Parser.php

Go to the documentation of this file.

<?php
class Parser {
    const VERSION = '1.6.4';

    const HALF_PARSED_VERSION = 2;

    # Flags for Parser::setFunctionHook
    # Also available as global constants from Defines.php
    const SFH_NO_HASH = 1;
    const SFH_OBJECT_ARGS = 2;

    # Constants needed for external link processing
    # Everything except bracket, space, or control characters
    # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
    # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
    const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
    const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
        \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';

    # State constants for the definition list colon extraction
    const COLON_STATE_TEXT = 0;
    const COLON_STATE_TAG = 1;
    const COLON_STATE_TAGSTART = 2;
    const COLON_STATE_CLOSETAG = 3;
    const COLON_STATE_TAGSLASH = 4;
    const COLON_STATE_COMMENT = 5;
    const COLON_STATE_COMMENTDASH = 6;
    const COLON_STATE_COMMENTDASHDASH = 7;

    # Flags for preprocessToDom
    const PTD_FOR_INCLUSION = 1;

    # Allowed values for $this->mOutputType
    # Parameter to startExternalParse().
    const OT_HTML = 1; # like parse()
    const OT_WIKI = 2; # like preSaveTransform()
    const OT_PREPROCESS = 3; # like preprocess()
    const OT_MSG = 3;
    const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.

    # Marker Suffix needs to be accessible staticly.
    const MARKER_SUFFIX = "-QINU\x7f";

    # Markers used for wrapping the table of contents
    const TOC_START = '<mw:toc>';
    const TOC_END = '</mw:toc>';

    # Persistent:
    public $mTagHooks = array();
    public $mTransparentTagHooks = array();
    public $mFunctionHooks = array();
    public $mFunctionSynonyms = array( 0 => array(), 1 => array() );
    public $mFunctionTagHooks = array();
    public $mStripList = array();
    public $mDefaultStripList = array();
    public $mVarCache = array();
    public $mImageParams = array();
    public $mImageParamsMagicArray = array();
    public $mMarkerIndex = 0;
    public $mFirstCall = true;

    # Initialised by initialiseVariables()

    public $mVariables;

    public $mSubstWords;
    public $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor

    # Cleared with clearState():

    public $mOutput;
    public $mAutonumber, $mDTopen;

    public $mStripState;

    public $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
    public $mLinkHolders;

    public $mLinkID;
    public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
    public $mDefaultSort;
    public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
    public $mExpensiveFunctionCount; # number of expensive parser function calls
    public $mShowToc, $mForceTocPosition;

    public $mUser; # User object; only used when doing pre-save transform

    # Temporary
    # These are variables reset at least once per parse regardless of $clearState

    public $mOptions;

    public $mTitle;        # Title context, used for self-link rendering and similar things
    public $mOutputType;   # Output type, one of the OT_xxx constants
    public $ot;            # Shortcut alias, see setOutputType()
    public $mRevisionObject; # The revision object of the specified revision ID
    public $mRevisionId;   # ID to display in {{REVISIONID}} tags
    public $mRevisionTimestamp; # The timestamp of the specified revision ID
    public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
    public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
    public $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
    public $mInputSize = false; # For {{PAGESIZE}} on current page.

    public $mUniqPrefix;

    public $mLangLinkLanguages;

    public $mInParse = false;

    public function __construct( $conf = array() ) {
        $this->mConf = $conf;
        $this->mUrlProtocols = wfUrlProtocols();
        $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
            self::EXT_LINK_URL_CLASS . '+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
        if ( isset( $conf['preprocessorClass'] ) ) {
            $this->mPreprocessorClass = $conf['preprocessorClass'];
        } elseif ( defined( 'HPHP_VERSION' ) ) {
            # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
            $this->mPreprocessorClass = 'Preprocessor_Hash';
        } elseif ( extension_loaded( 'domxml' ) ) {
            # PECL extension that conflicts with the core DOM extension (bug 13770)
            wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
            $this->mPreprocessorClass = 'Preprocessor_Hash';
        } elseif ( extension_loaded( 'dom' ) ) {
            $this->mPreprocessorClass = 'Preprocessor_DOM';
        } else {
            $this->mPreprocessorClass = 'Preprocessor_Hash';
        }
        wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
    }

    public function __destruct() {
        if ( isset( $this->mLinkHolders ) ) {
            unset( $this->mLinkHolders );
        }
        foreach ( $this as $name => $value ) {
            unset( $this->$name );
        }
    }

    public function __clone() {
        $this->mInParse = false;
        wfRunHooks( 'ParserCloned', array( $this ) );
    }

    public function firstCallInit() {
        if ( !$this->mFirstCall ) {
            return;
        }
        $this->mFirstCall = false;

        wfProfileIn( __METHOD__ );

        CoreParserFunctions::register( $this );
        CoreTagHooks::register( $this );
        $this->initialiseVariables();

        wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
        wfProfileOut( __METHOD__ );
    }

    public function clearState() {
        wfProfileIn( __METHOD__ );
        if ( $this->mFirstCall ) {
            $this->firstCallInit();
        }
        $this->mOutput = new ParserOutput;
        $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
        $this->mAutonumber = 0;
        $this->mLastSection = '';
        $this->mDTopen = false;
        $this->mIncludeCount = array();
        $this->mArgStack = false;
        $this->mInPre = false;
        $this->mLinkHolders = new LinkHolderArray( $this );
        $this->mLinkID = 0;
        $this->mRevisionObject = $this->mRevisionTimestamp =
            $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
        $this->mVarCache = array();
        $this->mUser = null;
        $this->mLangLinkLanguages = array();

        $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
        $this->mStripState = new StripState( $this->mUniqPrefix );

        # Clear these on every parse, bug 4549
        $this->mTplRedirCache = $this->mTplDomCache = array();

        $this->mShowToc = true;
        $this->mForceTocPosition = false;
        $this->mIncludeSizes = array(
            'post-expand' => 0,
            'arg' => 0,
        );
        $this->mPPNodeCount = 0;
        $this->mGeneratedPPNodeCount = 0;
        $this->mHighestExpansionDepth = 0;
        $this->mDefaultSort = false;
        $this->mHeadings = array();
        $this->mDoubleUnderscores = array();
        $this->mExpensiveFunctionCount = 0;

