4 * Base class for content handling.
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License as published by
8 * the Free Software Foundation; either version 2 of the License, or
9 * (at your option) any later version.
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
16 * You should have received a copy of the GNU General Public License along
17 * with this program; if not, write to the Free Software Foundation, Inc.,
18 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
19 * http://www.gnu.org/copyleft/gpl.html
26 * @author Daniel Kinzler
29 use MediaWiki\Storage\RevisionRecord
;
30 use Wikimedia\Assert\Assert
;
31 use MediaWiki\Logger\LoggerFactory
;
32 use MediaWiki\MediaWikiServices
;
33 use MediaWiki\Revision\SlotRenderingProvider
;
34 use MediaWiki\Search\ParserOutputSearchDataExtractor
;
37 * A content handler knows how do deal with a specific type of content on a wiki
38 * page. Content is stored in the database in a serialized form (using a
39 * serialization format a.k.a. MIME type) and is unserialized into its native
40 * PHP representation (the content model), which is wrapped in an instance of
41 * the appropriate subclass of Content.
43 * ContentHandler instances are stateless singletons that serve, among other
44 * things, as a factory for Content objects. Generally, there is one subclass
45 * of ContentHandler and one subclass of Content for every type of content model.
47 * Some content types have a flat model, that is, their native representation
48 * is the same as their serialized form. Examples would be JavaScript and CSS
49 * code. As of now, this also applies to wikitext (MediaWiki's default content
50 * type), but wikitext content may be represented by a DOM or AST structure in
55 abstract class ContentHandler
{
57 * Convenience function for getting flat text from a Content object. This
58 * should only be used in the context of backwards compatibility with code
59 * that is not yet able to handle Content objects!
61 * If $content is null, this method returns the empty string.
63 * If $content is an instance of TextContent, this method returns the flat
64 * text as returned by $content->getText().
66 * If $content is not a TextContent object, the behavior of this method
67 * depends on the global $wgContentHandlerTextFallback:
68 * - If $wgContentHandlerTextFallback is 'fail' and $content is not a
69 * TextContent object, an MWException is thrown.
70 * - If $wgContentHandlerTextFallback is 'serialize' and $content is not a
71 * TextContent object, $content->serialize() is called to get a string
72 * form of the content.
73 * - If $wgContentHandlerTextFallback is 'ignore' and $content is not a
74 * TextContent object, this method returns null.
75 * - otherwise, the behavior is undefined.
79 * @param Content|null $content
81 * @throws MWException If the content is not an instance of TextContent and
82 * wgContentHandlerTextFallback was set to 'fail'.
83 * @return string|null Textual form of the content, if available.
85 public static function getContentText( Content
$content = null ) {
86 global $wgContentHandlerTextFallback;
88 if ( is_null( $content ) ) {
92 if ( $content instanceof TextContent
) {
93 return $content->getText();
96 wfDebugLog( 'ContentHandler', 'Accessing ' . $content->getModel() . ' content as text!' );
98 if ( $wgContentHandlerTextFallback == 'fail' ) {
99 throw new MWException(
100 "Attempt to get text from Content with model " .
105 if ( $wgContentHandlerTextFallback == 'serialize' ) {
106 return $content->serialize();
113 * Convenience function for creating a Content object from a given textual
116 * $text will be deserialized into a Content object of the model specified
117 * by $modelId (or, if that is not given, $title->getContentModel()) using
122 * @param string $text The textual representation, will be
123 * unserialized to create the Content object
124 * @param Title|null $title The title of the page this text belongs to.
125 * Required if $modelId is not provided.
126 * @param string|null $modelId The model to deserialize to. If not provided,
127 * $title->getContentModel() is used.
128 * @param string|null $format The format to use for deserialization. If not
129 * given, the model's default format is used.
131 * @throws MWException If model ID or format is not supported or if the text can not be
132 * unserialized using the format.
133 * @return Content A Content object representing the text.
135 public static function makeContent( $text, Title
$title = null,
136 $modelId = null, $format = null ) {
137 if ( is_null( $modelId ) ) {
138 if ( is_null( $title ) ) {
139 throw new MWException( "Must provide a Title object or a content model ID." );
142 $modelId = $title->getContentModel();
145 $handler = self
::getForModelID( $modelId );
147 return $handler->unserializeContent( $text, $format );
151 * Returns the name of the default content model to be used for the page
152 * with the given title.
154 * Note: There should rarely be need to call this method directly.
155 * To determine the actual content model for a given page, use
156 * Title::getContentModel().
158 * Which model is to be used by default for the page is determined based
159 * on several factors:
160 * - The global setting $wgNamespaceContentModels specifies a content model
162 * - The hook ContentHandlerDefaultModelFor may be used to override the page's default
164 * - Pages in NS_MEDIAWIKI and NS_USER default to the CSS or JavaScript
165 * model if they end in .js or .css, respectively.
166 * - Pages in NS_MEDIAWIKI default to the wikitext model otherwise.
167 * - The hook TitleIsCssOrJsPage may be used to force a page to use the CSS
168 * or JavaScript model. This is a compatibility feature. The ContentHandlerDefaultModelFor
169 * hook should be used instead if possible.
170 * - The hook TitleIsWikitextPage may be used to force a page to use the
171 * wikitext model. This is a compatibility feature. The ContentHandlerDefaultModelFor
172 * hook should be used instead if possible.
174 * If none of the above applies, the wikitext model is used.
