $status = PermissionStatus::newEmpty(); |
| 206 | |
| 207 | $authorizer( 'edit', $this->source, $status ); |
| 208 | $authorizer( 'edit', $this->dest, $status ); |
| 209 | |
| 210 | // Anti-spam |
| 211 | if ( $this->spamChecker->checkSummary( $reason ) !== false ) { |
| 212 | // This is kind of lame, won't display nice |
| 213 | $status->fatal( 'spamprotectiontext' ); |
| 214 | } |
| 215 | |
| 216 | // Check mergehistory permission |
| 217 | if ( !$performer->isAllowed( 'mergehistory' ) ) { |
| 218 | // User doesn't have the right to merge histories |
| 219 | $status->fatal( 'mergehistory-fail-permission' ); |
| 220 | } |
| 221 | return $status; |
| 222 | } |
| 223 | |
| 224 | /** |
| 225 | * Check whether $performer can execute the merge. |
| 226 | * |
| 227 | * @note this method does not guarantee full permissions check, so it should |
| 228 | * only be used to to decide whether to show a merge form. To authorize the merge |
| 229 | * action use {@link self::authorizeMerge} instead. |
| 230 | * |
| 231 | * @param Authority $performer |
| 232 | * @param string|null $reason |
| 233 | * @return PermissionStatus |
| 234 | */ |
| 235 | public function probablyCanMerge( Authority $performer, ?string $reason = null ): PermissionStatus { |
| 236 | return $this->authorizeInternal( |
| 237 | static function ( string $action, PageIdentity $target, PermissionStatus $status ) use ( $performer ) { |
| 238 | return $performer->probablyCan( $action, $target, $status ); |
| 239 | }, |
| 240 | $performer, |
| 241 | $reason |
| 242 | ); |
| 243 | } |
| 244 | |
| 245 | /** |
| 246 | * Authorize the merge by $performer. |
| 247 | * |
| 248 | * @note this method should be used right before the actual merge is performed. |
| 249 | * To check whether a current performer has the potential to merge the history, |
| 250 | * use {@link self::probablyCanMerge} instead. |
| 251 | * |
| 252 | * @param Authority $performer |
| 253 | * @param string|null $reason |
| 254 | * @return PermissionStatus |
| 255 | */ |
| 256 | public function authorizeMerge( Authority $performer, ?string $reason = null ): PermissionStatus { |
| 257 | return $this->authorizeInternal( |
| 258 | static function ( string $action, PageIdentity $target, PermissionStatus $status ) use ( $performer ) { |
| 259 | return $performer->authorizeWrite( $action, $target, $status ); |
| 260 | }, |
| 261 | $performer, |
| 262 | $reason |
| 263 | ); |
| 264 | } |
| 265 | |
| 266 | /** |
| 267 | * Does various checks that the merge is |
| 268 | * valid. Only things based on the two pages |
| 269 | * should be checked here. |
| 270 | * |
| 271 | * @return Status |
| 272 | */ |
| 273 | public function isValidMerge() { |
| 274 | $status = new Status(); |
| 275 | |
| 276 | // If either article ID is 0, then revisions cannot be reliably selected |
| 277 | if ( $this->source->getId() === 0 ) { |
| 278 | $status->fatal( 'mergehistory-fail-invalid-source' ); |
| 279 | } |
| 280 | if ( $this->dest->getId() === 0 ) { |
| 281 | $status->fatal( 'mergehistory-fail-invalid-dest' ); |
| 282 | } |
| 283 | |
| 284 | // Make sure page aren't the same |
| 285 | if ( $this->source->isSamePageAs( $this->dest ) ) { |
| 286 | $status->fatal( 'mergehistory-fail-self-merge' ); |
| 287 | } |
| 288 | |
| 289 | // Make sure the timestamp is valid |
| 290 | $ts = $this->getTimestampLimit(); |
