WikiImporter.php

Go to the documentation of this file.

<?php
class WikiImporter {
    private $reader = null;
    private $foreignNamespaces = null;
    private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
    private $mSiteInfoCallback, $mPageOutCallback;
    private $mNoticeCallback, $mDebug;
    private $mImportUploads, $mImageBasePath;
    private $mNoUpdates = false;
    private $config;
    private $importTitleFactory;
    private $countableCache = [];

    function __construct( ImportSource $source, Config $config = null ) {
        if ( !class_exists( 'XMLReader' ) ) {
            throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
        }

        $this->reader = new XMLReader();
        if ( !$config ) {
            wfDeprecated( __METHOD__ . ' without a Config instance', '1.25' );
            $config = ConfigFactory::getDefaultInstance()->makeConfig( 'main' );
        }
        $this->config = $config;

        if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
            stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
        }
        $id = UploadSourceAdapter::registerSource( $source );

        // Enable the entity loader, as it is needed for loading external URLs via
        // XMLReader::open (T86036)
        $oldDisable = libxml_disable_entity_loader( false );
        if ( defined( 'LIBXML_PARSEHUGE' ) ) {
            $status = $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
        } else {
            $status = $this->reader->open( "uploadsource://$id" );
        }
        if ( !$status ) {
            $error = libxml_get_last_error();
            libxml_disable_entity_loader( $oldDisable );
            throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
                $error->message );
        }
        libxml_disable_entity_loader( $oldDisable );

        // 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();
    }

    public function getReader() {
        return $this->reader;
    }

    public function throwXmlError( $err ) {
        $this->debug( "FAILURE: $err" );
        wfDebug( "WikiImporter XML error: $err\n" );
    }

    public function debug( $data ) {
        if ( $this->mDebug ) {
            wfDebug( "IMPORT: $data\n" );
        }
    }

    public function warn( $data ) {
        wfDebug( "IMPORT: $data\n" );
    }

    public function notice( $msg /*, $param, ...*/ ) {
        $params = func_get_args();
        array_shift( $params );

        if ( is_callable( $this->mNoticeCallback ) ) {
            call_user_func( $this->mNoticeCallback, $msg, $params );
        } else { # No ImportReporter -> CLI
            echo wfMessage( $msg, $params )->text() . "\n";
        }
    }

    function setDebug( $debug ) {
        $this->mDebug = $debug;
    }

    function setNoUpdates( $noupdates ) {
        $this->mNoUpdates = $noupdates;
    }

    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 ( is_null( $namespace ) ) {
            // Don't override namespaces
            $this->setImportTitleFactory( new NaiveImportTitleFactory() );
            return true;
        } elseif (
            $namespace >= 0 &&
            MWNamespace::exists( intval( $namespace ) )
        ) {
            $namespace = intval( $namespace );
            $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
            return true;
        } else {
            return false;
        }
    }

    public function setTargetRootPage( $rootpage ) {
        $status = Status::newGood();
        if ( is_null( $rootpage ) ) {
            // No rootpage
            $this->setImportTitleFactory( new NaiveImportTitleFactory() );
        } elseif ( $rootpage !== '' ) {
            $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
            $title = Title::newFromText( $rootpage );

            if ( !$title || $title->isExternal() ) {
                $status->fatal( 'import-rootpage-invalid' );
            } else {
                if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
                    global $wgContLang;

                    $displayNSText = $title->getNamespace() == NS_MAIN
                        ? wfMessage( 'blanknamespace' )->text()
                        : $wgContLang->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( $title ) );
                }
            }
        }
        return $status;
    }

    public function setImageBasePath( $dir ) {
        $this->mImageBasePath = $dir;
    }

    public function setImportUploads( $import ) {
        $this->mImportUploads = $import;
    }

    public function beforeImportPage( $titleAndForeignTitle ) {
        $title = $titleAndForeignTitle[0];
        $page = WikiPage::factory( $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 {
            $dbw = wfGetDB( DB_MASTER );
            return $dbw->deadlockLoop( [ $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 ) {
        $dbw = wfGetDB( DB_MASTER );
        return $dbw->deadlockLoop( [ $revision, 'importLogItem' ] );
    }

