WikiImporter.php

Go to the documentation of this file.

<?php
use MediaWiki\Cache\CacheKeyHelper;
use MediaWiki\Content\IContentHandlerFactory;
use MediaWiki\HookContainer\HookContainer;
use MediaWiki\HookContainer\HookRunner;
use MediaWiki\MainConfigNames;
use MediaWiki\MediaWikiServices;
use MediaWiki\Page\PageIdentity;
use MediaWiki\Page\WikiPageFactory;
use MediaWiki\Permissions\PermissionManager;
use MediaWiki\Revision\SlotRecord;
use MediaWiki\Revision\SlotRoleRegistry;
use MediaWiki\Title\Title;
use MediaWiki\Title\TitleFactory;
use Wikimedia\AtEase\AtEase;
use Wikimedia\NormalizedException\NormalizedException;
 
class WikiImporter {
    private $reader;
 
    private $sourceAdapterId;
 
    private $foreignNamespaces = null;
 
    private $mLogItemCallback;
 
    private $mUploadCallback;
 
    private $mRevisionCallback;
 
    private $mPageCallback;
 
    private $mSiteInfoCallback;
 
    private $mPageOutCallback;
 
    private $mNoticeCallback;
 
    private $mDebug;
 
    private $mImportUploads;
 
    private $mImageBasePath;
 
    private $mNoUpdates = false;
 
    private $pageOffset = 0;
 
    private $config;
 
    private $importTitleFactory;
 
    private $hookRunner;
 
    private $countableCache = [];
 
    private $disableStatisticsUpdate = false;
 
    private $externalUserNames;
 
    private $contentLanguage;
 
    private $namespaceInfo;
 
    private $titleFactory;
 
    private $wikiPageFactory;
 
    private $uploadRevisionImporter;
 
    private $permissionManager;
 
    private $contentHandlerFactory;
 
    private $slotRoleRegistry;
 
    public function __construct(
        ImportSource $source,
        Config $config,
        HookContainer $hookContainer,
        Language $contentLanguage,
        NamespaceInfo $namespaceInfo,
        TitleFactory $titleFactory,
        WikiPageFactory $wikiPageFactory,
        UploadRevisionImporter $uploadRevisionImporter,
        PermissionManager $permissionManager,
        IContentHandlerFactory $contentHandlerFactory,
        SlotRoleRegistry $slotRoleRegistry
    ) {
        $this->config = $config;
        $this->hookRunner = new HookRunner( $hookContainer );
        $this->contentLanguage = $contentLanguage;
        $this->namespaceInfo = $namespaceInfo;
        $this->titleFactory = $titleFactory;
        $this->wikiPageFactory = $wikiPageFactory;
        $this->uploadRevisionImporter = $uploadRevisionImporter;
        $this->permissionManager = $permissionManager;
        $this->contentHandlerFactory = $contentHandlerFactory;
        $this->slotRoleRegistry = $slotRoleRegistry;
 
        if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
            stream_wrapper_register( 'uploadsource', UploadSourceAdapter::class );
        }
        $this->sourceAdapterId = UploadSourceAdapter::registerSource( $source );
 
        $this->openReader();
 
        // Default callbacks
        $this->setPageCallback( [ $this, 'beforeImportPage' ] );
        $this->setRevisionCallback( [ $this, "importRevision" ] );
        $this->setUploadCallback( [ $this, 'importUpload' ] );
        $this->setLogItemCallback( [ $this, 'importLogItem' ] );
        $this->setPageOutCallback( [ $this, 'finishImportPage' ] );
 
        $this->importTitleFactory = new NaiveImportTitleFactory(
            $this->contentLanguage,
            $this->namespaceInfo,
            $this->titleFactory
        );
        $this->externalUserNames = new ExternalUserNames( 'imported', false );
    }
 
    public function getReader() {
        return $this->reader;
    }
 
    public function throwXmlError( $err ) {
        $this->debug( "FAILURE: $err" );
        wfDebug( "WikiImporter XML error: $err" );
    }
 
    public function debug( $data ) {
        if ( $this->mDebug ) {
            wfDebug( "IMPORT: $data" );
        }
    }
 