| 291 | if ( !$ts ) { |
| 292 | $status->fatal( 'mergehistory-fail-bad-timestamp' ); |
| 293 | } |
| 294 | // Now all of the required params are correctly specified so test the trickier cases |
| 295 | if ( $status->isGood() ) { |
| 296 | $tss = $this->timestampStartLimit; |
| 297 | if ( $tss && $tss > $ts ) { |
| 298 | $status->fatal( 'mergehistory-fail-start-after-end', ); |
| 299 | } elseif ( $tss == $ts && $this->revidStart > $this->revidLimit ) { |
| 300 | $status->fatal( 'mergehistory-fail-start-after-end' ); |
| 301 | } |
| 302 | // Check that there are not too many revisions to move |
| 303 | if ( $this->getRevisionCount() > self::REVISION_LIMIT ) { |
| 304 | $status->fatal( 'mergehistory-fail-toobig', Message::numParam( self::REVISION_LIMIT ) ); |
| 305 | } |
| 306 | // Don't allow overlapping timestamps (destination page revisions that meet the timeWhere ) |
| 307 | if ( $this->hasOverlappingTimestamps() ) { |
| 308 | $status->fatal( 'mergehistory-fail-timestamps-overlap' ); |
| 309 | } |
| 310 | // Don't allow changing the current revision of a page via a merge |
| 311 | // (except for a full merge of the entire history) |
| 312 | // as that would require reparsing the page etc. and it's easier to not deal |
| 313 | if ( $this->wouldClobberDestLatest() ) { |
| 314 | $status->fatal( 'mergehistory-fail-change-current-revision' ); |
| 315 | } |
| 316 | if ( $this->wouldClobberSourceLatest() ) { |
| 317 | $status->fatal( 'mergehistory-fail-change-current-revision' ); |
| 318 | } |
| 319 | } |
| 320 | return $status; |
| 321 | } |
| 322 | |
| 323 | private function hasOverlappingTimestamps(): bool { |
| 324 | return $this->dbw->newSelectQueryBuilder() |
| 325 | ->select( '1' ) |
| 326 | ->from( 'revision' ) |
| 327 | ->where( [ 'rev_page' => $this->dest->getId(), ...$this->getTimeWhere() ] ) |
| 328 | ->caller( __METHOD__ )->fetchField(); |
| 329 | } |
| 330 | |
| 331 | private function wouldClobberDestLatest(): bool { |
| 332 | // Return whether page_latest of the source page meets the conditions to move |
| 333 | $row = $this->dbw->newSelectQueryBuilder() |
| 334 | ->select( [ 'rev_id', 'rev_timestamp' ] ) |
| 335 | ->from( 'page' ) |
| 336 | ->join( 'revision', null, 'rev_id = page_latest' ) |
| 337 | ->where( [ 'page_id' => $this->dest->getId() ] ) |
| 338 | ->caller( __METHOD__ )->fetchRow(); |
| 339 | $tsLimitString = $this->timestampLimit !== false ? $this->timestampLimit?->getTimestamp() : null; |
| 340 | $destTsString = ( new MWTimestamp( $row->rev_timestamp ) )->getTimestamp(); |
| 341 | if ( $tsLimitString > $destTsString ) { |
| 342 | // Easy case |
| 343 | return true; |
| 344 | } elseif ( $tsLimitString === $destTsString ) { |
| 345 | // Life is full of suffering; fetch the maximum rev ID that would be merged |
| 346 | $maxRevidMerged = $this->dbw->newSelectQueryBuilder() |
| 347 | ->select( 'MAX(rev_id)' ) |
| 348 | ->from( 'page' ) |
| 349 | ->join( 'revision', null, 'rev_page = page_id' ) |
| 350 | ->where( [ |
| 351 | 'page_id' => $this->source->getId(), |
| 352 | ...$this->getTimeWhere(), |
| 353 | 'rev_timestamp' => $this->dbw->timestamp( $destTsString ) |
| 354 | ] ) |
| 355 | ->caller( __METHOD__ )->fetchField(); |
| 356 | return $maxRevidMerged > $row->rev_id; |
| 357 | } else { |
| 358 | // Easy case |
| 359 | return false; |
| 360 | } |