176 * Note: this is used by, and may thus not use, Title::getContentModel()
179 * @deprecated since 1.33, use SlotRoleHandler::getDefaultModel() together with
180 * SlotRoleRegistry::getRoleHandler().
182 * @param Title $title
184 * @return string Default model name for the page given by $title
186 public static function getDefaultModelFor( Title
$title ) {
187 $slotRoleregistry = MediaWikiServices
::getInstance()->getSlotRoleRegistry();
188 $mainSlotHandler = $slotRoleregistry->getRoleHandler( 'main' );
189 return $mainSlotHandler->getDefaultModel( $title );
193 * Returns the appropriate ContentHandler singleton for the given title.
197 * @param Title $title
199 * @return ContentHandler
201 public static function getForTitle( Title
$title ) {
202 $modelId = $title->getContentModel();
204 return self
::getForModelID( $modelId );
208 * Returns the appropriate ContentHandler singleton for the given Content
213 * @param Content $content
215 * @return ContentHandler
217 public static function getForContent( Content
$content ) {
218 $modelId = $content->getModel();
220 return self
::getForModelID( $modelId );
224 * @var array A Cache of ContentHandler instances by model id
226 protected static $handlers;
229 * Returns the ContentHandler singleton for the given model ID. Use the
230 * CONTENT_MODEL_XXX constants to identify the desired content model.
232 * ContentHandler singletons are taken from the global $wgContentHandlers
233 * array. Keys in that array are model names, the values are either
234 * ContentHandler singleton objects, or strings specifying the appropriate
235 * subclass of ContentHandler.
237 * If a class name is encountered when looking up the singleton for a given
238 * model name, the class is instantiated and the class name is replaced by
239 * the resulting singleton in $wgContentHandlers.
241 * If no ContentHandler is defined for the desired $modelId, the
242 * ContentHandler may be provided by the ContentHandlerForModelID hook.
243 * If no ContentHandler can be determined, an MWException is raised.
247 * @param string $modelId The ID of the content model for which to get a
248 * handler. Use CONTENT_MODEL_XXX constants.
250 * @throws MWException For internal errors and problems in the configuration.
251 * @throws MWUnknownContentModelException If no handler is known for the model ID.
252 * @return ContentHandler The ContentHandler singleton for handling the model given by the ID.
254 public static function getForModelID( $modelId ) {
255 global $wgContentHandlers;
257 if ( isset( self
::$handlers[$modelId] ) ) {
258 return self
::$handlers[$modelId];
261 if ( empty( $wgContentHandlers[$modelId] ) ) {
264 Hooks
::run( 'ContentHandlerForModelID', [ $modelId, &$handler ] );
266 if ( $handler === null ) {
267 throw new MWUnknownContentModelException( $modelId );
270 if ( !( $handler instanceof ContentHandler
) ) {
271 throw new MWException( "ContentHandlerForModelID must supply a ContentHandler instance" );
274 $classOrCallback = $wgContentHandlers[$modelId];
276 if ( is_callable( $classOrCallback ) ) {
277 $handler = call_user_func( $classOrCallback, $modelId );
279 $handler = new $classOrCallback( $modelId );
282 if ( !( $handler instanceof ContentHandler
) ) {
283 throw new MWException(
284 var_export( $classOrCallback, true ) . " from \$wgContentHandlers is not " .
285 "compatible with ContentHandler"
290 wfDebugLog( 'ContentHandler', 'Created handler for ' . $modelId
291 . ': ' . get_class( $handler ) );
293 self
::$handlers[$modelId] = $handler;
295 return self
::$handlers[$modelId];
299 * Clean up handlers cache.
301 public static function cleanupHandlersCache() {
302 self
::$handlers = [];
306 * Returns the localized name for a given content model.
308 * Model names are localized using system messages. Message keys
309 * have the form content-model-$name, where $name is getContentModelName( $id ).
311 * @param string $name The content model ID, as given by a CONTENT_MODEL_XXX
312 * constant or returned by Revision::getContentModel().
313 * @param Language|null $lang The language to parse the message in (since 1.26)
315 * @throws MWException If the model ID isn't known.
316 * @return string The content model's localized name.
318 public static function getLocalizedName( $name, Language
$lang = null ) {
319 // Messages: content-model-wikitext, content-model-text,
320 // content-model-javascript, content-model-css
321 $key = "content-model-$name";
323 $msg = wfMessage( $key );
325 $msg->inLanguage( $lang );
328 return $msg->exists() ?
$msg->plain() : $name;
331 public static function getContentModels() {
332 global $wgContentHandlers;
334 $models = array_keys( $wgContentHandlers );
335 Hooks
::run( 'GetContentModels', [ &$models ] );
339 public static function getAllContentFormats() {
340 global $wgContentHandlers;
344 foreach ( $wgContentHandlers as $model => $class ) {
345 $handler = self
::getForModelID( $model );
346 $formats = array_merge( $formats, $handler->getSupportedFormats() );
349 $formats = array_unique( $formats );
354 // ------------------------------------------------------------------------
364 protected $mSupportedFormats;
367 * Constructor, initializing the ContentHandler instance with its model ID
368 * and a list of supported formats. Values for the parameters are typically
369 * provided as literals by subclass's constructors.
371 * @param string $modelId (use CONTENT_MODEL_XXX constants).
