Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
86.40% |
521 / 603 |
|
67.86% |
38 / 56 |
CRAP | |
0.00% |
0 / 1 |
| DerivedPageDataUpdater | |
86.40% |
521 / 603 |
|
67.86% |
38 / 56 |
296.59 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
21 / 21 |
|
100.00% |
1 / 1 |
1 | |||
| setLogger | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setCause | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| setPerformer | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getCauseForTracing | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| doTransition | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| assertTransition | |
50.00% |
1 / 2 |
|
0.00% |
0 / 1 |
2.50 | |||
| isReusableFor | |
91.67% |
22 / 24 |
|
0.00% |
0 / 1 |
24.33 | |||
| setForceEmptyRevision | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
| setArticleCountMethod | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getTitle | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getWikiPage | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getPage | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| pageExisted | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getParentRevision | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
5 | |||
| getOldRevision | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| grabCurrentRevision | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
4 | |||
| isContentPrepared | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| isUpdatePrepared | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
2 | |||
| getPageId | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| isContentDeleted | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
| getRawSlot | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getRawContent | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| usePrimary | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| isCountable | |
100.00% |
16 / 16 |
|
100.00% |
1 / 1 |
8 | |||
| isRedirect | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| revisionIsRedirect | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| prepareContent | |
82.98% |
78 / 94 |
|
0.00% |
0 / 1 |
20.78 | |||
| getRevision | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getRenderedRevision | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| assertHasPageState | |
20.00% |
1 / 5 |
|
0.00% |
0 / 1 |
4.05 | |||
| assertPrepared | |
25.00% |
1 / 4 |
|
0.00% |
0 / 1 |
3.69 | |||
| assertHasRevision | |
25.00% |
1 / 4 |
|
0.00% |
0 / 1 |
3.69 | |||
| isCreation | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| isChange | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| wasRedirect | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
3.03 | |||
| getSlots | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| getRevisionSlotsUpdate | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
3 | |||
| getTouchedSlotRoles | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getModifiedSlotRoles | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getRemovedSlotRoles | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| prepareUpdate | |
71.56% |
78 / 109 |
|
0.00% |
0 / 1 |
48.35 | |||
| getPreparedEdit | |
92.86% |
13 / 14 |
|
0.00% |
0 / 1 |
3.00 | |||
| getSlotParserOutput | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| getParserOutputForMetaData | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getCanonicalParserOutput | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getCanonicalParserOptions | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getSecondaryDataUpdates | |
95.35% |
41 / 43 |
|
0.00% |
0 / 1 |
8 | |||
| shouldGenerateHTMLOnEdit | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
3 | |||
| doUpdates | |
85.48% |
53 / 62 |
|
0.00% |
0 / 1 |
22.35 | |||
| emitEvents | |
80.00% |
4 / 5 |
|
0.00% |
0 / 1 |
2.03 | |||
| getPageRevisionUpdatedEvent | |
94.44% |
34 / 36 |
|
0.00% |
0 / 1 |
9.01 | |||
| triggerParserCacheUpdate | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
2 | |||
| maybeAddRecreateChangeTag | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
2 | |||
| doSecondaryDataUpdates | |
95.24% |
20 / 21 |
|
0.00% |
0 / 1 |
4 | |||
| doParserCacheUpdate | |
100.00% |
19 / 19 |
|
100.00% |
1 / 1 |
3 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * This program is free software; you can redistribute it and/or modify |
| 4 | * it under the terms of the GNU General Public License as published by |
| 5 | * the Free Software Foundation; either version 2 of the License, or |
| 6 | * (at your option) any later version. |
| 7 | * |
| 8 | * This program is distributed in the hope that it will be useful, |
| 9 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 10 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 11 | * GNU General Public License for more details. |
| 12 | * |
| 13 | * You should have received a copy of the GNU General Public License along |
| 14 | * with this program; if not, write to the Free Software Foundation, Inc., |
| 15 | * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. |
| 16 | * http://www.gnu.org/copyleft/gpl.html |
| 17 | * |
| 18 | * @file |
| 19 | */ |
| 20 | |
| 21 | namespace MediaWiki\Storage; |
| 22 | |
| 23 | use InvalidArgumentException; |
| 24 | use LogicException; |
| 25 | use MediaWiki\ChangeTags\ChangeTags; |
| 26 | use MediaWiki\ChangeTags\ChangeTagsStore; |
| 27 | use MediaWiki\Config\ServiceOptions; |
| 28 | use MediaWiki\Content\Content; |
| 29 | use MediaWiki\Content\IContentHandlerFactory; |
| 30 | use MediaWiki\Content\Transform\ContentTransformer; |
| 31 | use MediaWiki\Deferred\DeferrableUpdate; |
| 32 | use MediaWiki\Deferred\DeferredUpdates; |
| 33 | use MediaWiki\Deferred\LinksUpdate\LinksUpdate; |
| 34 | use MediaWiki\Deferred\RefreshSecondaryDataUpdate; |
| 35 | use MediaWiki\Deferred\SiteStatsUpdate; |
| 36 | use MediaWiki\DomainEvent\DomainEventDispatcher; |
| 37 | use MediaWiki\Edit\PreparedEdit; |
| 38 | use MediaWiki\HookContainer\HookContainer; |
| 39 | use MediaWiki\HookContainer\HookRunner; |
| 40 | use MediaWiki\JobQueue\JobQueueGroup; |
| 41 | use MediaWiki\JobQueue\Jobs\ParsoidCachePrewarmJob; |
| 42 | use MediaWiki\Language\Language; |
| 43 | use MediaWiki\MainConfigNames; |
| 44 | use MediaWiki\Page\Event\PageRevisionUpdatedEvent; |
| 45 | use MediaWiki\Page\PageIdentity; |
| 46 | use MediaWiki\Page\ParserOutputAccess; |
| 47 | use MediaWiki\Page\ProperPageIdentity; |
| 48 | use MediaWiki\Page\WikiPage; |
| 49 | use MediaWiki\Page\WikiPageFactory; |
| 50 | use MediaWiki\Parser\ParserCache; |
| 51 | use MediaWiki\Parser\ParserOptions; |
| 52 | use MediaWiki\Parser\ParserOutput; |
| 53 | use MediaWiki\Revision\MutableRevisionRecord; |
| 54 | use MediaWiki\Revision\RenderedRevision; |
| 55 | use MediaWiki\Revision\RevisionRecord; |
| 56 | use MediaWiki\Revision\RevisionRenderer; |
| 57 | use MediaWiki\Revision\RevisionSlots; |
| 58 | use MediaWiki\Revision\RevisionStore; |
| 59 | use MediaWiki\Revision\SlotRecord; |
| 60 | use MediaWiki\Revision\SlotRoleRegistry; |
| 61 | use MediaWiki\Title\Title; |
| 62 | use MediaWiki\User\UserIdentity; |
| 63 | use MediaWiki\Utils\MWTimestamp; |
| 64 | use Psr\Log\LoggerAwareInterface; |
| 65 | use Psr\Log\LoggerInterface; |
| 66 | use Psr\Log\NullLogger; |
| 67 | use Wikimedia\Assert\Assert; |
| 68 | use Wikimedia\ObjectCache\WANObjectCache; |
| 69 | use Wikimedia\Rdbms\IDBAccessObject; |
| 70 | use Wikimedia\Rdbms\ILBFactory; |
| 71 | |
| 72 | /** |
| 73 | * A handle for managing updates for derived page data on edit, import, purge, etc. |
| 74 | * |
| 75 | * @note Avoid direct usage of DerivedPageDataUpdater. |
| 76 | * |
| 77 | * @todo Define interfaces for the different use cases of DerivedPageDataUpdater, particularly |
| 78 | * providing access to post-PST content and ParserOutput to callbacks during revision creation, |
| 79 | * which currently use WikiPage::prepareContentForEdit, and allowing updates to be triggered on |
| 80 | * purge, import, and undeletion, which currently use WikiPage::doEditUpdates() and |
| 81 | * Content::getSecondaryDataUpdates(). |
| 82 | * |
| 83 | * DerivedPageDataUpdater instances are designed to be cached inside a WikiPage instance, |
| 84 | * and re-used by callback code over the course of an update operation. It's a stepping stone |
| 85 | * on the way to a more complete refactoring of WikiPage. |
| 86 | * |
| 87 | * When using a DerivedPageDataUpdater, the following life cycle must be observed: |
| 88 | * grabCurrentRevision (optional), prepareContent (optional), prepareUpdate (required |
| 89 | * for doUpdates). getCanonicalParserOutput, getSlots, and getSecondaryDataUpdates |
| 90 | * require prepareContent or prepareUpdate to have been called first, to initialize the |
| 91 | * DerivedPageDataUpdater. |
| 92 | * |
| 93 | * MCR migration note: this replaces the relevant methods in WikiPage, and covers the use cases |
| 94 | * of PreparedEdit. |
| 95 | * |
| 96 | * @see docs/pageupdater.md for more information. |
| 97 | * |
| 98 | * @internal |
| 99 | * @since 1.32 |
| 100 | * @ingroup Page |
| 101 | */ |
| 102 | class DerivedPageDataUpdater implements LoggerAwareInterface, PreparedUpdate { |
| 103 | |
| 104 | /** |
| 105 | * @var UserIdentity|null |
| 106 | */ |
| 107 | private $user = null; |
| 108 | |
| 109 | /** |
| 110 | * @var WikiPage |
| 111 | */ |
| 112 | private $wikiPage; |
| 113 | |
| 114 | /** |
| 115 | * @var ParserCache |
| 116 | */ |
| 117 | private $parserCache; |
| 118 | |
| 119 | /** |
| 120 | * @var RevisionStore |
| 121 | */ |
| 122 | private $revisionStore; |
| 123 | |
| 124 | /** |
| 125 | * @var Language |
| 126 | */ |
| 127 | private $contLang; |
| 128 | |
| 129 | /** |
| 130 | * @var JobQueueGroup |
| 131 | */ |
| 132 | private $jobQueueGroup; |
| 133 | |
| 134 | /** |
| 135 | * @var ILBFactory |
| 136 | */ |
| 137 | private $loadbalancerFactory; |
| 138 | |
| 139 | /** |
| 140 | * @var HookRunner |
| 141 | */ |
| 142 | private $hookRunner; |
| 143 | |
| 144 | /** |
| 145 | * @var DomainEventDispatcher |
| 146 | */ |
| 147 | private $eventDispatcher; |
| 148 | |
| 149 | /** |
| 150 | * @var LoggerInterface |
| 151 | */ |
| 152 | private $logger; |
| 153 | |
| 154 | /** |
| 155 | * @var string see $wgArticleCountMethod |
| 156 | */ |
| 157 | private $articleCountMethod; |
| 158 | |
| 159 | /** |
