ParserOutput.php

Go to the documentation of this file.

<?php
 
use MediaWiki\Json\JsonUnserializable;
use MediaWiki\Json\JsonUnserializableTrait;
use MediaWiki\Json\JsonUnserializer;
use MediaWiki\Logger\LoggerFactory;
use Wikimedia\Reflection\GhostFieldAccessTrait;
 
class ParserOutput extends CacheTime {
    use GhostFieldAccessTrait;
    use JsonUnserializableTrait;
 
    public const SUPPORTS_STATELESS_TRANSFORMS = 1;
 
    public const SUPPORTS_UNWRAP_TRANSFORM = 1;
 
    public $mText = null;
 
    public $mLanguageLinks;
 
    public $mCategories;
 
    public $mIndicators = [];
 
    public $mTitleText;
 
    public $mLinks = [];
 
    public $mLinksSpecial = [];
 
    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 $mModuleStyles = [];
 
    public $mJsConfigVars = [];
 
    public $mOutputHooks = [];
 
    public $mWarnings = [];
 
    public $mSections = [];
 
    public $mProperties = [];
 
    public $mTOCHTML = '';
 
    public $mTimestamp;
 
    public $mEnableOOUI = false;
 
    private $mIndexPolicy = '';
 
    private $mExtensionData = [];
 
    private $mLimitReportData = [];
 
    private $mLimitReportJSData = [];
 
    private $mParseStartTime = [];
 
    private $mPreventClickjacking = false;
 
    private $mExtraScriptSrcs = [];
 
    private $mExtraDefaultSrcs = [];
 
    private $mExtraStyleSrcs = [];
 
    private $mFlags = [];
 
    private const SPECULATIVE_FIELDS = [
        'speculativePageIdUsed',
        'mSpeculativeRevId',
        'revisionTimestampUsed'
    ];
 
    private $mSpeculativeRevId;
    private $speculativePageIdUsed;
    private $revisionTimestampUsed;
 
    private $revisionUsedSha1Base36;
 
    private $mWrapperDivClasses = [];
 
    private $mMaxAdaptiveExpiry = INF;
 
    private 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/
    private const PARSE_FAST_SEC = 0.100; // perceived "fast" page parse
    private const PARSE_SLOW_SEC = 1.0; // perceived "slow" page parse
    private const FAST_AR_TTL = 60; // adaptive TTL for "fast" pages
    private const SLOW_AR_TTL = 3600; // adaptive TTL for "slow" pages
    private 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 hasText() {
        return ( $this->mText !== null );
    }
 
    public function getRawText() {
        if ( $this->mText === null ) {
            throw new LogicException( 'This ParserOutput contains no text!' );
        }
 
        return $this->mText;
    }
 
    public function getText( $options = [] ) {
        $options += [
            'allowTOC' => true,
            'enableSectionEditLinks' => true,
            'skin' => null,
            'unwrap' => false,
            'deduplicateStyles' => true,
            'wrapperDivClass' => $this->getWrapperDivClass(),
        ];
        $text = $this->getRawText();
 
        Hooks::runner()->onParserOutputPostCacheTransform( $this, $text, $options );
 
        if ( $options['wrapperDivClass'] !== '' && !$options['unwrap'] ) {
            $text = Html::rawElement( 'div', [ 'class' => $options['wrapperDivClass'] ], $text );
        }
 
        if ( $options['enableSectionEditLinks'] ) {
            // TODO: Passing the skin should be required
            $skin = $options['skin'] ?: RequestContext::getMain()->getSkin();
 
            $text = preg_replace_callback(
                self::EDITSECTION_REGEX,
                function ( $m ) use ( $skin ) {
                    $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 ) ) {
                        LoggerFactory::getInstance( 'Parser' )
                            ->error(
                                'ParserOutput::getText(): bad title in editsection placeholder',
                                [
                                    'placeholder' => $m[0],
                                    'editsectionPage' => $m[1],
                                    'titletext' => $this->getTitleText(),
                                    'phab' => 'T261347'
                                ]
                            );
                        return '';
                    }
 
                    return $skin->doEditSectionLink(
                        $editsectionPage,
                        $editsectionSection,
                        $editsectionContent,
                        $skin->getLanguage()
                    );
                },
                $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',
                static 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
            );
        }
 
