Import.php

Go to the documentation of this file.

<?php
class WikiImporter {
    private $reader = null;
    private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
    private $mSiteInfoCallback, $mTargetNamespace, $mTargetRootPage, $mPageOutCallback;
    private $mNoticeCallback, $mDebug;
    private $mImportUploads, $mImageBasePath;
    private $mNoUpdates = false;

    function __construct( ImportStreamSource $source ) {
        $this->reader = new XMLReader();

        if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
            stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
        }
        $id = UploadSourceAdapter::registerSource( $source );
        if ( defined( 'LIBXML_PARSEHUGE' ) ) {
            $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
        } else {
            $this->reader->open( "uploadsource://$id" );
        }

        // Default callbacks
        $this->setRevisionCallback( array( $this, "importRevision" ) );
        $this->setUploadCallback( array( $this, 'importUpload' ) );
        $this->setLogItemCallback( array( $this, 'importLogItem' ) );
        $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
    }

    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 setTargetNamespace( $namespace ) {
        if ( is_null( $namespace ) ) {
            // Don't override namespaces
            $this->mTargetNamespace = null;
        } elseif ( $namespace >= 0 ) {
            // @todo FIXME: Check for validity
            $this->mTargetNamespace = intval( $namespace );
        } else {
            return false;
        }
    }

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

            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 passed
                    $this->setTargetNamespace( null );
                    $this->mTargetRootPage = $title->getPrefixedDBkey();
                }
            }
        }
        return $status;
    }

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

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

    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( array( $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( array( $revision, 'importLogItem' ) );
    }

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

    public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
        $args = func_get_args();
        return wfRunHooks( '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 );
    }

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

    private function pageOutCallback( $title, $origTitle, $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,
                    array( $revision, $this ) );
        } else {
            return false;
        }
    }

    private function logItemCallback( $revision ) {
        if ( isset( $this->mLogItemCallback ) ) {
            return call_user_func_array( $this->mLogItemCallback,
                    array( $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::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->name != 'mediawiki' ) {
            libxml_disable_entity_loader( $oldDisable );
            throw new MWException( "Expected <mediawiki> tag, got " .
                $this->reader->name );
        }
        $this->debug( "<mediawiki> tag is correct." );

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

        $keepReading = $this->reader->read();
        $skip = false;
        while ( $keepReading ) {
            $tag = $this->reader->name;
            $type = $this->reader->nodeType;

            if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', array( $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();
            }
        }

        libxml_disable_entity_loader( $oldDisable );
        return true;
    }

    private function handleSiteInfo() {
        // Site info is useful, but not actually used for dump imports.
        // Includes a quick short-circuit to save performance.
        if ( !$this->mSiteInfoCallback ) {
            $this->reader->next();
            return true;
        }
        throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
    }

    private function handleLogItem() {
        $this->debug( "Enter log item handler." );
        $logInfo = array();

        // Fields that can just be stuffed in the pageInfo object
        $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
                    'logtitle', 'params' );

        while ( $this->reader->read() ) {
            if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
                    $this->reader->name == 'logitem' ) {
                break;
            }

            $tag = $this->reader->name;

            if ( !wfRunHooks( 'ImportHandleLogItemXMLTag', array(
                $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;

        $revision->setID( $logInfo['id'] );
        $revision->setType( $logInfo['type'] );
        $revision->setAction( $logInfo['action'] );
        $revision->setTimestamp( $logInfo['timestamp'] );
        $revision->setParams( $logInfo['params'] );
        $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( $logInfo['contributor']['username'] );
        }

        return $this->logItemCallback( $revision );
    }

    private function handlePage() {
        // Handle page data.
        $this->debug( "Enter page handler." );
        $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );

        // Fields that can just be stuffed in the pageInfo object
        $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );

        $skip = false;
        $badTitle = false;

        while ( $skip ? $this->reader->next() : $this->reader->read() ) {
            if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
                    $this->reader->name == 'page' ) {
                break;
            }

            $tag = $this->reader->name;

            if ( $badTitle ) {
                // The title is invalid, bail out of this page
                $skip = true;
            } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
                        &$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();
                    if ( $tag == 'title' ) {
                        $title = $this->processTitle( $pageInfo['title'] );

                        if ( !$title ) {
                            $badTitle = true;
                            $skip = true;
                        }

                        $this->pageCallback( $title );
                        list( $pageInfo['_title'], $origTitle ) = $title;
                    }
                }
            } elseif ( $tag == 'revision' ) {
                $this->handleRevision( $pageInfo );
            } elseif ( $tag == 'upload' ) {
                $this->handleUpload( $pageInfo );
            } elseif ( $tag != '#text' ) {
                $this->warn( "Unhandled page XML tag $tag" );
                $skip = true;
            }
        }