| 361 | } |
| 362 | |
| 363 | private function wouldClobberSourceLatest(): bool { |
| 364 | if ( !$this->timestampStartLimit ) { |
| 365 | // We don't care; it would merge the entire history |
| 366 | return false; |
| 367 | } |
| 368 | // Return whether page_latest of the source page meets the conditions to move |
| 369 | return $this->dbw->newSelectQueryBuilder() |
| 370 | ->select( 'rev_id' ) |
| 371 | ->from( 'page' ) |
| 372 | ->join( 'revision', null, 'rev_id = page_latest' ) |
| 373 | ->where( [ 'page_id' => $this->source->getId(), |
| 374 | ...$this->getTimeWhere() |
| 375 | ] ) |
| 376 | ->caller( __METHOD__ )->fetchField(); |
| 377 | } |
| 378 | |
| 379 | /** |
| 380 | * Actually attempt the history move |
| 381 | * |
| 382 | * @todo if all versions of page A are moved to B and then a user |
| 383 | * tries to do a reverse-merge via the "unmerge" log link, then page |
| 384 | * A will still be a redirect (as it was after the original merge), |
| 385 | * though it will have the old revisions back from before (as expected). |
| 386 | * The user may have to "undo" the redirect manually to finish the "unmerge". |
| 387 | * Maybe this should delete redirects at the source page of merges? |
| 388 | * |
| 389 | * @param Authority $performer |
| 390 | * @param string $reason |
| 391 | * @return Status status of the history merge |
| 392 | */ |
| 393 | public function merge( Authority $performer, $reason = '' ) { |
| 394 | $status = new Status(); |
| 395 | |
| 396 | // Check validity and permissions required for merge |
| 397 | $validCheck = $this->isValidMerge(); // Check this first to check for null pages |
| 398 | if ( !$validCheck->isOK() ) { |
| 399 | return $validCheck; |
| 400 | } |
| 401 | $permCheck = $this->authorizeMerge( $performer, $reason ); |
| 402 | if ( !$permCheck->isOK() ) { |
| 403 | return Status::wrap( $permCheck ); |
| 404 | } |
| 405 | |
| 406 | $updater = $this->pageUpdaterFactory->newPageUpdater( |
| 407 | $this->source, |
| 408 | $performer->getUser() |
| 409 | ); |
| 410 | |
| 411 | $this->dbw->startAtomic( __METHOD__ ); |
| 412 | $updater->grabParentRevision(); // preserve latest revision for later |
| 413 | |
| 414 | $this->dbw->newUpdateQueryBuilder() |
| 415 | ->update( 'revision' ) |
| 416 | ->set( [ 'rev_page' => $this->dest->getId() ] ) |
| 417 | ->where( [ 'rev_page' => $this->source->getId(), ...$this->getTimeWhere() ] ) |
| 418 | ->caller( __METHOD__ )->execute(); |
| 419 | |
| 420 | // Check if this did anything |
| 421 | $this->revisionsMerged = $this->dbw->affectedRows(); |
| 422 | if ( $this->revisionsMerged < 1 ) { |
| 423 | $this->dbw->endAtomic( __METHOD__ ); |
| 424 | return $status->fatal( 'mergehistory-fail-no-change' ); |
| 425 | } |
| 426 | |
| 427 | $haveRevisions = $this->dbw->newSelectQueryBuilder() |
| 428 | ->from( 'revision' ) |
| 429 | ->where( [ 'rev_page' => $this->source->getId() ] ) |
| 430 | ->forUpdate() |
| 431 | ->caller( __METHOD__ ) |
| 432 | ->fetchRowCount(); |
| 433 | |
| 434 | $legacySource = $this->titleFactory->newFromPageIdentity( $this->source ); |
| 435 | $legacyDest = $this->titleFactory->newFromPageIdentity( $this->dest ); |
| 436 | |
| 437 | // Update source page, histories and invalidate caches |
| 438 | if ( !$haveRevisions ) { |
| 439 | if ( $reason ) { |
| 440 | $revisionComment = wfMessage( |