        # Fix cloning
        if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
            $this->mPreprocessor = null;
        }

        wfRunHooks( 'ParserClearState', array( &$this ) );
        wfProfileOut( __METHOD__ );
    }

    public function parse( $text, Title $title, ParserOptions $options,
        $linestart = true, $clearState = true, $revid = null
    ) {
        global $wgUseTidy, $wgAlwaysUseTidy, $wgShowHostnames;
        $fname = __METHOD__ . '-' . wfGetCaller();
        wfProfileIn( __METHOD__ );
        wfProfileIn( $fname );

        if ( $clearState ) {
            $magicScopeVariable = $this->lock();
        }

        $this->startParse( $title, $options, self::OT_HTML, $clearState );

        $this->mInputSize = strlen( $text );
        if ( $this->mOptions->getEnableLimitReport() ) {
            $this->mOutput->resetParseStartTime();
        }

        # Remove the strip marker tag prefix from the input, if present.
        if ( $clearState ) {
            $text = str_replace( $this->mUniqPrefix, '', $text );
        }

        $oldRevisionId = $this->mRevisionId;
        $oldRevisionObject = $this->mRevisionObject;
        $oldRevisionTimestamp = $this->mRevisionTimestamp;
        $oldRevisionUser = $this->mRevisionUser;
        $oldRevisionSize = $this->mRevisionSize;
        if ( $revid !== null ) {
            $this->mRevisionId = $revid;
            $this->mRevisionObject = null;
            $this->mRevisionTimestamp = null;
            $this->mRevisionUser = null;
            $this->mRevisionSize = null;
        }

        wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
        # No more strip!
        wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
        $text = $this->internalParse( $text );
        wfRunHooks( 'ParserAfterParse', array( &$this, &$text, &$this->mStripState ) );

        $text = $this->mStripState->unstripGeneral( $text );

        # Clean up special characters, only run once, next-to-last before doBlockLevels
        $fixtags = array(
            # french spaces, last one Guillemet-left
            # only if there is something before the space
            '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
            # french spaces, Guillemet-right
            '/(\\302\\253) /' => '\\1&#160;',
            '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
        );
        $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );

        $text = $this->doBlockLevels( $text, $linestart );

        $this->replaceLinkHolders( $text );

        if ( !( $options->getDisableContentConversion()
            || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
        ) {
            if ( !$this->mOptions->getInterfaceMessage() ) {
                # The position of the convert() call should not be changed. it
                # assumes that the links are all replaced and the only thing left
                # is the <nowiki> mark.
                $text = $this->getConverterLanguage()->convert( $text );
            }
        }

        if ( !( $options->getDisableTitleConversion()
            || isset( $this->mDoubleUnderscores['nocontentconvert'] )
            || isset( $this->mDoubleUnderscores['notitleconvert'] )
            || $this->mOutput->getDisplayTitle() !== false )
        ) {
            $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
            if ( $convruletitle ) {
                $this->mOutput->setTitleText( $convruletitle );
            } else {
                $titleText = $this->getConverterLanguage()->convertTitle( $title );
                $this->mOutput->setTitleText( $titleText );
            }
        }

        $text = $this->mStripState->unstripNoWiki( $text );

        wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );

        $text = $this->replaceTransparentTags( $text );
        $text = $this->mStripState->unstripGeneral( $text );

        $text = Sanitizer::normalizeCharReferences( $text );

        if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
            $text = MWTidy::tidy( $text );
        } else {
            # attempt to sanitize at least some nesting problems
            # (bug #2702 and quite a few others)
            $tidyregs = array(
                # ''Something [http://www.cool.com cool''] -->
                # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
                '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
                '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
                # fix up an anchor inside another anchor, only
                # at least for a single single nested link (bug 3695)
                '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
                '\\1\\2</a>\\3</a>\\1\\4</a>',
                # fix div inside inline elements- doBlockLevels won't wrap a line which
                # contains a div, so fix it up here; replace
                # div with escaped text
                '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
                '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
                # remove empty italic or bold tag pairs, some
                # introduced by rules above
                '/<([bi])><\/\\1>/' => '',
            );

            $text = preg_replace(
                array_keys( $tidyregs ),
                array_values( $tidyregs ),
                $text );
        }

        if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
            $this->limitationWarn( 'expensive-parserfunction',
                $this->mExpensiveFunctionCount,
                $this->mOptions->getExpensiveParserFunctionLimit()
            );
        }

        wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );

        # Information on include size limits, for the benefit of users who try to skirt them
        if ( $this->mOptions->getEnableLimitReport() ) {
            $max = $this->mOptions->getMaxIncludeSize();

            $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
            if ( $cpuTime !== null ) {
                $this->mOutput->setLimitReportData( 'limitreport-cputime',
                    sprintf( "%.3f", $cpuTime )
                );
            }

            $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
            $this->mOutput->setLimitReportData( 'limitreport-walltime',
                sprintf( "%.3f", $wallTime )
            );

            $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
                array( $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() )
            );
            $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
                array( $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() )
            );
            $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
                array( $this->mIncludeSizes['post-expand'], $max )
            );
            $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
                array( $this->mIncludeSizes['arg'], $max )
            );
            $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
                array( $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() )
            );
            $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
                array( $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() )
            );
            wfRunHooks( 'ParserLimitReportPrepare', array( $this, $this->mOutput ) );

            $limitReport = "NewPP limit report\n";
            if ( $wgShowHostnames ) {
                $limitReport .= 'Parsed by ' . wfHostname() . "\n";
            }
            foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
                if ( wfRunHooks( 'ParserLimitReportFormat',
                    array( $key, &$value, &$limitReport, false, false )
                ) ) {
                    $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
                    $valueMsg = wfMessage( array( "$key-value-text", "$key-value" ) )
                        ->inLanguage( 'en' )->useDatabase( false );
                    if ( !$valueMsg->exists() ) {
                        $valueMsg = new RawMessage( '$1' );
                    }
                    if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
                        $valueMsg->params( $value );
                        $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
                    }
                }
            }
            // Since we're not really outputting HTML, decode the entities and
            // then re-encode the things that need hiding inside HTML comments.
            $limitReport = htmlspecialchars_decode( $limitReport );
            wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );

