SpecialExport.php

Go to the documentation of this file.

<?php
class SpecialExport extends SpecialPage {
    private $curonly, $doExport, $pageLinkDepth, $templates;
    private $images;

    public function __construct() {
        parent::__construct( 'Export' );
    }

    public function execute( $par ) {
        $this->setHeaders();
        $this->outputHeader();
        $config = $this->getConfig();

        // Set some variables
        $this->curonly = true;
        $this->doExport = false;
        $request = $this->getRequest();
        $this->templates = $request->getCheck( 'templates' );
        $this->images = $request->getCheck( 'images' ); // Doesn't do anything yet
        $this->pageLinkDepth = $this->validateLinkDepth(
            $request->getIntOrNull( 'pagelink-depth' )
        );
        $nsindex = '';
        $exportall = false;

        if ( $request->getCheck( 'addcat' ) ) {
            $page = $request->getText( 'pages' );
            $catname = $request->getText( 'catname' );

            if ( $catname !== '' && $catname !== null && $catname !== false ) {
                $t = Title::makeTitleSafe( NS_MAIN, $catname );
                if ( $t ) {
                    $catpages = $this->getPagesFromCategory( $t );
                    if ( $catpages ) {
                        $page .= "\n" . implode( "\n", $catpages );
                    }
                }
            }
        } elseif ( $request->getCheck( 'addns' ) && $config->get( 'ExportFromNamespaces' ) ) {
            $page = $request->getText( 'pages' );
            $nsindex = $request->getText( 'nsindex', '' );

            if ( strval( $nsindex ) !== '' ) {
                $nspages = $this->getPagesFromNamespace( $nsindex );
                if ( $nspages ) {
                    $page .= "\n" . implode( "\n", $nspages );
                }
            }
        } elseif ( $request->getCheck( 'exportall' ) && $config->get( 'ExportAllowAll' ) ) {
            $this->doExport = true;
            $exportall = true;

            /* Although $page and $history are not used later on, we
            nevertheless set them to avoid that PHP notices about using
            undefined variables foul up our XML output (see call to
            doExport(...) further down) */
            $page = '';
            $history = '';
        } elseif ( $request->wasPosted() && $par == '' ) {
            $page = $request->getText( 'pages' );
            $this->curonly = $request->getCheck( 'curonly' );
            $rawOffset = $request->getVal( 'offset' );

            if ( $rawOffset ) {
                $offset = wfTimestamp( TS_MW, $rawOffset );
            } else {
                $offset = null;
            }

            $maxHistory = $config->get( 'ExportMaxHistory' );
            $limit = $request->getInt( 'limit' );
            $dir = $request->getVal( 'dir' );
            $history = array(
                'dir' => 'asc',
                'offset' => false,
                'limit' => $maxHistory,
            );
            $historyCheck = $request->getCheck( 'history' );

            if ( $this->curonly ) {
                $history = WikiExporter::CURRENT;
            } elseif ( !$historyCheck ) {
                if ( $limit > 0 && ( $maxHistory == 0 || $limit < $maxHistory ) ) {
                    $history['limit'] = $limit;
                }

                if ( !is_null( $offset ) ) {
                    $history['offset'] = $offset;
                }

                if ( strtolower( $dir ) == 'desc' ) {
                    $history['dir'] = 'desc';
                }
            }

            if ( $page != '' ) {
                $this->doExport = true;
            }
        } else {
            // Default to current-only for GET requests.
            $page = $request->getText( 'pages', $par );
            $historyCheck = $request->getCheck( 'history' );

            if ( $historyCheck ) {
                $history = WikiExporter::FULL;
            } else {
                $history = WikiExporter::CURRENT;
            }

            if ( $page != '' ) {
                $this->doExport = true;
            }
        }

        if ( !$config->get( 'ExportAllowHistory' ) ) {
            // Override
            $history = WikiExporter::CURRENT;
        }

        $list_authors = $request->getCheck( 'listauthors' );
        if ( !$this->curonly || !$config->get( 'ExportAllowListContributors' ) ) {
            $list_authors = false;
        }

        if ( $this->doExport ) {
            $this->getOutput()->disable();

            // Cancel output buffering and gzipping if set
            // This should provide safer streaming for pages with history
            wfResetOutputBuffers();
            $request->response()->header( "Content-type: application/xml; charset=utf-8" );

            if ( $request->getCheck( 'wpDownload' ) ) {
                // Provide a sane filename suggestion
                $filename = urlencode( $config->get( 'Sitename' ) . '-' . wfTimestampNow() . '.xml' );
                $request->response()->header( "Content-disposition: attachment;filename={$filename}" );
            }

            $this->doExport( $page, $history, $list_authors, $exportall );

            return;
        }

        $out = $this->getOutput();
        $out->addWikiMsg( 'exporttext' );