| 160 | * Stores (most of) the $options parameter of prepareUpdate(). |
| 161 | * @see prepareUpdate() |
| 162 | * |
| 163 | * @var array |
| 164 | * @phpcs:ignore Generic.Files.LineLength |
| 165 | * @phan-var array{changed:bool,created:bool,moved:bool,cause:string,oldrevision:null|RevisionRecord,triggeringUser:null|UserIdentity,oldredirect:bool|null|string,oldcountable:bool|null|string,causeAction:null|string,causeAgent:null|string,editResult:null|EditResult} |
| 166 | */ |
| 167 | private $options = [ |
| 168 | 'changed' => true, |
| 169 | // newrev is true if prepareUpdate is handling the creation of a new revision, |
| 170 | // as opposed to a null edit or a forced update. |
| 171 | 'newrev' => false, |
| 172 | 'created' => false, |
| 173 | 'oldtitle' => null, |
| 174 | 'oldrevision' => null, |
| 175 | 'oldcountable' => null, |
| 176 | 'oldredirect' => null, |
| 177 | 'triggeringUser' => null, |
| 178 | // causeAction/causeAgent default to 'unknown' but that's handled where it's read, |
| 179 | // to make the life of prepareUpdate() callers easier. |
| 180 | 'causeAction' => null, |
| 181 | 'causeAgent' => null, |
| 182 | 'editResult' => null, |
| 183 | 'rcPatrolStatus' => 0, |
| 184 | 'tags' => [], |
| 185 | 'cause' => 'edit', |
| 186 | 'emitEvents' => true, |
| 187 | ] + PageRevisionUpdatedEvent::DEFAULT_FLAGS; |
| 188 | |
| 189 | /** |
| 190 | * The state of the relevant row in page table before the edit. |
| 191 | * This is determined by the first call to grabCurrentRevision, prepareContent, |
| 192 | * or prepareUpdate (so it is only accessible in 'knows-current' or a later stage). |
| 193 | * If pageState was not initialized when prepareUpdate() is called, prepareUpdate() will |
| 194 | * attempt to emulate the state of the page table before the edit. |
| 195 | * |
| 196 | * Contains the following fields: |
| 197 | * - oldRevision (RevisionRecord|null): the revision that was current before the change |
| 198 | * associated with this update. Might not be set, use getParentRevision(). |
| 199 | * - oldId (int|null): the id of the above revision. 0 if there is no such revision (the change |
| 200 | * was about creating a new page); null if not known (that should not happen). |
| 201 | * - oldIsRedirect (bool|null): whether the page was a redirect before the change. Lazy-loaded, |
| 202 | * can be null; use wasRedirect() instead of direct access. |
| 203 | * - oldCountable (bool|null): whether the page was countable before the change (or null |
| 204 | * if we don't have that information) |
| 205 | * - oldRecord (ExistingPageRecord|null): the page record before the update (or null |
| 206 | * if the page didn't exist) |
| 207 | * |
| 208 | * @var array |
| 209 | */ |
| 210 | private $pageState = null; |
| 211 | |
| 212 | /** |
| 213 | * @var RevisionSlotsUpdate|null |
| 214 | */ |
| 215 | private $slotsUpdate = null; |
| 216 | |
| 217 | /** |
| 218 | * @var RevisionRecord|null |
| 219 | */ |
| 220 | private $parentRevision = null; |
| 221 | |
| 222 | /** |
| 223 | * @var RevisionRecord|null |
| 224 | */ |
| 225 | private $revision = null; |
| 226 | |
| 227 | /** |
| 228 | * @var RenderedRevision |
| 229 | */ |
| 230 | private $renderedRevision = null; |
| 231 | |
| 232 | /** @var ?PageRevisionUpdatedEvent */ |
| 233 | private $pageUpdatedEvent = null; |
| 234 | |
| 235 | /** |
| 236 | * @var RevisionRenderer |
| 237 | */ |
| 238 | private $revisionRenderer; |
| 239 | |
| 240 | /** @var SlotRoleRegistry */ |
| 241 | private $slotRoleRegistry; |
| 242 | |
| 243 | /** |
| 244 | * @var bool Whether null-edits create a revision. |
| 245 | */ |
| 246 | private $forceEmptyRevision = false; |
| 247 | |
| 248 | /** |
| 249 | * A stage identifier for managing the life cycle of this instance. |
| 250 | * Possible stages are 'new', 'knows-current', 'has-content', 'has-revision', and 'done'. |
| 251 | * |
| 252 | * @see docs/pageupdater.md for documentation of the life cycle. |
| 253 | * |
| 254 | * @var string |
| 255 | */ |
| 256 | private $stage = 'new'; |
| 257 | |
| 258 | /** |
| 259 | * Transition table for managing the life cycle of DerivedPageDateUpdater instances. |
| 260 | * |
| 261 | * XXX: Overkill. This is a linear order, we could just count. Names are nice though, |
| 262 | * and constants are also overkill... |
| 263 | * |
| 264 | * @see docs/pageupdater.md for documentation of the life cycle. |
| 265 | */ |
| 266 | private const TRANSITIONS = [ |
| 267 | 'new' => [ |
| 268 | 'new' => true, |
| 269 | 'knows-current' => true, |
| 270 | 'has-content' => true, |
| 271 | 'has-revision' => true, |
| 272 | ], |
| 273 | 'knows-current' => [ |
| 274 | 'knows-current' => true, |
| 275 | 'has-content' => true, |
| 276 | 'has-revision' => true, |
| 277 | ], |
| 278 | 'has-content' => [ |
| 279 | 'has-content' => true, |
| 280 | 'has-revision' => true, |
| 281 | ], |
| 282 | 'has-revision' => [ |
| 283 | 'has-revision' => true, |
| 284 | 'done' => true, |
| 285 | ], |
| 286 | ]; |
| 287 | |
| 288 | /** @var IContentHandlerFactory */ |
| 289 | private $contentHandlerFactory; |
| 290 | |
| 291 | /** @var EditResultCache */ |
| 292 | private $editResultCache; |
| 293 | |
| 294 | /** @var ContentTransformer */ |
| 295 | private $contentTransformer; |
| 296 | |
| 297 | /** @var PageEditStash */ |
| 298 | private $pageEditStash; |
| 299 | |
| 300 | /** @var WANObjectCache */ |
| 301 | private $mainWANObjectCache; |
| 302 | |
| 303 | /** @var bool */ |
| 304 | private $warmParsoidParserCache; |
| 305 | |
| 306 | /** @var bool */ |
| 307 | private $useRcPatrol; |
| 308 | |
| 309 | private ChangeTagsStore $changeTagsStore; |
| 310 | |
| 311 | public function __construct( |
| 312 | ServiceOptions $options, |
| 313 | PageIdentity $page, |
| 314 | RevisionStore $revisionStore, |
| 315 | RevisionRenderer $revisionRenderer, |
| 316 | SlotRoleRegistry $slotRoleRegistry, |
| 317 | ParserCache $parserCache, |
| 318 | JobQueueGroup $jobQueueGroup, |
| 319 | Language $contLang, |
| 320 | ILBFactory $loadbalancerFactory, |
| 321 | IContentHandlerFactory $contentHandlerFactory, |
| 322 | HookContainer $hookContainer, |
| 323 | DomainEventDispatcher $eventDispatcher, |
| 324 | EditResultCache $editResultCache, |
| 325 | ContentTransformer $contentTransformer, |
| 326 | PageEditStash $pageEditStash, |
| 327 | WANObjectCache $mainWANObjectCache, |
| 328 | WikiPageFactory $wikiPageFactory, |
| 329 | ChangeTagsStore $changeTagsStore |
| 330 | ) { |
| 331 | // TODO: Remove this cast eventually |
| 332 | $this->wikiPage = $wikiPageFactory->newFromTitle( $page ); |
| 333 | |
| 334 | $this->parserCache = $parserCache; |
| 335 | $this->revisionStore = $revisionStore; |
| 336 | $this->revisionRenderer = $revisionRenderer; |
| 337 | $this->slotRoleRegistry = $slotRoleRegistry; |
| 338 | $this->jobQueueGroup = $jobQueueGroup; |
| 339 | $this->contLang = $contLang; |
| 340 | // XXX only needed for waiting for replicas to catch up; there should be a narrower |
| 341 | // interface for that. |
| 342 | $this->loadbalancerFactory = $loadbalancerFactory; |
| 343 | $this->contentHandlerFactory = $contentHandlerFactory; |
| 344 | $this->hookRunner = new HookRunner( $hookContainer ); |
| 345 | $this->eventDispatcher = $eventDispatcher; |
| 346 | $this->editResultCache = $editResultCache; |
| 347 | $this->contentTransformer = $contentTransformer; |
| 348 | $this->pageEditStash = $pageEditStash; |
| 349 | $this->mainWANObjectCache = $mainWANObjectCache; |
| 350 | $this->changeTagsStore = $changeTagsStore; |
| 351 | |
| 352 | $this->logger = new NullLogger(); |
| 353 | |
| 354 | $this->warmParsoidParserCache = $options |
| 355 | ->get( MainConfigNames::ParsoidCacheConfig )['WarmParsoidParserCache']; |
| 356 | $this->useRcPatrol = $options |
| 357 | ->get( MainConfigNames::UseRCPatrol ); |
| 358 | } |
| 359 | |
| 360 | public function setLogger( LoggerInterface $logger ) { |
| 361 | $this->logger = $logger; |
| 362 | } |
| 363 | |
| 364 | /** |
| 365 | * Set the cause of the update. Will be used for the PageRevisionUpdatedEvent |
| 366 | * and for tracing/logging in jobs, etc. |
| 367 | * |
| 368 | * @param string $cause See PageRevisionUpdatedEvent::CAUSE_XXX |
| 369 | * |
| 370 | * @return void |
| 371 | */ |
| 372 | public function setCause( string $cause ) { |
| 373 | // 'cause' is for use in PageRevisionUpdatedEvent, 'causeAction' is for |
| 374 | // use in tracing in updates, jobs, and RevisionRenderer. |
| 375 | // Note that PageRevisionUpdatedEvent uses causes like "edit" and "move", but |
| 376 | // the convention for causeAction is to use "page-edit", etc. |
| 377 | $this->options['cause'] = $cause; |
| 378 | $this->options['causeAction'] = 'page-' . $cause; |
| 379 | } |
| 380 | |
| 381 | /** |
| 382 | * Set the performer of the action. |
| 383 | * |
| 384 | * @return void |
| 385 | */ |
| 386 | public function setPerformer( UserIdentity $performer ) { |
| 387 | $this->options['triggeringUser'] = $performer; |
| 388 | $this->options['causeAgent'] = $performer->getName(); |
| 389 | } |
| 390 | |
| 391 | /** |
| 392 | * @return string[] [ $causeAction, $causeAgent ] |
| 393 | */ |
| 394 | private function getCauseForTracing(): array { |
| 395 | return [ |
| 396 | $this->options['causeAction'] ?? 'unknown', |
| 397 | $this->options['causeAgent'] |
| 398 | ?? ( $this->user ? $this->user->getName() : 'unknown' ), |
| 399 | ]; |
| 400 | } |
| 401 | |
| 402 | /** |
| 403 | * Transition function for managing the life cycle of this instances. |
| 404 | * |
| 405 | * @see docs/pageupdater.md for documentation of the life cycle. |
| 406 | * |
| 407 | * @param string $newStage |
| 408 | * @return string the previous stage |
| 409 | */ |
| 410 | private function doTransition( $newStage ) { |
| 411 | $this->assertTransition( $newStage ); |
| 412 | |
| 413 | $oldStage = $this->stage; |
| 414 | $this->stage = $newStage; |
| 415 | |
| 416 | return $oldStage; |
| 417 | } |
| 418 | |
| 419 | /** |
| 420 | * Asserts that a transition to the given stage is possible, without performing it. |
| 421 | * |
| 422 | * @see docs/pageupdater.md for documentation of the life cycle. |
| 423 | * |
| 424 | * @param string $newStage |
| 425 | */ |
| 426 | private function assertTransition( $newStage ) { |
| 427 | if ( empty( self::TRANSITIONS[$this->stage][$newStage] ) ) { |
| 428 | throw new LogicException( "Cannot transition from {$this->stage} to $newStage" ); |