    public function warn( $data ) {
        wfDebug( "IMPORT: $data" );
    }
 
    public function notice( $msg, ...$params ) {
        if ( is_callable( $this->mNoticeCallback ) ) {
            call_user_func( $this->mNoticeCallback, $msg, $params );
        } else { # No ImportReporter -> CLI
            // T177997: the command line importers should call setNoticeCallback()
            // for their own custom callback to echo the notice
            wfDebug( wfMessage( $msg, $params )->text() );
        }
    }
 
    public function setDebug( $debug ) {
        $this->mDebug = $debug;
    }
 
    public function setNoUpdates( $noupdates ) {
        $this->mNoUpdates = $noupdates;
    }
 
    public function setPageOffset( $nthPage ) {
        $this->pageOffset = $nthPage;
    }
 
    public function setNoticeCallback( $callback ) {
        return wfSetVar( $this->mNoticeCallback, $callback );
    }
 
    public function setPageCallback( $callback ) {
        $previous = $this->mPageCallback;
        $this->mPageCallback = $callback;
        return $previous;
    }
 
    public function setPageOutCallback( $callback ) {
        $previous = $this->mPageOutCallback;
        $this->mPageOutCallback = $callback;
        return $previous;
    }
 
    public function setRevisionCallback( $callback ) {
        $previous = $this->mRevisionCallback;
        $this->mRevisionCallback = $callback;
        return $previous;
    }
 
    public function setUploadCallback( $callback ) {
        $previous = $this->mUploadCallback;
        $this->mUploadCallback = $callback;
        return $previous;
    }
 
    public function setLogItemCallback( $callback ) {
        $previous = $this->mLogItemCallback;
        $this->mLogItemCallback = $callback;
        return $previous;
    }
 
    public function setSiteInfoCallback( $callback ) {
        $previous = $this->mSiteInfoCallback;
        $this->mSiteInfoCallback = $callback;
        return $previous;
    }
 
    public function setImportTitleFactory( $factory ) {
        $this->importTitleFactory = $factory;
    }
 
    public function setTargetNamespace( $namespace ) {
        if ( $namespace === null ) {
            // Don't override namespaces
            $this->setImportTitleFactory(
                new NaiveImportTitleFactory(
                    $this->contentLanguage,
                    $this->namespaceInfo,
                    $this->titleFactory
                )
            );
            return true;
        } elseif (
            $namespace >= 0 &&
            $this->namespaceInfo->exists( intval( $namespace ) )
        ) {
            $namespace = intval( $namespace );
            $this->setImportTitleFactory(
                new NamespaceImportTitleFactory(
                    $this->namespaceInfo,
                    $this->titleFactory,
                    $namespace
                )
            );
            return true;
        } else {
            return false;
        }
    }
 
    public function setTargetRootPage( $rootpage ) {
        $status = Status::newGood();
        $nsInfo = $this->namespaceInfo;
        if ( $rootpage === null ) {
            // No rootpage
            $this->setImportTitleFactory(
                new NaiveImportTitleFactory(
                    $this->contentLanguage,
                    $nsInfo,
                    $this->titleFactory
                )
            );
        } elseif ( $rootpage !== '' ) {
            $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
            $title = Title::newFromText( $rootpage );
 
            if ( !$title || $title->isExternal() ) {
                $status->fatal( 'import-rootpage-invalid' );
            } elseif ( !$nsInfo->hasSubpages( $title->getNamespace() ) ) {
                $displayNSText = $title->getNamespace() === NS_MAIN
                    ? wfMessage( 'blanknamespace' )->text()
                    : $this->contentLanguage->getNsText( $title->getNamespace() );
                $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
            } else {
                // set namespace to 'all', so the namespace check in processTitle() can pass
                $this->setTargetNamespace( null );
                $this->setImportTitleFactory(
                    new SubpageImportTitleFactory(
                        $nsInfo,
                        $this->titleFactory,
                        $title
                    )
                );
            }
        }
        return $status;
    }
 
    public function setImageBasePath( $dir ) {
        $this->mImageBasePath = $dir;
    }
 
    public function setImportUploads( $import ) {
        $this->mImportUploads = $import;
    }
 
    public function setUsernamePrefix( $usernamePrefix, $assignKnownUsers ) {
        $this->externalUserNames = new ExternalUserNames( $usernamePrefix, $assignKnownUsers );
    }
 
    public function disableStatisticsUpdate() {
        $this->disableStatisticsUpdate = true;
    }
 
    public function beforeImportPage( $titleAndForeignTitle ) {
        $title = $titleAndForeignTitle[0];
        $page = $this->wikiPageFactory->newFromTitle( $title );
        $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
        return true;
    }
 