    public function importUpload( $revision ) {
        $dbw = wfGetDB( DB_MASTER );
        return $dbw->deadlockLoop( [ $revision, 'importUpload' ] );
    }

    public function finishImportPage( $title, $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 slave lag. We let WikiPage handle the total page
        // and revision count, and we implement our own custom logic for the
        // article (content page) count.
        $page = WikiPage::factory( $title );
        $page->loadPageData( 'fromdbmaster' );
        $content = $page->getContent();
        if ( $content === null ) {
            wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
                ' because WikiPage::getContent() returned null' );
        } else {
            $editInfo = $page->prepareContentForEdit( $content );
            $countKey = 'title_' . $title->getPrefixedText();
            $countable = $page->isCountable( $editInfo );
            if ( array_key_exists( $countKey, $this->countableCache ) &&
                $countable != $this->countableCache[$countKey] ) {
                DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
                    'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
                ] ) );
            }
        }

        $args = func_get_args();
        return Hooks::run( 'AfterImportPage', $args );
    }

    public function debugRevisionHandler( &$revision ) {
        $this->debug( "Got revision:" );
        if ( is_object( $revision->title ) ) {
            $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
        } else {
            $this->debug( "-- Title: <invalid>" );
        }
        $this->debug( "-- User: " . $revision->user_text );
        $this->debug( "-- Timestamp: " . $revision->timestamp );
        $this->debug( "-- Comment: " . $revision->comment );
        $this->debug( "-- Text: " . $revision->text );
    }

    private function siteInfoCallback( $siteInfo ) {
        if ( isset( $this->mSiteInfoCallback ) ) {
            return call_user_func_array( $this->mSiteInfoCallback,
                    [ $siteInfo, $this ] );
        } else {
            return false;
        }
    }

    function pageCallback( $title ) {
        if ( isset( $this->mPageCallback ) ) {
            call_user_func( $this->mPageCallback, $title );
        }
    }

    private function pageOutCallback( $title, $foreignTitle, $revCount,
            $sucCount, $pageInfo ) {
        if ( isset( $this->mPageOutCallback ) ) {
            $args = func_get_args();
            call_user_func_array( $this->mPageOutCallback, $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() {
        // Calls to reader->read need to be wrapped in calls to
        // libxml_disable_entity_loader() to avoid local file
        // inclusion attacks (bug 46932).
        $oldDisable = libxml_disable_entity_loader( true );
        $this->reader->read();

        if ( $this->reader->localName != 'mediawiki' ) {
            libxml_disable_entity_loader( $oldDisable );
            throw new MWException( "Expected <mediawiki> tag, got " .
                $this->reader->localName );
        }
        $this->debug( "<mediawiki> tag is correct." );

        $this->debug( "Starting primary dump processing loop." );

        $keepReading = $this->reader->read();
        $skip = false;
        $rethrow = null;
        try {
            while ( $keepReading ) {
                $tag = $this->reader->localName;
                $type = $this->reader->nodeType;

                if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $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();
                }
            }
        } catch ( Exception $ex ) {
            $rethrow = $ex;
        }

        // finally
        libxml_disable_entity_loader( $oldDisable );
        $this->reader->close();

        if ( $rethrow ) {
            throw $rethrow;
        }

        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 ( !Hooks::run( 'ImportHandleLogItemXMLTag', [
                $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( $this->config );

        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']['ip'] ) ) {
            $revision->setUserIP( $logInfo['contributor']['ip'] );
        }

        if ( !isset( $logInfo['contributor']['username'] ) ) {
            $revision->setUsername( 'Unknown user' );
        } else {
            $revision->setUsername( $logInfo['contributor']['username'] );
        }

        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 ( !Hooks::run( 'ImportHandlePageXMLTag', [ $this,
                        &$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'],
                        isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );

                    // $title is either an array of two titles or false.
                    if ( is_array( $title ) ) {
                        $this->pageCallback( $title );
                        list( $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 ) ) {
            $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
                    $pageInfo['revisionCount'],
                    $pageInfo['successfulRevisionCount'],
                    $pageInfo );
        }
    }

    private function handleRevision( &$pageInfo ) {
        $this->debug( "Enter revision handler" );
        $revisionInfo = [];