| 429 | } |
| 430 | } |
| 431 | |
| 432 | /** |
| 433 | * Checks whether this DerivedPageDataUpdater can be re-used for running updates targeting |
| 434 | * the given revision. |
| 435 | * |
| 436 | * @param UserIdentity|null $user The user creating the revision in question |
| 437 | * @param RevisionRecord|null $revision New revision (after save, if already saved) |
| 438 | * @param RevisionSlotsUpdate|null $slotsUpdate New content (before PST) |
| 439 | * @param null|int $parentId Parent revision of the edit (use 0 for page creation) |
| 440 | * |
| 441 | * @return bool |
| 442 | */ |
| 443 | public function isReusableFor( |
| 444 | ?UserIdentity $user = null, |
| 445 | ?RevisionRecord $revision = null, |
| 446 | ?RevisionSlotsUpdate $slotsUpdate = null, |
| 447 | $parentId = null |
| 448 | ) { |
| 449 | if ( $revision |
| 450 | && $parentId |
| 451 | && $revision->getParentId() !== $parentId |
| 452 | ) { |
| 453 | throw new InvalidArgumentException( '$parentId should match the parent of $revision' ); |
| 454 | } |
| 455 | |
| 456 | // NOTE: For null revisions, $user may be different from $this->revision->getUser |
| 457 | // and also from $revision->getUser. |
| 458 | // But $user should always match $this->user. |
| 459 | if ( $user && $this->user && $user->getName() !== $this->user->getName() ) { |
| 460 | return false; |
| 461 | } |
| 462 | |
| 463 | if ( $revision && $this->revision && $this->revision->getId() |
| 464 | && $this->revision->getId() !== $revision->getId() |
| 465 | ) { |
| 466 | return false; |
| 467 | } |
| 468 | |
| 469 | if ( $this->pageState |
| 470 | && $revision |
| 471 | && $revision->getParentId() !== null |
| 472 | && $this->pageState['oldId'] !== $revision->getParentId() |
| 473 | ) { |
| 474 | return false; |
| 475 | } |
| 476 | |
| 477 | if ( $this->pageState |
| 478 | && $parentId !== null |
| 479 | && $this->pageState['oldId'] !== $parentId |
| 480 | ) { |
| 481 | return false; |
| 482 | } |
| 483 | |
| 484 | // NOTE: this check is the primary reason for having the $this->slotsUpdate field! |
| 485 | if ( $this->slotsUpdate |
| 486 | && $slotsUpdate |
| 487 | && !$this->slotsUpdate->hasSameUpdates( $slotsUpdate ) |
| 488 | ) { |
| 489 | return false; |
| 490 | } |
| 491 | |
| 492 | if ( $revision |
| 493 | && $this->revision |
| 494 | && !$this->revision->getSlots()->hasSameContent( $revision->getSlots() ) |
| 495 | ) { |
| 496 | return false; |
| 497 | } |
| 498 | |
| 499 | return true; |
| 500 | } |
| 501 | |
| 502 | /** |
| 503 | * Set whether null-edits should create a revision. Enabling this allows the creation of dummy |
| 504 | * revisions ("null revisions") to mark events such as renaming in the page history. |
| 505 | * |
| 506 | * Must not be called once prepareContent() or prepareUpdate() have been called. |
| 507 | * |
| 508 | * @since 1.38 |
| 509 | * @see PageUpdater setForceEmptyRevision |
| 510 | * |
| 511 | * @param bool $forceEmptyRevision |
| 512 | */ |
| 513 | public function setForceEmptyRevision( bool $forceEmptyRevision ) { |
| 514 | if ( $this->revision ) { |
| 515 | throw new LogicException( 'prepareContent() or prepareUpdate() was already called.' ); |
| 516 | } |
| 517 | |
| 518 | $this->forceEmptyRevision = $forceEmptyRevision; |
| 519 | } |
| 520 | |
| 521 | /** |
| 522 | * @param string $articleCountMethod "any" or "link". |
| 523 | * @see $wgArticleCountMethod |
| 524 | */ |
| 525 | public function setArticleCountMethod( $articleCountMethod ) { |
| 526 | $this->articleCountMethod = $articleCountMethod; |
| 527 | } |
| 528 | |
| 529 | /** |
| 530 | * @return Title |
| 531 | */ |
| 532 | private function getTitle() { |
| 533 | // NOTE: eventually, this won't use WikiPage any more |
| 534 | return $this->wikiPage->getTitle(); |
| 535 | } |
| 536 | |
| 537 | /** |
| 538 | * @return WikiPage |
| 539 | */ |
| 540 | private function getWikiPage() { |
| 541 | // NOTE: eventually, this won't use WikiPage any more |
| 542 | return $this->wikiPage; |
| 543 | } |
| 544 | |
| 545 | /** |
| 546 | * Returns the page being updated. |
| 547 | * @since 1.37 |
| 548 | * @return ProperPageIdentity (narrowed to ProperPageIdentity in 1.44) |
| 549 | */ |
| 550 | public function getPage(): ProperPageIdentity { |
| 551 | return $this->wikiPage; |
| 552 | } |
| 553 | |
| 554 | /** |
| 555 | * Determines whether the page being edited already existed. |
| 556 | * Only defined after calling grabCurrentRevision() or prepareContent() or prepareUpdate()! |
| 557 | * |
| 558 | * @return bool |
| 559 | * @throws LogicException if called before grabCurrentRevision |
| 560 | */ |
| 561 | public function pageExisted() { |
| 562 | $this->assertHasPageState( __METHOD__ ); |
| 563 | |
| 564 | return $this->pageState['oldId'] > 0; |
| 565 | } |
| 566 | |
| 567 | /** |
| 568 | * Returns the parent revision of the new revision wrapped by this update. |
| 569 | * If the update is a null-edit, this will return the parent of the current (and new) revision. |
| 570 | * This will return null if the revision wrapped by this update created the page. |
| 571 | * Only defined after calling prepareContent() or prepareUpdate()! |
| 572 | * |
| 573 | * @return RevisionRecord|null the parent revision of the new revision, or null if |
| 574 | * the update created the page. |
| 575 | */ |
| 576 | private function getParentRevision() { |
| 577 | $this->assertPrepared( __METHOD__ ); |
| 578 | |
| 579 | if ( $this->parentRevision ) { |
| 580 | return $this->parentRevision; |
| 581 | } |
| 582 | |
| 583 | if ( !$this->pageState['oldId'] ) { |
| 584 | // If there was no current revision, there is no parent revision, |
| 585 | // since the page didn't exist. |
| 586 | return null; |
| 587 | } |
| 588 | |
| 589 | $oldId = $this->revision->getParentId(); |
| 590 | $flags = $this->usePrimary() ? IDBAccessObject::READ_LATEST : 0; |
| 591 | $this->parentRevision = $oldId |
| 592 | ? $this->revisionStore->getRevisionById( $oldId, $flags ) |
| 593 | : null; |
| 594 | |
| 595 | return $this->parentRevision; |
| 596 | } |
| 597 | |
| 598 | /** |
| 599 | * Returns the revision that was the page's current revision when grabCurrentRevision() |
| 600 | * was first called. |
| 601 | * |
| 602 | * @return RevisionRecord|null the original revision before the update, or null |
| 603 | * if the page did not yet exist. |
| 604 | */ |
| 605 | private function getOldRevision() { |
| 606 | $this->assertPrepared( __METHOD__ ); |
| 607 | return $this->pageState['oldRevision']; |
| 608 | } |
| 609 | |
| 610 | /** |
| 611 | * Returns the revision that was the page's current revision when grabCurrentRevision() |
| 612 | * was first called. |
| 613 | * |
| 614 | * During an edit, that revision will act as the logical parent of the new revision. |
| 615 | * |
| 616 | * Some updates are performed based on the difference between the database state at the |
| 617 | * moment this method is first called, and the state after the edit. |
| 618 | * |
| 619 | * @see docs/pageupdater.md for more information on when thie method can and should be called. |
| 620 | * |
| 621 | * @note After prepareUpdate() was called, grabCurrentRevision() will throw an exception |
| 622 | * to avoid confusion, since the page's current revision is then the new revision after |
| 623 | * the edit, which was presumably passed to prepareUpdate() as the $revision parameter. |
| 624 | * Use getParentRevision() instead to access the revision that is the parent of the |
| 625 | * new revision. |
| 626 | * |
| 627 | * @return RevisionRecord|null the page's current revision, or null if the page does not |
| 628 | * yet exist. |
| 629 | */ |
| 630 | public function grabCurrentRevision() { |
| 631 | if ( $this->pageState ) { |
| 632 | return $this->pageState['oldRevision']; |
| 633 | } |
| 634 | |
| 635 | $this->assertTransition( 'knows-current' ); |
| 636 | |
| 637 | // NOTE: eventually, this won't use WikiPage any more |
| 638 | $wikiPage = $this->getWikiPage(); |
| 639 | |
| 640 | // Do not call WikiPage::clear(), since the caller may already have caused page data |
| 641 | // to be loaded with SELECT FOR UPDATE. Just assert it's loaded now. |
| 642 | $wikiPage->loadPageData( IDBAccessObject::READ_LATEST ); |
| 643 | $current = $wikiPage->getRevisionRecord(); |
| 644 | |
| 645 | $this->pageState = [ |
| 646 | 'oldRevision' => $current, |
| 647 | 'oldId' => $current ? $current->getId() : 0, |
| 648 | 'oldIsRedirect' => $wikiPage->isRedirect(), // NOTE: uses page table |
| 649 | 'oldCountable' => $wikiPage->isCountable(), // NOTE: uses pagelinks table |
| 650 | 'oldRecord' => $wikiPage->exists() ? $wikiPage->toPageRecord() : null, |
| 651 | ]; |
| 652 | |
| 653 | $this->doTransition( 'knows-current' ); |
| 654 | |
| 655 | return $this->pageState['oldRevision']; |
| 656 | } |
| 657 | |
| 658 | /** |
| 659 | * Whether prepareUpdate() or prepareContent() have been called on this instance. |
| 660 | * |
| 661 | * @return bool |
| 662 | */ |
| 663 | public function isContentPrepared() { |
| 664 | return $this->revision !== null; |
| 665 | } |
| 666 | |
| 667 | /** |
| 668 | * Whether prepareUpdate() has been called on this instance. |
| 669 | * |
| 670 | * @note will also return null in case of a null-edit! |
| 671 | * |
| 672 | * @return bool |
| 673 | */ |
| 674 | public function isUpdatePrepared() { |
| 675 | return $this->revision !== null && $this->revision->getId() !== null; |
| 676 | } |
| 677 | |
| 678 | /** |
| 679 | * @return int |
| 680 | */ |
| 681 | private function getPageId() { |
| 682 | // NOTE: eventually, this won't use WikiPage any more |
| 683 | return $this->wikiPage->getId(); |
| 684 | } |
| 685 | |
| 686 | /** |
| 687 | * Whether the content is deleted and thus not visible to the public. |
| 688 | * |
| 689 | * @return bool |
| 690 | */ |
| 691 | public function isContentDeleted() { |
| 692 | if ( $this->revision ) { |
| 693 | return $this->revision->isDeleted( RevisionRecord::DELETED_TEXT ); |
| 694 | } else { |
| 695 | // If the content has not been saved yet, it cannot have been deleted yet. |
| 696 | return false; |
| 697 | } |
| 698 | } |
| 699 | |
| 700 | /** |
| 701 | * Returns the slot, modified or inherited, after PST, with no audience checks applied. |
| 702 | * |
| 703 | * @param string $role slot role name |
| 704 | * |
| 705 | * @throws PageUpdateException If the slot is neither set for update nor inherited from the |
