MediaWikiSite.php

Go to the documentation of this file.

<?php
class MediaWikiSite extends Site {
    const PATH_FILE = 'file_path';
    const PATH_PAGE = 'page_path';

    public static function newFromGlobalId( $globalId ) {
        $site = new static();
        $site->setGlobalId( $globalId );
        return $site;
    }

    public function __construct( $type = self::TYPE_MEDIAWIKI ) {
        parent::__construct( $type );
    }

    public function toDBKey( $title ) {
        return str_replace( ' ', '_', $title );
    }

    public function normalizePageName( $pageName ) {

        // Check if we have strings as arguments.
        if ( !is_string( $pageName ) ) {
            throw new MWException( '$pageName must be a string' );
        }

        // Go on call the external site
        if ( defined( 'MW_PHPUNIT_TEST' ) ) {
            // If the code is under test, don't call out to other sites, just
            // normalize locally.
            // Note: this may cause results to be inconsistent with the actual
            // normalization used by the respective remote site!

            $t = Title::newFromText( $pageName );
            return $t->getPrefixedText();
        } else {

            // Make sure the string is normalized into NFC (due to the bug 40017)
            // but do nothing to the whitespaces, that should work appropriately.
            // @see https://bugzilla.wikimedia.org/show_bug.cgi?id=40017
            $pageName = UtfNormal::cleanUp( $pageName );

            // Build the args for the specific call
            $args = array(
                'action' => 'query',
                'prop' => 'info',
                'redirects' => true,
                'converttitles' => true,
                'format' => 'json',
                'titles' => $pageName,
                // @todo options for maxlag and maxage
                // Note that maxlag will lead to a long delay before a reply is made,
                // but that maxage can avoid the extreme delay. On the other hand
                // maxage could be nice to use anyhow as it stops unnecessary requests.
                // Also consider smaxage if maxage is used.
            );

            $url = wfAppendQuery( $this->getFileUrl( 'api.php' ), $args );

            // Go on call the external site
            // @todo we need a good way to specify a timeout here.
            $ret = Http::get( $url );
        }

        if ( $ret === false ) {
            wfDebugLog( "MediaWikiSite", "call to external site failed: $url" );
            return false;
        }

        $data = FormatJson::decode( $ret, true );

        if ( !is_array( $data ) ) {
            wfDebugLog( "MediaWikiSite", "call to <$url> returned bad json: " . $ret );
            return false;
        }

        $page = static::extractPageRecord( $data, $pageName );

        if ( isset( $page['missing'] ) ) {
            wfDebugLog( "MediaWikiSite", "call to <$url> returned a marker for a missing page title! "
                . $ret );
            return false;
        }

        if ( isset( $page['invalid'] ) ) {
            wfDebugLog( "MediaWikiSite", "call to <$url> returned a marker for an invalid page title! "
                . $ret );
            return false;
        }

        if ( !isset( $page['title'] ) ) {
            wfDebugLog( "MediaWikiSite", "call to <$url> did not return a page title! " . $ret );
            return false;
        }

        return $page['title'];
    }

    private static function extractPageRecord( $externalData, $pageTitle ) {
        // If there is a special case with only one returned page
        // we can cheat, and only return
        // the single page in the "pages" substructure.
        if ( isset( $externalData['query']['pages'] ) ) {
            $pages = array_values( $externalData['query']['pages'] );
            if ( count( $pages ) === 1 ) {
                return $pages[0];
            }
        }
        // This is only used during internal testing, as it is assumed
        // a more optimal (and lossfree) storage.
        // Make initial checks and return if prerequisites are not meet.
        if ( !is_array( $externalData ) || !isset( $externalData['query'] ) ) {
            return false;
        }
        // Loop over the tree different named structures, that otherwise are similar
        $structs = array(
            'normalized' => 'from',
            'converted' => 'from',
            'redirects' => 'from',
            'pages' => 'title'
        );
        foreach ( $structs as $listId => $fieldId ) {
            // Check if the substructure exist at all.
            if ( !isset( $externalData['query'][$listId] ) ) {
                continue;
            }
            // Filter the substructure down to what we actually are using.
            $collectedHits = array_filter(
                array_values( $externalData['query'][$listId] ),
                function ( $a ) use ( $fieldId, $pageTitle ) {
                    return $a[$fieldId] === $pageTitle;
                }
            );
            // If still looping over normalization, conversion or redirects,
            // then we need to keep the new page title for later rounds.
            if ( $fieldId === 'from' && is_array( $collectedHits ) ) {
                switch ( count( $collectedHits ) ) {
                    case 0:
                        break;
                    case 1:
                        $pageTitle = $collectedHits[0]['to'];
                        break;
                    default:
                        return false;
                }
            }
            // If on the pages structure we should prepare for returning.
            elseif ( $fieldId === 'title' && is_array( $collectedHits ) ) {
                switch ( count( $collectedHits ) ) {
                    case 0:
                        return false;
                    case 1:
                        return array_shift( $collectedHits );
                    default:
                        return false;
                }
            }
        }
        // should never be here
        return false;
    }

    public function getLinkPathType() {
        return self::PATH_PAGE;
    }

    public function getRelativePagePath() {
        return parse_url( $this->getPath( self::PATH_PAGE ), PHP_URL_PATH );
    }

    public function getRelativeFilePath() {
        return parse_url( $this->getPath( self::PATH_FILE ), PHP_URL_PATH );
    }

    public function setPagePath( $path ) {
        $this->setPath( self::PATH_PAGE, $path );
    }

    public function setFilePath( $path ) {
        $this->setPath( self::PATH_FILE, $path );
    }

    public function getPageUrl( $pageName = false ) {
        $url = $this->getLinkPath();

        if ( $url === false ) {
            return false;
        }

        if ( $pageName !== false ) {
            $pageName = $this->toDBKey( trim( $pageName ) );
            $url = str_replace( '$1', wfUrlencode( $pageName ), $url );
        }

        return $url;
    }

    public function getFileUrl( $path = false ) {
        $filePath = $this->getPath( self::PATH_FILE );

        if ( $filePath !== false ) {
            $filePath = str_replace( '$1', $path, $filePath );
        }

        return $filePath;
    }
}