        $this->pageOutCallback( $pageInfo['_title'], $origTitle,
                    $pageInfo['revisionCount'],
                    $pageInfo['successfulRevisionCount'],
                    $pageInfo );
    }

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

        $normalFields = array( '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->name == 'revision' ) {
                break;
            }

            $tag = $this->reader->name;

            if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', array(
                $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 ) {
        $revision = new WikiRevision;

        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'] );
        }
        if ( isset( $revisionInfo['contributor']['username'] ) ) {
            $revision->setUserName( $revisionInfo['contributor']['username'] );
        }
        $revision->setNoUpdates( $this->mNoUpdates );

        return $this->revisionCallback( $revision );
    }

    private function handleUpload( &$pageInfo ) {
        $this->debug( "Enter upload handler" );
        $uploadInfo = array();

        $normalFields = array( '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->name == 'upload' ) {
                break;
            }

            $tag = $this->reader->name;

            if ( !wfRunHooks( 'ImportHandleUploadXMLTag', array(
                $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;
        $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 = array( 'id', 'ip', 'username' );
        $info = array();

        while ( $this->reader->read() ) {
            if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
                    $this->reader->name == 'contributor' ) {
                break;
            }

            $tag = $this->reader->name;

            if ( in_array( $tag, $fields ) ) {
                $info[$tag] = $this->nodeContents();
            }
        }

        return $info;
    }

    private function processTitle( $text ) {
        global $wgCommandLineMode;

        $workTitle = $text;
        $origTitle = Title::newFromText( $workTitle );

        if ( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
            # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
            # and than dbKey can begin with a lowercase char
            $title = Title::makeTitleSafe( $this->mTargetNamespace,
                $origTitle->getDBkey() );
        } else {
            if ( !is_null( $this->mTargetRootPage ) ) {
                $workTitle = $this->mTargetRootPage . '/' . $workTitle;
            }
            $title = Title::newFromText( $workTitle );
        }

        if ( is_null( $title ) ) {
            # Invalid page title? Ignore the page
            $this->notice( 'import-error-invalid', $workTitle );
            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' ) && !$wgCommandLineMode ) {
            # Do not import if the importing wiki user cannot edit this page
            $this->notice( 'import-error-edit', $title->getPrefixedText() );
            return false;
        } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
            # Do not import if the importing wiki user cannot create this page
            $this->notice( 'import-error-create', $title->getPrefixedText() );
            return false;
        }

        return array( $title, $origTitle );
    }
}

class UploadSourceAdapter {
    public static $sourceRegistrations = array();

    private $mSource;

    private $mBuffer;

    private $mPosition;

    static function registerSource( ImportStreamSource $source ) {
        $id = wfRandomString();

        self::$sourceRegistrations[$id] = $source;

        return $id;
    }

    function stream_open( $path, $mode, $options, &$opened_path ) {
        $url = parse_url( $path );
        $id = $url['host'];

        if ( !isset( self::$sourceRegistrations[$id] ) ) {
            return false;
        }

        $this->mSource = self::$sourceRegistrations[$id];

        return true;
    }

    function stream_read( $count ) {
        $return = '';
        $leave = false;

        while ( !$leave && !$this->mSource->atEnd() &&
                strlen( $this->mBuffer ) < $count ) {
            $read = $this->mSource->readChunk();

            if ( !strlen( $read ) ) {
                $leave = true;
            }

            $this->mBuffer .= $read;
        }

        if ( strlen( $this->mBuffer ) ) {
            $return = substr( $this->mBuffer, 0, $count );
            $this->mBuffer = substr( $this->mBuffer, $count );
        }

        $this->mPosition += strlen( $return );

        return $return;
    }

    function stream_write( $data ) {
        return false;
    }

    function stream_tell() {
        return $this->mPosition;
    }

    function stream_eof() {
        return $this->mSource->atEnd();
    }

    function url_stat() {
        $result = array();