        // Hydrate slot section header placeholders generated by RevisionRenderer.
        $text = preg_replace_callback(
            '#<mw:slotheader>(.*?)</mw:slotheader>#',
            static function ( $m ) {
                $role = htmlspecialchars_decode( $m[1] );
                // TODO: map to message, using the interface language. Set lang="xyz" accordingly.
                $headerText = $role;
                return $headerText;
            },
            $text
        );
        return $text;
    }
 
    public function addCacheMessage( string $msg ) {
        $this->mText .= "\n<!-- $msg\n -->\n";
    }
 
    public function addWrapperDivClass( $class ) {
        $this->mWrapperDivClasses[$class] = true;
    }
 
    public function clearWrapperDivClass() {
        $this->mWrapperDivClasses = [];
    }
 
    public function getWrapperDivClass() {
        return implode( ' ', array_keys( $this->mWrapperDivClasses ) );
    }
 
    public function setSpeculativeRevIdUsed( $id ) {
        $this->mSpeculativeRevId = $id;
    }
 
    public function getSpeculativeRevIdUsed() {
        return $this->mSpeculativeRevId;
    }
 
    public function setSpeculativePageIdUsed( $id ) {
        $this->speculativePageIdUsed = $id;
    }
 
    public function getSpeculativePageIdUsed() {
        return $this->speculativePageIdUsed;
    }
 
    public function setRevisionTimestampUsed( $timestamp ) {
        $this->revisionTimestampUsed = $timestamp;
    }
 
    public function getRevisionTimestampUsed() {
        return $this->revisionTimestampUsed;
    }
 
    public function setRevisionUsedSha1Base36( $hash ) {
        if ( $hash === null ) {
            return; // e.g. RevisionRecord::getSha1() returned null
        }
 
        if (
            $this->revisionUsedSha1Base36 !== null &&
            $this->revisionUsedSha1Base36 !== $hash
        ) {
            $this->revisionUsedSha1Base36 = ''; // mismatched
        } else {
            $this->revisionUsedSha1Base36 = $hash;
        }
    }
 
    public function getRevisionUsedSha1Base36() {
        return $this->revisionUsedSha1Base36;
    }
 
    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 &getLinks() {
        return $this->mLinks;
    }
 
    public function &getLinksSpecial() {
        return $this->mLinksSpecial;
    }
 
    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 setNoGallery( $value ) {
        $this->mNoGallery = (bool)$value;
    }
 
    public function getNoGallery() {
        return $this->mNoGallery;
    }
 
    public function getHeadItems() {
        return $this->mHeadItems;
    }
 
    public function getModules() {
        return $this->mModules;
    }
 
    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 getEnableOOUI() {
        return $this->mEnableOOUI;
    }
 
    public function getExtraCSPDefaultSrcs() {
        return $this->mExtraDefaultSrcs;
    }
 
    public function getExtraCSPScriptSrcs() {
        return $this->mExtraScriptSrcs;
    }
 
    public function getExtraCSPStyleSrcs() {
        return $this->mExtraStyleSrcs;
    }
 
    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 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 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 want to record Special: links in the database, so put them in a separate place.
            // It might actually be wise to, but we'd need to do some normalization.
            $this->mLinksSpecial[$dbk] = 1;
            return;
        } elseif ( $dbk === '' ) {
            // Don't record self links -  [[#Foo]]
            return;
        }
        if ( !isset( $this->mLinks[$ns] ) ) {
            $this->mLinks[$ns] = [];
        }
        if ( $id === null ) {
            $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 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->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!" );
            return false;
        }
 
        // Important to parse with correct title (T33469)
        $cat = wfMessage( $msg )
            ->page( $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!" );
            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 getAllFlags() {
        return array_keys( $this->mFlags );
    }
 
    public function setProperty( $name, $value ) {
        $this->mProperties[$name] = $value;
    }
 
    public function getProperty( $name ) {
        return $this->mProperties[$name] ?? false;
    }
 
    public function unsetProperty( $name ) {
        unset( $this->mProperties[$name] );
    }
 
    public function getProperties() {
        if ( !isset( $this->mProperties ) ) {
            $this->mProperties = [];
        }
        return $this->mProperties;
    }
 
    public function setExtensionData( $key, $value ) {
        if ( $value === null ) {
            unset( $this->mExtensionData[$key] );
        } else {
            $this->mExtensionData[$key] = $value;
        }
    }
 