            // Sanitize for comment. Note '‐' in the replacement is U+2010,
            // which looks much like the problematic '-'.
            $limitReport = str_replace( array( '-', '&' ), array( '‐', '&amp;' ), $limitReport );
            $text .= "\n<!-- \n$limitReport-->\n";

            if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
                wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
                    $this->mTitle->getPrefixedDBkey() );
            }
        }
        $this->mOutput->setText( $text );

        $this->mRevisionId = $oldRevisionId;
        $this->mRevisionObject = $oldRevisionObject;
        $this->mRevisionTimestamp = $oldRevisionTimestamp;
        $this->mRevisionUser = $oldRevisionUser;
        $this->mRevisionSize = $oldRevisionSize;
        $this->mInputSize = false;
        wfProfileOut( $fname );
        wfProfileOut( __METHOD__ );

        return $this->mOutput;
    }

    public function recursiveTagParse( $text, $frame = false ) {
        wfProfileIn( __METHOD__ );
        wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
        wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
        $text = $this->internalParse( $text, false, $frame );
        wfProfileOut( __METHOD__ );
        return $text;
    }

    public function preprocess( $text, Title $title = null, ParserOptions $options, $revid = null, $frame = false ) {
        wfProfileIn( __METHOD__ );
        $magicScopeVariable = $this->lock();
        $this->startParse( $title, $options, self::OT_PREPROCESS, true );
        if ( $revid !== null ) {
            $this->mRevisionId = $revid;
        }
        wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
        wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
        $text = $this->replaceVariables( $text, $frame );
        $text = $this->mStripState->unstripBoth( $text );
        wfProfileOut( __METHOD__ );
        return $text;
    }

    public function recursivePreprocess( $text, $frame = false ) {
        wfProfileIn( __METHOD__ );
        $text = $this->replaceVariables( $text, $frame );
        $text = $this->mStripState->unstripBoth( $text );
        wfProfileOut( __METHOD__ );
        return $text;
    }

    public function getPreloadText( $text, Title $title, ParserOptions $options, $params = array() ) {
        $msg = new RawMessage( $text );
        $text = $msg->params( $params )->plain();

        # Parser (re)initialisation
        $magicScopeVariable = $this->lock();
        $this->startParse( $title, $options, self::OT_PLAIN, true );

        $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
        $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
        $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
        $text = $this->mStripState->unstripBoth( $text );
        return $text;
    }

    public static function getRandomString() {
        return wfRandomString( 16 );
    }

    public function setUser( $user ) {
        $this->mUser = $user;
    }

    public function uniqPrefix() {
        if ( !isset( $this->mUniqPrefix ) ) {
            # @todo FIXME: This is probably *horribly wrong*
            # LanguageConverter seems to want $wgParser's uniqPrefix, however
            # if this is called for a parser cache hit, the parser may not
            # have ever been initialized in the first place.
            # Not really sure what the heck is supposed to be going on here.
            return '';
            # throw new MWException( "Accessing uninitialized mUniqPrefix" );
        }
        return $this->mUniqPrefix;
    }

    public function setTitle( $t ) {
        if ( !$t ) {
            $t = Title::newFromText( 'NO TITLE' );
        }

        if ( $t->hasFragment() ) {
            # Strip the fragment to avoid various odd effects
            $this->mTitle = clone $t;
            $this->mTitle->setFragment( '' );
        } else {
            $this->mTitle = $t;
        }
    }

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

    public function Title( $x = null ) {
        return wfSetVar( $this->mTitle, $x );
    }

    public function setOutputType( $ot ) {
        $this->mOutputType = $ot;
        # Shortcut alias
        $this->ot = array(
            'html' => $ot == self::OT_HTML,
            'wiki' => $ot == self::OT_WIKI,
            'pre' => $ot == self::OT_PREPROCESS,
            'plain' => $ot == self::OT_PLAIN,
        );
    }

    public function OutputType( $x = null ) {
        return wfSetVar( $this->mOutputType, $x );
    }

    public function getOutput() {
        return $this->mOutput;
    }

    public function getOptions() {
        return $this->mOptions;
    }

    public function Options( $x = null ) {
        return wfSetVar( $this->mOptions, $x );
    }

    public function nextLinkID() {
        return $this->mLinkID++;
    }

    public function setLinkID( $id ) {
        $this->mLinkID = $id;
    }

    public function getFunctionLang() {
        return $this->getTargetLanguage();
    }

    public function getTargetLanguage() {
        $target = $this->mOptions->getTargetLanguage();

        if ( $target !== null ) {
            return $target;
        } elseif ( $this->mOptions->getInterfaceMessage() ) {
            return $this->mOptions->getUserLangObj();
        } elseif ( is_null( $this->mTitle ) ) {
            throw new MWException( __METHOD__ . ': $this->mTitle is null' );
        }

        return $this->mTitle->getPageLanguage();
    }

    public function getConverterLanguage() {
        return $this->getTargetLanguage();
    }

    public function getUser() {
        if ( !is_null( $this->mUser ) ) {
            return $this->mUser;
        }
        return $this->mOptions->getUser();
    }

    public function getPreprocessor() {
        if ( !isset( $this->mPreprocessor ) ) {
            $class = $this->mPreprocessorClass;
            $this->mPreprocessor = new $class( $this );
        }
        return $this->mPreprocessor;
    }

    public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
        static $n = 1;
        $stripped = '';
        $matches = array();

        $taglist = implode( '|', $elements );
        $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";

        while ( $text != '' ) {
            $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
            $stripped .= $p[0];
            if ( count( $p ) < 5 ) {
                break;
            }
            if ( count( $p ) > 5 ) {
                # comment
                $element = $p[4];
                $attributes = '';
                $close = '';
                $inside = $p[5];
            } else {
                # tag
                $element = $p[1];
                $attributes = $p[2];
                $close = $p[3];
                $inside = $p[4];
            }

            $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
            $stripped .= $marker;

            if ( $close === '/>' ) {
                # Empty element tag, <tag />
                $content = null;
                $text = $inside;
                $tail = null;
            } else {
                if ( $element === '!--' ) {
                    $end = '/(-->)/';
                } else {
                    $end = "/(<\\/$element\\s*>)/i";
                }
                $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
                $content = $q[0];
                if ( count( $q ) < 3 ) {
                    # No end tag -- let it run out to the end of the text.
                    $tail = '';
                    $text = '';
                } else {
                    $tail = $q[1];
                    $text = $q[2];
                }
            }

