ParserOutput.php

Go to the documentation of this file.

<?php
 
class ParserOutput extends CacheTime {
    const SUPPORTS_STATELESS_TRANSFORMS = 1;
    const SUPPORTS_UNWRAP_TRANSFORM = 1;
 
    public $mText;
 
    public $mLanguageLinks;
 
    public $mCategories;
 
    public $mIndicators = [];
 
    public $mTitleText;
 
    public $mLinks = [];
 
    public $mTemplates = [];
 
    public $mTemplateIds = [];
 
    public $mImages = [];
 
    public $mFileSearchOptions = [];
 
    public $mExternalLinks = [];
 
    public $mInterwikiLinks = [];
 
    public $mNewSection = false;
 
    public $mHideNewSection = false;
 
    public $mNoGallery = false;
 
    public $mHeadItems = [];
 
    public $mModules = [];
 
    public $mModuleScripts = [];
 
    public $mModuleStyles = [];
 
    public $mJsConfigVars = [];
 
    public $mOutputHooks = [];
 
    public $mWarnings = [];
 
    public $mSections = [];
 
    public $mProperties = [];
 
    public $mTOCHTML = '';
 
    public $mTimestamp;
 
    public $mEnableOOUI = false;
 
    private $mIndexPolicy = '';
 
    private $mAccessedOptions = [];
 
    private $mExtensionData = [];
 
    private $mLimitReportData = [];
 
    private $mLimitReportJSData = [];
 
    private $mParseStartTime = [];
 
    private $mPreventClickjacking = false;
 
    private $mFlags = [];
 
    private $mSpeculativeRevId;
 
    private $mMaxAdaptiveExpiry = INF;
 
    const EDITSECTION_REGEX =
        '#<(?:mw:)?editsection page="(.*?)" section="(.*?)"(?:/>|>(.*?)(</(?:mw:)?editsection>))#s';
 
    // finalizeAdaptiveCacheExpiry() uses TTL = MAX( m * PARSE_TIME + b, MIN_AR_TTL)
    // Current values imply that m=3933.333333 and b=-333.333333
    // See https://www.nngroup.com/articles/website-response-times/
    const PARSE_FAST_SEC = 0.100; // perceived "fast" page parse
    const PARSE_SLOW_SEC = 1.0; // perceived "slow" page parse
    const FAST_AR_TTL = 60; // adaptive TTL for "fast" pages
    const SLOW_AR_TTL = 3600; // adaptive TTL for "slow" pages
    const MIN_AR_TTL = 15; // min adaptive TTL (for sanity, pool counter, and edit stashing)
 
    public function __construct( $text = '', $languageLinks = [], $categoryLinks = [],
        $unused = false, $titletext = ''
    ) {
        $this->mText = $text;
        $this->mLanguageLinks = $languageLinks;
        $this->mCategories = $categoryLinks;
        $this->mTitleText = $titletext;
    }
 
    public function getRawText() {
        return $this->mText;
    }
 
    public function getText( $options = [] ) {
        $options += [
            'allowTOC' => true,
            'enableSectionEditLinks' => true,
            'unwrap' => false,
            'deduplicateStyles' => true,
        ];
        $text = $this->mText;
 
        Hooks::runWithoutAbort( 'ParserOutputPostCacheTransform', [ $this, &$text, &$options ] );
 
        if ( $options['unwrap'] !== false ) {
            $start = Html::openElement( 'div', [
                'class' => 'mw-parser-output'
            ] );
            $startLen = strlen( $start );
            $end = Html::closeElement( 'div' );
            $endPos = strrpos( $text, $end );
            $endLen = strlen( $end );
 
            if ( substr( $text, 0, $startLen ) === $start && $endPos !== false
                // if the closing div is followed by real content, bail out of unwrapping
                && preg_match( '/^(?>\s*<!--.*?-->)*\s*$/s', substr( $text, $endPos + $endLen ) )
            ) {
                $text = substr( $text, $startLen );
                $text = substr( $text, 0, $endPos - $startLen )
                    . substr( $text, $endPos - $startLen + $endLen );
            }
        }
 
        if ( $options['enableSectionEditLinks'] ) {
            $text = preg_replace_callback(
                self::EDITSECTION_REGEX,
                function ( $m ) {
                    global $wgOut, $wgLang;
                    $editsectionPage = Title::newFromText( htmlspecialchars_decode( $m[1] ) );
                    $editsectionSection = htmlspecialchars_decode( $m[2] );
                    $editsectionContent = isset( $m[4] ) ? Sanitizer::decodeCharReferences( $m[3] ) : null;
 
                    if ( !is_object( $editsectionPage ) ) {
                        throw new MWException( "Bad parser output text." );
                    }
 