| 706 | * parent revision. |
| 707 | * @return SlotRecord |
| 708 | */ |
| 709 | public function getRawSlot( $role ) { |
| 710 | return $this->getSlots()->getSlot( $role ); |
| 711 | } |
| 712 | |
| 713 | /** |
| 714 | * Returns the content of the given slot, with no audience checks. |
| 715 | * |
| 716 | * @throws PageUpdateException If the slot is neither set for update nor inherited from the |
| 717 | * parent revision. |
| 718 | * @param string $role slot role name |
| 719 | * @return Content |
| 720 | */ |
| 721 | public function getRawContent( string $role ): Content { |
| 722 | return $this->getRawSlot( $role )->getContent(); |
| 723 | } |
| 724 | |
| 725 | private function usePrimary(): bool { |
| 726 | // TODO: can we just set a flag to true in prepareContent()? |
| 727 | return $this->wikiPage->wasLoadedFrom( IDBAccessObject::READ_LATEST ); |
| 728 | } |
| 729 | |
| 730 | public function isCountable(): bool { |
| 731 | // NOTE: Keep in sync with WikiPage::isCountable. |
| 732 | |
| 733 | if ( !$this->getTitle()->isContentPage() ) { |
| 734 | return false; |
| 735 | } |
| 736 | |
| 737 | if ( $this->isContentDeleted() ) { |
| 738 | // This should be irrelevant: countability only applies to the current revision, |
| 739 | // and the current revision is never suppressed. |
| 740 | return false; |
| 741 | } |
| 742 | |
| 743 | if ( $this->isRedirect() ) { |
| 744 | return false; |
| 745 | } |
| 746 | |
| 747 | $hasLinks = null; |
| 748 | |
| 749 | if ( $this->articleCountMethod === 'link' ) { |
| 750 | // NOTE: it would be more appropriate to determine for each slot separately |
| 751 | // whether it has links, and use that information with that slot's |
| 752 | // isCountable() method. However, that would break parity with |
| 753 | // WikiPage::isCountable, which uses the pagelinks table to determine |
| 754 | // whether the current revision has links. |
| 755 | $hasLinks = $this->getParserOutputForMetaData()->hasLinks(); |
| 756 | } |
| 757 | |
| 758 | foreach ( $this->getSlots()->getSlotRoles() as $role ) { |
| 759 | $roleHandler = $this->slotRoleRegistry->getRoleHandler( $role ); |
| 760 | if ( $roleHandler->supportsArticleCount() ) { |
| 761 | $content = $this->getRawContent( $role ); |
| 762 | |
| 763 | if ( $content->isCountable( $hasLinks ) ) { |
| 764 | return true; |
| 765 | } |
| 766 | } |
| 767 | } |
| 768 | |
| 769 | return false; |
| 770 | } |
| 771 | |
| 772 | public function isRedirect(): bool { |
| 773 | // NOTE: main slot determines redirect status |
| 774 | // TODO: MCR: this should be controlled by a PageTypeHandler |
| 775 | $mainContent = $this->getRawContent( SlotRecord::MAIN ); |
| 776 | |
| 777 | return $mainContent->isRedirect(); |
| 778 | } |
| 779 | |
| 780 | /** |
| 781 | * @param RevisionRecord $rev |
| 782 | * |
| 783 | * @return bool |
| 784 | */ |
| 785 | private function revisionIsRedirect( RevisionRecord $rev ) { |
| 786 | // NOTE: main slot determines redirect status |
| 787 | $mainContent = $rev->getMainContentRaw(); |
| 788 | |
| 789 | return $mainContent->isRedirect(); |
| 790 | } |
| 791 | |
| 792 | /** |
| 793 | * Prepare updates based on an update which has not yet been saved. |
| 794 | * |
| 795 | * This may be used to create derived data that is needed when creating a new revision; |
| 796 | * particularly, this makes available the slots of the new revision via the getSlots() |
| 797 | * method, after applying PST and slot inheritance. |
| 798 | * |
| 799 | * The derived data prepared for revision creation may then later be re-used by doUpdates(), |
| 800 | * without the need to re-calculate. |
| 801 | * |
| 802 | * @see docs/pageupdater.md for more information on when thie method can and should be called. |
| 803 | * |
| 804 | * @note Calling this method more than once with the same $slotsUpdate |
| 805 | * has no effect. Calling this method multiple times with different content will cause |
| 806 | * an exception. |
| 807 | * |
| 808 | * @note Calling this method after prepareUpdate() has been called will cause an exception. |
| 809 | * |
| 810 | * @param UserIdentity $user The user to act as context for pre-save transformation (PST). |
| 811 | * @param RevisionSlotsUpdate $slotsUpdate The new content of the slots to be updated |
| 812 | * by this edit, before PST. |
| 813 | * @param bool $useStash Whether to use stashed ParserOutput |
| 814 | */ |
| 815 | public function prepareContent( |
| 816 | UserIdentity $user, |
| 817 | RevisionSlotsUpdate $slotsUpdate, |
| 818 | $useStash = true |
| 819 | ) { |
| 820 | if ( $this->slotsUpdate ) { |
| 821 | if ( !$this->user ) { |
| 822 | throw new LogicException( |
| 823 | 'Unexpected state: $this->slotsUpdate was initialized, ' |
| 824 | . 'but $this->user was not.' |
| 825 | ); |
| 826 | } |
| 827 | |
| 828 | if ( $this->user->getName() !== $user->getName() ) { |
| 829 | throw new LogicException( 'Can\'t call prepareContent() again for different user! ' |
| 830 | . 'Expected ' . $this->user->getName() . ', got ' . $user->getName() |
| 831 | ); |
| 832 | } |
| 833 | |
| 834 | if ( !$this->slotsUpdate->hasSameUpdates( $slotsUpdate ) ) { |
| 835 | throw new LogicException( |
| 836 | 'Can\'t call prepareContent() again with different slot content!' |
| 837 | ); |
| 838 | } |
| 839 | |
| 840 | return; // prepareContent() already done, nothing to do |
| 841 | } |
| 842 | |
| 843 | $this->assertTransition( 'has-content' ); |
| 844 | |
| 845 | $wikiPage = $this->getWikiPage(); // TODO: use only for legacy hooks! |
| 846 | $title = $this->getTitle(); |
| 847 | |
| 848 | $parentRevision = $this->grabCurrentRevision(); |
| 849 | |
| 850 | // The edit may have already been prepared via api.php?action=stashedit |
| 851 | $stashedEdit = false; |
| 852 | |
| 853 | // TODO: MCR: allow output for all slots to be stashed. |
| 854 | if ( $useStash && $slotsUpdate->isModifiedSlot( SlotRecord::MAIN ) ) { |
| 855 | $stashedEdit = $this->pageEditStash->checkCache( |
| 856 | $title, |
| 857 | $slotsUpdate->getModifiedSlot( SlotRecord::MAIN )->getContent(), |
| 858 | $user |
| 859 | ); |
| 860 | } |
| 861 | |
| 862 | $userPopts = ParserOptions::newFromUserAndLang( $user, $this->contLang ); |
| 863 | $userPopts->setRenderReason( $this->options['causeAgent'] ?? 'unknown' ); |
| 864 | |
| 865 | $this->hookRunner->onArticlePrepareTextForEdit( $wikiPage, $userPopts ); |
| 866 | |
| 867 | $this->user = $user; |
| 868 | $this->slotsUpdate = $slotsUpdate; |
| 869 | |
| 870 | if ( $parentRevision ) { |
| 871 | $this->revision = MutableRevisionRecord::newFromParentRevision( $parentRevision ); |
| 872 | } else { |
| 873 | $this->revision = new MutableRevisionRecord( $title ); |
| 874 | } |
| 875 | |
| 876 | // NOTE: user and timestamp must be set, so they can be used for |
| 877 | // {{subst:REVISIONUSER}} and {{subst:REVISIONTIMESTAMP}} in PST! |
| 878 | $this->revision->setTimestamp( MWTimestamp::now( TS_MW ) ); |
| 879 | $this->revision->setUser( $user ); |
| 880 | |
| 881 | // Set up ParserOptions to operate on the new revision |
| 882 | $oldCallback = $userPopts->getCurrentRevisionRecordCallback(); |
| 883 | $userPopts->setCurrentRevisionRecordCallback( |
| 884 | function ( Title $parserTitle, $parser = null ) use ( $title, $oldCallback ) { |
| 885 | if ( $parserTitle->equals( $title ) ) { |
| 886 | return $this->revision; |
| 887 | } else { |
| 888 | return $oldCallback( $parserTitle, $parser ); |
| 889 | } |
| 890 | } |
| 891 | ); |
| 892 | |
| 893 | $pstContentSlots = $this->revision->getSlots(); |
| 894 | |
| 895 | foreach ( $slotsUpdate->getModifiedRoles() as $role ) { |
| 896 | $slot = $slotsUpdate->getModifiedSlot( $role ); |
| 897 | |
| 898 | if ( $slot->isInherited() ) { |
| 899 | // No PST for inherited slots! Note that "modified" slots may still be inherited |
| 900 | // from an earlier version, e.g. for rollbacks. |
| 901 | $pstSlot = $slot; |
| 902 | } elseif ( $role === SlotRecord::MAIN && $stashedEdit ) { |
| 903 | // TODO: MCR: allow PST content for all slots to be stashed. |
| 904 | $pstSlot = SlotRecord::newUnsaved( $role, $stashedEdit->pstContent ); |
| 905 | } else { |
| 906 | $pstContent = $this->contentTransformer->preSaveTransform( |
| 907 | $slot->getContent(), |
| 908 | $title, |
| 909 | $user, |
| 910 | $userPopts |
| 911 | ); |
| 912 | |
| 913 | $pstSlot = SlotRecord::newUnsaved( $role, $pstContent ); |
| 914 | } |
| 915 | |
| 916 | $pstContentSlots->setSlot( $pstSlot ); |
| 917 | } |
| 918 | |
| 919 | foreach ( $slotsUpdate->getRemovedRoles() as $role ) { |
| 920 | $pstContentSlots->removeSlot( $role ); |
| 921 | } |
| 922 | |
| 923 | $this->options['created'] = ( $parentRevision === null ); |
| 924 | $this->options['changed'] = ( $parentRevision === null |
| 925 | || !$pstContentSlots->hasSameContent( $parentRevision->getSlots() ) ); |
| 926 | |
| 927 | $this->doTransition( 'has-content' ); |
| 928 | |
| 929 | if ( !$this->options['changed'] ) { |
| 930 | if ( $this->forceEmptyRevision ) { |
| 931 | // dummy revision, inherit all slots |
| 932 | foreach ( $parentRevision->getSlotRoles() as $role ) { |
| 933 | $this->revision->inheritSlot( $parentRevision->getSlot( $role ) ); |
| 934 | } |
| 935 | } else { |
| 936 | // null-edit, the new revision *is* the old revision. |
| 937 | |
| 938 | // TODO: move this into MutableRevisionRecord |
| 939 | $this->revision->setId( $parentRevision->getId() ); |
| 940 | $this->revision->setTimestamp( $parentRevision->getTimestamp() ); |
| 941 | $this->revision->setPageId( $parentRevision->getPageId() ); |
| 942 | $this->revision->setParentId( $parentRevision->getParentId() ); |
| 943 | $this->revision->setUser( $parentRevision->getUser( RevisionRecord::RAW ) ); |
| 944 | $this->revision->setComment( $parentRevision->getComment( RevisionRecord::RAW ) ); |
| 945 | $this->revision->setMinorEdit( $parentRevision->isMinor() ); |
| 946 | $this->revision->setVisibility( $parentRevision->getVisibility() ); |
| 947 | |
| 948 | // prepareUpdate() is redundant for null-edits (but not for dummy revisions) |
| 949 | $this->doTransition( 'has-revision' ); |
| 950 | } |
| 951 | } else { |
| 952 | $this->parentRevision = $parentRevision; |
| 953 | } |
| 954 | |
| 955 | $renderHints = [ 'use-master' => $this->usePrimary(), 'audience' => RevisionRecord::RAW ]; |
| 956 | |
| 957 | if ( $stashedEdit ) { |
| 958 | /** @var ParserOutput $output */ |
| 959 | $output = $stashedEdit->output; |