    public function getExtensionData( $key ) {
        return $this->mExtensionData[$key] ?? null;
    }
 
    private static function getTimes( $clock = null ) {
        $ret = [];
        if ( !$clock || $clock === 'wall' ) {
            $ret['wall'] = microtime( true );
        }
        if ( !$clock || $clock === 'cpu' ) {
            $ru = getrusage( 0 /* RUSAGE_SELF */ );
            $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 hasReducedExpiry() : bool {
        global $wgParserCacheExpireTime;
 
        return $this->getCacheExpiry() < $wgParserCacheExpireTime;
    }
 
    public function hasDynamicContent() {
        return $this->hasReducedExpiry();
    }
 
    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 addExtraCSPDefaultSrc( $src ) {
        $this->mExtraDefaultSrcs[] = $src;
    }
 
    public function addExtraCSPStyleSrc( $src ) {
        $this->mExtraStyleSrcs[] = $src;
    }
 
    public function addExtraCSPScriptSrc( $src ) {
        $this->mExtraScriptSrcs[] = $src;
    }
 
    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_filter( array_keys( get_object_vars( $this ) ),
            static function ( $field ) {
                if ( $field === 'mParseStartTime' ) {
                    return false;
                } elseif ( strpos( $field, "\0" ) !== false ) {
                    // Unserializing unknown private fields in HHVM causes
                    // member variables with nulls in their names (T229366)
                    return false;
                } else {
                    return true;
                }
            }
        );
    }
 
    public function mergeInternalMetaDataFrom( ParserOutput $source ) {
        $this->mOutputHooks = self::mergeList( $this->mOutputHooks, $source->getOutputHooks() );
        $this->mWarnings = self::mergeMap( $this->mWarnings, $source->mWarnings ); // don't use getter
        $this->mTimestamp = $this->useMaxValue( $this->mTimestamp, $source->getTimestamp() );
 
        foreach ( self::SPECULATIVE_FIELDS as $field ) {
            if ( $this->$field && $source->$field && $this->$field !== $source->$field ) {
                wfLogWarning( __METHOD__ . ": inconsistent '$field' properties!" );
            }
            $this->$field = $this->useMaxValue( $this->$field, $source->$field );
        }
 
        $this->mParseStartTime = $this->useEachMinValue(
            $this->mParseStartTime,
            $source->mParseStartTime
        );
 
        $this->mFlags = self::mergeMap( $this->mFlags, $source->mFlags );
        $this->mParseUsedOptions = self::mergeMap( $this->mParseUsedOptions, $source->mParseUsedOptions );
 
        // TODO: maintain per-slot limit reports!
        if ( empty( $this->mLimitReportData ) ) {
            $this->mLimitReportData = $source->mLimitReportData;
        }
        if ( empty( $this->mLimitReportJSData ) ) {
            $this->mLimitReportJSData = $source->mLimitReportJSData;
        }
    }
 
    public function mergeHtmlMetaDataFrom( ParserOutput $source ) {
        // HTML and HTTP
        $this->mHeadItems = self::mergeMixedList( $this->mHeadItems, $source->getHeadItems() );
        $this->mModules = self::mergeList( $this->mModules, $source->getModules() );
        $this->mModuleStyles = self::mergeList( $this->mModuleStyles, $source->getModuleStyles() );
        $this->mJsConfigVars = self::mergeMap( $this->mJsConfigVars, $source->getJsConfigVars() );
        $this->mMaxAdaptiveExpiry = min( $this->mMaxAdaptiveExpiry, $source->mMaxAdaptiveExpiry );
        $this->mExtraStyleSrcs = self::mergeList(
            $this->mExtraStyleSrcs,
            $source->getExtraCSPStyleSrcs()
        );
        $this->mExtraScriptSrcs = self::mergeList(
            $this->mExtraScriptSrcs,
            $source->getExtraCSPScriptSrcs()
        );
        $this->mExtraDefaultSrcs = self::mergeList(
            $this->mExtraDefaultSrcs,
            $source->getExtraCSPDefaultSrcs()
        );
 
        // "noindex" always wins!
        if ( $this->mIndexPolicy === 'noindex' || $source->mIndexPolicy === 'noindex' ) {
            $this->mIndexPolicy = 'noindex';
        } elseif ( $this->mIndexPolicy !== 'index' ) {
            $this->mIndexPolicy = $source->mIndexPolicy;
        }
 