    public function importRevision( $revision ) {
        if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
            $this->notice( 'import-error-bad-location',
                $revision->getTitle()->getPrefixedText(),
                $revision->getID(),
                $revision->getModel(),
                $revision->getFormat()
            );
 
            return false;
        }
 
        try {
            return $revision->importOldRevision();
        } catch ( MWContentSerializationException $ex ) {
            $this->notice( 'import-error-unserialize',
                $revision->getTitle()->getPrefixedText(),
                $revision->getID(),
                $revision->getModel(),
                $revision->getFormat()
            );
        }
 
        return false;
    }
 
    public function importLogItem( $revision ) {
        return $revision->importLogItem();
    }
 
    public function importUpload( $revision ) {
        $status = $this->uploadRevisionImporter->import( $revision );
        return $status->isGood();
    }
 
    public function finishImportPage( PageIdentity $pageIdentity, $foreignTitle, $revCount,
        $sRevCount, $pageInfo
    ) {
        // Update article count statistics (T42009)
        // The normal counting logic in WikiPage->doEditUpdates() is designed for
        // one-revision-at-a-time editing, not bulk imports. In this situation it
        // suffers from issues of replica DB lag. We let WikiPage handle the total page
        // and revision count, and we implement our own custom logic for the
        // article (content page) count.
        if ( !$this->disableStatisticsUpdate ) {
            $page = $this->wikiPageFactory->newFromTitle( $pageIdentity );
 
            $page->loadPageData( WikiPage::READ_LATEST );
            $rev = $page->getRevisionRecord();
            if ( $rev === null ) {
 
                wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $pageIdentity .
                    ' because WikiPage::getRevisionRecord() returned null' );
            } else {
                $user = RequestContext::getMain()->getUser();
                $update = $page->newPageUpdater( $user )->prepareUpdate();
                $countKey = 'title_' . CacheKeyHelper::getKeyForPage( $pageIdentity );
                $countable = $update->isCountable();
                if ( array_key_exists( $countKey, $this->countableCache ) &&
                    $countable != $this->countableCache[$countKey] ) {
                    DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
                        'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
                    ] ) );
                }
            }
        }
 
        $title = Title::castFromPageIdentity( $pageIdentity );
        // @phan-suppress-next-line PhanTypeMismatchArgumentNullable castFrom does not return null here
        return $this->hookRunner->onAfterImportPage( $title, $foreignTitle,
            $revCount, $sRevCount, $pageInfo );
    }
 
    private function siteInfoCallback( $siteInfo ) {
        if ( isset( $this->mSiteInfoCallback ) ) {
            return call_user_func_array(
                $this->mSiteInfoCallback,
                [ $siteInfo, $this ]
            );
        } else {
            return false;
        }
    }
 
    public function pageCallback( $title ) {
        if ( isset( $this->mPageCallback ) ) {
            call_user_func( $this->mPageCallback, $title );
        }
    }
 
    private function pageOutCallback( PageIdentity $pageIdentity, $foreignTitle, $revCount,
            $sucCount, $pageInfo ) {
        if ( isset( $this->mPageOutCallback ) ) {
            call_user_func_array( $this->mPageOutCallback, func_get_args() );
        }
    }
 
    private function revisionCallback( $revision ) {
        if ( isset( $this->mRevisionCallback ) ) {
            return call_user_func_array(
                $this->mRevisionCallback,
                [ $revision, $this ]
            );
        } else {
            return false;
        }
    }
 
    private function logItemCallback( $revision ) {
        if ( isset( $this->mLogItemCallback ) ) {
            return call_user_func_array(
                $this->mLogItemCallback,
                [ $revision, $this ]
            );
        } else {
            return false;
        }
    }
 
    public function nodeAttribute( $attr ) {
        return $this->reader->getAttribute( $attr ) ?? '';
    }
 
    public function nodeContents() {
        if ( $this->reader->isEmptyElement ) {
            return "";
        }
        $buffer = "";
        while ( $this->reader->read() ) {
            switch ( $this->reader->nodeType ) {
                case XMLReader::TEXT:
                case XMLReader::CDATA:
                case XMLReader::SIGNIFICANT_WHITESPACE:
                    $buffer .= $this->reader->value;
                    break;
                case XMLReader::END_ELEMENT:
                    return $buffer;
            }
        }
 