372 * @param string[] $formats List for supported serialization formats
373 * (typically as MIME types)
375 public function __construct( $modelId, $formats ) {
376 $this->mModelID
= $modelId;
377 $this->mSupportedFormats
= $formats;
381 * Serializes a Content object of the type supported by this ContentHandler.
385 * @param Content $content The Content object to serialize
386 * @param string|null $format The desired serialization format
388 * @return string Serialized form of the content
390 abstract public function serializeContent( Content
$content, $format = null );
393 * Applies transformations on export (returns the blob unchanged per default).
394 * Subclasses may override this to perform transformations such as conversion
395 * of legacy formats or filtering of internal meta-data.
397 * @param string $blob The blob to be exported
398 * @param string|null $format The blob's serialization format
402 public function exportTransform( $blob, $format = null ) {
407 * Unserializes a Content object of the type supported by this ContentHandler.
411 * @param string $blob Serialized form of the content
412 * @param string|null $format The format used for serialization
414 * @return Content The Content object created by deserializing $blob
416 abstract public function unserializeContent( $blob, $format = null );
419 * Apply import transformation (per default, returns $blob unchanged).
420 * This gives subclasses an opportunity to transform data blobs on import.
424 * @param string $blob
425 * @param string|null $format
429 public function importTransform( $blob, $format = null ) {
434 * Creates an empty Content object of the type supported by this
441 abstract public function makeEmptyContent();
444 * Creates a new Content object that acts as a redirect to the given page,
445 * or null if redirects are not supported by this content model.
447 * This default implementation always returns null. Subclasses supporting redirects
448 * must override this method.
450 * Note that subclasses that override this method to return a Content object
451 * should also override supportsRedirects() to return true.
455 * @param Title $destination The page to redirect to.
456 * @param string $text Text to include in the redirect, if possible.
458 * @return Content Always null.
460 public function makeRedirectContent( Title
$destination, $text = '' ) {
465 * Returns the model id that identifies the content model this
466 * ContentHandler can handle. Use with the CONTENT_MODEL_XXX constants.
470 * @return string The model ID
472 public function getModelID() {
473 return $this->mModelID
;
479 * @param string $model_id The model to check
481 * @throws MWException If the model ID is not the ID of the content model supported by this
484 protected function checkModelID( $model_id ) {
485 if ( $model_id !== $this->mModelID
) {
486 throw new MWException( "Bad content model: " .
487 "expected {$this->mModelID} " .
488 "but got $model_id." );
493 * Returns a list of serialization formats supported by the
494 * serializeContent() and unserializeContent() methods of this
499 * @return string[] List of serialization formats as MIME type like strings
501 public function getSupportedFormats() {
502 return $this->mSupportedFormats
;
506 * The format used for serialization/deserialization by default by this
509 * This default implementation will return the first element of the array
510 * of formats that was passed to the constructor.
514 * @return string The name of the default serialization format as a MIME type
516 public function getDefaultFormat() {
517 return $this->mSupportedFormats
[0];
521 * Returns true if $format is a serialization format supported by this
522 * ContentHandler, and false otherwise.
524 * Note that if $format is null, this method always returns true, because
525 * null means "use the default format".
529 * @param string $format The serialization format to check
533 public function isSupportedFormat( $format ) {
535 return true; // this means "use the default"
538 return in_array( $format, $this->mSupportedFormats
);
542 * Convenient for checking whether a format provided as a parameter is actually supported.
544 * @param string $format The serialization format to check
546 * @throws MWException If the format is not supported by this content handler.
548 protected function checkFormat( $format ) {
549 if ( !$this->isSupportedFormat( $format ) ) {
550 throw new MWException(
551 "Format $format is not supported for content model "
552 . $this->getModelID()
558 * Returns overrides for action handlers.
559 * Classes listed here will be used instead of the default one when
560 * (and only when) $wgActions[$action] === true. This allows subclasses
561 * to override the default action handlers.
565 * @return array An array mapping action names (typically "view", "edit", "history" etc.) to
566 * either the full qualified class name of an Action class, a callable taking ( Page $page,
567 * IContextSource $context = null ) as parameters and returning an Action object, or an actual
568 * Action object. An empty array in this default implementation.
570 * @see Action::factory
572 public function getActionOverrides() {
577 * Factory for creating an appropriate DifferenceEngine for this content model.
578 * Since 1.32, this is only used for page-level diffs; to diff two content objects,
579 * use getSlotDiffRenderer.
581 * The DifferenceEngine subclass to use is selected in getDiffEngineClass(). The
582 * GetDifferenceEngine hook will receive the DifferenceEngine object and can replace or
584 * (Note that in older versions of MediaWiki the hook documentation instructed extensions
585 * to return false from the hook; you should not rely on always being able to decorate
586 * the DifferenceEngine instance from the hook. If the owner of the content type wants to
587 * decorare the instance, overriding this method is a safer approach.)
589 * @todo This is page-level functionality so it should not belong to ContentHandler.
590 * Move it to a better place once one exists (e.g. PageTypeHandler).
594 * @param IContextSource $context Context to use, anything else will be ignored.
595 * @param int $old Revision ID we want to show and diff with.
596 * @param int|string $new Either a revision ID or one of the strings 'cur', 'prev' or 'next'.
597 * @param int $rcid FIXME: Deprecated, no longer used. Defaults to 0.
598 * @param bool $refreshCache If set, refreshes the diff cache. Defaults to false.
599 * @param bool $unhide If set, allow viewing deleted revs. Defaults to false.