            $matches[$marker] = array( $element,
                $content,
                Sanitizer::decodeTagAttributes( $attributes ),
                "<$element$attributes$close$content$tail" );
        }
        return $stripped;
    }

    public function getStripList() {
        return $this->mStripList;
    }

    public function insertStripItem( $text ) {
        $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
        $this->mMarkerIndex++;
        $this->mStripState->addGeneral( $rnd, $text );
        return $rnd;
    }

    public function doTableStuff( $text ) {
        wfProfileIn( __METHOD__ );

        $lines = StringUtils::explode( "\n", $text );
        $out = '';
        $td_history = array(); # Is currently a td tag open?
        $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
        $tr_history = array(); # Is currently a tr tag open?
        $tr_attributes = array(); # history of tr attributes
        $has_opened_tr = array(); # Did this table open a <tr> element?
        $indent_level = 0; # indent level of the table

        foreach ( $lines as $outLine ) {
            $line = trim( $outLine );

            if ( $line === '' ) { # empty line, go to next line
                $out .= $outLine . "\n";
                continue;
            }

            $first_character = $line[0];
            $matches = array();

            if ( preg_match( '/^(:*)\{\|(.*)$/', $line, $matches ) ) {
                # First check if we are starting a new table
                $indent_level = strlen( $matches[1] );

                $attributes = $this->mStripState->unstripBoth( $matches[2] );
                $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );

                $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
                array_push( $td_history, false );
                array_push( $last_tag_history, '' );
                array_push( $tr_history, false );
                array_push( $tr_attributes, '' );
                array_push( $has_opened_tr, false );
            } elseif ( count( $td_history ) == 0 ) {
                # Don't do any of the following
                $out .= $outLine . "\n";
                continue;
            } elseif ( substr( $line, 0, 2 ) === '|}' ) {
                # We are ending a table
                $line = '</table>' . substr( $line, 2 );
                $last_tag = array_pop( $last_tag_history );

                if ( !array_pop( $has_opened_tr ) ) {
                    $line = "<tr><td></td></tr>{$line}";
                }

                if ( array_pop( $tr_history ) ) {
                    $line = "</tr>{$line}";
                }

                if ( array_pop( $td_history ) ) {
                    $line = "</{$last_tag}>{$line}";
                }
                array_pop( $tr_attributes );
                $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
            } elseif ( substr( $line, 0, 2 ) === '|-' ) {
                # Now we have a table row
                $line = preg_replace( '#^\|-+#', '', $line );

                # Whats after the tag is now only attributes
                $attributes = $this->mStripState->unstripBoth( $line );
                $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
                array_pop( $tr_attributes );
                array_push( $tr_attributes, $attributes );

                $line = '';
                $last_tag = array_pop( $last_tag_history );
                array_pop( $has_opened_tr );
                array_push( $has_opened_tr, true );

                if ( array_pop( $tr_history ) ) {
                    $line = '</tr>';
                }

                if ( array_pop( $td_history ) ) {
                    $line = "</{$last_tag}>{$line}";
                }

                $outLine = $line;
                array_push( $tr_history, false );
                array_push( $td_history, false );
                array_push( $last_tag_history, '' );
            } elseif ( $first_character === '|'
                || $first_character === '!'
                || substr( $line, 0, 2 ) === '|+'
            ) {
                # This might be cell elements, td, th or captions
                if ( substr( $line, 0, 2 ) === '|+' ) {
                    $first_character = '+';
                    $line = substr( $line, 1 );
                }

                $line = substr( $line, 1 );

                if ( $first_character === '!' ) {
                    $line = str_replace( '!!', '||', $line );
                }

                # Split up multiple cells on the same line.
                # FIXME : This can result in improper nesting of tags processed
                # by earlier parser steps, but should avoid splitting up eg
                # attribute values containing literal "||".
                $cells = StringUtils::explodeMarkup( '||', $line );

                $outLine = '';

                # Loop through each table cell
                foreach ( $cells as $cell ) {
                    $previous = '';
                    if ( $first_character !== '+' ) {
                        $tr_after = array_pop( $tr_attributes );
                        if ( !array_pop( $tr_history ) ) {
                            $previous = "<tr{$tr_after}>\n";
                        }
                        array_push( $tr_history, true );
                        array_push( $tr_attributes, '' );
                        array_pop( $has_opened_tr );
                        array_push( $has_opened_tr, true );
                    }

                    $last_tag = array_pop( $last_tag_history );

                    if ( array_pop( $td_history ) ) {
                        $previous = "</{$last_tag}>\n{$previous}";
                    }

                    if ( $first_character === '|' ) {
                        $last_tag = 'td';
                    } elseif ( $first_character === '!' ) {
                        $last_tag = 'th';
                    } elseif ( $first_character === '+' ) {
                        $last_tag = 'caption';
                    } else {
                        $last_tag = '';
                    }

                    array_push( $last_tag_history, $last_tag );

                    # A cell could contain both parameters and data
                    $cell_data = explode( '|', $cell, 2 );

                    # Bug 553: Note that a '|' inside an invalid link should not
                    # be mistaken as delimiting cell parameters
                    if ( strpos( $cell_data[0], '[[' ) !== false ) {
                        $cell = "{$previous}<{$last_tag}>{$cell}";
                    } elseif ( count( $cell_data ) == 1 ) {
                        $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
                    } else {
                        $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
                        $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
                        $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
                    }

                    $outLine .= $cell;
                    array_push( $td_history, true );
                }
            }
            $out .= $outLine . "\n";
        }

        # Closing open td, tr && table
        while ( count( $td_history ) > 0 ) {
            if ( array_pop( $td_history ) ) {
                $out .= "</td>\n";
            }
            if ( array_pop( $tr_history ) ) {
                $out .= "</tr>\n";
            }
            if ( !array_pop( $has_opened_tr ) ) {
                $out .= "<tr><td></td></tr>\n";
            }