        $this->reader->close();
        return '';
    }
 
    public function doImport() {
        $this->syntaxCheckXML();
 
        // Calls to reader->read need to be wrapped in calls to
        // libxml_disable_entity_loader() to avoid local file
        // inclusion attacks (T48932).
        // phpcs:ignore Generic.PHP.NoSilencedErrors -- suppress deprecation per T268847
        $oldDisable = @libxml_disable_entity_loader( true );
        try {
            $this->reader->read();
 
            if ( $this->reader->localName != 'mediawiki' ) {
                // phpcs:ignore Generic.PHP.NoSilencedErrors
                @libxml_disable_entity_loader( $oldDisable );
                $error = libxml_get_last_error();
                if ( $error ) {
                    throw new NormalizedException( "XML error at line {line}: {message}", [
                        'line' => $error->line,
                        'message' => $error->message,
                    ] );
                } else {
                    throw new MWException(
                        "Expected '<mediawiki>' tag, got '<{$this->reader->localName}>' tag."
                    );
                }
            }
            $this->debug( "<mediawiki> tag is correct." );
 
            $this->debug( "Starting primary dump processing loop." );
 
            $keepReading = $this->reader->read();
            $skip = false;
            $pageCount = 0;
            while ( $keepReading ) {
                $tag = $this->reader->localName;
                if ( $this->pageOffset ) {
                    if ( $tag === 'page' ) {
                        $pageCount++;
                    }
                    if ( $pageCount < $this->pageOffset ) {
                        $keepReading = $this->reader->next();
                        continue;
                    }
                }
                $type = $this->reader->nodeType;
 
                if ( !$this->hookRunner->onImportHandleToplevelXMLTag( $this ) ) {
                    // Do nothing
                } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
                    break;
                } elseif ( $tag == 'siteinfo' ) {
                    $this->handleSiteInfo();
                } elseif ( $tag == 'page' ) {
                    $this->handlePage();
                } elseif ( $tag == 'logitem' ) {
                    $this->handleLogItem();
                } elseif ( $tag != '#text' ) {
                    $this->warn( "Unhandled top-level XML tag $tag" );
 
                    $skip = true;
                }
 
                if ( $skip ) {
                    $keepReading = $this->reader->next();
                    $skip = false;
                    $this->debug( "Skip" );
                } else {
                    $keepReading = $this->reader->read();
                }
            }
        } finally {
            // phpcs:ignore Generic.PHP.NoSilencedErrors
            @libxml_disable_entity_loader( $oldDisable );
            $this->reader->close();
        }
 
        return true;
    }
 
    private function handleSiteInfo() {
        $this->debug( "Enter site info handler." );
        $siteInfo = [];
 
        // Fields that can just be stuffed in the siteInfo object
        $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
 
        while ( $this->reader->read() ) {
            if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
                    $this->reader->localName == 'siteinfo' ) {
                break;
            }
 
            $tag = $this->reader->localName;
 
            if ( $tag == 'namespace' ) {
                $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
                    $this->nodeContents();
            } elseif ( in_array( $tag, $normalFields ) ) {
                $siteInfo[$tag] = $this->nodeContents();
            }
        }
 
        $siteInfo['_namespaces'] = $this->foreignNamespaces;
        $this->siteInfoCallback( $siteInfo );
    }
 
    private function handleLogItem() {
        $this->debug( "Enter log item handler." );
        $logInfo = [];
 
        // Fields that can just be stuffed in the pageInfo object
        $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
            'logtitle', 'params' ];
 
        while ( $this->reader->read() ) {
            if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
                    $this->reader->localName == 'logitem' ) {
                break;
            }
 
            $tag = $this->reader->localName;
 
            if ( !$this->hookRunner->onImportHandleLogItemXMLTag( $this, $logInfo ) ) {
                // Do nothing
            } elseif ( in_array( $tag, $normalFields ) ) {
                $logInfo[$tag] = $this->nodeContents();
            } elseif ( $tag == 'contributor' ) {
                $logInfo['contributor'] = $this->handleContributor();
            } elseif ( $tag != '#text' ) {
                $this->warn( "Unhandled log-item XML tag $tag" );
            }
        }
 