| 441 | 'mergehistory-comment', |
| 442 | $this->titleFormatter->getPrefixedText( $this->source ), |
| 443 | $this->titleFormatter->getPrefixedText( $this->dest ), |
| 444 | $reason |
| 445 | )->inContentLanguage()->text(); |
| 446 | } else { |
| 447 | $revisionComment = wfMessage( |
| 448 | 'mergehistory-autocomment', |
| 449 | $this->titleFormatter->getPrefixedText( $this->source ), |
| 450 | $this->titleFormatter->getPrefixedText( $this->dest ) |
| 451 | )->inContentLanguage()->text(); |
| 452 | } |
| 453 | |
| 454 | $this->updateSourcePage( $status, $performer, $revisionComment, $updater ); |
| 455 | |
| 456 | // Duplicate watchers of the old article to the new article |
| 457 | $this->watchedItemStore->duplicateAllAssociatedEntries( $this->source, $this->dest ); |
| 458 | } else { |
| 459 | $legacySource->invalidateCache(); |
| 460 | } |
| 461 | $legacyDest->invalidateCache(); |
| 462 | |
| 463 | // Update our logs |
| 464 | $logEntry = new ManualLogEntry( 'merge', 'merge' ); |
| 465 | $logEntry->setPerformer( $performer->getUser() ); |
| 466 | $logEntry->setComment( $reason ); |
| 467 | $logEntry->setTarget( $this->source ); |
| 468 | $srcParams = [ |
| 469 | '4::dest' => $this->titleFormatter->getPrefixedText( $this->dest ), |
| 470 | '5::mergepoint' => $this->getTimestampLimit()->getTimestamp( TS::MW ), |
| 471 | '6::mergerevid' => $this->revidLimit |
| 472 | ]; |
| 473 | if ( $this->timestampStartLimit ) { |
| 474 | $srcParams['7::mergestart'] = $this->timestampStartLimit->getTimestamp( TS::MW ); |
| 475 | $srcParams['8::mergestartid'] = $this->revidStart; |
| 476 | } |
| 477 | $logEntry->setParameters( $srcParams ); |
| 478 | $logId = $logEntry->insert(); |
| 479 | $logEntry->publish( $logId ); |
| 480 | |
| 481 | // And at the destination |
| 482 | // https://phabricator.wikimedia.org/T118132 |
| 483 | $destLog = new ManualLogEntry( 'merge', 'merge-into' ); |
| 484 | $destLog->setPerformer( $performer->getUser() ); |
| 485 | $destLog->setComment( $reason ); |
| 486 | $destLog->setTarget( $this->dest ); |
| 487 | |
| 488 | $destParams = [ |
| 489 | '4::src' => $this->titleFormatter->getPrefixedText( $this->source ), |
| 490 | '5::mergepoint' => $this->getTimestampLimit()->getTimestamp( TS::MW ), |
| 491 | '6::mergerevid' => $this->revidLimit |
| 492 | ]; |
| 493 | if ( $this->timestampStartLimit ) { |
| 494 | $destParams['7::mergestart'] = $this->timestampStartLimit->getTimestamp( TS::MW ); |
| 495 | $destParams['8::mergestartid'] = $this->revidStart; |
| 496 | } |
| 497 | |
| 498 | $destLog->setParameters( $destParams ); |
| 499 | $destId = $destLog->insert(); |
| 500 | $destLog->publish( $destId ); |
| 501 | |
| 502 | $this->hookRunner->onArticleMergeComplete( $legacySource, $legacyDest ); |
| 503 | |
| 504 | $this->dbw->endAtomic( __METHOD__ ); |
| 505 | |
| 506 | return $status; |
| 507 | } |
| 508 | |
| 509 | /** |
| 510 | * Do various cleanup work and updates to the source page. This method |
| 511 | * will only be called if no revision is remaining on the page. |
| 512 | * |
| 513 | * At the end, there would be either a redirect page or a deleted page, |
| 514 | * depending on whether the content model of the page supports redirects or not. |
| 515 | * |
| 516 | * @param Status $status |
| 517 | * @param Authority $performer |