        $result['dev'] = $result[0] = 0;
        $result['ino'] = $result[1] = 0;
        $result['mode'] = $result[2] = 0;
        $result['nlink'] = $result[3] = 0;
        $result['uid'] = $result[4] = 0;
        $result['gid'] = $result[5] = 0;
        $result['rdev'] = $result[6] = 0;
        $result['size'] = $result[7] = 0;
        $result['atime'] = $result[8] = 0;
        $result['mtime'] = $result[9] = 0;
        $result['ctime'] = $result[10] = 0;
        $result['blksize'] = $result[11] = 0;
        $result['blocks'] = $result[12] = 0;

        return $result;
    }
}

class WikiRevision {
    public $importer = null;

    public $title = null;

    public $id = 0;

    public $timestamp = "20010115000000";

    public $user = 0;

    public $user_text = "";

    public $model = null;

    public $format = null;

    public $text = "";

    protected $size;

    public $content = null;

    protected $contentHandler = null;

    public $comment = "";

    public $minor = false;

    public $type = "";

    public $action = "";

    public $params = "";

    public $fileSrc = '';

    public $sha1base36 = false;

    public $isTemp = false;

    public $archiveName = '';

    protected $filename;

    protected $src;

    public $fileIsTemp;

    private $mNoUpdates = false;

    function setTitle( $title ) {
        if ( is_object( $title ) ) {
            $this->title = $title;
        } elseif ( is_null( $title ) ) {
            throw new MWException( "WikiRevision given a null title in import. "
                . "You may need to adjust \$wgLegalTitleChars." );
        } else {
            throw new MWException( "WikiRevision given non-object title in import." );
        }
    }

    function setID( $id ) {
        $this->id = $id;
    }

    function setTimestamp( $ts ) {
        # 2003-08-05T18:30:02Z
        $this->timestamp = wfTimestamp( TS_MW, $ts );
    }

    function setUsername( $user ) {
        $this->user_text = $user;
    }

    function setUserIP( $ip ) {
        $this->user_text = $ip;
    }

    function setModel( $model ) {
        $this->model = $model;
    }

    function setFormat( $format ) {
        $this->format = $format;
    }

    function setText( $text ) {
        $this->text = $text;
    }

    function setComment( $text ) {
        $this->comment = $text;
    }

    function setMinor( $minor ) {
        $this->minor = (bool)$minor;
    }

    function setSrc( $src ) {
        $this->src = $src;
    }

    function setFileSrc( $src, $isTemp ) {
        $this->fileSrc = $src;
        $this->fileIsTemp = $isTemp;
    }

    function setSha1Base36( $sha1base36 ) {
        $this->sha1base36 = $sha1base36;
    }

    function setFilename( $filename ) {
        $this->filename = $filename;
    }

    function setArchiveName( $archiveName ) {
        $this->archiveName = $archiveName;
    }

    function setSize( $size ) {
        $this->size = intval( $size );
    }

    function setType( $type ) {
        $this->type = $type;
    }

    function setAction( $action ) {
        $this->action = $action;
    }

    function setParams( $params ) {
        $this->params = $params;
    }

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

    function getTitle() {
        return $this->title;
    }

    function getID() {
        return $this->id;
    }

    function getTimestamp() {
        return $this->timestamp;
    }

    function getUser() {
        return $this->user_text;
    }

    function getText() {
        ContentHandler::deprecated( __METHOD__, '1.21' );

        return $this->text;
    }

    function getContentHandler() {
        if ( is_null( $this->contentHandler ) ) {
            $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
        }

        return $this->contentHandler;
    }

    function getContent() {
        if ( is_null( $this->content ) ) {
            $handler = $this->getContentHandler();
            $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
        }

        return $this->content;
    }

    function getModel() {
        if ( is_null( $this->model ) ) {
            $this->model = $this->getTitle()->getContentModel();
        }

        return $this->model;
    }

    function getFormat() {
        if ( is_null( $this->format ) ) {
            $this->format = $this->getContentHandler()->getDefaultFormat();
        }

        return $this->format;
    }

    function getComment() {
        return $this->comment;
    }

    function getMinor() {
        return $this->minor;
    }

    function getSrc() {
        return $this->src;
    }

    function getSha1() {
        if ( $this->sha1base36 ) {
            return wfBaseConvert( $this->sha1base36, 36, 16 );
        }
        return false;
    }

    function getFileSrc() {
        return $this->fileSrc;
    }

    function isTempSrc() {
        return $this->isTemp;
    }

    function getFilename() {
        return $this->filename;
    }

    function getArchiveName() {
        return $this->archiveName;
    }

    function getSize() {
        return $this->size;
    }

    function getType() {
        return $this->type;
    }

    function getAction() {
        return $this->action;
    }

    function getParams() {
        return $this->params;
    }

    function importOldRevision() {
        $dbw = wfGetDB( DB_MASTER );

        # Sneak a single revision into place
        $user = User::newFromName( $this->getUser() );
        if ( $user ) {
            $userId = intval( $user->getId() );
            $userText = $user->getName();
            $userObj = $user;
        } else {
            $userId = 0;
            $userText = $this->getUser();
            $userObj = new User;
        }