                    $skin = $wgOut->getSkin();
                    return call_user_func_array(
                        [ $skin, 'doEditSectionLink' ],
                        [ $editsectionPage, $editsectionSection,
                            $editsectionContent, $wgLang->getCode() ]
                    );
                },
                $text
            );
        } else {
            $text = preg_replace( self::EDITSECTION_REGEX, '', $text );
        }
 
        if ( $options['allowTOC'] ) {
            $text = str_replace( [ Parser::TOC_START, Parser::TOC_END ], '', $text );
        } else {
            $text = preg_replace(
                '#' . preg_quote( Parser::TOC_START, '#' ) . '.*?' . preg_quote( Parser::TOC_END, '#' ) . '#s',
                '',
                $text
            );
        }
 
        if ( $options['deduplicateStyles'] ) {
            $seen = [];
            $text = preg_replace_callback(
                '#<style\s+([^>]*data-mw-deduplicate\s*=[^>]*)>.*?</style>#s',
                function ( $m ) use ( &$seen ) {
                    $attr = Sanitizer::decodeTagAttributes( $m[1] );
                    if ( !isset( $attr['data-mw-deduplicate'] ) ) {
                        return $m[0];
                    }
 
                    $key = $attr['data-mw-deduplicate'];
                    if ( !isset( $seen[$key] ) ) {
                        $seen[$key] = true;
                        return $m[0];
                    }
 
                    // We were going to use an empty <style> here, but there
                    // was concern that would be too much overhead for browsers.
                    // So let's hope a <link> with a non-standard rel and href isn't
                    // going to be misinterpreted or mangled by any subsequent processing.
                    return Html::element( 'link', [
                        'rel' => 'mw-deduplicated-inline-style',
                        'href' => "mw-data:" . wfUrlencode( $key ),
                    ] );
                },
                $text
            );
        }
 
        return $text;
    }
 
    public function setSpeculativeRevIdUsed( $id ) {
        $this->mSpeculativeRevId = $id;
    }
 
    public function getSpeculativeRevIdUsed() {
        return $this->mSpeculativeRevId;
    }
 
    public function &getLanguageLinks() {
        return $this->mLanguageLinks;
    }
 
    public function getInterwikiLinks() {
        return $this->mInterwikiLinks;
    }
 
    public function getCategoryLinks() {
        return array_keys( $this->mCategories );
    }
 
    public function &getCategories() {
        return $this->mCategories;
    }
 
    public function getIndicators() {
        return $this->mIndicators;
    }
 
    public function getTitleText() {
        return $this->mTitleText;
    }
 
    public function getSections() {
        return $this->mSections;
    }
 
    public function getEditSectionTokens() {
        wfDeprecated( __METHOD__, '1.31' );
        return true;
    }
 
    public function &getLinks() {
        return $this->mLinks;
    }
 
    public function &getTemplates() {
        return $this->mTemplates;
    }
 
    public function &getTemplateIds() {
        return $this->mTemplateIds;
    }
 
    public function &getImages() {
        return $this->mImages;
    }
 
    public function &getFileSearchOptions() {
        return $this->mFileSearchOptions;
    }
 
    public function &getExternalLinks() {
        return $this->mExternalLinks;
    }
 
    public function getNoGallery() {
        return $this->mNoGallery;
    }
 
    public function getHeadItems() {
        return $this->mHeadItems;
    }
 
    public function getModules() {
        return $this->mModules;
    }
 
    public function getModuleScripts() {
        return $this->mModuleScripts;
    }
 
    public function getModuleStyles() {
        return $this->mModuleStyles;
    }
 
    public function getJsConfigVars() {
        return $this->mJsConfigVars;
    }
 
    public function getOutputHooks() {
        return (array)$this->mOutputHooks;
    }
 
    public function getWarnings() {
        return array_keys( $this->mWarnings );
    }
 
    public function getIndexPolicy() {
        return $this->mIndexPolicy;
    }
 
    public function getTOCHTML() {
        return $this->mTOCHTML;
    }
 
    public function getTimestamp() {
        return $this->mTimestamp;
    }
 
    public function getLimitReportData() {
        return $this->mLimitReportData;
    }
 