| 960 | // TODO: this should happen when stashing the ParserOutput, not now! |
| 961 | $output->setCacheTime( $stashedEdit->timestamp ); |
| 962 | |
| 963 | $renderHints['known-revision-output'] = $output; |
| 964 | |
| 965 | $this->logger->debug( __METHOD__ . ': using stashed edit output...' ); |
| 966 | } |
| 967 | |
| 968 | $renderHints['generate-html'] = $this->shouldGenerateHTMLOnEdit(); |
| 969 | |
| 970 | [ $causeAction, ] = $this->getCauseForTracing(); |
| 971 | $renderHints['causeAction'] = $causeAction; |
| 972 | |
| 973 | // NOTE: we want a canonical rendering, so don't pass $this->user or ParserOptions |
| 974 | // NOTE: the revision is either new or current, so we can bypass audience checks. |
| 975 | $this->renderedRevision = $this->revisionRenderer->getRenderedRevision( |
| 976 | $this->revision, |
| 977 | null, |
| 978 | null, |
| 979 | $renderHints |
| 980 | ); |
| 981 | } |
| 982 | |
| 983 | /** |
| 984 | * Returns the update's target revision - that is, the revision that will be the current |
| 985 | * revision after the update. |
| 986 | * |
| 987 | * @note Callers must treat the returned RevisionRecord's content as immutable, even |
| 988 | * if it is a MutableRevisionRecord instance. Other aspects of a MutableRevisionRecord |
| 989 | * returned from here, such as the user or the comment, may be changed, but may not |
| 990 | * be reflected in ParserOutput until after prepareUpdate() has been called. |
| 991 | * |
| 992 | * @todo This is currently used by PageUpdater::makeNewRevision() to construct an unsaved |
| 993 | * MutableRevisionRecord instance. Introduce something like an UnsavedRevisionFactory service |
| 994 | * for that purpose instead! |
| 995 | * |
| 996 | * @return RevisionRecord |
| 997 | */ |
| 998 | public function getRevision(): RevisionRecord { |
| 999 | $this->assertPrepared( __METHOD__ ); |
| 1000 | return $this->revision; |
| 1001 | } |
| 1002 | |
| 1003 | public function getRenderedRevision(): RenderedRevision { |
| 1004 | $this->assertPrepared( __METHOD__ ); |
| 1005 | |
| 1006 | return $this->renderedRevision; |
| 1007 | } |
| 1008 | |
| 1009 | private function assertHasPageState( string $method ) { |
| 1010 | if ( !$this->pageState ) { |
| 1011 | throw new LogicException( |
| 1012 | 'Must call grabCurrentRevision() or prepareContent() ' |
| 1013 | . 'or prepareUpdate() before calling ' . $method |
| 1014 | ); |
| 1015 | } |
| 1016 | } |
| 1017 | |
| 1018 | private function assertPrepared( string $method ) { |
| 1019 | if ( !$this->revision ) { |
| 1020 | throw new LogicException( |
| 1021 | 'Must call prepareContent() or prepareUpdate() before calling ' . $method |
| 1022 | ); |
| 1023 | } |
| 1024 | } |
| 1025 | |
| 1026 | private function assertHasRevision( string $method ) { |
| 1027 | if ( !$this->revision->getId() ) { |
| 1028 | throw new LogicException( |
| 1029 | 'Must call prepareUpdate() before calling ' . $method |
| 1030 | ); |
| 1031 | } |
| 1032 | } |
| 1033 | |
| 1034 | /** |
| 1035 | * Whether the edit creates the page. |
| 1036 | * |
| 1037 | * @return bool |
| 1038 | */ |
| 1039 | public function isCreation() { |
| 1040 | $this->assertPrepared( __METHOD__ ); |
| 1041 | return $this->options['created']; |
| 1042 | } |
| 1043 | |
| 1044 | /** |
| 1045 | * Whether the content of the current revision after the edit is different from the content of the |
| 1046 | * current revision before the edit. This will return false for a null-edit (no revision created), |
| 1047 | * as well as for a dummy revision (a "null-revision" that has the same content as its parent). |
| 1048 | * |
| 1049 | * @warning at present, dummy revision would return false after prepareContent(), |
| 1050 | * but true after prepareUpdate()! |
| 1051 | * |
| 1052 | * @todo This should probably be fixed. |
| 1053 | * |
| 1054 | * @return bool |
| 1055 | */ |
| 1056 | public function isChange() { |
| 1057 | $this->assertPrepared( __METHOD__ ); |
| 1058 | return $this->options['changed']; |
| 1059 | } |
| 1060 | |
| 1061 | /** |
| 1062 | * Whether the page was a redirect before the edit. |
| 1063 | * |
| 1064 | * @return bool |
| 1065 | */ |
| 1066 | public function wasRedirect() { |
| 1067 | $this->assertHasPageState( __METHOD__ ); |
| 1068 | |
| 1069 | if ( $this->pageState['oldIsRedirect'] === null ) { |
| 1070 | /** @var RevisionRecord $rev */ |
| 1071 | $rev = $this->pageState['oldRevision']; |
| 1072 | if ( $rev ) { |
| 1073 | $this->pageState['oldIsRedirect'] = $this->revisionIsRedirect( $rev ); |
| 1074 | } else { |
| 1075 | $this->pageState['oldIsRedirect'] = false; |
| 1076 | } |
| 1077 | } |
| 1078 | |
| 1079 | return $this->pageState['oldIsRedirect']; |
| 1080 | } |
| 1081 | |
| 1082 | /** |
| 1083 | * Returns the slots of the target revision, after PST. |
| 1084 | * |
| 1085 | * @note Callers must treat the returned RevisionSlots instance as immutable, even |
| 1086 | * if it is a MutableRevisionSlots instance. |
| 1087 | * |
| 1088 | * @return RevisionSlots |
| 1089 | */ |
| 1090 | public function getSlots() { |
| 1091 | $this->assertPrepared( __METHOD__ ); |
| 1092 | return $this->revision->getSlots(); |
| 1093 | } |
| 1094 | |
| 1095 | /** |
| 1096 | * Returns the RevisionSlotsUpdate for this updater. |
| 1097 | * |
| 1098 | * @return RevisionSlotsUpdate |
| 1099 | */ |
| 1100 | private function getRevisionSlotsUpdate() { |
| 1101 | $this->assertPrepared( __METHOD__ ); |
| 1102 | |
| 1103 | if ( !$this->slotsUpdate ) { |
| 1104 | $old = $this->getParentRevision(); |
| 1105 | $this->slotsUpdate = RevisionSlotsUpdate::newFromRevisionSlots( |
| 1106 | $this->revision->getSlots(), |
| 1107 | $old ? $old->getSlots() : null |
| 1108 | ); |
| 1109 | } |
| 1110 | return $this->slotsUpdate; |
| 1111 | } |
| 1112 | |
| 1113 | /** |
| 1114 | * Returns the role names of the slots touched by the new revision, |
| 1115 | * including removed roles. |
| 1116 | * |
| 1117 | * @return string[] |
| 1118 | */ |
| 1119 | public function getTouchedSlotRoles() { |
| 1120 | return $this->getRevisionSlotsUpdate()->getTouchedRoles(); |
| 1121 | } |
| 1122 | |
| 1123 | /** |
| 1124 | * Returns the role names of the slots modified by the new revision, |
| 1125 | * not including removed roles. |
| 1126 | * |
| 1127 | * @return string[] |
| 1128 | */ |
| 1129 | public function getModifiedSlotRoles(): array { |
| 1130 | return $this->getRevisionSlotsUpdate()->getModifiedRoles(); |
| 1131 | } |
| 1132 | |
| 1133 | /** |
| 1134 | * Returns the role names of the slots removed by the new revision. |
| 1135 | * |
| 1136 | * @return string[] |
| 1137 | */ |
| 1138 | public function getRemovedSlotRoles(): array { |
| 1139 | return $this->getRevisionSlotsUpdate()->getRemovedRoles(); |
| 1140 | } |
| 1141 | |
| 1142 | /** |
| 1143 | * Prepare derived data updates targeting the given RevisionRecord. |
| 1144 | * |
| 1145 | * Calling this method requires the given revision to be present in the database. |
| 1146 | * This may be right after a new revision has been created, or when re-generating |
| 1147 | * derived data e.g. in ApiPurge, RefreshLinksJob, and the refreshLinks |
| 1148 | * script. |
| 1149 | * |
| 1150 | * @see docs/pageupdater.md for more information on when thie method can and should be called. |
| 1151 | * |
| 1152 | * @note Calling this method more than once with the same revision has no effect. |
| 1153 | * $options are only used for the first call. Calling this method multiple times with |
| 1154 | * different revisions will cause an exception. |
| 1155 | * |
| 1156 | * @note If grabCurrentRevision() (or prepareContent()) has been called before |
| 1157 | * calling this method, $revision->getParentRevision() has to refer to the revision that |
| 1158 | * was the current revision at the time grabCurrentRevision() was called. |
| 1159 | * |
| 1160 | * @param RevisionRecord $revision |
| 1161 | * @param array $options Array of options. Supports the flags defined by |
| 1162 | * PageRevisionUpdatedEvent. In addition, the following keys are supported used: |
| 1163 | * - oldtitle: PageIdentity, if the page was moved this is the source title (default null) |
| 1164 | * - oldrevision: RevisionRecord object for the pre-update revision (default null) |
| 1165 | * - triggeringUser: The user triggering the update (UserIdentity, defaults to the |
| 1166 | * user who created the revision) |
| 1167 | * - oldredirect: bool, null, or string 'no-change' (default null): |
| 1168 | * - bool: whether the page was counted as a redirect before that |
| 1169 | * revision, only used in changed is true and created is false |
| 1170 | * - null or 'no-change': don't update the redirect status. |
| 1171 | * - oldcountable: bool, null, or string 'no-change' (default null): |
| 1172 | * - bool: whether the page was counted as an article before that |
| 1173 | * revision, only used in changed is true and created is false |
| 1174 | * - null: if created is false, don't update the article count; if created |
| 1175 | * is true, do update the article count |
| 1176 | * - 'no-change': don't update the article count, ever |
| 1177 | * When set to null, pageState['oldCountable'] will be used instead if available. |
| 1178 | * - cause: the reason for the update, see PageRevisionUpdatedEvent::CAUSE_XXX. |
| 1179 | * - known-revision-output: a combined canonical ParserOutput for the revision, perhaps |
| 1180 | * from some cache. The caller is responsible for ensuring that the ParserOutput indeed |
| 1181 | * matched the $rev and $options. This mechanism is intended as a temporary stop-gap, |
| 1182 | * for the time until caches have been changed to store RenderedRevision states instead |
| 1183 | * of ParserOutput objects. (default: null) (since 1.33) |
| 1184 | * - editResult: EditResult object created during the update. Required to perform reverted |
| 1185 | * tag update using RevertedTagUpdateJob. (default: null) (since 1.36) |
| 1186 | */ |
| 1187 | public function prepareUpdate( RevisionRecord $revision, array $options = [] ) { |
| 1188 | Assert::parameter( |
| 1189 | !isset( $options['oldrevision'] ) |
| 1190 | || $options['oldrevision'] instanceof RevisionRecord, |
| 1191 | '$options["oldrevision"]', |
| 1192 | 'must be a RevisionRecord' |
| 1193 | ); |
| 1194 | Assert::parameter( |
| 1195 | !isset( $options['triggeringUser'] ) |
| 1196 | || $options['triggeringUser'] instanceof UserIdentity, |
