RenderedRevision.php

Go to the documentation of this file.

<?php
namespace MediaWiki\Revision;
 
use Content;
use InvalidArgumentException;
use LogicException;
use MediaWiki\Content\Renderer\ContentRenderer;
use MediaWiki\Page\PageReference;
use MediaWiki\Parser\ParserOutputFlags;
use MediaWiki\Permissions\Authority;
use ParserOptions;
use ParserOutput;
use Psr\Log\LoggerInterface;
use Psr\Log\NullLogger;
use Wikimedia\Assert\Assert;
 
class RenderedRevision implements SlotRenderingProvider {
 
    private $revision;
 
    private $options;
 
    private $audience = RevisionRecord::FOR_PUBLIC;
 
    private $performer = null;
 
    private $revisionOutput = null;
 
    private $slotsOutput = [];
 
    private $combineOutput;
 
    private $saveParseLogger;
 
    private $contentRenderer;
 
    public function __construct(
        RevisionRecord $revision,
        ParserOptions $options,
        ContentRenderer $contentRenderer,
        callable $combineOutput,
        $audience = RevisionRecord::FOR_PUBLIC,
        Authority $performer = null
    ) {
        $this->options = $options;
 
        $this->setRevisionInternal( $revision );
 
        $this->contentRenderer = $contentRenderer;
        $this->combineOutput = $combineOutput;
        $this->saveParseLogger = new NullLogger();
 
        if ( $audience === RevisionRecord::FOR_THIS_USER && !$performer ) {
            throw new InvalidArgumentException(
                'User must be specified when setting audience to FOR_THIS_USER'
            );
        }
 
        $this->audience = $audience;
        $this->performer = $performer;
    }
 
    public function setSaveParseLogger( LoggerInterface $saveParseLogger ) {
        $this->saveParseLogger = $saveParseLogger;
    }
 
    public function isContentDeleted() {
        return $this->revision->isDeleted( RevisionRecord::DELETED_TEXT );
    }
 
    public function getRevision() {
        return $this->revision;
    }
 
    public function getOptions() {
        return $this->options;
    }
 
    public function setRevisionParserOutput( ParserOutput $output ) {
        $this->revisionOutput = $output;
 
        // If there is only one slot, we assume that the combined output is identical
        // with the main slot's output. This is intended to prevent a redundant re-parse of
        // the content in case getSlotParserOutput( SlotRecord::MAIN ) is called, for instance
        // from ContentHandler::getSecondaryDataUpdates.
        if ( $this->revision->getSlotRoles() === [ SlotRecord::MAIN ] ) {
            $this->slotsOutput[ SlotRecord::MAIN ] = $output;
        }
    }
 
    public function getRevisionParserOutput( array $hints = [] ) {
        $withHtml = $hints['generate-html'] ?? true;
 
        if ( !$this->revisionOutput
            || ( $withHtml && !$this->revisionOutput->hasText() )
        ) {
            $output = call_user_func( $this->combineOutput, $this, $hints );
 
            Assert::postcondition(
                $output instanceof ParserOutput,
                'Callback did not return a ParserOutput object!'
            );
 
            $this->revisionOutput = $output;
        }
 
        return $this->revisionOutput;
    }
 
    public function getSlotParserOutput( $role, array $hints = [] ) {
        $withHtml = $hints['generate-html'] ?? true;
 
        if ( !isset( $this->slotsOutput[ $role ] )
            || ( $withHtml && !$this->slotsOutput[ $role ]->hasText() )
        ) {
            $content = $this->revision->getContentOrThrow( $role, $this->audience, $this->performer );
 
            // XXX: allow SlotRoleHandler to control the ParserOutput?
            $output = $this->getSlotParserOutputUncached( $content, $withHtml );
 
            if ( $withHtml && !$output->hasText() ) {
                throw new LogicException(
                    'HTML generation was requested, but '
                    . get_class( $content )
                    . ' that passed to '
                    . 'ContentRenderer::getParserOutput() returns a ParserOutput with no text set.'
                );
            }
 