        // avoid memory leak...?
        $linkCache = LinkCache::singleton();
        $linkCache->clear();

        $page = WikiPage::factory( $this->title );
        $page->loadPageData( 'fromdbmaster' );
        if ( !$page->exists() ) {
            # must create the page...
            $pageId = $page->insertOn( $dbw );
            $created = true;
            $oldcountable = null;
        } else {
            $pageId = $page->getId();
            $created = false;

            $prior = $dbw->selectField( 'revision', '1',
                array( 'rev_page' => $pageId,
                    'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
                    'rev_user_text' => $userText,
                    'rev_comment' => $this->getComment() ),
                __METHOD__
            );
            if ( $prior ) {
                // @todo FIXME: This could fail slightly for multiple matches :P
                wfDebug( __METHOD__ . ": skipping existing revision for [[" .
                    $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
                return false;
            }
            $oldcountable = $page->isCountable();
        }

        # @todo FIXME: Use original rev_id optionally (better for backups)
        # Insert the row
        $revision = new Revision( array(
            'title' => $this->title,
            'page' => $pageId,
            'content_model' => $this->getModel(),
            'content_format' => $this->getFormat(),
            //XXX: just set 'content' => $this->getContent()?
            'text' => $this->getContent()->serialize( $this->getFormat() ),
            'comment' => $this->getComment(),
            'user' => $userId,
            'user_text' => $userText,
            'timestamp' => $this->timestamp,
            'minor_edit' => $this->minor,
            ) );
        $revision->insertOn( $dbw );
        $changed = $page->updateIfNewerOn( $dbw, $revision );

        if ( $changed !== false && !$this->mNoUpdates ) {
            wfDebug( __METHOD__ . ": running updates\n" );
            $page->doEditUpdates(
                $revision,
                $userObj,
                array( 'created' => $created, 'oldcountable' => $oldcountable )
            );
        }

        return true;
    }

    function importLogItem() {
        $dbw = wfGetDB( DB_MASTER );
        # @todo FIXME: This will not record autoblocks
        if ( !$this->getTitle() ) {
            wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
                $this->timestamp . "\n" );
            return;
        }
        # Check if it exists already
        // @todo FIXME: Use original log ID (better for backups)
        $prior = $dbw->selectField( 'logging', '1',
            array( 'log_type' => $this->getType(),
                'log_action' => $this->getAction(),
                'log_timestamp' => $dbw->timestamp( $this->timestamp ),
                'log_namespace' => $this->getTitle()->getNamespace(),
                'log_title' => $this->getTitle()->getDBkey(),
                'log_comment' => $this->getComment(),
                #'log_user_text' => $this->user_text,
                'log_params' => $this->params ),
            __METHOD__
        );
        // @todo FIXME: This could fail slightly for multiple matches :P
        if ( $prior ) {
            wfDebug( __METHOD__
                . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
                . $this->timestamp . "\n" );
            return;
        }
        $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
        $data = array(
            'log_id' => $log_id,
            'log_type' => $this->type,
            'log_action' => $this->action,
            'log_timestamp' => $dbw->timestamp( $this->timestamp ),
            'log_user' => User::idFromName( $this->user_text ),
            #'log_user_text' => $this->user_text,
            'log_namespace' => $this->getTitle()->getNamespace(),
            'log_title' => $this->getTitle()->getDBkey(),
            'log_comment' => $this->getComment(),
            'log_params' => $this->params
        );
        $dbw->insert( 'logging', $data, __METHOD__ );
    }

    function importUpload() {
        # Construct a file
        $archiveName = $this->getArchiveName();
        if ( $archiveName ) {
            wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
            $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
                RepoGroup::singleton()->getLocalRepo(), $archiveName );
        } else {
            $file = wfLocalFile( $this->getTitle() );
            wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
            if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
                $archiveName = $file->getTimestamp() . '!' . $file->getName();
                $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
                    RepoGroup::singleton()->getLocalRepo(), $archiveName );
                wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
            }
        }
        if ( !$file ) {
            wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
            return false;
        }

        # Get the file source or download if necessary
        $source = $this->getFileSrc();
        $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
        if ( !$source ) {
            $source = $this->downloadSource();
            $flags |= File::DELETE_SOURCE;
        }
        if ( !$source ) {
            wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
            return false;
        }
        $sha1 = $this->getSha1();
        if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
            if ( $flags & File::DELETE_SOURCE ) {
                # Broken file; delete it if it is a temporary file
                unlink( $source );
            }
            wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
            return false;
        }

        $user = User::newFromName( $this->user_text );