        $this->processLogItem( $logInfo );
    }
 
    private function processLogItem( $logInfo ) {
        $revision = new WikiRevision();
 
        if ( isset( $logInfo['id'] ) ) {
            $revision->setID( $logInfo['id'] );
        }
        $revision->setType( $logInfo['type'] );
        $revision->setAction( $logInfo['action'] );
        if ( isset( $logInfo['timestamp'] ) ) {
            $revision->setTimestamp( $logInfo['timestamp'] );
        }
        if ( isset( $logInfo['params'] ) ) {
            $revision->setParams( $logInfo['params'] );
        }
        if ( isset( $logInfo['logtitle'] ) ) {
            // @todo Using Title for non-local titles is a recipe for disaster.
            // We should use ForeignTitle here instead.
            $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
        }
 
        $revision->setNoUpdates( $this->mNoUpdates );
 
        if ( isset( $logInfo['comment'] ) ) {
            $revision->setComment( $logInfo['comment'] );
        }
 
        if ( isset( $logInfo['contributor']['username'] ) ) {
            $revision->setUsername(
                $this->externalUserNames->applyPrefix( $logInfo['contributor']['username'] )
            );
        } elseif ( isset( $logInfo['contributor']['ip'] ) ) {
            $revision->setUserIP( $logInfo['contributor']['ip'] );
        } else {
            $revision->setUsername( $this->externalUserNames->addPrefix( 'Unknown user' ) );
        }
 
        return $this->logItemCallback( $revision );
    }
 
    private function handlePage() {
        // Handle page data.
        $this->debug( "Enter page handler." );
        $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
 
        // Fields that can just be stuffed in the pageInfo object
        $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
 
        $skip = false;
        $badTitle = false;
 
        while ( $skip ? $this->reader->next() : $this->reader->read() ) {
            if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
                    $this->reader->localName == 'page' ) {
                break;
            }
 
            $skip = false;
 
            $tag = $this->reader->localName;
 
            if ( $badTitle ) {
                // The title is invalid, bail out of this page
                $skip = true;
            } elseif ( !$this->hookRunner->onImportHandlePageXMLTag( $this, $pageInfo ) ) {
                // Do nothing
            } elseif ( in_array( $tag, $normalFields ) ) {
                // An XML snippet:
                // <page>
                //     <id>123</id>
                //     <title>Page</title>
                //     <redirect title="NewTitle"/>
                //     ...
                // Because the redirect tag is built differently, we need special handling for that case.
                if ( $tag == 'redirect' ) {
                    $pageInfo[$tag] = $this->nodeAttribute( 'title' );
                } else {
                    $pageInfo[$tag] = $this->nodeContents();
                }
            } elseif ( $tag == 'revision' || $tag == 'upload' ) {
                if ( !isset( $title ) ) {
                    $title = $this->processTitle( $pageInfo['title'],
                        $pageInfo['ns'] ?? null );
 
                    // $title is either an array of two titles or false.
                    if ( is_array( $title ) ) {
                        $this->pageCallback( $title );
                        [ $pageInfo['_title'], $foreignTitle ] = $title;
                    } else {
                        $badTitle = true;
                        $skip = true;
                    }
                }
 
                if ( $title ) {
                    if ( $tag == 'revision' ) {
                        $this->handleRevision( $pageInfo );
                    } else {
                        $this->handleUpload( $pageInfo );
                    }
                }
            } elseif ( $tag != '#text' ) {
                $this->warn( "Unhandled page XML tag $tag" );
                $skip = true;
            }
        }
 
        // @note $pageInfo is only set if a valid $title is processed above with
        //       no error. If we have a valid $title, then pageCallback is called
        //       above, $pageInfo['title'] is set and we do pageOutCallback here.
        //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
        //       set since they both come from $title above.
        if ( array_key_exists( '_title', $pageInfo ) ) {
            $title = $pageInfo['_title'];
            $this->pageOutCallback(
                $title,
                // @phan-suppress-next-line PhanPossiblyUndeclaredVariable Set together with _title key
                $foreignTitle,
                $pageInfo['revisionCount'],
                $pageInfo['successfulRevisionCount'],
                $pageInfo
            );
        }
    }
 
    private function handleRevision( &$pageInfo ) {
        $this->debug( "Enter revision handler" );
        $revisionInfo = [];
 
        $normalFields = [ 'id', 'parentid', 'timestamp', 'comment', 'minor', 'origin',
            'model', 'format', 'text', 'sha1' ];
 