| 1197 | '$options["triggeringUser"]', |
| 1198 | 'must be a UserIdentity' |
| 1199 | ); |
| 1200 | Assert::parameter( |
| 1201 | !isset( $options['editResult'] ) |
| 1202 | || $options['editResult'] instanceof EditResult, |
| 1203 | '$options["editResult"]', |
| 1204 | 'must be an EditResult' |
| 1205 | ); |
| 1206 | |
| 1207 | if ( !$revision->getId() ) { |
| 1208 | throw new InvalidArgumentException( |
| 1209 | 'Revision must have an ID set for it to be used with prepareUpdate()!' |
| 1210 | ); |
| 1211 | } |
| 1212 | |
| 1213 | if ( !$this->wikiPage->exists() ) { |
| 1214 | // If the ongoing edit is creating the page, the state of $this->wikiPage |
| 1215 | // may be out of whack. This would only happen if the page creation was |
| 1216 | // done using a different WikiPage instance, which shouldn't be the case. |
| 1217 | $this->logger->warning( |
| 1218 | __METHOD__ . ': Reloading page meta-data after page creation', |
| 1219 | [ |
| 1220 | 'page' => (string)$this->wikiPage, |
| 1221 | 'rev_id' => $revision->getId(), |
| 1222 | ] |
| 1223 | ); |
| 1224 | |
| 1225 | $this->wikiPage->clear(); |
| 1226 | $this->wikiPage->loadPageData( IDBAccessObject::READ_LATEST ); |
| 1227 | } |
| 1228 | |
| 1229 | if ( $this->revision && $this->revision->getId() ) { |
| 1230 | if ( $this->revision->getId() === $revision->getId() ) { |
| 1231 | $this->options['changed'] = false; // null-edit |
| 1232 | } else { |
| 1233 | throw new LogicException( |
| 1234 | 'Trying to re-use DerivedPageDataUpdater with revision ' |
| 1235 | . $revision->getId() |
| 1236 | . ', but it\'s already bound to revision ' |
| 1237 | . $this->revision->getId() |
| 1238 | ); |
| 1239 | } |
| 1240 | } |
| 1241 | |
| 1242 | if ( $this->revision |
| 1243 | && !$this->revision->getSlots()->hasSameContent( $revision->getSlots() ) |
| 1244 | ) { |
| 1245 | throw new LogicException( |
| 1246 | 'The revision provided has mismatching content!' |
| 1247 | ); |
| 1248 | } |
| 1249 | |
| 1250 | // Override fields defined in $this->options with values from $options. |
| 1251 | $this->options = array_intersect_key( $options, $this->options ) + $this->options; |
| 1252 | |
| 1253 | if ( $this->revision ) { |
| 1254 | $oldId = $this->pageState['oldId'] ?? 0; |
| 1255 | $this->options['newrev'] = ( $revision->getId() !== $oldId ); |
| 1256 | } elseif ( isset( $this->options['oldrevision'] ) ) { |
| 1257 | /** @var RevisionRecord $oldRev */ |
| 1258 | $oldRev = $this->options['oldrevision']; |
| 1259 | $oldId = $oldRev->getId(); |
| 1260 | $this->options['newrev'] = ( $revision->getId() !== $oldId ); |
| 1261 | } else { |
| 1262 | $oldId = $revision->getParentId(); |
| 1263 | } |
| 1264 | |
| 1265 | if ( $oldId !== null ) { |
| 1266 | // XXX: what if $options['changed'] disagrees? |
| 1267 | // MovePage creates a dummy revision with changed = false! |
| 1268 | // We may want to explicitly distinguish between "no new revision" (null-edit) |
| 1269 | // and "new revision without new content" (dummy revision). |
| 1270 | |
| 1271 | if ( $oldId === $revision->getParentId() ) { |
| 1272 | // NOTE: this may still be a NullRevision! |
| 1273 | // New revision! |
| 1274 | $this->options['changed'] = true; |
| 1275 | } elseif ( $oldId === $revision->getId() ) { |
| 1276 | // Null-edit! |
| 1277 | $this->options['changed'] = false; |
| 1278 | } else { |
| 1279 | // This indicates that calling code has given us the wrong RevisionRecord object |
| 1280 | throw new LogicException( |
| 1281 | 'The RevisionRecord mismatches old revision ID: ' |
| 1282 | . 'Old ID is ' . $oldId |
| 1283 | . ', parent ID is ' . $revision->getParentId() |
| 1284 | . ', revision ID is ' . $revision->getId() |
| 1285 | ); |
| 1286 | } |
| 1287 | } |
| 1288 | |
| 1289 | // If prepareContent() was used to generate the PST content (which is indicated by |
| 1290 | // $this->slotsUpdate being set), and this is not a null-edit, then the given |
| 1291 | // revision must have the acting user as the revision author. Otherwise, user |
| 1292 | // signatures generated by PST would mismatch the user in the revision record. |
| 1293 | if ( $this->user !== null && $this->options['changed'] && $this->slotsUpdate ) { |
| 1294 | $user = $revision->getUser(); |
| 1295 | if ( !$this->user->equals( $user ) ) { |
| 1296 | throw new LogicException( |
| 1297 | 'The RevisionRecord provided has a mismatching actor: expected ' |
| 1298 | . $this->user->getName() |
| 1299 | . ', got ' |
| 1300 | . $user->getName() |
| 1301 | ); |
| 1302 | } |
| 1303 | } |
| 1304 | |
| 1305 | // If $this->pageState was not yet initialized by grabCurrentRevision or prepareContent, |
| 1306 | // emulate the state of the page table before the edit, as good as we can. |
| 1307 | if ( !$this->pageState ) { |
| 1308 | $this->pageState = [ |
| 1309 | 'oldIsRedirect' => isset( $this->options['oldredirect'] ) |
| 1310 | && is_bool( $this->options['oldredirect'] ) |
| 1311 | ? $this->options['oldredirect'] |
| 1312 | : null, |
| 1313 | 'oldCountable' => isset( $this->options['oldcountable'] ) |
| 1314 | && is_bool( $this->options['oldcountable'] ) |
| 1315 | ? $this->options['oldcountable'] |
| 1316 | : null, |
| 1317 | ]; |
| 1318 | |
| 1319 | if ( $this->options['changed'] ) { |
| 1320 | // The edit created a new revision |
| 1321 | $this->pageState['oldId'] = $revision->getParentId(); |
| 1322 | // Old revision is null if this is a page creation |
| 1323 | $this->pageState['oldRevision'] = $this->options['oldrevision'] ?? null; |
| 1324 | } else { |
| 1325 | // This is a null-edit, so the old revision IS the new revision! |
| 1326 | $this->pageState['oldId'] = $revision->getId(); |
| 1327 | $this->pageState['oldRevision'] = $revision; |
| 1328 | } |
| 1329 | } |
| 1330 | |
| 1331 | // "created" is forced here |
| 1332 | $this->options['created'] = ( $this->options['created'] || |
| 1333 | ( $this->pageState['oldId'] === 0 ) ); |
| 1334 | |
| 1335 | $this->revision = $revision; |
| 1336 | |
| 1337 | $this->doTransition( 'has-revision' ); |
| 1338 | |
| 1339 | // NOTE: in case we have a User object, don't override with a UserIdentity. |
| 1340 | // We already checked that $revision->getUser() matches $this->user; |
| 1341 | if ( !$this->user ) { |
| 1342 | $this->user = $revision->getUser( RevisionRecord::RAW ); |
| 1343 | } |
| 1344 | |
| 1345 | // Prune any output that depends on the revision ID. |
| 1346 | if ( $this->renderedRevision ) { |
| 1347 | $this->renderedRevision->updateRevision( $revision ); |
| 1348 | } else { |
| 1349 | [ $causeAction, ] = $this->getCauseForTracing(); |
| 1350 | // NOTE: we want a canonical rendering, so don't pass $this->user or ParserOptions |
| 1351 | // NOTE: the revision is either new or current, so we can bypass audience checks. |
| 1352 | $this->renderedRevision = $this->revisionRenderer->getRenderedRevision( |
| 1353 | $this->revision, |
| 1354 | null, |
| 1355 | null, |
| 1356 | [ |
| 1357 | 'use-master' => $this->usePrimary(), |
| 1358 | 'audience' => RevisionRecord::RAW, |
| 1359 | 'known-revision-output' => $options['known-revision-output'] ?? null, |
| 1360 | 'causeAction' => $causeAction |
| 1361 | ] |
| 1362 | ); |
| 1363 | |
| 1364 | // XXX: Since we presumably are dealing with the current revision, |
| 1365 | // we could try to get the ParserOutput from the parser cache. |
| 1366 | } |
| 1367 | |
| 1368 | // TODO: optionally get ParserOutput from the ParserCache here. |
| 1369 | // Move the logic used by RefreshLinksJob here! |
| 1370 | } |
| 1371 | |
| 1372 | /** |
| 1373 | * @deprecated since 1.43; This only exists for B/C, use the getters on DerivedPageDataUpdater directly! |
| 1374 | * @return PreparedEdit |
| 1375 | */ |
| 1376 | public function getPreparedEdit() { |
| 1377 | $this->assertPrepared( __METHOD__ ); |
| 1378 | |
| 1379 | $slotsUpdate = $this->getRevisionSlotsUpdate(); |
| 1380 | $preparedEdit = new PreparedEdit(); |
| 1381 | |
| 1382 | $preparedEdit->popts = $this->getCanonicalParserOptions(); |
| 1383 | $preparedEdit->parserOutputCallback = [ $this, 'getCanonicalParserOutput' ]; |
| 1384 | $preparedEdit->pstContent = $this->revision->getContent( SlotRecord::MAIN ); |
| 1385 | $preparedEdit->newContent = |
| 1386 | $slotsUpdate->isModifiedSlot( SlotRecord::MAIN ) |
| 1387 | ? $slotsUpdate->getModifiedSlot( SlotRecord::MAIN )->getContent() |
| 1388 | : $this->revision->getContent( SlotRecord::MAIN ); // XXX: can we just remove this? |
| 1389 | $preparedEdit->oldContent = null; // unused. // XXX: could get this from the parent revision |
| 1390 | $preparedEdit->revid = $this->revision ? $this->revision->getId() : null; |
| 1391 | $preparedEdit->format = $preparedEdit->pstContent->getDefaultFormat(); |
| 1392 | |
| 1393 | return $preparedEdit; |
| 1394 | } |
| 1395 | |
| 1396 | /** |
| 1397 | * @param string $role |
| 1398 | * @param bool $generateHtml |
| 1399 | * @return ParserOutput |
| 1400 | */ |
| 1401 | public function getSlotParserOutput( $role, $generateHtml = true ) { |
| 1402 | return $this->getRenderedRevision()->getSlotParserOutput( |
| 1403 | $role, |
| 1404 | [ 'generate-html' => $generateHtml ] |
| 1405 | ); |
| 1406 | } |
| 1407 | |
| 1408 | /** |
| 1409 | * @since 1.37 |
| 1410 | * @return ParserOutput |
| 1411 | */ |
| 1412 | public function getParserOutputForMetaData(): ParserOutput { |
| 1413 | return $this->getRenderedRevision()->getRevisionParserOutput( [ 'generate-html' => false ] ); |
| 1414 | } |
| 1415 | |
| 1416 | /** |
| 1417 | * @inheritDoc |
| 1418 | * @return ParserOutput |
| 1419 | */ |
| 1420 | public function getCanonicalParserOutput(): ParserOutput { |
| 1421 | return $this->getRenderedRevision()->getRevisionParserOutput(); |
| 1422 | } |
| 1423 | |
| 1424 | public function getCanonicalParserOptions(): ParserOptions { |
| 1425 | return $this->getRenderedRevision()->getOptions(); |
| 1426 | } |
| 1427 | |
| 1428 | /** |
| 1429 | * @param bool $recursive |
| 1430 | * |
| 1431 | * @return DeferrableUpdate[] |
| 1432 | */ |
| 1433 | public function getSecondaryDataUpdates( $recursive = false ) { |
| 1434 | if ( $this->isContentDeleted() ) { |
| 1435 | // This shouldn't happen, since the current content is always public, |
| 1436 | // and DataUpdates are only needed for current content. |
| 1437 | return []; |
| 1438 | } |
| 1439 | |
| 1440 | $wikiPage = $this->getWikiPage(); |
| 1441 | $wikiPage->loadPageData( IDBAccessObject::READ_LATEST ); |