| 518 | * @param string $revisionComment Edit summary for the redirect or empty revision |
| 519 | * to be created in place of the source page |
| 520 | * @param PageUpdater $updater For turning the source page into a redirect |
| 521 | */ |
| 522 | private function updateSourcePage( $status, $performer, $revisionComment, PageUpdater $updater ): void { |
| 523 | $deleteSource = false; |
| 524 | $legacySourceTitle = $this->titleFactory->newFromPageIdentity( $this->source ); |
| 525 | $legacyDestTitle = $this->titleFactory->newFromPageIdentity( $this->dest ); |
| 526 | $sourceModel = $legacySourceTitle->getContentModel(); |
| 527 | $contentHandler = $this->contentHandlerFactory->getContentHandler( $sourceModel ); |
| 528 | |
| 529 | if ( !$contentHandler->supportsRedirects() || ( |
| 530 | // Do not create redirects for wikitext message overrides (T376399). |
| 531 | // Maybe one day they will have a custom content model and this special case won't be needed. |
| 532 | $legacySourceTitle->getNamespace() === NS_MEDIAWIKI && |
| 533 | $legacySourceTitle->getContentModel() === CONTENT_MODEL_WIKITEXT |
| 534 | ) ) { |
| 535 | $deleteSource = true; |
| 536 | $newContent = $contentHandler->makeEmptyContent(); |
| 537 | } else { |
| 538 | $msg = wfMessage( 'mergehistory-redirect-text' )->inContentLanguage()->plain(); |
| 539 | $newContent = $contentHandler->makeRedirectContent( $legacyDestTitle, $msg ); |
| 540 | } |
| 541 | |
| 542 | if ( !$newContent instanceof Content ) { |
| 543 | // Handler supports redirect but cannot create new redirect content? |
| 544 | // Not possible to proceed without Content. |
| 545 | |
| 546 | // @todo. Remove this once there's no evidence it's happening or if it's |
| 547 | // determined all violating handlers have been fixed. |
| 548 | // This is mostly kept because previous code was also blindly checking |
| 549 | // existing of the Content for both content models that supports redirects |
| 550 | // and those that that don't, so it's hard to know what it was masking. |
| 551 | $logger = MediaWiki\Logger\LoggerFactory::getInstance( 'ContentHandler' ); |
| 552 | $logger->warning( |
| 553 | 'ContentHandler for {model} says it supports redirects but failed ' |
| 554 | . 'to return Content object from ContentHandler::makeRedirectContent().' |
| 555 | . ' {value} returned instead.', |
| 556 | [ |
| 557 | 'value' => get_debug_type( $newContent ), |
| 558 | 'model' => $sourceModel |
| 559 | ] |
| 560 | ); |
| 561 | |
| 562 | throw new InvalidArgumentException( |
| 563 | "ContentHandler for '$sourceModel' supports redirects" . |
| 564 | ' but cannot create redirect content during History merge.' |
| 565 | ); |
| 566 | } |
| 567 | |
| 568 | // T263340/T93469: Create revision record to also serve as the page revision. |
| 569 | // This revision will be used to create page content. If the source page's |
| 570 | // content model supports redirects, then it will be the redirect content. |
| 571 | // If the content model does not support redirects, this content will aid |
| 572 | // proper deletion of the page below. |
| 573 | |
| 574 | $updater->setContent( SlotRecord::MAIN, $newContent ) |
| 575 | ->setHints( [ 'suppressDerivedDataUpdates' => $deleteSource ] ) |
| 576 | ->saveRevision( $revisionComment, EDIT_INTERNAL | EDIT_IMPLICIT | EDIT_SILENT ); |
| 577 | |
| 578 | if ( $deleteSource ) { |