601 * @return DifferenceEngine
603 public function createDifferenceEngine( IContextSource
$context, $old = 0, $new = 0,
604 $rcid = 0, // FIXME: Deprecated, no longer used
605 $refreshCache = false, $unhide = false
607 $diffEngineClass = $this->getDiffEngineClass();
608 $differenceEngine = new $diffEngineClass( $context, $old, $new, $rcid, $refreshCache, $unhide );
609 Hooks
::run( 'GetDifferenceEngine', [ $context, $old, $new, $refreshCache, $unhide,
610 &$differenceEngine ] );
611 return $differenceEngine;
615 * Get an appropriate SlotDiffRenderer for this content model.
617 * @param IContextSource $context
618 * @return SlotDiffRenderer
620 final public function getSlotDiffRenderer( IContextSource
$context ) {
621 $slotDiffRenderer = $this->getSlotDiffRendererInternal( $context );
622 if ( get_class( $slotDiffRenderer ) === TextSlotDiffRenderer
::class ) {
623 // To keep B/C, when SlotDiffRenderer is not overridden for a given content type
624 // but DifferenceEngine is, use that instead.
625 $differenceEngine = $this->createDifferenceEngine( $context );
626 if ( get_class( $differenceEngine ) !== DifferenceEngine
::class ) {
627 // TODO turn this into a deprecation warning in a later release
628 LoggerFactory
::getInstance( 'diff' )->info(
629 'Falling back to DifferenceEngineSlotDiffRenderer', [
630 'modelID' => $this->getModelID(),
631 'DifferenceEngine' => get_class( $differenceEngine ),
633 $slotDiffRenderer = new DifferenceEngineSlotDiffRenderer( $differenceEngine );
636 Hooks
::run( 'GetSlotDiffRenderer', [ $this, &$slotDiffRenderer, $context ] );
637 return $slotDiffRenderer;
641 * Return the SlotDiffRenderer appropriate for this content handler.
642 * @param IContextSource $context
643 * @return SlotDiffRenderer
645 protected function getSlotDiffRendererInternal( IContextSource
$context ) {
646 $contentLanguage = MediaWikiServices
::getInstance()->getContentLanguage();
647 $statsdDataFactory = MediaWikiServices
::getInstance()->getStatsdDataFactory();
648 $slotDiffRenderer = new TextSlotDiffRenderer();
649 $slotDiffRenderer->setStatsdDataFactory( $statsdDataFactory );
650 // XXX using the page language would be better, but it's unclear how that should be injected
651 $slotDiffRenderer->setLanguage( $contentLanguage );
653 $engine = DifferenceEngine
::getEngine();
654 if ( $engine === false ) {
655 $slotDiffRenderer->setEngine( TextSlotDiffRenderer
::ENGINE_PHP
);
656 } elseif ( $engine === 'wikidiff2' ) {
657 $slotDiffRenderer->setEngine( TextSlotDiffRenderer
::ENGINE_WIKIDIFF2
);
659 $slotDiffRenderer->setEngine( TextSlotDiffRenderer
::ENGINE_EXTERNAL
, $engine );
662 return $slotDiffRenderer;
666 * Get the language in which the content of the given page is written.
668 * This default implementation just returns the content language (except for pages
669 * in the MediaWiki namespace)
671 * Note that the page's language is not cacheable, since it may in some
672 * cases depend on user settings.
674 * Also note that the page language may or may not depend on the actual content of the page,
675 * that is, this method may load the content in order to determine the language.
679 * @param Title $title The page to determine the language for.
680 * @param Content|null $content The page's content, if you have it handy, to avoid reloading it.
682 * @return Language The page's language
684 public function getPageLanguage( Title
$title, Content
$content = null ) {
686 $pageLang = MediaWikiServices
::getInstance()->getContentLanguage();
688 if ( $title->inNamespace( NS_MEDIAWIKI
) ) {
689 // Parse mediawiki messages with correct target language
690 list( /* $unused */, $lang ) = MessageCache
::singleton()->figureMessage( $title->getText() );
691 $pageLang = Language
::factory( $lang );
694 // Simplify hook handlers by only passing objects of one type, in case nothing
695 // else has unstubbed the StubUserLang object by now.
696 StubObject
::unstub( $wgLang );
698 Hooks
::run( 'PageContentLanguage', [ $title, &$pageLang, $wgLang ] );
700 return wfGetLangObj( $pageLang );
704 * Get the language in which the content of this page is written when
705 * viewed by user. Defaults to $this->getPageLanguage(), but if the user
706 * specified a preferred variant, the variant will be used.
708 * This default implementation just returns $this->getPageLanguage( $title, $content ) unless
709 * the user specified a preferred variant.
711 * Note that the pages view language is not cacheable, since it depends on user settings.
713 * Also note that the page language may or may not depend on the actual content of the page,
714 * that is, this method may load the content in order to determine the language.
718 * @param Title $title The page to determine the language for.
719 * @param Content|null $content The page's content, if you have it handy, to avoid reloading it.
721 * @return Language The page's language for viewing
723 public function getPageViewLanguage( Title
$title, Content
$content = null ) {
724 $pageLang = $this->getPageLanguage( $title, $content );
726 if ( $title->getNamespace() !== NS_MEDIAWIKI
) {
727 // If the user chooses a variant, the content is actually
728 // in a language whose code is the variant code.