| 1442 | if ( !$wikiPage->exists() ) { |
| 1443 | // page deleted while deferring the update |
| 1444 | return []; |
| 1445 | } |
| 1446 | |
| 1447 | $title = $wikiPage->getTitle(); |
| 1448 | $allUpdates = []; |
| 1449 | $parserOutput = $this->shouldGenerateHTMLOnEdit() ? |
| 1450 | $this->getCanonicalParserOutput() : $this->getParserOutputForMetaData(); |
| 1451 | |
| 1452 | // Construct a LinksUpdate for the combined canonical output. |
| 1453 | $linksUpdate = new LinksUpdate( |
| 1454 | $title, |
| 1455 | $parserOutput, |
| 1456 | $recursive, |
| 1457 | // Redirect target may have changed if the page is or was a redirect. |
| 1458 | // (We can't check if it was definitely changed without additional queries.) |
| 1459 | $this->isRedirect() || $this->wasRedirect() |
| 1460 | ); |
| 1461 | if ( $this->options['cause'] === PageRevisionUpdatedEvent::CAUSE_MOVE ) { |
| 1462 | // @phan-suppress-next-line PhanTypeMismatchArgument Oldtitle is set along with moved |
| 1463 | $linksUpdate->setMoveDetails( $this->options['oldtitle'] ); |
| 1464 | } |
| 1465 | |
| 1466 | $allUpdates[] = $linksUpdate; |
| 1467 | // NOTE: Run updates for all slots, not just the modified slots! Otherwise, |
| 1468 | // info for an inherited slot may end up being removed. This is also needed |
| 1469 | // to ensure that purges are effective. |
| 1470 | $renderedRevision = $this->getRenderedRevision(); |
| 1471 | |
| 1472 | foreach ( $this->getSlots()->getSlotRoles() as $role ) { |
| 1473 | $slot = $this->getRawSlot( $role ); |
| 1474 | $content = $slot->getContent(); |
| 1475 | $handler = $content->getContentHandler(); |
| 1476 | |
| 1477 | $updates = $handler->getSecondaryDataUpdates( |
| 1478 | $title, |
| 1479 | $content, |
| 1480 | $role, |
| 1481 | $renderedRevision |
| 1482 | ); |
| 1483 | |
| 1484 | $allUpdates = array_merge( $allUpdates, $updates ); |
| 1485 | } |
| 1486 | |
| 1487 | // XXX: if a slot was removed by an earlier edit, but deletion updates failed to run at |
| 1488 | // that time, we don't know for which slots to run deletion updates when purging a page. |
| 1489 | // We'd have to examine the entire history of the page to determine that. Perhaps there |
| 1490 | // could be a "try extra hard" mode for that case that would run a DB query to find all |
| 1491 | // roles/models ever used on the page. On the other hand, removing slots should be quite |
| 1492 | // rare, so perhaps this isn't worth the trouble. |
| 1493 | |
| 1494 | // TODO: consolidate with similar logic in WikiPage::getDeletionUpdates() |
| 1495 | $parentRevision = $this->getParentRevision(); |
| 1496 | foreach ( $this->getRemovedSlotRoles() as $role ) { |
| 1497 | // HACK: we should get the content model of the removed slot from a SlotRoleHandler! |
| 1498 | // For now, find the slot in the parent revision - if the slot was removed, it should |
| 1499 | // always exist in the parent revision. |
| 1500 | $parentSlot = $parentRevision->getSlot( $role, RevisionRecord::RAW ); |
| 1501 | $content = $parentSlot->getContent(); |
| 1502 | $handler = $content->getContentHandler(); |
| 1503 | |
| 1504 | $updates = $handler->getDeletionUpdates( |
| 1505 | $title, |
| 1506 | $role |
| 1507 | ); |
| 1508 | |
| 1509 | $allUpdates = array_merge( $allUpdates, $updates ); |
| 1510 | } |
| 1511 | |
| 1512 | // TODO: hard deprecate SecondaryDataUpdates in favor of RevisionDataUpdates in 1.33! |
| 1513 | $this->hookRunner->onRevisionDataUpdates( $title, $renderedRevision, $allUpdates ); |
| 1514 | |
| 1515 | return $allUpdates; |
| 1516 | } |
| 1517 | |
| 1518 | /** |
| 1519 | * @return bool true if at least one of slots require rendering HTML on edit, false otherwise. |
| 1520 | * This is needed for example in populating ParserCache. |
| 1521 | */ |
| 1522 | private function shouldGenerateHTMLOnEdit(): bool { |
| 1523 | foreach ( $this->getSlots()->getSlotRoles() as $role ) { |
| 1524 | $slot = $this->getRawSlot( $role ); |
| 1525 | $contentHandler = $this->contentHandlerFactory->getContentHandler( $slot->getModel() ); |
| 1526 | if ( $contentHandler->generateHTMLOnEdit() ) { |
| 1527 | return true; |
| 1528 | } |
| 1529 | } |
| 1530 | return false; |
| 1531 | } |
| 1532 | |
| 1533 | /** |
| 1534 | * Do standard updates after page edit, purge, or import. |
| 1535 | * Update links tables and other derived data. |
| 1536 | * Purges pages that depend on this page when appropriate. |
| 1537 | * With a 10% chance, triggers pruning the recent changes table. |
| 1538 | * |
| 1539 | * Further updates may be triggered by core components and extensions |
| 1540 | * that listen to the PageRevisionUpdated event. Search for method names starting |
| 1541 | * with "handlePageUpdatedEvent" to find listeners. |
| 1542 | * |
| 1543 | * @note prepareUpdate() must be called before calling this method! |
| 1544 | * |
| 1545 | * MCR migration note: this replaces WikiPage::doEditUpdates. |
| 1546 | */ |
| 1547 | public function doUpdates() { |
| 1548 | $this->assertTransition( 'done' ); |
| 1549 | |
| 1550 | if ( $this->options['emitEvents'] ) { |
| 1551 | $this->emitEvents(); |
| 1552 | } |
| 1553 | |
| 1554 | // TODO: move more logic into ingress objects subscribed to PageRevisionUpdatedEvent! |
| 1555 | $event = $this->getPageRevisionUpdatedEvent(); |
| 1556 | |
| 1557 | if ( $this->shouldGenerateHTMLOnEdit() ) { |
| 1558 | $this->triggerParserCacheUpdate(); |
| 1559 | } |
| 1560 | |
| 1561 | $this->doSecondaryDataUpdates( [ |
| 1562 | // T52785 do not update any other pages on dummy revisions and null edits |
| 1563 | 'recursive' => $event->isEffectiveContentChange(), |
| 1564 | // Defer the getCanonicalParserOutput() call made by getSecondaryDataUpdates() |
| 1565 | 'defer' => DeferredUpdates::POSTSEND |
| 1566 | ] ); |
| 1567 | |
| 1568 | $id = $this->getPageId(); |
| 1569 | $title = $this->getTitle(); |
| 1570 | $wikiPage = $this->getWikiPage(); |
| 1571 | |
| 1572 | if ( !$title->exists() ) { |
| 1573 | wfDebug( __METHOD__ . ": Page doesn't exist any more, bailing out" ); |
| 1574 | |
| 1575 | $this->doTransition( 'done' ); |
| 1576 | return; |
| 1577 | } |
| 1578 | |
| 1579 | DeferredUpdates::addCallableUpdate( function () use ( $event ) { |
| 1580 | if ( |
| 1581 | $this->options['oldcountable'] === 'no-change' || |
| 1582 | ( !$event->isEffectiveContentChange() && !$event->hasCause( PageRevisionUpdatedEvent::CAUSE_MOVE ) ) |
| 1583 | ) { |
| 1584 | $good = 0; |
| 1585 | } elseif ( $event->isCreation() ) { |
| 1586 | $good = (int)$this->isCountable(); |
| 1587 | } elseif ( $this->options['oldcountable'] !== null ) { |
| 1588 | $good = (int)$this->isCountable() |
| 1589 | - (int)$this->options['oldcountable']; |
| 1590 | } elseif ( isset( $this->pageState['oldCountable'] ) ) { |
| 1591 | $good = (int)$this->isCountable() |
| 1592 | - (int)$this->pageState['oldCountable']; |
| 1593 | } else { |
| 1594 | $good = 0; |
| 1595 | } |
| 1596 | $edits = $event->isEffectiveContentChange() ? 1 : 0; |
| 1597 | $pages = $event->isCreation() ? 1 : 0; |
| 1598 | |
| 1599 | DeferredUpdates::addUpdate( SiteStatsUpdate::factory( |
| 1600 | [ 'edits' => $edits, 'articles' => $good, 'pages' => $pages ] |
| 1601 | ) ); |
| 1602 | } ); |
| 1603 | |
| 1604 | // TODO: move onArticleCreate and onArticleEdit into a PageEventEmitter service |
| 1605 | if ( $event->isCreation() ) { |
| 1606 | // Deferred update that adds a mw-recreated tag to edits that create new pages |
| 1607 | // which have an associated deletion log entry for the specific namespace/title combination |
| 1608 | // and which are not undeletes |
| 1609 | if ( !( $event->hasCause( PageRevisionUpdatedEvent::CAUSE_UNDELETE ) ) ) { |
| 1610 | $revision = $this->revision; |
| 1611 | DeferredUpdates::addCallableUpdate( function () use ( $revision, $wikiPage ) { |
| 1612 | $this->maybeAddRecreateChangeTag( $wikiPage, $revision->getId() ); |
| 1613 | } ); |
| 1614 | } |
| 1615 | WikiPage::onArticleCreate( $title, $this->isRedirect() ); |
| 1616 | } elseif ( $event->isEffectiveContentChange() ) { // T52785 |
| 1617 | // TODO: Check $event->isNominalContentChange() instead so we still |
| 1618 | // trigger updates on null edits, but pass a flag to suppress |
| 1619 | // backlink purges through queueBacklinksJobs() id |
| 1620 | // $event->changedLatestRevisionId() returns false. |
| 1621 | WikiPage::onArticleEdit( |
| 1622 | $title, |
| 1623 | $this->revision, |
| 1624 | $this->getTouchedSlotRoles(), |
| 1625 | // Redirect target may have changed if the page is or was a redirect. |
| 1626 | // (We can't check if it was definitely changed without additional queries.) |
| 1627 | $this->isRedirect() || $this->wasRedirect() |
| 1628 | ); |
| 1629 | } |
| 1630 | |
| 1631 | if ( $event->hasCause( PageRevisionUpdatedEvent::CAUSE_UNDELETE ) ) { |
| 1632 | $this->mainWANObjectCache->touchCheckKey( |
| 1633 | "DerivedPageDataUpdater:restore:page:$id" |
| 1634 | ); |
| 1635 | } |
| 1636 | |
| 1637 | $editResult = $event->getEditResult(); |
| 1638 | |
| 1639 | if ( $editResult && !$editResult->isNullEdit() ) { |
| 1640 | // Cache EditResult for future use, via |
| 1641 | // RevertTagUpdateManager::approveRevertedTagForRevision(). |
| 1642 | // This drives RevertedTagUpdateManager::approveRevertedTagForRevision. |
| 1643 | // It is only needed if RCPatrolling is enabled and the edit is a revert. |
| 1644 | // Skip in other cases to avoid flooding the cache, see T386217 and T388573. |
| 1645 | if ( $editResult->isRevert() && $this->useRcPatrol ) { |
| 1646 | $this->editResultCache->set( |
| 1647 | $this->revision->getId(), |
| 1648 | $editResult |
| 1649 | ); |
| 1650 | } |
| 1651 | } |
| 1652 | |
| 1653 | $this->doTransition( 'done' ); |
| 1654 | } |
| 1655 | |
| 1656 | /** |
| 1657 | * @internal |
| 1658 | */ |
| 1659 | public function emitEvents(): void { |
| 1660 | if ( !$this->options['emitEvents'] ) { |
| 1661 | throw new LogicException( 'emitEvents was disabled on this updater' ); |
| 1662 | } |
| 1663 | |
| 1664 | $event = $this->getPageRevisionUpdatedEvent(); |
| 1665 | |
| 1666 | // don't dispatch again! |
| 1667 | $this->options['emitEvents'] = false; |
| 1668 | |
| 1669 | $this->eventDispatcher->dispatch( $event, $this->loadbalancerFactory ); |
| 1670 | } |
| 1671 | |
| 1672 | private function getPageRevisionUpdatedEvent(): PageRevisionUpdatedEvent { |
| 1673 | if ( $this->pageUpdatedEvent ) { |
| 1674 | return $this->pageUpdatedEvent; |