        $form = Xml::openElement( 'form', array( 'method' => 'post',
            'action' => $this->getPageTitle()->getLocalURL( 'action=submit' ) ) );
        $form .= Xml::inputLabel(
            $this->msg( 'export-addcattext' )->text(),
            'catname',
            'catname',
            40
        ) . '&#160;';
        $form .= Xml::submitButton(
            $this->msg( 'export-addcat' )->text(),
            array( 'name' => 'addcat' )
        ) . '<br />';

        if ( $config->get( 'ExportFromNamespaces' ) ) {
            $form .= Html::namespaceSelector(
                array(
                    'selected' => $nsindex,
                    'label' => $this->msg( 'export-addnstext' )->text()
                ), array(
                    'name' => 'nsindex',
                    'id' => 'namespace',
                    'class' => 'namespaceselector',
                )
            ) . '&#160;';
            $form .= Xml::submitButton(
                $this->msg( 'export-addns' )->text(),
                array( 'name' => 'addns' )
            ) . '<br />';
        }

        if ( $config->get( 'ExportAllowAll' ) ) {
            $form .= Xml::checkLabel(
                $this->msg( 'exportall' )->text(),
                'exportall',
                'exportall',
                $request->wasPosted() ? $request->getCheck( 'exportall' ) : false
            ) . '<br />';
        }

        $form .= Xml::element(
            'textarea',
            array( 'name' => 'pages', 'cols' => 40, 'rows' => 10 ),
            $page,
            false
        );
        $form .= '<br />';

        if ( $config->get( 'ExportAllowHistory' ) ) {
            $form .= Xml::checkLabel(
                $this->msg( 'exportcuronly' )->text(),
                'curonly',
                'curonly',
                $request->wasPosted() ? $request->getCheck( 'curonly' ) : true
            ) . '<br />';
        } else {
            $out->addWikiMsg( 'exportnohistory' );
        }

        $form .= Xml::checkLabel(
            $this->msg( 'export-templates' )->text(),
            'templates',
            'wpExportTemplates',
            $request->wasPosted() ? $request->getCheck( 'templates' ) : false
        ) . '<br />';

        if ( $config->get( 'ExportMaxLinkDepth' ) || $this->userCanOverrideExportDepth() ) {
            $form .= Xml::inputLabel(
                $this->msg( 'export-pagelinks' )->text(),
                'pagelink-depth',
                'pagelink-depth',
                20,
                0
            ) . '<br />';
        }

        /* Enable this when we can do something useful exporting/importing image information.
        $form .= Xml::checkLabel(
                $this->msg( 'export-images' )->text(),
                'images',
                'wpExportImages',
                false
            ) . '<br />';
        */
        $form .= Xml::checkLabel(
            $this->msg( 'export-download' )->text(),
            'wpDownload',
            'wpDownload',
            $request->wasPosted() ? $request->getCheck( 'wpDownload' ) : true
        ) . '<br />';

        if ( $config->get( 'ExportAllowListContributors' ) ) {
            $form .= Xml::checkLabel(
                $this->msg( 'exportlistauthors' )->text(),
                'listauthors',
                'listauthors',
                $request->wasPosted() ? $request->getCheck( 'listauthors' ) : false
            ) . '<br />';
        }

        $form .= Xml::submitButton(
            $this->msg( 'export-submit' )->text(),
            Linker::tooltipAndAccesskeyAttribs( 'export' )
        );
        $form .= Xml::closeElement( 'form' );

        $out->addHTML( $form );
    }

    private function userCanOverrideExportDepth() {
        return $this->getUser()->isAllowed( 'override-export-depth' );
    }

    private function doExport( $page, $history, $list_authors, $exportall ) {

        // If we are grabbing everything, enable full history and ignore the rest
        if ( $exportall ) {
            $history = WikiExporter::FULL;
        } else {

            $pageSet = array(); // Inverted index of all pages to look up

            // Split up and normalize input
            foreach ( explode( "\n", $page ) as $pageName ) {
                $pageName = trim( $pageName );
                $title = Title::newFromText( $pageName );
                if ( $title && !$title->isExternal() && $title->getText() !== '' ) {
                    // Only record each page once!
                    $pageSet[$title->getPrefixedText()] = true;
                }
            }

            // Set of original pages to pass on to further manipulation...
            $inputPages = array_keys( $pageSet );

            // Look up any linked pages if asked...
            if ( $this->templates ) {
                $pageSet = $this->getTemplates( $inputPages, $pageSet );
            }
            $linkDepth = $this->pageLinkDepth;
            if ( $linkDepth ) {
                $pageSet = $this->getPageLinks( $inputPages, $pageSet, $linkDepth );
            }

            // Enable this when we can do something useful exporting/importing image information.
            // if( $this->images ) ) {
            // $pageSet = $this->getImages( $inputPages, $pageSet );
            // }

            $pages = array_keys( $pageSet );

            // Normalize titles to the same format and remove dupes, see bug 17374
            foreach ( $pages as $k => $v ) {
                $pages[$k] = str_replace( " ", "_", $v );
            }

            $pages = array_unique( $pages );
        }

        /* Ok, let's get to it... */
        if ( $history == WikiExporter::CURRENT ) {
            $lb = false;
            $db = wfGetDB( DB_SLAVE );
            $buffer = WikiExporter::BUFFER;
        } else {
            // Use an unbuffered query; histories may be very long!
            $lb = wfGetLBFactory()->newMainLB();
            $db = $lb->getConnection( DB_SLAVE );
            $buffer = WikiExporter::STREAM;