        $skip = false;
 
        while ( $skip ? $this->reader->next() : $this->reader->read() ) {
            if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
                    $this->reader->localName == 'revision' ) {
                break;
            }
 
            $tag = $this->reader->localName;
 
            if ( !$this->hookRunner->onImportHandleRevisionXMLTag(
                $this, $pageInfo, $revisionInfo )
            ) {
                // Do nothing
            } elseif ( in_array( $tag, $normalFields ) ) {
                $revisionInfo[$tag] = $this->nodeContents();
            } elseif ( $tag == 'content' ) {
                // We can have multiple content tags, so make this an array.
                $revisionInfo[$tag][] = $this->handleContent();
            } elseif ( $tag == 'contributor' ) {
                $revisionInfo['contributor'] = $this->handleContributor();
            } elseif ( $tag != '#text' ) {
                $this->warn( "Unhandled revision XML tag $tag" );
                $skip = true;
            }
        }
 
        $pageInfo['revisionCount']++;
        if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
            $pageInfo['successfulRevisionCount']++;
        }
    }
 
    private function handleContent() {
        $this->debug( "Enter content handler" );
        $contentInfo = [];
 
        $normalFields = [ 'role', 'origin', 'model', 'format', 'text' ];
 
        $skip = false;
 
        while ( $skip ? $this->reader->next() : $this->reader->read() ) {
            if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
                $this->reader->localName == 'content' ) {
                break;
            }
 
            $tag = $this->reader->localName;
 
            if ( !$this->hookRunner->onImportHandleContentXMLTag(
                $this, $contentInfo )
            ) {
                // Do nothing
            } elseif ( in_array( $tag, $normalFields ) ) {
                $contentInfo[$tag] = $this->nodeContents();
            } elseif ( $tag != '#text' ) {
                $this->warn( "Unhandled content XML tag $tag" );
                $skip = true;
            }
        }
 
        return $contentInfo;
    }
 
    private function makeContent( Title $title, $revisionId, $contentInfo ) {
        $maxArticleSize = MediaWikiServices::getInstance()->getMainConfig()->get(
            MainConfigNames::MaxArticleSize );
 
        if ( !isset( $contentInfo['text'] ) ) {
            throw new MWException( 'Missing text field in import.' );
        }
 
        // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
        // database errors and instability. Testing for revisions with only listed
        // content models, as other content models might use serialization formats
        // which aren't checked against $wgMaxArticleSize.
        if ( ( !isset( $contentInfo['model'] ) ||
                in_array( $contentInfo['model'], [
                    'wikitext',
                    'css',
                    'json',
                    'javascript',
                    'text',
                    ''
                ] ) ) &&
            strlen( $contentInfo['text'] ) > $maxArticleSize * 1024
        ) {
            throw new MWException( 'The text of ' .
                ( $revisionId ?
                    "the revision with ID $revisionId" :
                    'a revision'
                ) . " exceeds the maximum allowable size ({$maxArticleSize} KiB)" );
        }
 
        $role = $contentInfo['role'] ?? SlotRecord::MAIN;
        $model = $contentInfo['model'] ?? $this->getDefaultContentModel( $title, $role );
        $handler = $this->getContentHandler( $model );
 
        $text = $handler->importTransform( $contentInfo['text'] );
 
        return $handler->unserializeContent( $text );
    }
 
    private function processRevision( $pageInfo, $revisionInfo ) {
        $revision = new WikiRevision();
 
        $revId = $revisionInfo['id'] ?? 0;
        if ( $revId ) {
            $revision->setID( $revisionInfo['id'] );
        }
 
        $title = $pageInfo['_title'];
        $revision->setTitle( $title );
 
        $content = $this->makeContent( $title, $revId, $revisionInfo );
        $revision->setContent( SlotRecord::MAIN, $content );
 
        foreach ( $revisionInfo['content'] ?? [] as $slotInfo ) {
            if ( !isset( $slotInfo['role'] ) ) {
                throw new MWException( "Missing role for imported slot." );
            }
 