| 1675 | } |
| 1676 | |
| 1677 | $this->assertHasRevision( __METHOD__ ); |
| 1678 | |
| 1679 | $flags = array_intersect_key( |
| 1680 | $this->options, |
| 1681 | PageRevisionUpdatedEvent::DEFAULT_FLAGS |
| 1682 | ); |
| 1683 | |
| 1684 | $pageRecordBefore = $this->pageState['oldRecord'] ?? null; |
| 1685 | $pageRecordAfter = $this->getWikiPage()->toPageRecord(); |
| 1686 | |
| 1687 | $revisionBefore = $this->getOldRevision(); |
| 1688 | $revisionAfter = $this->getRevision(); |
| 1689 | |
| 1690 | if ( $this->options['created'] ) { |
| 1691 | // Page creation. No prior state. |
| 1692 | // Force null to make sure we don't get confused during imports when |
| 1693 | // updates are triggered after importing the last revision of several. |
| 1694 | // In that case, the page and older revisions do already exist when |
| 1695 | // the DerivedPageDataUpdater is initialized, because they were |
| 1696 | // created during the import. But they didn't exist prior to the |
| 1697 | // import (based on the fact that the 'created' flag is set). |
| 1698 | $pageRecordBefore = null; |
| 1699 | $revisionBefore = null; |
| 1700 | } elseif ( !$this->options['changed'] ) { |
| 1701 | // Null edit. Should already be the same, just make sure. |
| 1702 | $pageRecordBefore = $pageRecordAfter; |
| 1703 | } |
| 1704 | |
| 1705 | if ( $revisionBefore && $revisionAfter->getId() === $revisionBefore->getId() ) { |
| 1706 | // This is a null edit, flag it as a reconciliation request. |
| 1707 | $flags[ PageRevisionUpdatedEvent::FLAG_RECONCILIATION_REQUEST ] = true; |
| 1708 | } |
| 1709 | |
| 1710 | if ( $pageRecordBefore === null && !$this->options['created'] ) { |
| 1711 | // If the page wasn't just created, we need the state before. |
| 1712 | // If we are not actually emitting the event, we can ignore the issue. |
| 1713 | // This is needed to support the deprecated WikiPage::doEditUpdates() |
| 1714 | // method. Once that is gone, we can remove this conditional. |
| 1715 | if ( $this->options['emitEvents'] ) { |
| 1716 | throw new LogicException( 'Missing page state before update' ); |
| 1717 | } |
| 1718 | } |
| 1719 | |
| 1720 | /** @var UserIdentity $performer */ |
| 1721 | $performer = $this->options['triggeringUser'] ?? $this->user; |
| 1722 | '@phan-var UserIdentity $performer'; |
| 1723 | |
| 1724 | $this->pageUpdatedEvent = new PageRevisionUpdatedEvent( |
| 1725 | $this->options['cause'] ?? PageUpdateCauses::CAUSE_EDIT, |
| 1726 | $pageRecordBefore, |
| 1727 | $pageRecordAfter, |
| 1728 | $revisionBefore, |
| 1729 | $revisionAfter, |
| 1730 | $this->getRevisionSlotsUpdate(), |
| 1731 | $this->options['editResult'] ?? null, |
| 1732 | $performer, |
| 1733 | $this->options['tags'] ?? [], |
| 1734 | $flags, |
| 1735 | $this->options['rcPatrolStatus'] ?? 0, |
| 1736 | ); |
| 1737 | |
| 1738 | return $this->pageUpdatedEvent; |
| 1739 | } |
| 1740 | |
| 1741 | private function triggerParserCacheUpdate() { |
| 1742 | $this->assertHasRevision( __METHOD__ ); |
| 1743 | |
| 1744 | $userParserOptions = ParserOptions::newFromUser( $this->user ); |
| 1745 | |
| 1746 | // Decide whether to save the final canonical parser output based on the fact that |
| 1747 | // users are typically redirected to viewing pages right after they edit those pages. |
| 1748 | // Due to vary-revision-id, getting/saving that output here might require a reparse. |
| 1749 | if ( $userParserOptions->matchesForCacheKey( $this->getCanonicalParserOptions() ) ) { |
| 1750 | // Whether getting the final output requires a reparse or not, the user will |
| 1751 | // need canonical output anyway, since that is what their parser options use. |
| 1752 | // A reparse now at least has the benefit of various warm process caches. |
| 1753 | $this->doParserCacheUpdate(); |
| 1754 | } else { |
| 1755 | // If the user does not have canonical parse options, then don't risk another parse |
| 1756 | // to make output they cannot use on the page refresh that typically occurs after |
| 1757 | // editing. Doing the parser output save post-send will still benefit *other* users. |
| 1758 | DeferredUpdates::addCallableUpdate( function () { |
| 1759 | $this->doParserCacheUpdate(); |
| 1760 | } ); |
| 1761 | } |
| 1762 | } |
| 1763 | |
| 1764 | /** |
| 1765 | * Checks deletion logs for the specific article title and namespace combination |
| 1766 | * if a deletion log exists, we can assume this is a new page recreation and are tagging it with `mw-recreated`. |
| 1767 | * This does not consider deletions that were suppressed and therefore will not tag those. |
| 1768 | * |
| 1769 | * @param WikiPage $wikiPage |
| 1770 | * @param int $revisionId |
| 1771 | */ |
| 1772 | private function maybeAddRecreateChangeTag( WikiPage $wikiPage, int $revisionId ) { |
| 1773 | if ( $this->loadbalancerFactory->getReplicaDatabase()->newSelectQueryBuilder() |
| 1774 | ->select( [ '1' ] ) |
| 1775 | ->from( 'logging' ) |
| 1776 | ->where( [ |
| 1777 | 'log_type' => 'delete', |
| 1778 | 'log_title' => $wikiPage->getTitle()->getDBkey(), |
| 1779 | 'log_namespace' => $wikiPage->getNamespace(), |
| 1780 | ] )->caller( __METHOD__ )->limit( 1 )->fetchField() ) { |
| 1781 | $this->changeTagsStore->addTags( |
| 1782 | [ ChangeTags::TAG_RECREATE ], |
| 1783 | null, |
| 1784 | $revisionId ); |
| 1785 | } |
| 1786 | } |
| 1787 | |
| 1788 | /** |
| 1789 | * Do secondary data updates (e.g. updating link tables) or schedule them as deferred updates |
| 1790 | * |
| 1791 | * @note This does not update the parser cache. Use doParserCacheUpdate() for that. |
| 1792 | * @note Application logic should use Wikipage::doSecondaryDataUpdates instead. |
| 1793 | * |
| 1794 | * @param array $options |
| 1795 | * - recursive: make the update recursive, i.e. also update pages which transclude the |
| 1796 | * current page or otherwise depend on it (default: false) |
| 1797 | * - defer: one of the DeferredUpdates constants, or false to run immediately after waiting |
| 1798 | * for replication of the changes from the SecondaryDataUpdates hooks (default: false) |
| 1799 | * - freshness: used with 'defer'; forces an update if the last update was before the given timestamp, |
| 1800 | * even if the page and its dependencies didn't change since then (TS_MW; default: false) |
| 1801 | * @since 1.32 |
| 1802 | */ |
| 1803 | public function doSecondaryDataUpdates( array $options = [] ) { |
| 1804 | $this->assertHasRevision( __METHOD__ ); |
| 1805 | $options += [ 'recursive' => false, 'defer' => false, 'freshness' => false ]; |
| 1806 | $deferValues = [ false, DeferredUpdates::PRESEND, DeferredUpdates::POSTSEND ]; |
| 1807 | if ( !in_array( $options['defer'], $deferValues, true ) ) { |
| 1808 | throw new InvalidArgumentException( 'Invalid value for defer: ' . $options['defer'] ); |
| 1809 | } |
| 1810 | |
| 1811 | $triggeringUser = $this->options['triggeringUser'] ?? $this->user; |
| 1812 | [ $causeAction, $causeAgent ] = $this->getCauseForTracing(); |
| 1813 | if ( isset( $options['known-revision-output'] ) ) { |
| 1814 | $this->getRenderedRevision()->setRevisionParserOutput( $options['known-revision-output'] ); |
| 1815 | } |
| 1816 | |
| 1817 | // Bundle all of the data updates into a single deferred update wrapper so that |
| 1818 | // any failure will cause at most one refreshLinks job to be enqueued by |
| 1819 | // DeferredUpdates::doUpdates(). This is hard to do when there are many separate |
| 1820 | // updates that are not defined as being related. |
| 1821 | $update = new RefreshSecondaryDataUpdate( |
| 1822 | $this->loadbalancerFactory, |
| 1823 | // @phan-suppress-next-line PhanTypeMismatchArgumentNullable Already checked |
| 1824 | $triggeringUser, |
| 1825 | $this->wikiPage, |
| 1826 | $this->revision, |
| 1827 | $this, |
| 1828 | [ 'recursive' => $options['recursive'], 'freshness' => $options['freshness'] ] |
| 1829 | ); |
| 1830 | $update->setCause( $causeAction, $causeAgent ); |
| 1831 | |
| 1832 | if ( $options['defer'] === false ) { |
| 1833 | DeferredUpdates::attemptUpdate( $update ); |
| 1834 | } else { |
| 1835 | DeferredUpdates::addUpdate( $update, $options['defer'] ); |
| 1836 | } |
| 1837 | } |
| 1838 | |
| 1839 | /** |
| 1840 | * Causes parser cache entries to be updated. |
| 1841 | * |
| 1842 | * @note This does not update links tables. Use doSecondaryDataUpdates() for that. |
| 1843 | * @note Application logic should use Wikipage::updateParserCache instead. |
| 1844 | */ |
| 1845 | public function doParserCacheUpdate() { |
| 1846 | $this->assertHasRevision( __METHOD__ ); |
| 1847 | |
| 1848 | $wikiPage = $this->getWikiPage(); // TODO: ParserCache should accept a RevisionRecord instead |
| 1849 | |
| 1850 | // NOTE: this may trigger the first parsing of the new content after an edit (when not |
| 1851 | // using pre-generated stashed output). |
| 1852 | // XXX: we may want to use the PoolCounter here. This would perhaps allow the initial parse |
| 1853 | // to be performed post-send. The client could already follow a HTTP redirect to the |
| 1854 | // page view, but would then have to wait for a response until rendering is complete. |
| 1855 | $output = $this->getCanonicalParserOutput(); |
| 1856 | |
| 1857 | // Save it to the parser cache. Use the revision timestamp in the case of a |
| 1858 | // freshly saved edit, as that matches page_touched and a mismatch would trigger an |
| 1859 | // unnecessary reparse. |
| 1860 | $timestamp = $this->options['newrev'] ? $this->revision->getTimestamp() |
| 1861 | : $output->getCacheTime(); |
| 1862 | $this->parserCache->save( |
| 1863 | $output, $wikiPage, $this->getCanonicalParserOptions(), |
| 1864 | $timestamp, $this->revision->getId() |
| 1865 | ); |
| 1866 | |
| 1867 | // If we enable cache warming with parsoid outputs, let's do it at the same |
| 1868 | // time we're populating the parser cache with pre-generated HTML. |
| 1869 | // Use OPT_FORCE_PARSE to avoid a useless cache lookup. |
| 1870 | if ( $this->warmParsoidParserCache ) { |
| 1871 | $cacheWarmingParams = $this->getCauseForTracing(); |
| 1872 | $cacheWarmingParams['options'] = ParserOutputAccess::OPT_FORCE_PARSE; |
| 1873 | |
| 1874 | $this->jobQueueGroup->lazyPush( |
| 1875 | ParsoidCachePrewarmJob::newSpec( |
| 1876 | $this->revision->getId(), |
| 1877 | $wikiPage->toPageRecord(), |
| 1878 | $cacheWarmingParams |
| 1879 | ) |
| 1880 | ); |
| 1881 | } |
| 1882 | } |
| 1883 | |
| 1884 | } |