            $out .= "</table>\n";
        }

        # Remove trailing line-ending (b/c)
        if ( substr( $out, -1 ) === "\n" ) {
            $out = substr( $out, 0, -1 );
        }

        # special case: don't return empty table
        if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
            $out = '';
        }

        wfProfileOut( __METHOD__ );

        return $out;
    }

    public function internalParse( $text, $isMain = true, $frame = false ) {
        wfProfileIn( __METHOD__ );

        $origText = $text;

        # Hook to suspend the parser in this state
        if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
            wfProfileOut( __METHOD__ );
            return $text;
        }

        # if $frame is provided, then use $frame for replacing any variables
        if ( $frame ) {
            # use frame depth to infer how include/noinclude tags should be handled
            # depth=0 means this is the top-level document; otherwise it's an included document
            if ( !$frame->depth ) {
                $flag = 0;
            } else {
                $flag = Parser::PTD_FOR_INCLUSION;
            }
            $dom = $this->preprocessToDom( $text, $flag );
            $text = $frame->expand( $dom );
        } else {
            # if $frame is not provided, then use old-style replaceVariables
            $text = $this->replaceVariables( $text );
        }

        wfRunHooks( 'InternalParseBeforeSanitize', array( &$this, &$text, &$this->mStripState ) );
        $text = Sanitizer::removeHTMLtags(
            $text,
            array( &$this, 'attributeStripCallback' ),
            false,
            array_keys( $this->mTransparentTagHooks )
        );
        wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );

        # Tables need to come after variable replacement for things to work
        # properly; putting them before other transformations should keep
        # exciting things like link expansions from showing up in surprising
        # places.
        $text = $this->doTableStuff( $text );

        $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );

        $text = $this->doDoubleUnderscore( $text );

        $text = $this->doHeadings( $text );
        $text = $this->replaceInternalLinks( $text );
        $text = $this->doAllQuotes( $text );
        $text = $this->replaceExternalLinks( $text );

        # replaceInternalLinks may sometimes leave behind
        # absolute URLs, which have to be masked to hide them from replaceExternalLinks
        $text = str_replace( $this->mUniqPrefix . 'NOPARSE', '', $text );

        $text = $this->doMagicLinks( $text );
        $text = $this->formatHeadings( $text, $origText, $isMain );

        wfProfileOut( __METHOD__ );
        return $text;
    }

    public function doMagicLinks( $text ) {
        wfProfileIn( __METHOD__ );
        $prots = wfUrlProtocolsWithoutProtRel();
        $urlChar = self::EXT_LINK_URL_CLASS;
        $text = preg_replace_callback(
            '!(?:                           # Start cases
                (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
                (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
                (\\b(?i:$prots)$urlChar+) |  # m[3]: Free external links" . '
                (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
                ISBN\s+(\b                  # m[5]: ISBN, capture number
                    (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
                    (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
                    [0-9Xx]                 # check digit
                    \b)
            )!xu', array( &$this, 'magicLinkCallback' ), $text );
        wfProfileOut( __METHOD__ );
        return $text;
    }

    public function magicLinkCallback( $m ) {
        if ( isset( $m[1] ) && $m[1] !== '' ) {
            # Skip anchor
            return $m[0];
        } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
            # Skip HTML element
            return $m[0];
        } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
            # Free external link
            return $this->makeFreeExternalLink( $m[0] );
        } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
            # RFC or PMID
            if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
                $keyword = 'RFC';
                $urlmsg = 'rfcurl';
                $cssClass = 'mw-magiclink-rfc';
                $id = $m[4];
            } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
                $keyword = 'PMID';
                $urlmsg = 'pubmedurl';
                $cssClass = 'mw-magiclink-pmid';
                $id = $m[4];
            } else {
                throw new MWException( __METHOD__ . ': unrecognised match type "' .
                    substr( $m[0], 0, 20 ) . '"' );
            }
            $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
            return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass );
        } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
            # ISBN
            $isbn = $m[5];
            $num = strtr( $isbn, array(
                '-' => '',
                ' ' => '',
                'x' => 'X',
            ));
            $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
            return '<a href="' .
                htmlspecialchars( $titleObj->getLocalURL() ) .
                "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
        } else {
            return $m[0];
        }
    }

    public function makeFreeExternalLink( $url ) {
        wfProfileIn( __METHOD__ );

        $trail = '';

        # The characters '<' and '>' (which were escaped by
        # removeHTMLtags()) should not be included in
        # URLs, per RFC 2396.
        $m2 = array();
        if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
            $trail = substr( $url, $m2[0][1] ) . $trail;
            $url = substr( $url, 0, $m2[0][1] );
        }

        # Move trailing punctuation to $trail
        $sep = ',;\.:!?';
        # If there is no left bracket, then consider right brackets fair game too
        if ( strpos( $url, '(' ) === false ) {
            $sep .= ')';
        }

        $numSepChars = strspn( strrev( $url ), $sep );
        if ( $numSepChars ) {
            $trail = substr( $url, -$numSepChars ) . $trail;
            $url = substr( $url, 0, -$numSepChars );
        }

        $url = Sanitizer::cleanUrl( $url );

        # Is this an external image?
        $text = $this->maybeMakeExternalImage( $url );
        if ( $text === false ) {
            # Not an image, make a link
            $text = Linker::makeExternalLink( $url,
                $this->getConverterLanguage()->markNoConversion( $url, true ),
                true, 'free',
                $this->getExternalLinkAttribs( $url ) );
            # Register it in the output object...
            # Replace unnecessary URL escape codes with their equivalent characters
            $pasteurized = self::normalizeLinkUrl( $url );
            $this->mOutput->addExternalLink( $pasteurized );
        }
        wfProfileOut( __METHOD__ );
        return $text . $trail;
    }

    public function doHeadings( $text ) {
        wfProfileIn( __METHOD__ );
        for ( $i = 6; $i >= 1; --$i ) {
            $h = str_repeat( '=', $i );
            $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
        }
        wfProfileOut( __METHOD__ );
        return $text;
    }

    public function doAllQuotes( $text ) {
        wfProfileIn( __METHOD__ );
        $outtext = '';
        $lines = StringUtils::explode( "\n", $text );
        foreach ( $lines as $line ) {
            $outtext .= $this->doQuotes( $line ) . "\n";
        }
        $outtext = substr( $outtext, 0, -1 );
        wfProfileOut( __METHOD__ );
        return $outtext;
    }

    public function doQuotes( $text ) {
        $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
        $countarr = count( $arr );
        if ( $countarr == 1 ) {
            return $text;
        }