            // This might take a while... :D
            wfSuppressWarnings();
            set_time_limit( 0 );
            wfRestoreWarnings();
        }

        $exporter = new WikiExporter( $db, $history, $buffer );
        $exporter->list_authors = $list_authors;
        $exporter->openStream();

        if ( $exportall ) {
            $exporter->allPages();
        } else {
            foreach ( $pages as $page ) {
                #Bug 8824: Only export pages the user can read
                $title = Title::newFromText( $page );
                if ( is_null( $title ) ) {
                    // @todo Perhaps output an <error> tag or something.
                    continue;
                }

                if ( !$title->userCan( 'read', $this->getUser() ) ) {
                    // @todo Perhaps output an <error> tag or something.
                    continue;
                }

                $exporter->pageByTitle( $title );
            }
        }

        $exporter->closeStream();

        if ( $lb ) {
            $lb->closeAll();
        }
    }

    private function getPagesFromCategory( $title ) {
        global $wgContLang;

        $name = $title->getDBkey();

        $dbr = wfGetDB( DB_SLAVE );
        $res = $dbr->select(
            array( 'page', 'categorylinks' ),
            array( 'page_namespace', 'page_title' ),
            array( 'cl_from=page_id', 'cl_to' => $name ),
            __METHOD__,
            array( 'LIMIT' => '5000' )
        );

        $pages = array();

        foreach ( $res as $row ) {
            $n = $row->page_title;
            if ( $row->page_namespace ) {
                $ns = $wgContLang->getNsText( $row->page_namespace );
                $n = $ns . ':' . $n;
            }

            $pages[] = $n;
        }

        return $pages;
    }

    private function getPagesFromNamespace( $nsindex ) {
        global $wgContLang;

        $dbr = wfGetDB( DB_SLAVE );
        $res = $dbr->select(
            'page',
            array( 'page_namespace', 'page_title' ),
            array( 'page_namespace' => $nsindex ),
            __METHOD__,
            array( 'LIMIT' => '5000' )
        );

        $pages = array();

        foreach ( $res as $row ) {
            $n = $row->page_title;

            if ( $row->page_namespace ) {
                $ns = $wgContLang->getNsText( $row->page_namespace );
                $n = $ns . ':' . $n;
            }

            $pages[] = $n;
        }

        return $pages;
    }

    private function getTemplates( $inputPages, $pageSet ) {
        return $this->getLinks( $inputPages, $pageSet,
            'templatelinks',
            array( 'namespace' => 'tl_namespace', 'title' => 'tl_title' ),
            array( 'page_id=tl_from' )
        );
    }

    private function validateLinkDepth( $depth ) {
        if ( $depth < 0 ) {
            return 0;
        }

        if ( !$this->userCanOverrideExportDepth() ) {
            $maxLinkDepth = $this->getConfig()->get( 'ExportMaxLinkDepth' );
            if ( $depth > $maxLinkDepth ) {
                return $maxLinkDepth;
            }
        }

        /*
         * There's a HARD CODED limit of 5 levels of recursion here to prevent a
         * crazy-big export from being done by someone setting the depth
         * number too high. In other words, last resort safety net.
         */

        return intval( min( $depth, 5 ) );
    }

    private function getPageLinks( $inputPages, $pageSet, $depth ) {
        // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
        for ( ; $depth > 0; --$depth ) {
            // @codingStandardsIgnoreEnd
            $pageSet = $this->getLinks(
                $inputPages, $pageSet, 'pagelinks',
                array( 'namespace' => 'pl_namespace', 'title' => 'pl_title' ),
                array( 'page_id=pl_from' )
            );
            $inputPages = array_keys( $pageSet );
        }

        return $pageSet;
    }

    private function getImages( $inputPages, $pageSet ) {
        return $this->getLinks(
            $inputPages,
            $pageSet,
            'imagelinks',
            array( 'namespace' => NS_FILE, 'title' => 'il_to' ),
            array( 'page_id=il_from' )
        );
    }

    private function getLinks( $inputPages, $pageSet, $table, $fields, $join ) {
        $dbr = wfGetDB( DB_SLAVE );

        foreach ( $inputPages as $page ) {
            $title = Title::newFromText( $page );

            if ( $title ) {
                $pageSet[$title->getPrefixedText()] = true;
                $result = $dbr->select(
                    array( 'page', $table ),
                    $fields,
                    array_merge(
                        $join,
                        array(
                            'page_namespace' => $title->getNamespace(),
                            'page_title' => $title->getDBkey()
                        )
                    ),
                    __METHOD__
                );

                foreach ( $result as $row ) {
                    $template = Title::makeTitle( $row->namespace, $row->title );
                    $pageSet[$template->getPrefixedText()] = true;
                }
            }
        }

        return $pageSet;
    }

    protected function getGroupName() {
        return 'pagetools';
    }
}