        # Do the actual upload
        if ( $archiveName ) {
            $status = $file->uploadOld( $source, $archiveName,
                $this->getTimestamp(), $this->getComment(), $user, $flags );
        } else {
            $status = $file->upload( $source, $this->getComment(), $this->getComment(),
                $flags, false, $this->getTimestamp(), $user );
        }

        if ( $status->isGood() ) {
            wfDebug( __METHOD__ . ": Successful\n" );
            return true;
        } else {
            wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
            return false;
        }
    }

    function downloadSource() {
        global $wgEnableUploads;
        if ( !$wgEnableUploads ) {
            return false;
        }

        $tempo = tempnam( wfTempDir(), 'download' );
        $f = fopen( $tempo, 'wb' );
        if ( !$f ) {
            wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
            return false;
        }

        // @todo FIXME!
        $src = $this->getSrc();
        $data = Http::get( $src );
        if ( !$data ) {
            wfDebug( "IMPORT: couldn't fetch source $src\n" );
            fclose( $f );
            unlink( $tempo );
            return false;
        }

        fwrite( $f, $data );
        fclose( $f );

        return $tempo;
    }

}

class ImportStringSource {
    function __construct( $string ) {
        $this->mString = $string;
        $this->mRead = false;
    }

    function atEnd() {
        return $this->mRead;
    }

    function readChunk() {
        if ( $this->atEnd() ) {
            return false;
        }
        $this->mRead = true;
        return $this->mString;
    }
}

class ImportStreamSource {
    function __construct( $handle ) {
        $this->mHandle = $handle;
    }

    function atEnd() {
        return feof( $this->mHandle );
    }

    function readChunk() {
        return fread( $this->mHandle, 32768 );
    }

    static function newFromFile( $filename ) {
        wfSuppressWarnings();
        $file = fopen( $filename, 'rt' );
        wfRestoreWarnings();
        if ( !$file ) {
            return Status::newFatal( "importcantopen" );
        }
        return Status::newGood( new ImportStreamSource( $file ) );
    }

    static function newFromUpload( $fieldname = "xmlimport" ) {
        $upload =& $_FILES[$fieldname];

        if ( $upload === null || !$upload['name'] ) {
            return Status::newFatal( 'importnofile' );
        }
        if ( !empty( $upload['error'] ) ) {
            switch ( $upload['error'] ) {
                case 1:
                    # The uploaded file exceeds the upload_max_filesize directive in php.ini.
                    return Status::newFatal( 'importuploaderrorsize' );
                case 2:
                    # The uploaded file exceeds the MAX_FILE_SIZE directive that
                    # was specified in the HTML form.
                    return Status::newFatal( 'importuploaderrorsize' );
                case 3:
                    # The uploaded file was only partially uploaded
                    return Status::newFatal( 'importuploaderrorpartial' );
                case 6:
                    # Missing a temporary folder.
                    return Status::newFatal( 'importuploaderrortemp' );
                # case else: # Currently impossible
            }

        }
        $fname = $upload['tmp_name'];
        if ( is_uploaded_file( $fname ) ) {
            return ImportStreamSource::newFromFile( $fname );
        } else {
            return Status::newFatal( 'importnofile' );
        }
    }

    static function newFromURL( $url, $method = 'GET' ) {
        wfDebug( __METHOD__ . ": opening $url\n" );
        # Use the standard HTTP fetch function; it times out
        # quicker and sorts out user-agent problems which might
        # otherwise prevent importing from large sites, such
        # as the Wikimedia cluster, etc.
        $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
        if ( $data !== false ) {
            $file = tmpfile();
            fwrite( $file, $data );
            fflush( $file );
            fseek( $file, 0 );
            return Status::newGood( new ImportStreamSource( $file ) );
        } else {
            return Status::newFatal( 'importcantopen' );
        }
    }

    public static function newFromInterwiki( $interwiki, $page, $history = false,
        $templates = false, $pageLinkDepth = 0
    ) {
        if ( $page == '' ) {
            return Status::newFatal( 'import-noarticle' );
        }
        $link = Title::newFromText( "$interwiki:Special:Export/$page" );
        if ( is_null( $link ) || !$link->isExternal() ) {
            return Status::newFatal( 'importbadinterwiki' );
        } else {
            $params = array();
            if ( $history ) {
                $params['history'] = 1;
            }
            if ( $templates ) {
                $params['templates'] = 1;
            }
            if ( $pageLinkDepth ) {
                $params['pagelink-depth'] = $pageLinkDepth;
            }
            $url = $link->getFullURL( $params );
            # For interwikis, use POST to avoid redirects.
            return ImportStreamSource::newFromURL( $url, "POST" );
        }
    }
}