        // First, do some preliminary work. This may shift some apostrophes from
        // being mark-up to being text. It also counts the number of occurrences
        // of bold and italics mark-ups.
        $numbold = 0;
        $numitalics = 0;
        for ( $i = 1; $i < $countarr; $i += 2 ) {
            $thislen = strlen( $arr[$i] );
            // If there are ever four apostrophes, assume the first is supposed to
            // be text, and the remaining three constitute mark-up for bold text.
            // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
            if ( $thislen == 4 ) {
                $arr[$i - 1] .= "'";
                $arr[$i] = "'''";
                $thislen = 3;
            } elseif ( $thislen > 5 ) {
                // If there are more than 5 apostrophes in a row, assume they're all
                // text except for the last 5.
                // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
                $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
                $arr[$i] = "'''''";
                $thislen = 5;
            }
            // Count the number of occurrences of bold and italics mark-ups.
            if ( $thislen == 2 ) {
                $numitalics++;
            } elseif ( $thislen == 3 ) {
                $numbold++;
            } elseif ( $thislen == 5 ) {
                $numitalics++;
                $numbold++;
            }
        }

        // If there is an odd number of both bold and italics, it is likely
        // that one of the bold ones was meant to be an apostrophe followed
        // by italics. Which one we cannot know for certain, but it is more
        // likely to be one that has a single-letter word before it.
        if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
            $firstsingleletterword = -1;
            $firstmultiletterword = -1;
            $firstspace = -1;
            for ( $i = 1; $i < $countarr; $i += 2 ) {
                if ( strlen( $arr[$i] ) == 3 ) {
                    $x1 = substr( $arr[$i - 1], -1 );
                    $x2 = substr( $arr[$i - 1], -2, 1 );
                    if ( $x1 === ' ' ) {
                        if ( $firstspace == -1 ) {
                            $firstspace = $i;
                        }
                    } elseif ( $x2 === ' ' ) {
                        if ( $firstsingleletterword == -1 ) {
                            $firstsingleletterword = $i;
                            // if $firstsingleletterword is set, we don't
                            // look at the other options, so we can bail early.
                            break;
                        }
                    } else {
                        if ( $firstmultiletterword == -1 ) {
                            $firstmultiletterword = $i;
                        }
                    }
                }
            }

            // If there is a single-letter word, use it!
            if ( $firstsingleletterword > -1 ) {
                $arr[$firstsingleletterword] = "''";
                $arr[$firstsingleletterword - 1] .= "'";
            } elseif ( $firstmultiletterword > -1 ) {
                // If not, but there's a multi-letter word, use that one.
                $arr[$firstmultiletterword] = "''";
                $arr[$firstmultiletterword - 1] .= "'";
            } elseif ( $firstspace > -1 ) {
                // ... otherwise use the first one that has neither.
                // (notice that it is possible for all three to be -1 if, for example,
                // there is only one pentuple-apostrophe in the line)
                $arr[$firstspace] = "''";
                $arr[$firstspace - 1] .= "'";
            }
        }

        // Now let's actually convert our apostrophic mush to HTML!
        $output = '';
        $buffer = '';
        $state = '';
        $i = 0;
        foreach ( $arr as $r ) {
            if ( ( $i % 2 ) == 0 ) {
                if ( $state === 'both' ) {
                    $buffer .= $r;
                } else {
                    $output .= $r;
                }
            } else {
                $thislen = strlen( $r );
                if ( $thislen == 2 ) {
                    if ( $state === 'i' ) {
                        $output .= '</i>';
                        $state = '';
                    } elseif ( $state === 'bi' ) {
                        $output .= '</i>';
                        $state = 'b';
                    } elseif ( $state === 'ib' ) {
                        $output .= '</b></i><b>';
                        $state = 'b';
                    } elseif ( $state === 'both' ) {
                        $output .= '<b><i>' . $buffer . '</i>';
                        $state = 'b';
                    } else { // $state can be 'b' or ''
                        $output .= '<i>';
                        $state .= 'i';
                    }
                } elseif ( $thislen == 3 ) {
                    if ( $state === 'b' ) {
                        $output .= '</b>';
                        $state = '';
                    } elseif ( $state === 'bi' ) {
                        $output .= '</i></b><i>';
                        $state = 'i';
                    } elseif ( $state === 'ib' ) {
                        $output .= '</b>';
                        $state = 'i';
                    } elseif ( $state === 'both' ) {
                        $output .= '<i><b>' . $buffer . '</b>';
                        $state = 'i';
                    } else { // $state can be 'i' or ''
                        $output .= '<b>';
                        $state .= 'b';
                    }
                } elseif ( $thislen == 5 ) {
                    if ( $state === 'b' ) {
                        $output .= '</b><i>';
                        $state = 'i';
                    } elseif ( $state === 'i' ) {
                        $output .= '</i><b>';
                        $state = 'b';
                    } elseif ( $state === 'bi' ) {
                        $output .= '</i></b>';
                        $state = '';
                    } elseif ( $state === 'ib' ) {
                        $output .= '</b></i>';
                        $state = '';
                    } elseif ( $state === 'both' ) {
                        $output .= '<i><b>' . $buffer . '</b></i>';
                        $state = '';
                    } else { // ($state == '')
                        $buffer = '';
                        $state = 'both';
                    }
                }
            }
            $i++;
        }
        // Now close all remaining tags.  Notice that the order is important.
        if ( $state === 'b' || $state === 'ib' ) {
            $output .= '</b>';
        }
        if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
            $output .= '</i>';
        }
        if ( $state === 'bi' ) {
            $output .= '</b>';
        }
        // There might be lonely ''''', so make sure we have a buffer
        if ( $state === 'both' && $buffer ) {
            $output .= '<b><i>' . $buffer . '</i></b>';
        }
        return $output;
    }

    public function replaceExternalLinks( $text ) {
        wfProfileIn( __METHOD__ );