        // Skin control
        $this->mNewSection = $this->mNewSection || $source->getNewSection();
        $this->mHideNewSection = $this->mHideNewSection || $source->getHideNewSection();
        $this->mNoGallery = $this->mNoGallery || $source->getNoGallery();
        $this->mEnableOOUI = $this->mEnableOOUI || $source->getEnableOOUI();
        $this->mPreventClickjacking = $this->mPreventClickjacking || $source->preventClickjacking();
 
        // TODO: we'll have to be smarter about this!
        $this->mSections = array_merge( $this->mSections, $source->getSections() );
        $this->mTOCHTML .= $source->mTOCHTML;
 
        // XXX: we don't want to concatenate title text, so first write wins.
        // We should use the first *modified* title text, but we don't have the original to check.
        if ( $this->mTitleText === null || $this->mTitleText === '' ) {
            $this->mTitleText = $source->mTitleText;
        }
 
        // class names are stored in array keys
        $this->mWrapperDivClasses = self::mergeMap(
            $this->mWrapperDivClasses,
            $source->mWrapperDivClasses
        );
 
        // NOTE: last write wins, same as within one ParserOutput
        $this->mIndicators = self::mergeMap( $this->mIndicators, $source->getIndicators() );
 
        // NOTE: include extension data in "tracking meta data" as well as "html meta data"!
        // TODO: add a $mergeStrategy parameter to setExtensionData to allow different
        // kinds of extension data to be merged in different ways.
        $this->mExtensionData = self::mergeMap(
            $this->mExtensionData,
            $source->mExtensionData
        );
    }
 
    public function mergeTrackingMetaDataFrom( ParserOutput $source ) {
        $this->mLanguageLinks = self::mergeList( $this->mLanguageLinks, $source->getLanguageLinks() );
        $this->mCategories = self::mergeMap( $this->mCategories, $source->getCategories() );
        $this->mLinks = self::merge2D( $this->mLinks, $source->getLinks() );
        $this->mTemplates = self::merge2D( $this->mTemplates, $source->getTemplates() );
        $this->mTemplateIds = self::merge2D( $this->mTemplateIds, $source->getTemplateIds() );
        $this->mImages = self::mergeMap( $this->mImages, $source->getImages() );
        $this->mFileSearchOptions = self::mergeMap(
            $this->mFileSearchOptions,
            $source->getFileSearchOptions()
        );
        $this->mExternalLinks = self::mergeMap( $this->mExternalLinks, $source->getExternalLinks() );
        $this->mInterwikiLinks = self::merge2D(
            $this->mInterwikiLinks,
            $source->getInterwikiLinks()
        );
 
        // TODO: add a $mergeStrategy parameter to setProperty to allow different
        // kinds of properties to be merged in different ways.
        $this->mProperties = self::mergeMap( $this->mProperties, $source->getProperties() );
 
        // NOTE: include extension data in "tracking meta data" as well as "html meta data"!
        // TODO: add a $mergeStrategy parameter to setExtensionData to allow different
        // kinds of extension data to be merged in different ways.
        $this->mExtensionData = self::mergeMap(
            $this->mExtensionData,
            $source->mExtensionData
        );
    }
 
    private static function mergeMixedList( array $a, array $b ) {
        return array_unique( array_merge( $a, $b ), SORT_REGULAR );
    }
 
    private static function mergeList( array $a, array $b ) {
        return array_values( array_unique( array_merge( $a, $b ), SORT_REGULAR ) );
    }
 
    private static function mergeMap( array $a, array $b ) {
        return array_replace( $a, $b );
    }
 
    private static function merge2D( array $a, array $b ) {
        $values = [];
        $keys = array_merge( array_keys( $a ), array_keys( $b ) );
 
        foreach ( $keys as $k ) {
            if ( empty( $a[$k] ) ) {
                $values[$k] = $b[$k];
            } elseif ( empty( $b[$k] ) ) {
                $values[$k] = $a[$k];
            } elseif ( is_array( $a[$k] ) && is_array( $b[$k] ) ) {
                $values[$k] = array_replace( $a[$k], $b[$k] );
            } else {
                $values[$k] = $b[$k];
            }
        }
 