            $content = $this->makeContent( $title, $revId, $slotInfo );
            $revision->setContent( $slotInfo['role'], $content );
        }
        $revision->setTimestamp( $revisionInfo['timestamp'] ?? wfTimestampNow() );
 
        if ( isset( $revisionInfo['comment'] ) ) {
            $revision->setComment( $revisionInfo['comment'] );
        }
 
        if ( isset( $revisionInfo['minor'] ) ) {
            $revision->setMinor( true );
        }
        if ( isset( $revisionInfo['contributor']['username'] ) ) {
            $revision->setUsername(
                $this->externalUserNames->applyPrefix( $revisionInfo['contributor']['username'] )
            );
        } elseif ( isset( $revisionInfo['contributor']['ip'] ) ) {
            $revision->setUserIP( $revisionInfo['contributor']['ip'] );
        } else {
            $revision->setUsername( $this->externalUserNames->addPrefix( 'Unknown user' ) );
        }
        if ( isset( $revisionInfo['sha1'] ) ) {
            $revision->setSha1Base36( $revisionInfo['sha1'] );
        }
        $revision->setNoUpdates( $this->mNoUpdates );
 
        return $this->revisionCallback( $revision );
    }
 
    private function handleUpload( &$pageInfo ) {
        $this->debug( "Enter upload handler" );
        $uploadInfo = [];
 
        $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
            'src', 'size', 'sha1base36', 'archivename', 'rel' ];
 
        $skip = false;
 
        while ( $skip ? $this->reader->next() : $this->reader->read() ) {
            if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
                    $this->reader->localName == 'upload' ) {
                break;
            }
 
            $tag = $this->reader->localName;
 
            if ( !$this->hookRunner->onImportHandleUploadXMLTag( $this, $pageInfo ) ) {
                // Do nothing
            } elseif ( in_array( $tag, $normalFields ) ) {
                $uploadInfo[$tag] = $this->nodeContents();
            } elseif ( $tag == 'contributor' ) {
                $uploadInfo['contributor'] = $this->handleContributor();
            } elseif ( $tag == 'contents' ) {
                $contents = $this->nodeContents();
                $encoding = $this->reader->getAttribute( 'encoding' );
                if ( $encoding === 'base64' ) {
                    $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
                    $uploadInfo['isTempSrc'] = true;
                }
            } elseif ( $tag != '#text' ) {
                $this->warn( "Unhandled upload XML tag $tag" );
                $skip = true;
            }
        }
 
        if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
            $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
            if ( file_exists( $path ) ) {
                $uploadInfo['fileSrc'] = $path;
                $uploadInfo['isTempSrc'] = false;
            }
        }
 
        if ( $this->mImportUploads ) {
            return $this->processUpload( $pageInfo, $uploadInfo );
        }
    }
 
    private function dumpTemp( $contents ) {
        $filename = tempnam( wfTempDir(), 'importupload' );
        file_put_contents( $filename, $contents );
        return $filename;
    }
 
    private function processUpload( $pageInfo, $uploadInfo ) {
        $revision = new WikiRevision();
        $revId = $pageInfo['id'];
        $title = $pageInfo['_title'];
        // T292348: text key may be absent, force addition if null
        $uploadInfo['text'] = $uploadInfo['text'] ?? '';
        $content = $this->makeContent( $title, $revId, $uploadInfo );
 
        $revision->setTitle( $title );
        $revision->setID( $revId );
        $revision->setTimestamp( $uploadInfo['timestamp'] );
        $revision->setContent( SlotRecord::MAIN, $content );
        $revision->setFilename( $uploadInfo['filename'] );
        if ( isset( $uploadInfo['archivename'] ) ) {
            $revision->setArchiveName( $uploadInfo['archivename'] );
        }
        $revision->setSrc( $uploadInfo['src'] );
        if ( isset( $uploadInfo['fileSrc'] ) ) {
            $revision->setFileSrc( $uploadInfo['fileSrc'],
                !empty( $uploadInfo['isTempSrc'] )
            );
        }
        if ( isset( $uploadInfo['sha1base36'] ) ) {
            $revision->setSha1Base36( $uploadInfo['sha1base36'] );
        }
        $revision->setSize( intval( $uploadInfo['size'] ) );
        $revision->setComment( $uploadInfo['comment'] );
 
        if ( isset( $uploadInfo['contributor']['username'] ) ) {
            $revision->setUsername(
                $this->externalUserNames->applyPrefix( $uploadInfo['contributor']['username'] )
            );
        } elseif ( isset( $uploadInfo['contributor']['ip'] ) ) {
            $revision->setUserIP( $uploadInfo['contributor']['ip'] );
        }
        $revision->setNoUpdates( $this->mNoUpdates );
 
        return call_user_func( $this->mUploadCallback, $revision );
    }
 
    private function handleContributor() {
        $this->debug( "Enter contributor handler." );
 
        if ( $this->reader->isEmptyElement ) {
            return [];
        }
 
        $fields = [ 'id', 'ip', 'username' ];
        $info = [];
 
        while ( $this->reader->read() ) {
            if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
                    $this->reader->localName == 'contributor' ) {
                break;
            }
 