        $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
        if ( $bits === false ) {
            wfProfileOut( __METHOD__ );
            throw new MWException( "PCRE needs to be compiled with "
                . "--enable-unicode-properties in order for MediaWiki to function" );
        }
        $s = array_shift( $bits );

        $i = 0;
        while ( $i < count( $bits ) ) {
            $url = $bits[$i++];
            $i++; // protocol
            $text = $bits[$i++];
            $trail = $bits[$i++];

            # The characters '<' and '>' (which were escaped by
            # removeHTMLtags()) should not be included in
            # URLs, per RFC 2396.
            $m2 = array();
            if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
                $text = substr( $url, $m2[0][1] ) . ' ' . $text;
                $url = substr( $url, 0, $m2[0][1] );
            }

            # If the link text is an image URL, replace it with an <img> tag
            # This happened by accident in the original parser, but some people used it extensively
            $img = $this->maybeMakeExternalImage( $text );
            if ( $img !== false ) {
                $text = $img;
            }

            $dtrail = '';

            # Set linktype for CSS - if URL==text, link is essentially free
            $linktype = ( $text === $url ) ? 'free' : 'text';

            # No link text, e.g. [http://domain.tld/some.link]
            if ( $text == '' ) {
                # Autonumber
                $langObj = $this->getTargetLanguage();
                $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
                $linktype = 'autonumber';
            } else {
                # Have link text, e.g. [http://domain.tld/some.link text]s
                # Check for trail
                list( $dtrail, $trail ) = Linker::splitTrail( $trail );
            }

            $text = $this->getConverterLanguage()->markNoConversion( $text );

            $url = Sanitizer::cleanUrl( $url );

            # Use the encoded URL
            # This means that users can paste URLs directly into the text
            # Funny characters like ö aren't valid in URLs anyway
            # This was changed in August 2004
            $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
                $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;

            # Register link in the output object.
            # Replace unnecessary URL escape codes with the referenced character
            # This prevents spammers from hiding links from the filters
            $pasteurized = self::normalizeLinkUrl( $url );
            $this->mOutput->addExternalLink( $pasteurized );
        }

        wfProfileOut( __METHOD__ );
        return $s;
    }

    public static function getExternalLinkRel( $url = false, $title = null ) {
        global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
        $ns = $title ? $title->getNamespace() : false;
        if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
            && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
        ) {
            return 'nofollow';
        }
        return null;
    }

    public function getExternalLinkAttribs( $url = false ) {
        $attribs = array();
        $attribs['rel'] = self::getExternalLinkRel( $url, $this->mTitle );

        if ( $this->mOptions->getExternalLinkTarget() ) {
            $attribs['target'] = $this->mOptions->getExternalLinkTarget();
        }
        return $attribs;
    }

    public static function replaceUnusualEscapes( $url ) {
        wfDeprecated( __METHOD__, '1.24' );
        return self::normalizeLinkUrl( $url );
    }

    public static function normalizeLinkUrl( $url ) {
        # First, make sure unsafe characters are encoded
        $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
            function ( $m ) {
                return rawurlencode( $m[0] );
            },
            $url
        );

        $ret = '';
        $end = strlen( $url );

        # Fragment part - 'fragment'
        $start = strpos( $url, '#' );
        if ( $start !== false && $start < $end ) {
            $ret = self::normalizeUrlComponent(
                substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
            $end = $start;
        }

        # Query part - 'query' minus &=+;
        $start = strpos( $url, '?' );
        if ( $start !== false && $start < $end ) {
            $ret = self::normalizeUrlComponent(
                substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
            $end = $start;
        }

        # Scheme and path part - 'pchar'
        # (we assume no userinfo or encoded colons in the host)
        $ret = self::normalizeUrlComponent(
            substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;

        return $ret;
    }

    private static function normalizeUrlComponent( $component, $unsafe ) {
        $callback = function ( $matches ) use ( $unsafe ) {
            $char = urldecode( $matches[0] );
            $ord = ord( $char );
            if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
                # Unescape it
                return $char;
            } else {
                # Leave it escaped, but use uppercase for a-f
                return strtoupper( $matches[0] );
            }
        };
        return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
    }

    private function maybeMakeExternalImage( $url ) {
        $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
        $imagesexception = !empty( $imagesfrom );
        $text = false;
        # $imagesfrom could be either a single string or an array of strings, parse out the latter
        if ( $imagesexception && is_array( $imagesfrom ) ) {
            $imagematch = false;
            foreach ( $imagesfrom as $match ) {
                if ( strpos( $url, $match ) === 0 ) {
                    $imagematch = true;
                    break;
                }
            }
        } elseif ( $imagesexception ) {
            $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
        } else {
            $imagematch = false;
        }

        if ( $this->mOptions->getAllowExternalImages()
            || ( $imagesexception && $imagematch )
        ) {
            if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
                # Image found
                $text = Linker::makeExternalImage( $url );
            }
        }
        if ( !$text && $this->mOptions->getEnableImageWhitelist()
            && preg_match( self::EXT_IMAGE_REGEX, $url )
        ) {
            $whitelist = explode(
                "\n",
                wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
            );

            foreach ( $whitelist as $entry ) {
                # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
                if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
                    continue;
                }
                if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
                    # Image matches a whitelist entry
                    $text = Linker::makeExternalImage( $url );
                    break;
                }
            }
        }
        return $text;
    }

    public function replaceInternalLinks( $s ) {
        $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
        return $s;
    }

    public function replaceInternalLinks2( &$s ) {
        global $wgExtraInterlanguageLinkPrefixes;
        wfProfileIn( __METHOD__ );

        wfProfileIn( __METHOD__ . '-setup' );
        static $tc = false, $e1, $e1_img;
        # the % is needed to support urlencoded titles as well
        if ( !$tc ) {
            $tc = Title::legalChars() . '#%';
            # Match a link having the form [[namespace:link|alternate]]trail
            $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
            # Match cases where there is no "]]", which might still be images
            $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
        }

        $holders = new LinkHolderArray( $this );

        # split the entire text string on occurrences of [[
        $a = StringUtils::explode( '[[', ' ' . $s );
        # get the first element (all text up to first [[), and remove the space we added
        $s = $a->current();
        $a->next();
        $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
        $s = substr( $s, 1 );