        return $values;
    }
 
    private static function useEachMinValue( array $a, array $b ) {
        $values = [];
        $keys = array_merge( array_keys( $a ), array_keys( $b ) );
 
        foreach ( $keys as $k ) {
            if ( is_array( $a[$k] ?? null ) && is_array( $b[$k] ?? null ) ) {
                $values[$k] = self::useEachMinValue( $a[$k], $b[$k] );
            } else {
                $values[$k] = self::useMinValue( $a[$k] ?? null, $b[$k] ?? null );
            }
        }
 
        return $values;
    }
 
    private static function useMinValue( $a, $b ) {
        if ( $a === null ) {
            return $b;
        }
 
        if ( $b === null ) {
            return $a;
        }
 
        return min( $a, $b );
    }
 
    private static function useMaxValue( $a, $b ) {
        if ( $a === null ) {
            return $b;
        }
 
        if ( $b === null ) {
            return $a;
        }
 
        return max( $a, $b );
    }
 
    protected function toJsonArray(): array {
        $data = [
            'Text' => $this->mText,
            'LanguageLinks' => $this->mLanguageLinks,
            'Categories' => $this->mCategories,
            'Indicators' => $this->mIndicators,
            'TitleText' => $this->mTitleText,
            'Links' => $this->mLinks,
            'LinksSpecial' => $this->mLinksSpecial,
            'Templates' => $this->mTemplates,
            'TemplateIds' => $this->mTemplateIds,
            'Images' => $this->mImages,
            'FileSearchOptions' => $this->mFileSearchOptions,
            'ExternalLinks' => $this->mExternalLinks,
            'InterwikiLinks' => $this->mInterwikiLinks,
            'NewSection' => $this->mNewSection,
            'HideNewSection' => $this->mHideNewSection,
            'NoGallery' => $this->mNoGallery,
            'HeadItems' => $this->mHeadItems,
            'Modules' => $this->mModules,
            'ModuleStyles' => $this->mModuleStyles,
            'JsConfigVars' => $this->mJsConfigVars,
            'OutputHooks' => $this->mOutputHooks,
            'Warnings' => $this->mWarnings,
            'Sections' => $this->mSections,
            'Properties' => self::detectAndEncodeBinary( $this->mProperties ),
            'TOCHTML' => $this->mTOCHTML,
            'Timestamp' => $this->mTimestamp,
            'EnableOOUI' => $this->mEnableOOUI,
            'IndexPolicy' => $this->mIndexPolicy,
            // may contain arbitrary structures!
            'ExtensionData' => $this->mExtensionData,
            'LimitReportData' => $this->mLimitReportData,
            'LimitReportJSData' => $this->mLimitReportJSData,
            'ParseStartTime' => $this->mParseStartTime,
            'PreventClickjacking' => $this->mPreventClickjacking,
            'ExtraScriptSrcs' => $this->mExtraScriptSrcs,
            'ExtraDefaultSrcs' => $this->mExtraDefaultSrcs,
            'ExtraStyleSrcs' => $this->mExtraStyleSrcs,
            'Flags' => $this->mFlags,
            'SpeculativeRevId' => $this->mSpeculativeRevId,
            'SpeculativePageIdUsed' => $this->speculativePageIdUsed,
            'RevisionTimestampUsed' => $this->revisionTimestampUsed,
            'RevisionUsedSha1Base36' => $this->revisionUsedSha1Base36,
            'WrapperDivClasses' => $this->mWrapperDivClasses,
        ];
 
        // Fill in missing fields from parents. Array addition does not override existing fields.
        $data += parent::toJsonArray();
 
        // TODO: make more fields optional!
 
        if ( $this->mMaxAdaptiveExpiry !== INF ) {
            // NOTE: JSON can't encode infinity!
            $data['MaxAdaptiveExpiry'] = $this->mMaxAdaptiveExpiry;
        }
 
        return $data;
    }
 
    public static function newFromJsonArray( JsonUnserializer $unserializer, array $json ) {
        $parserOutput = new ParserOutput();
        $parserOutput->initFromJson( $unserializer, $json );
        return $parserOutput;
    }
 
    protected function initFromJson( JsonUnserializer $unserializer, array $jsonData ) {
        parent::initFromJson( $unserializer, $jsonData );
 