            // Detach watcher, to ensure option use is not recorded in the wrong ParserOutput.
            $this->options->registerWatcher( null );
 
            $this->slotsOutput[ $role ] = $output;
        }
 
        return $this->slotsOutput[$role];
    }
 
    private function getSlotParserOutputUncached( Content $content, $withHtml ) {
        $parserOutput = $this->contentRenderer->getParserOutput(
            $content,
            $this->revision->getPage(),
            $this->revision->getId(),
            $this->options,
            $withHtml
        );
        // Save the rev_id and timestamp so that we don't have to load the revision row on view
        $parserOutput->setCacheRevisionId( $this->revision->getId() );
        $parserOutput->setTimestamp( $this->revision->getTimestamp() );
        return $parserOutput;
    }
 
    public function updateRevision( RevisionRecord $rev ) {
        if ( $rev->getId() === $this->revision->getId() ) {
            return;
        }
 
        if ( $this->revision->getId() ) {
            throw new LogicException( 'RenderedRevision already has a revision with ID '
                . $this->revision->getId() . ', can\'t update to revision with ID ' . $rev->getId() );
        }
 
        if ( !$this->revision->getSlots()->hasSameContent( $rev->getSlots() ) ) {
            throw new LogicException( 'Cannot update to a revision with different content!' );
        }
 
        $this->setRevisionInternal( $rev );
 
        $this->pruneRevisionSensitiveOutput(
            $this->revision->getPageId(),
            $this->revision->getId(),
            $this->revision->getTimestamp()
        );
    }
 
    private function pruneRevisionSensitiveOutput(
        $actualPageId,
        $actualRevId,
        $actualRevTimestamp
    ) {
        if ( $this->revisionOutput ) {
            if ( $this->outputVariesOnRevisionMetaData(
                $this->revisionOutput,
                $actualPageId,
                $actualRevId,
                $actualRevTimestamp
            ) ) {
                $this->revisionOutput = null;
            }
        } else {
            $this->saveParseLogger->debug( __METHOD__ . ": no prepared revision output" );
        }
 
        foreach ( $this->slotsOutput as $role => $output ) {
            if ( $this->outputVariesOnRevisionMetaData(
                $output,
                $actualPageId,
                $actualRevId,
                $actualRevTimestamp
            ) ) {
                unset( $this->slotsOutput[$role] );
            }
        }
    }
 
    private function setRevisionInternal( RevisionRecord $revision ) {
        $this->revision = $revision;
 
        // Force the parser to use  $this->revision to resolve magic words like {{REVISIONUSER}}
        // if the revision is either known to be complete, or it doesn't have a revision ID set.
        // If it's incomplete and we have a revision ID, the parser can do better by loading
        // the revision from the database if needed to handle a magic word.
        //
        // The following considerations inform the logic described above:
        //
        // 1) If we have a saved revision already loaded, we want the parser to use it, instead of
        // loading it again.
        //
        // 2) If the revision is a fake that wraps some kind of synthetic content, such as an
        // error message from Article, it should be used directly and things like {{REVISIONUSER}}
        // should not expected to work, since there may not even be an actual revision to
        // refer to.
        //
        // 3) If the revision is a fake constructed around a page, a Content object, and
        // a revision ID, to provide backwards compatibility to code that has access to those
        // but not to a complete RevisionRecord for rendering, then we want the Parser to
        // load the actual revision from the database when it encounters a magic word like
        // {{REVISIONUSER}}, but we don't want to load that revision ahead of time just in case.
        //
        // 4) Previewing an edit to a template should use the submitted unsaved
        // MutableRevisionRecord for self-transclusions in the template's documentation (see T7278).
        // That revision would be complete except for the ID field.
        //
        // 5) Pre-save transform would provide a RevisionRecord that has all meta-data but is
        // incomplete due to not yet having content set. However, since it doesn't have a revision
        // ID either, the below code would still force it to be used, allowing
        // {{subst::REVISIONUSER}} to function as expected.
 