729 $variant = $pageLang->getPreferredVariant();
730 if ( $pageLang->getCode() !== $variant ) {
731 $pageLang = Language
::factory( $variant );
739 * Determines whether the content type handled by this ContentHandler
740 * can be used for the main slot of the given page.
742 * This default implementation always returns true.
743 * Subclasses may override this to restrict the use of this content model to specific locations,
744 * typically based on the namespace or some other aspect of the title, such as a special suffix
745 * (e.g. ".svg" for SVG content).
747 * @note this calls the ContentHandlerCanBeUsedOn hook which may be used to override which
748 * content model can be used where.
750 * @see SlotRoleHandler::isAllowedModel
752 * @param Title $title The page's title.
754 * @return bool True if content of this kind can be used on the given page, false otherwise.
756 public function canBeUsedOn( Title
$title ) {
759 Hooks
::run( 'ContentModelCanBeUsedOn', [ $this->getModelID(), $title, &$ok ] );
765 * Returns the name of the diff engine to use.
771 protected function getDiffEngineClass() {
772 return DifferenceEngine
::class;
776 * Attempts to merge differences between three versions. Returns a new
777 * Content object for a clean merge and false for failure or a conflict.
779 * This default implementation always returns false.
783 * @param Content $oldContent The page's previous content.
784 * @param Content $myContent One of the page's conflicting contents.
785 * @param Content $yourContent One of the page's conflicting contents.
787 * @return Content|bool Always false.
789 public function merge3( Content
$oldContent, Content
$myContent, Content
$yourContent ) {
794 * Return type of change if one exists for the given edit.
798 * @param Content|null $oldContent The previous text of the page.
799 * @param Content|null $newContent The submitted text of the page.
800 * @param int $flags Bit mask: a bit mask of flags submitted for the edit.
802 * @return string|null String key representing type of change, or null.
804 private function getChangeType(
805 Content
$oldContent = null,
806 Content
$newContent = null,
809 $oldTarget = $oldContent !== null ?
$oldContent->getRedirectTarget() : null;
810 $newTarget = $newContent !== null ?
$newContent->getRedirectTarget() : null;
812 // We check for the type of change in the given edit, and return string key accordingly
814 // Blanking of a page
815 if ( $oldContent && $oldContent->getSize() > 0 &&
816 $newContent && $newContent->getSize() === 0
824 // New redirect page (by creating new page or by changing content page)
825 return 'new-redirect';
826 } elseif ( !$newTarget->equals( $oldTarget ) ||
827 $oldTarget->getFragment() !== $newTarget->getFragment()
829 // Redirect target changed
830 return 'changed-redirect-target';
832 } elseif ( $oldTarget ) {
833 // Changing an existing redirect into a non-redirect
834 return 'removed-redirect';
838 if ( $flags & EDIT_NEW
&& $newContent ) {
839 if ( $newContent->getSize() === 0 ) {
847 // Removing more than 90% of the page
848 if ( $oldContent && $newContent && $oldContent->getSize() > 10 * $newContent->getSize() ) {
852 // Content model changed
853 if ( $oldContent && $newContent && $oldContent->getModel() !== $newContent->getModel() ) {
854 return 'contentmodelchange';
861 * Return an applicable auto-summary if one exists for the given edit.
865 * @param Content|null $oldContent The previous text of the page.
866 * @param Content|null $newContent The submitted text of the page.
867 * @param int $flags Bit mask: a bit mask of flags submitted for the edit.
869 * @return string An appropriate auto-summary, or an empty string.
871 public function getAutosummary(
872 Content
$oldContent = null,
873 Content
$newContent = null,
876 $changeType = $this->getChangeType( $oldContent, $newContent, $flags );
878 // There's no applicable auto-summary for our case, so our auto-summary is empty.
879 if ( !$changeType ) {
883 // Decide what kind of auto-summary is needed.
884 switch ( $changeType ) {
886 $newTarget = $newContent->getRedirectTarget();
887 $truncatedtext = $newContent->getTextForSummary(
889 - strlen( wfMessage( 'autoredircomment' )->inContentLanguage()->text() )
890 - strlen( $newTarget->getFullText() )
893 return wfMessage( 'autoredircomment', $newTarget->getFullText() )
894 ->plaintextParams( $truncatedtext )->inContentLanguage()->text();
895 case 'changed-redirect-target':
896 $oldTarget = $oldContent->getRedirectTarget();
897 $newTarget = $newContent->getRedirectTarget();
899 $truncatedtext = $newContent->getTextForSummary(
901 - strlen( wfMessage( 'autosumm-changed-redirect-target' )
902 ->inContentLanguage()->text() )
903 - strlen( $oldTarget->getFullText() )
904 - strlen( $newTarget->getFullText() )
907 return wfMessage( 'autosumm-changed-redirect-target',
908 $oldTarget->getFullText(),
909 $newTarget->getFullText() )
910 ->rawParams( $truncatedtext )->inContentLanguage()->text();
911 case 'removed-redirect':
912 $oldTarget = $oldContent->getRedirectTarget();
913 $truncatedtext = $newContent->getTextForSummary(
915 - strlen( wfMessage( 'autosumm-removed-redirect' )
916 ->inContentLanguage()->text() )
917 - strlen( $oldTarget->getFullText() ) );
919 return wfMessage( 'autosumm-removed-redirect', $oldTarget->getFullText() )
920 ->rawParams( $truncatedtext )->inContentLanguage()->text();
922 // If they're making a new article, give its text, truncated, in the summary.