        $this->mText = $jsonData['Text'];
        $this->mLanguageLinks = $jsonData['LanguageLinks'];
        $this->mCategories = $jsonData['Categories'];
        $this->mIndicators = $jsonData['Indicators'];
        $this->mTitleText = $jsonData['TitleText'];
        $this->mLinks = $jsonData['Links'];
        $this->mLinksSpecial = $jsonData['LinksSpecial'];
        $this->mTemplates = $jsonData['Templates'];
        $this->mTemplateIds = $jsonData['TemplateIds'];
        $this->mImages = $jsonData['Images'];
        $this->mFileSearchOptions = $jsonData['FileSearchOptions'];
        $this->mExternalLinks = $jsonData['ExternalLinks'];
        $this->mInterwikiLinks = $jsonData['InterwikiLinks'];
        $this->mNewSection = $jsonData['NewSection'];
        $this->mHideNewSection = $jsonData['HideNewSection'];
        $this->mNoGallery = $jsonData['NoGallery'];
        $this->mHeadItems = $jsonData['HeadItems'];
        $this->mModules = $jsonData['Modules'];
        $this->mModuleStyles = $jsonData['ModuleStyles'];
        $this->mJsConfigVars = $jsonData['JsConfigVars'];
        $this->mOutputHooks = $jsonData['OutputHooks'];
        $this->mWarnings = $jsonData['Warnings'];
        $this->mSections = $jsonData['Sections'];
        $this->mProperties = self::detectAndDecodeBinary( $jsonData['Properties'] );
        $this->mTOCHTML = $jsonData['TOCHTML'];
        $this->mTimestamp = $jsonData['Timestamp'];
        $this->mEnableOOUI = $jsonData['EnableOOUI'];
        $this->mIndexPolicy = $jsonData['IndexPolicy'];
        $this->mExtensionData = $unserializer->unserializeArray( $jsonData['ExtensionData'] ?? [] );
        $this->mExtensionData = $jsonData['ExtensionData'];
        $this->mLimitReportData = $jsonData['LimitReportData'];
        $this->mLimitReportJSData = $jsonData['LimitReportJSData'];
        $this->mParseStartTime = $jsonData['ParseStartTime'];
        $this->mPreventClickjacking = $jsonData['PreventClickjacking'];
        $this->mExtraScriptSrcs = $jsonData['ExtraScriptSrcs'];
        $this->mExtraDefaultSrcs = $jsonData['ExtraDefaultSrcs'];
        $this->mExtraStyleSrcs = $jsonData['ExtraStyleSrcs'];
        $this->mFlags = $jsonData['Flags'];
        $this->mSpeculativeRevId = $jsonData['SpeculativeRevId'];
        $this->speculativePageIdUsed = $jsonData['SpeculativePageIdUsed'];
        $this->revisionTimestampUsed = $jsonData['RevisionTimestampUsed'];
        $this->revisionUsedSha1Base36 = $jsonData['RevisionUsedSha1Base36'];
        $this->mWrapperDivClasses = $jsonData['WrapperDivClasses'];
        $this->mMaxAdaptiveExpiry = $jsonData['MaxAdaptiveExpiry'] ?? INF;
    }
 
    private static function detectAndEncodeBinary( array $properties ) {
        foreach ( $properties as $key => $value ) {
            if ( is_string( $value ) ) {
                if ( !mb_detect_encoding( $value, 'UTF-8', true ) ) {
                    $properties[$key] = [
                        '_type_' => 'string',
                        '_encoding_' => 'base64',
                        '_data_' => base64_encode( $value ),
                    ];
                }
            }
        }
 
        return $properties;
    }
 
    private static function detectAndDecodeBinary( array $properties ) {
        foreach ( $properties as $key => $value ) {
            if ( is_array( $value ) && isset( $value['_encoding_'] ) ) {
                if ( $value['_encoding_'] === 'base64' ) {
                    $properties[$key] = base64_decode( $value['_data_'] );
                }
            }
        }
 
        return $properties;
    }
 
    public function __wakeup() {
        // Backwards compatibility, pre 1.36
        $priorAccessedOptions = $this->getGhostFieldValue( 'mAccessedOptions' );
        if ( $priorAccessedOptions ) {
            $this->mParseUsedOptions = $priorAccessedOptions;
        }
    }
}