    public function getLimitReportJSData() {
        return $this->mLimitReportJSData;
    }
 
    public function getTOCEnabled() {
        wfDeprecated( __METHOD__, '1.31' );
        return true;
    }
 
    public function getEnableOOUI() {
        return $this->mEnableOOUI;
    }
 
    public function setText( $text ) {
        return wfSetVar( $this->mText, $text );
    }
 
    public function setLanguageLinks( $ll ) {
        return wfSetVar( $this->mLanguageLinks, $ll );
    }
 
    public function setCategoryLinks( $cl ) {
        return wfSetVar( $this->mCategories, $cl );
    }
 
    public function setTitleText( $t ) {
        return wfSetVar( $this->mTitleText, $t );
    }
 
    public function setSections( $toc ) {
        return wfSetVar( $this->mSections, $toc );
    }
 
    public function setEditSectionTokens( $t ) {
        wfDeprecated( __METHOD__, '1.31' );
        return true;
    }
 
    public function setIndexPolicy( $policy ) {
        return wfSetVar( $this->mIndexPolicy, $policy );
    }
 
    public function setTOCHTML( $tochtml ) {
        return wfSetVar( $this->mTOCHTML, $tochtml );
    }
 
    public function setTimestamp( $timestamp ) {
        return wfSetVar( $this->mTimestamp, $timestamp );
    }
 
    public function setTOCEnabled( $flag ) {
        wfDeprecated( __METHOD__, '1.31' );
        return true;
    }
 
    public function addCategory( $c, $sort ) {
        $this->mCategories[$c] = $sort;
    }
 
    public function setIndicator( $id, $content ) {
        $this->mIndicators[$id] = $content;
    }
 
    public function setEnableOOUI( $enable = false ) {
        $this->mEnableOOUI = $enable;
    }
 
    public function addLanguageLink( $t ) {
        $this->mLanguageLinks[] = $t;
    }
 
    public function addWarning( $s ) {
        $this->mWarnings[$s] = 1;
    }
 
    public function addOutputHook( $hook, $data = false ) {
        $this->mOutputHooks[] = [ $hook, $data ];
    }
 
    public function setNewSection( $value ) {
        $this->mNewSection = (bool)$value;
    }
    public function hideNewSection( $value ) {
        $this->mHideNewSection = (bool)$value;
    }
    public function getHideNewSection() {
        return (bool)$this->mHideNewSection;
    }
    public function getNewSection() {
        return (bool)$this->mNewSection;
    }
 
    public static function isLinkInternal( $internal, $url ) {
        return (bool)preg_match( '/^' .
            # If server is proto relative, check also for http/https links
            ( substr( $internal, 0, 2 ) === '//' ? '(?:https?:)?' : '' ) .
            preg_quote( $internal, '/' ) .
            # check for query/path/anchor or end of link in each case
            '(?:[\?\/\#]|$)/i',
            $url
        );
    }
 
    public function addExternalLink( $url ) {
        # We don't register links pointing to our own server, unless... :-)
        global $wgServer, $wgRegisterInternalExternals;
 
        # Replace unnecessary URL escape codes with the referenced character
        # This prevents spammers from hiding links from the filters
        $url = Parser::normalizeLinkUrl( $url );
 
        $registerExternalLink = true;
        if ( !$wgRegisterInternalExternals ) {
            $registerExternalLink = !self::isLinkInternal( $wgServer, $url );
        }
        if ( $registerExternalLink ) {
            $this->mExternalLinks[$url] = 1;
        }
    }
 
    public function addLink( Title $title, $id = null ) {
        if ( $title->isExternal() ) {
            // Don't record interwikis in pagelinks
            $this->addInterwikiLink( $title );
            return;
        }
        $ns = $title->getNamespace();
        $dbk = $title->getDBkey();
        if ( $ns == NS_MEDIA ) {
            // Normalize this pseudo-alias if it makes it down here...
            $ns = NS_FILE;
        } elseif ( $ns == NS_SPECIAL ) {
            // We don't record Special: links currently
            // It might actually be wise to, but we'd need to do some normalization.
            return;
        } elseif ( $dbk === '' ) {
            // Don't record self links -  [[#Foo]]
            return;
        }
        if ( !isset( $this->mLinks[$ns] ) ) {
            $this->mLinks[$ns] = [];
        }
        if ( is_null( $id ) ) {
            $id = $title->getArticleID();
        }
        $this->mLinks[$ns][$dbk] = $id;
    }
 