923 $truncatedtext = $newContent->getTextForSummary(
924 200 - strlen( wfMessage( 'autosumm-new' )->inContentLanguage()->text() ) );
926 return wfMessage( 'autosumm-new' )->rawParams( $truncatedtext )
927 ->inContentLanguage()->text();
929 return wfMessage( 'autosumm-blank' )->inContentLanguage()->text();
931 $truncatedtext = $newContent->getTextForSummary(
932 200 - strlen( wfMessage( 'autosumm-replace' )->inContentLanguage()->text() ) );
934 return wfMessage( 'autosumm-replace' )->rawParams( $truncatedtext )
935 ->inContentLanguage()->text();
937 return wfMessage( 'autosumm-newblank' )->inContentLanguage()->text();
944 * Return an applicable tag if one exists for the given edit or return null.
948 * @param Content|null $oldContent The previous text of the page.
949 * @param Content|null $newContent The submitted text of the page.
950 * @param int $flags Bit mask: a bit mask of flags submitted for the edit.
952 * @return string|null An appropriate tag, or null.
954 public function getChangeTag(
955 Content
$oldContent = null,
956 Content
$newContent = null,
959 $changeType = $this->getChangeType( $oldContent, $newContent, $flags );
961 // There's no applicable tag for this change.
962 if ( !$changeType ) {
966 // Core tags use the same keys as ones returned from $this->getChangeType()
967 // but prefixed with pseudo namespace 'mw-', so we add the prefix before checking
968 // if this type of change should be tagged
969 $tag = 'mw-' . $changeType;
971 // Not all change types are tagged, so we check against the list of defined tags.
972 if ( in_array( $tag, ChangeTags
::getSoftwareTags() ) ) {
980 * Auto-generates a deletion reason
984 * @param Title $title The page's title
985 * @param bool &$hasHistory Whether the page has a history
987 * @return mixed String containing deletion reason or empty string, or
988 * boolean false if no revision occurred
990 * @todo &$hasHistory is extremely ugly, it's here because
991 * WikiPage::getAutoDeleteReason() and Article::generateReason()
994 public function getAutoDeleteReason( Title
$title, &$hasHistory ) {
995 $dbr = wfGetDB( DB_REPLICA
);
997 // Get the last revision
998 $rev = Revision
::newFromTitle( $title );
1000 if ( is_null( $rev ) ) {
1004 // Get the article's contents
1005 $content = $rev->getContent();
1008 // If the page is blank, use the text from the previous revision,
1009 // which can only be blank if there's a move/import/protect dummy
1010 // revision involved
1011 if ( !$content ||
$content->isEmpty() ) {
1012 $prev = $rev->getPrevious();
1016 $content = $rev->getContent();
1021 $this->checkModelID( $rev->getContentModel() );
1023 // Find out if there was only one contributor
1024 // Only scan the last 20 revisions
1025 $revQuery = Revision
::getQueryInfo();
1026 $res = $dbr->select(
1027 $revQuery['tables'],
1028 [ 'rev_user_text' => $revQuery['fields']['rev_user_text'] ],
1030 'rev_page' => $title->getArticleID(),
1031 $dbr->bitAnd( 'rev_deleted', RevisionRecord
::DELETED_USER
) . ' = 0'
1038 if ( $res === false ) {
1039 // This page has no revisions, which is very weird
1043 $hasHistory = ( $res->numRows() > 1 );
1044 $row = $dbr->fetchObject( $res );
1046 if ( $row ) { // $row is false if the only contributor is hidden
1047 $onlyAuthor = $row->rev_user_text
;
1048 // Try to find a second contributor
1049 foreach ( $res as $row ) {
1050 if ( $row->rev_user_text
!= $onlyAuthor ) { // T24999
1051 $onlyAuthor = false;
1056 $onlyAuthor = false;
1059 // Generate the summary with a '$1' placeholder
1061 // The current revision is blank and the one before is also
1062 // blank. It's just not our lucky day
1063 $reason = wfMessage( 'exbeforeblank', '$1' )->inContentLanguage()->text();
1065 if ( $onlyAuthor ) {
1066 $reason = wfMessage(
1070 )->inContentLanguage()->text();
1072 $reason = wfMessage( 'excontent', '$1' )->inContentLanguage()->text();
1076 if ( $reason == '-' ) {
1077 // Allow these UI messages to be blanked out cleanly
1081 // Max content length = max comment length - length of the comment (excl. $1)
1082 $maxLength = CommentStore
::COMMENT_CHARACTER_LIMIT
- ( strlen( $reason ) - 2 );
1083 $text = $content ?
$content->getTextForSummary( $maxLength ) : '';
1085 // Now replace the '$1' placeholder
1086 $reason = str_replace( '$1', $text, $reason );
1092 * Get the Content object that needs to be saved in order to undo all revisions
1093 * between $undo and $undoafter. Revisions must belong to the same page,
1094 * must exist and must not be deleted.
1097 * @since 1.32 accepts Content objects for all parameters instead of Revision objects.
1098 * Passing Revision objects is deprecated.