            $tag = $this->reader->localName;
 
            if ( in_array( $tag, $fields ) ) {
                $info[$tag] = $this->nodeContents();
            }
        }
 
        return $info;
    }
 
    private function processTitle( $text, $ns = null ) {
        if ( $this->foreignNamespaces === null ) {
            $foreignTitleFactory = new NaiveForeignTitleFactory(
                $this->contentLanguage
            );
        } else {
            $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
                $this->foreignNamespaces );
        }
 
        $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
            intval( $ns ) );
 
        $title = $this->importTitleFactory->createTitleFromForeignTitle(
            $foreignTitle );
 
        $commandLineMode = $this->config->get( 'CommandLineMode' );
        if ( $title === null ) {
            # Invalid page title? Ignore the page
            $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
            return false;
        } elseif ( $title->isExternal() ) {
            $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
            return false;
        } elseif ( !$title->canExist() ) {
            $this->notice( 'import-error-special', $title->getPrefixedText() );
            return false;
        } elseif ( !$commandLineMode ) {
            $user = RequestContext::getMain()->getUser();
 
            if ( !$this->permissionManager->userCan( 'edit', $user, $title ) ) {
                # Do not import if the importing wiki user cannot edit this page
                $this->notice( 'import-error-edit', $title->getPrefixedText() );
 
                return false;
            }
        }
 
        return [ $title, $foreignTitle ];
    }
 
    private function getContentHandler( $model ) {
        return $this->contentHandlerFactory->getContentHandler( $model );
    }
 
    private function getDefaultContentModel( $title, $role ) {
        return $this->slotRoleRegistry
            ->getRoleHandler( $role )
            ->getDefaultModel( $title );
    }
 
    private function openReader() {
        // Enable the entity loader, as it is needed for loading external URLs via
        // XMLReader::open (T86036)
        // phpcs:ignore Generic.PHP.NoSilencedErrors -- suppress deprecation per T268847
        $oldDisable = @libxml_disable_entity_loader( false );
 
        if ( PHP_VERSION_ID >= 80000 ) {
            // A static call is now preferred, and avoids https://github.com/php/php-src/issues/11548
            $reader = XMLReader::open(
                'uploadsource://' . $this->sourceAdapterId, null, LIBXML_PARSEHUGE );
            if ( $reader instanceof XMLReader ) {
                $this->reader = $reader;
                $status = true;
            } else {
                $status = false;
            }
        } else {
            // A static call generated a deprecation warning prior to PHP 8.0
            $this->reader = new XMLReader;
            $status = $this->reader->open(
                'uploadsource://' . $this->sourceAdapterId, null, LIBXML_PARSEHUGE );
        }
        if ( !$status ) {
            $error = libxml_get_last_error();
            // phpcs:ignore Generic.PHP.NoSilencedErrors
            @libxml_disable_entity_loader( $oldDisable );
            throw new MWException(
                'Encountered an internal error while initializing WikiImporter object: ' . $error->message
            );
        }
        // phpcs:ignore Generic.PHP.NoSilencedErrors
        @libxml_disable_entity_loader( $oldDisable );
    }
 
    private function syntaxCheckXML() {
        if ( !UploadSourceAdapter::isSeekableSource( $this->sourceAdapterId ) ) {
            return;
        }
        AtEase::suppressWarnings();
        $oldDisable = libxml_disable_entity_loader( false );
        try {
            while ( $this->reader->read() );
            $error = libxml_get_last_error();
            if ( $error ) {
                $errorMessage = 'XML error at line ' . $error->line . ': ' . $error->message;
                wfDebug( __METHOD__ . ': Invalid xml found - ' . $errorMessage );
                throw new MWException( $errorMessage );
            }
        } finally {
            libxml_disable_entity_loader( $oldDisable );
            AtEase::restoreWarnings();
            $this->reader->close();
        }
 
        // Reopen for the real import
        UploadSourceAdapter::seekSource( $this->sourceAdapterId, 0 );
        $this->openReader();
    }
}