    public function addImage( $name, $timestamp = null, $sha1 = null ) {
        $this->mImages[$name] = 1;
        if ( $timestamp !== null && $sha1 !== null ) {
            $this->mFileSearchOptions[$name] = [ 'time' => $timestamp, 'sha1' => $sha1 ];
        }
    }
 
    public function addTemplate( $title, $page_id, $rev_id ) {
        $ns = $title->getNamespace();
        $dbk = $title->getDBkey();
        if ( !isset( $this->mTemplates[$ns] ) ) {
            $this->mTemplates[$ns] = [];
        }
        $this->mTemplates[$ns][$dbk] = $page_id;
        if ( !isset( $this->mTemplateIds[$ns] ) ) {
            $this->mTemplateIds[$ns] = [];
        }
        $this->mTemplateIds[$ns][$dbk] = $rev_id; // For versioning
    }
 
    public function addInterwikiLink( $title ) {
        if ( !$title->isExternal() ) {
            throw new MWException( 'Non-interwiki link passed, internal parser error.' );
        }
        $prefix = $title->getInterwiki();
        if ( !isset( $this->mInterwikiLinks[$prefix] ) ) {
            $this->mInterwikiLinks[$prefix] = [];
        }
        $this->mInterwikiLinks[$prefix][$title->getDBkey()] = 1;
    }
 
    public function addHeadItem( $section, $tag = false ) {
        if ( $tag !== false ) {
            $this->mHeadItems[$tag] = $section;
        } else {
            $this->mHeadItems[] = $section;
        }
    }
 
    public function addModules( $modules ) {
        $this->mModules = array_merge( $this->mModules, (array)$modules );
    }
 
    public function addModuleScripts( $modules ) {
        $this->mModuleScripts = array_merge( $this->mModuleScripts, (array)$modules );
    }
 
    public function addModuleStyles( $modules ) {
        $this->mModuleStyles = array_merge( $this->mModuleStyles, (array)$modules );
    }
 
    public function addJsConfigVars( $keys, $value = null ) {
        if ( is_array( $keys ) ) {
            foreach ( $keys as $key => $value ) {
                $this->mJsConfigVars[$key] = $value;
            }
            return;
        }
 
        $this->mJsConfigVars[$keys] = $value;
    }
 
    public function addOutputPageMetadata( OutputPage $out ) {
        $this->addModules( $out->getModules() );
        $this->addModuleScripts( $out->getModuleScripts() );
        $this->addModuleStyles( $out->getModuleStyles() );
        $this->addJsConfigVars( $out->getJsConfigVars() );
 
        $this->mHeadItems = array_merge( $this->mHeadItems, $out->getHeadItemsArray() );
        $this->mPreventClickjacking = $this->mPreventClickjacking || $out->getPreventClickjacking();
    }
 
    public function addTrackingCategory( $msg, $title ) {
        if ( $title->isSpecialPage() ) {
            wfDebug( __METHOD__ . ": Not adding tracking category $msg to special page!\n" );
            return false;
        }
 
        // Important to parse with correct title (T33469)
        $cat = wfMessage( $msg )
            ->title( $title )
            ->inContentLanguage()
            ->text();
 
        # Allow tracking categories to be disabled by setting them to "-"
        if ( $cat === '-' ) {
            return false;
        }
 
        $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
        if ( $containerCategory ) {
            $this->addCategory( $containerCategory->getDBkey(), $this->getProperty( 'defaultsort' ) ?: '' );
            return true;
        } else {
            wfDebug( __METHOD__ . ": [[MediaWiki:$msg]] is not a valid title!\n" );
            return false;
        }
    }
 
    public function setDisplayTitle( $text ) {
        $this->setTitleText( $text );
        $this->setProperty( 'displaytitle', $text );
    }
 
    public function getDisplayTitle() {
        $t = $this->getTitleText();
        if ( $t === '' ) {
            return false;
        }
        return $t;
    }
 
    public function setFlag( $flag ) {
        $this->mFlags[$flag] = true;
    }
 
    public function getFlag( $flag ) {
        return isset( $this->mFlags[$flag] );
    }
 
    public function setProperty( $name, $value ) {
        $this->mProperties[$name] = $value;
    }
 