1100 * @param Revision|Content $current The current text
1101 * @param Revision|Content $undo The content of the revision to undo
1102 * @param Revision|Content $undoafter Must be from an earlier revision than $undo
1103 * @param bool $undoIsLatest Set true if $undo is from the current revision (since 1.32)
1105 * @return mixed Content on success, false on failure
1107 public function getUndoContent( $current, $undo, $undoafter, $undoIsLatest = false ) {
1108 Assert
::parameterType( Revision
::class . '|' . Content
::class, $current, '$current' );
1109 if ( $current instanceof Content
) {
1110 Assert
::parameter( $undo instanceof Content
, '$undo',
1111 'Must be Content when $current is Content' );
1112 Assert
::parameter( $undoafter instanceof Content
, '$undoafter',
1113 'Must be Content when $current is Content' );
1114 $cur_content = $current;
1115 $undo_content = $undo;
1116 $undoafter_content = $undoafter;
1118 Assert
::parameter( $undo instanceof Revision
, '$undo',
1119 'Must be Revision when $current is Revision' );
1120 Assert
::parameter( $undoafter instanceof Revision
, '$undoafter',
1121 'Must be Revision when $current is Revision' );
1123 $cur_content = $current->getContent();
1125 if ( empty( $cur_content ) ) {
1126 return false; // no page
1129 $undo_content = $undo->getContent();
1130 $undoafter_content = $undoafter->getContent();
1132 if ( !$undo_content ||
!$undoafter_content ) {
1133 return false; // no content to undo
1136 $undoIsLatest = $current->getId() === $undo->getId();
1140 $this->checkModelID( $cur_content->getModel() );
1141 $this->checkModelID( $undo_content->getModel() );
1142 if ( !$undoIsLatest ) {
1143 // If we are undoing the most recent revision,
1144 // its ok to revert content model changes. However
1145 // if we are undoing a revision in the middle, then
1146 // doing that will be confusing.
1147 $this->checkModelID( $undoafter_content->getModel() );
1149 } catch ( MWException
$e ) {
1150 // If the revisions have different content models
1151 // just return false
1155 if ( $cur_content->equals( $undo_content ) ) {
1156 // No use doing a merge if it's just a straight revert.
1157 return $undoafter_content;
1160 $undone_content = $this->merge3( $undo_content, $undoafter_content, $cur_content );
1162 return $undone_content;
1166 * Get parser options suitable for rendering and caching the article
1168 * @deprecated since 1.32, use WikiPage::makeParserOptions() or
1169 * ParserOptions::newCanonical() instead.
1170 * @param IContextSource|User|string $context One of the following:
1171 * - IContextSource: Use the User and the Language of the provided
1173 * - User: Use the provided User object and $wgLang for the language,
1174 * so use an IContextSource object if possible.
1175 * - 'canonical': Canonical options (anonymous user with default
1176 * preferences and content language).
1178 * @throws MWException
1179 * @return ParserOptions
1181 public function makeParserOptions( $context ) {
1182 wfDeprecated( __METHOD__
, '1.32' );
1183 return ParserOptions
::newCanonical( $context );
1187 * Returns true for content models that support caching using the
1188 * ParserCache mechanism. See WikiPage::shouldCheckParserCache().
1192 * @return bool Always false.
1194 public function isParserCacheSupported() {
1199 * Returns true if this content model supports sections.
1200 * This default implementation returns false.
1202 * Content models that return true here should also implement
1203 * Content::getSection, Content::replaceSection, etc. to handle sections..
1205 * @return bool Always false.
1207 public function supportsSections() {
1212 * Returns true if this content model supports categories.
1213 * The default implementation returns true.
1215 * @return bool Always true.
1217 public function supportsCategories() {
1222 * Returns true if this content model supports redirects.
1223 * This default implementation returns false.
1225 * Content models that return true here should also implement
1226 * ContentHandler::makeRedirectContent to return a Content object.
1228 * @return bool Always false.
1230 public function supportsRedirects() {
1235 * Return true if this content model supports direct editing, such as via EditPage.
1237 * @return bool Default is false, and true for TextContent and it's derivatives.
1239 public function supportsDirectEditing() {
1244 * Whether or not this content model supports direct editing via ApiEditPage
1246 * @return bool Default is false, and true for TextContent and derivatives.
1248 public function supportsDirectApiEditing() {
1249 return $this->supportsDirectEditing();
1253 * Get fields definition for search index
1255 * @todo Expose title, redirect, namespace, text, source_text, text_bytes
1256 * field mappings here. (see T142670 and T143409)
1258 * @param SearchEngine $engine
1259 * @return SearchIndexField[] List of fields this content handler can provide.
1262 public function getFieldsForSearchIndex( SearchEngine
$engine ) {
1263 $fields['category'] = $engine->makeSearchFieldMapping(
1265 SearchIndexField
::INDEX_TYPE_TEXT
1267 $fields['category']->setFlag( SearchIndexField
::FLAG_CASEFOLD
);
1269 $fields['external_link'] = $engine->makeSearchFieldMapping(
1271 SearchIndexField
::INDEX_TYPE_KEYWORD
1274 $fields['outgoing_link'] = $engine->makeSearchFieldMapping(
1276 SearchIndexField
::INDEX_TYPE_KEYWORD
1279 $fields['template'] = $engine->makeSearchFieldMapping(
1281 SearchIndexField
::INDEX_TYPE_KEYWORD
1283 $fields['template']->setFlag( SearchIndexField
::FLAG_CASEFOLD
);
1285 $fields['content_model'] = $engine->makeSearchFieldMapping(
1287 SearchIndexField
::INDEX_TYPE_KEYWORD
1294 * Add new field definition to array.