| 579 | // T263340/T93469: Delete the source page to prevent errors because its |
| 580 | // revisions are now tied to a different title and its content model |
| 581 | // does not support redirects, so we cannot leave a new revision on it. |
| 582 | // This deletion does not depend on userright but may still fail. If it |
| 583 | // fails, it will be communicated in the status response. |
| 584 | $reason = wfMessage( 'mergehistory-source-deleted-reason' )->inContentLanguage()->plain(); |
| 585 | $delPage = $this->deletePageFactory->newDeletePage( $this->source, $performer ); |
| 586 | $deletionStatus = $delPage->deleteUnsafe( $reason ); |
| 587 | if ( $deletionStatus->isGood() && $delPage->deletionsWereScheduled()[DeletePage::PAGE_BASE] ) { |
| 588 | $deletionStatus->warning( |
| 589 | 'delete-scheduled', |
| 590 | wfEscapeWikiText( $this->titleFormatter->getPrefixedText( $this->source ) ) |
| 591 | ); |
| 592 | } |
| 593 | // Notify callers that the source page has been deleted. |
| 594 | $status->value = 'source-deleted'; |
| 595 | $status->merge( $deletionStatus ); |
| 596 | } |
| 597 | } |
| 598 | |
| 599 | /** |
| 600 | * Get the timestamp upto which history from the source will be merged, |
| 601 | * or null if something went wrong |
| 602 | */ |
| 603 | private function getTimestampLimit(): ?MWTimestamp { |
| 604 | if ( $this->timestampLimit === false ) { |
| 605 | $this->initTimestampLimits(); |
| 606 | } |
| 607 | return $this->timestampLimit; |
| 608 | } |
| 609 | |
| 610 | /** |
| 611 | * Get the SQL WHERE condition that selects source revisions to insert into destination, |
| 612 | * or null if something went wrong |
| 613 | */ |
| 614 | private function getTimeWhere(): array { |
| 615 | if ( $this->timeWhere === false ) { |
| 616 | $this->initTimestampLimits(); |
| 617 | } |
| 618 | return $this->timeWhere; |
| 619 | } |
| 620 | |
| 621 | /** |
| 622 | * Lazily initializes timestamp (and possibly revid) limits and conditions. |
| 623 | */ |
| 624 | private function initTimestampLimits() { |
| 625 | $this->revidLimit = null; |
| 626 | // Get the timestamp pivot condition |
| 627 | try { |
| 628 | if ( $this->timestamp ) { |
| 629 | $parts = explode( '|', $this->timestamp ); |
| 630 | if ( count( $parts ) == 2 ) { |
| 631 | $timestamp = $parts[0]; |
| 632 | $this->revidLimit = $parts[1]; |
| 633 | } else { |
| 634 | $timestamp = $this->timestamp; |
| 635 | } |
| 636 | |
| 637 | $lastWorkingTimestamp = $this->dbw->newSelectQueryBuilder() |
| 638 | ->select( 'MAX(rev_timestamp)' ) |
| 639 | ->from( 'revision' ) |
| 640 | ->where( [ |
| 641 | // If we have a requested timestamp, use the |
| 642 | // latest revision up to that point as the insertion point |
| 643 | $this->dbw->expr( 'rev_timestamp', '<=', $this->dbw->timestamp( $timestamp ) ), |
| 644 | 'rev_page' => $this->source->getId() |
| 645 | ] ) |
| 646 | ->caller( __METHOD__ )->fetchField(); |
| 647 | $mwLastWorkingTimestamp = new MWTimestamp( $lastWorkingTimestamp ); |
| 648 | |
| 649 | $timeInsert = $mwLastWorkingTimestamp; |
| 650 | $this->timestampLimit = $mwLastWorkingTimestamp; |
| 651 | } else { |
| 652 | // If we don't, merge entire source page history into the |
| 653 | // beginning of destination page history |
| 654 | $firstDestTimestamp = $this->dbw->newSelectQueryBuilder() |
| 655 | ->select( 'MIN(rev_timestamp)' ) |