    public function getProperty( $name ) {
        return isset( $this->mProperties[$name] ) ? $this->mProperties[$name] : false;
    }
 
    public function unsetProperty( $name ) {
        unset( $this->mProperties[$name] );
    }
 
    public function getProperties() {
        if ( !isset( $this->mProperties ) ) {
            $this->mProperties = [];
        }
        return $this->mProperties;
    }
 
    public function getUsedOptions() {
        if ( !isset( $this->mAccessedOptions ) ) {
            return [];
        }
        return array_keys( $this->mAccessedOptions );
    }
 
    public function recordOption( $option ) {
        $this->mAccessedOptions[$option] = true;
    }
 
    public function setExtensionData( $key, $value ) {
        if ( $value === null ) {
            unset( $this->mExtensionData[$key] );
        } else {
            $this->mExtensionData[$key] = $value;
        }
    }
 
    public function getExtensionData( $key ) {
        if ( isset( $this->mExtensionData[$key] ) ) {
            return $this->mExtensionData[$key];
        }
 
        return null;
    }
 
    private static function getTimes( $clock = null ) {
        $ret = [];
        if ( !$clock || $clock === 'wall' ) {
            $ret['wall'] = microtime( true );
        }
        if ( !$clock || $clock === 'cpu' ) {
            $ru = wfGetRusage();
            if ( $ru ) {
                $ret['cpu'] = $ru['ru_utime.tv_sec'] + $ru['ru_utime.tv_usec'] / 1e6;
                $ret['cpu'] += $ru['ru_stime.tv_sec'] + $ru['ru_stime.tv_usec'] / 1e6;
            }
        }
        return $ret;
    }
 
    public function resetParseStartTime() {
        $this->mParseStartTime = self::getTimes();
    }
 
    public function getTimeSinceStart( $clock ) {
        if ( !isset( $this->mParseStartTime[$clock] ) ) {
            return null;
        }
 
        $end = self::getTimes( $clock );
        return $end[$clock] - $this->mParseStartTime[$clock];
    }
 
    public function setLimitReportData( $key, $value ) {
        $this->mLimitReportData[$key] = $value;
 
        if ( is_array( $value ) ) {
            if ( array_keys( $value ) === [ 0, 1 ]
                && is_numeric( $value[0] )
                && is_numeric( $value[1] )
            ) {
                $data = [ 'value' => $value[0], 'limit' => $value[1] ];
            } else {
                $data = $value;
            }
        } else {
            $data = $value;
        }
 
        if ( strpos( $key, '-' ) ) {
            list( $ns, $name ) = explode( '-', $key, 2 );
            $this->mLimitReportJSData[$ns][$name] = $data;
        } else {
            $this->mLimitReportJSData[$key] = $data;
        }
    }
 
    public function hasDynamicContent() {
        global $wgParserCacheExpireTime;
 
        return $this->getCacheExpiry() < $wgParserCacheExpireTime;
    }
 
    public function preventClickjacking( $flag = null ) {
        return wfSetVar( $this->mPreventClickjacking, $flag );
    }
 
    public function updateRuntimeAdaptiveExpiry( $ttl ) {
        $this->mMaxAdaptiveExpiry = min( $ttl, $this->mMaxAdaptiveExpiry );
        $this->updateCacheExpiry( $ttl );
    }
 
    public function finalizeAdaptiveCacheExpiry() {
        if ( is_infinite( $this->mMaxAdaptiveExpiry ) ) {
            return; // not set
        }
 
        $runtime = $this->getTimeSinceStart( 'wall' );
        if ( is_float( $runtime ) ) {
            $slope = ( self::SLOW_AR_TTL - self::FAST_AR_TTL )
                / ( self::PARSE_SLOW_SEC - self::PARSE_FAST_SEC );
            // SLOW_AR_TTL = PARSE_SLOW_SEC * $slope + $point
            $point = self::SLOW_AR_TTL - self::PARSE_SLOW_SEC * $slope;
 
            $adaptiveTTL = min(
                max( $slope * $runtime + $point, self::MIN_AR_TTL ),
                $this->mMaxAdaptiveExpiry
            );
            $this->updateCacheExpiry( $adaptiveTTL );
        }
    }
 
    public function __sleep() {
        return array_diff(
            array_keys( get_object_vars( $this ) ),
            [ 'mParseStartTime' ]
        );
    }
}