1295 * @param SearchIndexField[] &$fields
1296 * @param SearchEngine $engine
1297 * @param string $name
1299 * @return SearchIndexField[] new field defs
1302 protected function addSearchField( &$fields, SearchEngine
$engine, $name, $type ) {
1303 $fields[$name] = $engine->makeSearchFieldMapping( $name, $type );
1308 * Return fields to be indexed by search engine
1309 * as representation of this document.
1310 * Overriding class should call parent function or take care of calling
1311 * the SearchDataForIndex hook.
1312 * @param WikiPage $page Page to index
1313 * @param ParserOutput $output
1314 * @param SearchEngine $engine Search engine for which we are indexing
1315 * @return array Map of name=>value for fields
1318 public function getDataForSearchIndex(
1320 ParserOutput
$output,
1321 SearchEngine
$engine
1324 $content = $page->getContent();
1327 $searchDataExtractor = new ParserOutputSearchDataExtractor();
1329 $fieldData['category'] = $searchDataExtractor->getCategories( $output );
1330 $fieldData['external_link'] = $searchDataExtractor->getExternalLinks( $output );
1331 $fieldData['outgoing_link'] = $searchDataExtractor->getOutgoingLinks( $output );
1332 $fieldData['template'] = $searchDataExtractor->getTemplates( $output );
1334 $text = $content->getTextForSearchIndex();
1336 $fieldData['text'] = $text;
1337 $fieldData['source_text'] = $text;
1338 $fieldData['text_bytes'] = $content->getSize();
1339 $fieldData['content_model'] = $content->getModel();
1342 Hooks
::run( 'SearchDataForIndex', [ &$fieldData, $this, $page, $output, $engine ] );
1347 * Produce page output suitable for indexing.
1349 * Specific content handlers may override it if they need different content handling.
1351 * @param WikiPage $page
1352 * @param ParserCache|null $cache
1353 * @return ParserOutput
1355 public function getParserOutputForIndexing( WikiPage
$page, ParserCache
$cache = null ) {
1356 // TODO: MCR: ContentHandler should be called per slot, not for the whole page.
1358 $parserOptions = $page->makeParserOptions( 'canonical' );
1360 $parserOutput = $cache->get( $page, $parserOptions );
1363 if ( empty( $parserOutput ) ) {
1364 $renderer = MediaWikiServices
::getInstance()->getRevisionRenderer();
1366 $renderer->getRenderedRevision(
1367 $page->getRevision()->getRevisionRecord(),
1369 )->getRevisionParserOutput();
1371 $cache->save( $parserOutput, $page, $parserOptions );
1374 return $parserOutput;
1378 * Returns a list of DeferrableUpdate objects for recording information about the
1379 * given Content in some secondary data store.
1381 * Application logic should not call this method directly. Instead, it should call
1382 * DerivedPageDataUpdater::getSecondaryDataUpdates().
1384 * @note Implementations must not return a LinksUpdate instance. Instead, a LinksUpdate
1385 * is created by the calling code in DerivedPageDataUpdater, on the combined ParserOutput
1386 * of all slots, not for each slot individually. This is in contrast to the old
1387 * getSecondaryDataUpdates method defined by AbstractContent, which returned a LinksUpdate.
1389 * @note Implementations should not call $content->getParserOutput, they should call
1390 * $slotOutput->getSlotRendering( $role, false ) instead if they need to access a ParserOutput
1391 * of $content. This allows existing ParserOutput objects to be re-used, while avoiding
1392 * creating a ParserOutput when none is needed.
1394 * @param Title $title The title of the page to supply the updates for
1395 * @param Content $content The content to generate data updates for.
1396 * @param string $role The role (slot) in which the content is being used. Which updates
1397 * are performed should generally not depend on the role the content has, but the
1398 * DeferrableUpdates themselves may need to know the role, to track to which slot the
1399 * data refers, and to avoid overwriting data of the same kind from another slot.
1400 * @param SlotRenderingProvider $slotOutput A provider that can be used to gain access to
1401 * a ParserOutput of $content by calling $slotOutput->getSlotParserOutput( $role, false ).
1402 * @return DeferrableUpdate[] A list of DeferrableUpdate objects for putting information
1403 * about this content object somewhere. The default implementation returns an empty
1407 public function getSecondaryDataUpdates(
1411 SlotRenderingProvider
$slotOutput
1417 * Returns a list of DeferrableUpdate objects for removing information about content
1418 * in some secondary data store. This is used when a page is deleted, and also when
1419 * a slot is removed from a page.
1421 * Application logic should not call this method directly. Instead, it should call
1422 * WikiPage::getSecondaryDataUpdates().
1424 * @note Implementations must not return a LinksDeletionUpdate instance. Instead, a
1425 * LinksDeletionUpdate is created by the calling code in WikiPage.
1426 * This is in contrast to the old getDeletionUpdates method defined by AbstractContent,
1427 * which returned a LinksUpdate.
1429 * @note Implementations should not rely on the page's current content, but rather the current
1430 * state of the secondary data store.
1432 * @param Title $title The title of the page to supply the updates for
1433 * @param string $role The role (slot) in which the content is being used. Which updates
1434 * are performed should generally not depend on the role the content has, but the
1435 * DeferrableUpdates themselves may need to know the role, to track to which slot the
1436 * data refers, and to avoid overwriting data of the same kind from another slot.
1438 * @return DeferrableUpdate[] A list of DeferrableUpdate objects for putting information
1439 * about this content object somewhere. The default implementation returns an empty
1444 public function getDeletionUpdates( Title
$title, $role ) {