| 656 | ->from( 'revision' ) |
| 657 | ->where( [ 'rev_page' => $this->dest->getId() ] ) |
| 658 | ->caller( __METHOD__ )->fetchField(); |
| 659 | // Get the latest timestamp of the source |
| 660 | $row = $this->dbw->newSelectQueryBuilder() |
| 661 | ->select( [ 'rev_timestamp', 'rev_id' ] ) |
| 662 | ->from( 'page' ) |
| 663 | ->join( 'revision', null, 'page_latest = rev_id' ) |
| 664 | ->where( [ 'page_id' => $this->source->getId() ] ) |
| 665 | ->caller( __METHOD__ )->fetchRow(); |
| 666 | $timeInsert = new MWTimestamp( $firstDestTimestamp ); |
| 667 | if ( $row ) { |
| 668 | $lasttimestamp = new MWTimestamp( $row->rev_timestamp ); |
| 669 | $this->timestampLimit = $lasttimestamp; |
| 670 | $this->revidLimit = $row->rev_id; |
| 671 | } else { |
| 672 | $this->timestampLimit = null; |
| 673 | } |
| 674 | } |
| 675 | // Now parse start time |
| 676 | if ( $this->timestampStart ) { |
| 677 | $parts = explode( '|', $this->timestampStart ); |
| 678 | if ( count( $parts ) == 2 ) { |
| 679 | $timestampStart = $parts[0]; |
| 680 | $this->revidStart = $parts[1]; |
| 681 | } else { |
| 682 | $timestampStart = $this->timestampStart; |
| 683 | } |
| 684 | |
| 685 | $lastWorkingTimestamp = $this->dbw->newSelectQueryBuilder() |
| 686 | ->select( 'MIN(rev_timestamp)' ) |
| 687 | ->from( 'revision' ) |
| 688 | ->where( [ |
| 689 | // If we have a requested timestamp, use the |
| 690 | // earliest revision after that point as the insertion point |
| 691 | $this->dbw->expr( 'rev_timestamp', '>=', $this->dbw->timestamp( $timestampStart ) ), |
| 692 | 'rev_page' => $this->source->getId() |
| 693 | ] ) |
| 694 | ->caller( __METHOD__ )->fetchField(); |
| 695 | // If there is no such revision, then this returns the current time |
| 696 | // which eventually fails with "start-after-end"; not ideal but workable |
| 697 | $this->timestampStartLimit = new MWTimestamp( $lastWorkingTimestamp ); |
| 698 | } else { |
| 699 | $this->timestampStartLimit = null; |
| 700 | } |
| 701 | $dbLimit = $this->dbw->timestamp( $timeInsert->getTimestamp() ); |
| 702 | if ( $this->revidLimit ) { |
| 703 | $endQuery = $this->dbw->buildComparison( '<=', |
| 704 | [ 'rev_timestamp' => $dbLimit, 'rev_id' => $this->revidLimit ] |
| 705 | ); |
| 706 | } else { |
| 707 | $endQuery = $this->dbw->buildComparison( '<=', |
| 708 | [ 'rev_timestamp' => $dbLimit ] |
| 709 | ); |
| 710 | } |
| 711 | if ( $this->timestampStartLimit ) { |
| 712 | $dbLimitStart = $this->dbw->timestamp( $this->timestampStartLimit->getTimestamp() ); |
| 713 | if ( $this->revidStart ) { |
| 714 | $startQuery = $this->dbw->buildComparison( '>=', |
| 715 | [ 'rev_timestamp' => $dbLimitStart, 'rev_id' => $this->revidStart ] |
| 716 | ); |
| 717 | } else { |
| 718 | $startQuery = $this->dbw->buildComparison( '>=', |
| 719 | [ 'rev_timestamp' => $dbLimitStart ] |
| 720 | ); |
| 721 | } |
| 722 | $this->timeWhere = [ $endQuery, $startQuery ]; |
| 723 | } else { |
| 724 | $this->timeWhere = [ $endQuery ]; |
| 725 | } |
| 726 | } catch ( TimestampException ) { |
| 727 | // The timestamp we got is screwed up and merge cannot continue |
| 728 | // This should be detected by $this->isValidMerge() |
| 729 | $this->timestampLimit = null; |
| 730 | $this->timeWhere = null; |
| 731 | $this->timestampStartLimit = null; |
| 732 | } |
| 733 | } |
| 734 | } |