        if ( $this->revision->isReadyForInsertion() || !$this->revision->getId() ) {
            $oldCallback = $this->options->getCurrentRevisionRecordCallback();
            $this->options->setCurrentRevisionRecordCallback(
                function ( PageReference $parserPage, $parser = null ) use ( $oldCallback ) {
                    if ( $this->revision->getPage()->isSamePageAs( $parserPage ) ) {
                        return $this->revision;
                    } else {
                        return call_user_func( $oldCallback, $parserPage, $parser );
                    }
                }
            );
        }
    }
 
    private function outputVariesOnRevisionMetaData(
        ParserOutput $parserOutput,
        $actualPageId,
        $actualRevId,
        $actualRevTimestamp
    ) {
        $logger = $this->saveParseLogger;
        $varyMsg = __METHOD__ . ": cannot use prepared output for '{title}'";
        $context = [ 'title' => (string)$this->revision->getPage() ];
 
        if ( $parserOutput->getOutputFlag( ParserOutputFlags::VARY_REVISION ) ) {
            // If {{PAGEID}} resolved to 0, then that word need to resolve to the actual page ID
            $logger->info( "$varyMsg (vary-revision)", $context );
            return true;
        } elseif (
            $parserOutput->getOutputFlag( ParserOutputFlags::VARY_REVISION_ID )
            && $actualRevId !== false
            && ( $actualRevId === true || $parserOutput->getSpeculativeRevIdUsed() !== $actualRevId )
        ) {
            $logger->info( "$varyMsg (vary-revision-id and wrong ID)", $context );
            return true;
        } elseif (
            $parserOutput->getOutputFlag( ParserOutputFlags::VARY_REVISION_TIMESTAMP )
            && $actualRevTimestamp !== false
            && ( $actualRevTimestamp === true ||
                $parserOutput->getRevisionTimestampUsed() !== $actualRevTimestamp )
        ) {
            $logger->info( "$varyMsg (vary-revision-timestamp and wrong timestamp)", $context );
            return true;
        } elseif (
            $parserOutput->getOutputFlag( ParserOutputFlags::VARY_PAGE_ID )
            && $actualPageId !== false
            && ( $actualPageId === true || $parserOutput->getSpeculativePageIdUsed() !== $actualPageId )
        ) {
            $logger->info( "$varyMsg (vary-page-id and wrong ID)", $context );
            return true;
        } elseif ( $parserOutput->getOutputFlag( ParserOutputFlags::VARY_REVISION_EXISTS ) ) {
            // If {{REVISIONID}} resolved to '', it now needs to resolve to '-'.
            // Note that edit stashing always uses '-', which can be used for both
            // edit filter checks and canonical parser cache.
            $logger->info( "$varyMsg (vary-revision-exists)", $context );
            return true;
        } elseif (
            $parserOutput->getOutputFlag( ParserOutputFlags::VARY_REVISION_SHA1 ) &&
            $parserOutput->getRevisionUsedSha1Base36() !== $this->revision->getSha1()
        ) {
            // If a self-transclusion used the proposed page text, it must match the final
            // page content after PST transformations and automatically merged edit conflicts
            $logger->info( "$varyMsg (vary-revision-sha1 with wrong SHA-1)", $context );
            return true;
        }
 
        // NOTE: In the original fix for T135261, the output was discarded if ParserOutputFlags::VARY_USER was
        // set for a null-edit. The reason was that the original rendering in that case was
        // targeting the user making the null-edit, not the user who made the original edit,
        // causing {{REVISIONUSER}} to return the wrong name.
        // This case is now expected to be handled by the code in RevisionRenderer that
        // constructs the ParserOptions: For a null-edit, setCurrentRevisionRecordCallback is
        // called with the old, existing revision.
        $logger->debug( __METHOD__ . ": reusing prepared output for '{title}'", $context );
        return false;
    }
}