        $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' ];

        $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 ( !Hooks::run( 'ImportHandleRevisionXMLTag', [
                $this, $pageInfo, $revisionInfo
            ] ) ) {
                // Do nothing
            } elseif ( in_array( $tag, $normalFields ) ) {
                $revisionInfo[$tag] = $this->nodeContents();
            } 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 processRevision( $pageInfo, $revisionInfo ) {
        global $wgMaxArticleSize;

        // 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( $revisionInfo['model'] ) ||
            in_array( $revisionInfo['model'], [
                'wikitext',
                'css',
                'json',
                'javascript',
                'text',
                ''
            ] ) ) &&
            (int)( strlen( $revisionInfo['text'] ) / 1024 ) > $wgMaxArticleSize
        ) {
            throw new MWException( 'The text of ' .
                ( isset( $revisionInfo['id'] ) ?
                    "the revision with ID $revisionInfo[id]" :
                    'a revision'
                ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
        }

        $revision = new WikiRevision( $this->config );

        if ( isset( $revisionInfo['id'] ) ) {
            $revision->setID( $revisionInfo['id'] );
        }
        if ( isset( $revisionInfo['model'] ) ) {
            $revision->setModel( $revisionInfo['model'] );
        }
        if ( isset( $revisionInfo['format'] ) ) {
            $revision->setFormat( $revisionInfo['format'] );
        }
        $revision->setTitle( $pageInfo['_title'] );

        if ( isset( $revisionInfo['text'] ) ) {
            $handler = $revision->getContentHandler();
            $text = $handler->importTransform(
                $revisionInfo['text'],
                $revision->getFormat() );

            $revision->setText( $text );
        }
        if ( isset( $revisionInfo['timestamp'] ) ) {
            $revision->setTimestamp( $revisionInfo['timestamp'] );
        } else {
            $revision->setTimestamp( wfTimestampNow() );
        }

        if ( isset( $revisionInfo['comment'] ) ) {
            $revision->setComment( $revisionInfo['comment'] );
        }

        if ( isset( $revisionInfo['minor'] ) ) {
            $revision->setMinor( true );
        }
        if ( isset( $revisionInfo['contributor']['ip'] ) ) {
            $revision->setUserIP( $revisionInfo['contributor']['ip'] );
        } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
            $revision->setUsername( $revisionInfo['contributor']['username'] );
        } else {
            $revision->setUsername( 'Unknown user' );
        }
        $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 ( !Hooks::run( 'ImportHandleUploadXMLTag', [
                $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( $this->config );
        $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';

        $revision->setTitle( $pageInfo['_title'] );
        $revision->setID( $pageInfo['id'] );
        $revision->setTimestamp( $uploadInfo['timestamp'] );
        $revision->setText( $text );
        $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']['ip'] ) ) {
            $revision->setUserIP( $uploadInfo['contributor']['ip'] );
        }
        if ( isset( $uploadInfo['contributor']['username'] ) ) {
            $revision->setUsername( $uploadInfo['contributor']['username'] );
        }
        $revision->setNoUpdates( $this->mNoUpdates );

        return call_user_func( $this->mUploadCallback, $revision );
    }

    private function handleContributor() {
        $fields = [ 'id', 'ip', 'username' ];
        $info = [];

        if ( $this->reader->isEmptyElement ) {
            return $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 ( is_null( $this->foreignNamespaces ) ) {
            $foreignTitleFactory = new NaiveForeignTitleFactory();
        } else {
            $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
                $this->foreignNamespaces );
        }

        $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
            intval( $ns ) );

        $title = $this->importTitleFactory->createTitleFromForeignTitle(
            $foreignTitle );

        $commandLineMode = $this->config->get( 'CommandLineMode' );
        if ( is_null( $title ) ) {
            # 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', $title->getPrefixedText() );
            return false;
        } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
            # Do not import if the importing wiki user cannot edit this page
            $this->notice( 'import-error-edit', $title->getPrefixedText() );
            return false;
        } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
            # Do not import if the importing wiki user cannot create this page
            $this->notice( 'import-error-create', $title->getPrefixedText() );
            return false;
        }

        return [ $title, $foreignTitle ];
    }
}