        $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
        $e2 = null;
        if ( $useLinkPrefixExtension ) {
            # Match the end of a line for a word that's not followed by whitespace,
            # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
            global $wgContLang;
            $charset = $wgContLang->linkPrefixCharset();
            $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
        }

        if ( is_null( $this->mTitle ) ) {
            wfProfileOut( __METHOD__ . '-setup' );
            wfProfileOut( __METHOD__ );
            throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
        }
        $nottalk = !$this->mTitle->isTalkPage();

        if ( $useLinkPrefixExtension ) {
            $m = array();
            if ( preg_match( $e2, $s, $m ) ) {
                $first_prefix = $m[2];
            } else {
                $first_prefix = false;
            }
        } else {
            $prefix = '';
        }

        $useSubpages = $this->areSubpagesAllowed();
        wfProfileOut( __METHOD__ . '-setup' );

        // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
        # Loop for each link
        for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
            // @codingStandardsIgnoreStart

            # Check for excessive memory usage
            if ( $holders->isBig() ) {
                # Too big
                # Do the existence check, replace the link holders and clear the array
                $holders->replace( $s );
                $holders->clear();
            }

            if ( $useLinkPrefixExtension ) {
                wfProfileIn( __METHOD__ . '-prefixhandling' );
                if ( preg_match( $e2, $s, $m ) ) {
                    $prefix = $m[2];
                    $s = $m[1];
                } else {
                    $prefix = '';
                }
                # first link
                if ( $first_prefix ) {
                    $prefix = $first_prefix;
                    $first_prefix = false;
                }
                wfProfileOut( __METHOD__ . '-prefixhandling' );
            }

            $might_be_img = false;

            wfProfileIn( __METHOD__ . "-e1" );
            if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
                $text = $m[2];
                # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
                # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
                # the real problem is with the $e1 regex
                # See bug 1300.
                #
                # Still some problems for cases where the ] is meant to be outside punctuation,
                # and no image is in sight. See bug 2095.
                #
                if ( $text !== ''
                    && substr( $m[3], 0, 1 ) === ']'
                    && strpos( $text, '[' ) !== false
                ) {
                    $text .= ']'; # so that replaceExternalLinks($text) works later
                    $m[3] = substr( $m[3], 1 );
                }
                # fix up urlencoded title texts
                if ( strpos( $m[1], '%' ) !== false ) {
                    # Should anchors '#' also be rejected?
                    $m[1] = str_replace( array( '<', '>' ), array( '&lt;', '&gt;' ), rawurldecode( $m[1] ) );
                }
                $trail = $m[3];
            } elseif ( preg_match( $e1_img, $line, $m ) ) {
                # Invalid, but might be an image with a link in its caption
                $might_be_img = true;
                $text = $m[2];
                if ( strpos( $m[1], '%' ) !== false ) {
                    $m[1] = rawurldecode( $m[1] );
                }
                $trail = "";
            } else { # Invalid form; output directly
                $s .= $prefix . '[[' . $line;
                wfProfileOut( __METHOD__ . "-e1" );
                continue;
            }
            wfProfileOut( __METHOD__ . "-e1" );
            wfProfileIn( __METHOD__ . "-misc" );

            $origLink = $m[1];

            # Don't allow internal links to pages containing
            # PROTO: where PROTO is a valid URL protocol; these
            # should be external links.
            if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
                $s .= $prefix . '[[' . $line;
                wfProfileOut( __METHOD__ . "-misc" );
                continue;
            }

            # Make subpage if necessary
            if ( $useSubpages ) {
                $link = $this->maybeDoSubpageLink( $origLink, $text );
            } else {
                $link = $origLink;
            }

            $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
            if ( !$noforce ) {
                # Strip off leading ':'
                $link = substr( $link, 1 );
            }

            wfProfileOut( __METHOD__ . "-misc" );
            wfProfileIn( __METHOD__ . "-title" );
            $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
            if ( $nt === null ) {
                $s .= $prefix . '[[' . $line;
                wfProfileOut( __METHOD__ . "-title" );
                continue;
            }

            $ns = $nt->getNamespace();
            $iw = $nt->getInterwiki();
            wfProfileOut( __METHOD__ . "-title" );

            if ( $might_be_img ) { # if this is actually an invalid link
                wfProfileIn( __METHOD__ . "-might_be_img" );
                if ( $ns == NS_FILE && $noforce ) { # but might be an image
                    $found = false;
                    while ( true ) {
                        # look at the next 'line' to see if we can close it there
                        $a->next();
                        $next_line = $a->current();
                        if ( $next_line === false || $next_line === null ) {
                            break;
                        }
                        $m = explode( ']]', $next_line, 3 );
                        if ( count( $m ) == 3 ) {
                            # the first ]] closes the inner link, the second the image
                            $found = true;
                            $text .= "[[{$m[0]}]]{$m[1]}";
                            $trail = $m[2];
                            break;
                        } elseif ( count( $m ) == 2 ) {
                            # if there's exactly one ]] that's fine, we'll keep looking
                            $text .= "[[{$m[0]}]]{$m[1]}";
                        } else {
                            # if $next_line is invalid too, we need look no further
                            $text .= '[[' . $next_line;
                            break;
                        }
                    }
                    if ( !$found ) {
                        # we couldn't find the end of this imageLink, so output it raw
                        # but don't ignore what might be perfectly normal links in the text we've examined
                        $holders->merge( $this->replaceInternalLinks2( $text ) );
                        $s .= "{$prefix}[[$link|$text";
                        # note: no $trail, because without an end, there *is* no trail
                        wfProfileOut( __METHOD__ . "-might_be_img" );
                        continue;
                    }
                } else { # it's not an image, so output it raw
                    $s .= "{$prefix}[[$link|$text";
                    # note: no $trail, because without an end, there *is* no trail
                    wfProfileOut( __METHOD__ . "-might_be_img" );
                    continue;
                }
                wfProfileOut( __METHOD__ . "-might_be_img" );
            }

            $wasblank = ( $text == '' );
            if ( $wasblank ) {
                $text = $link;
            } else {
                # Bug 4598 madness. Handle the quotes only if they come from the alternate part
                # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
                # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
                #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
                $text = $this->doQuotes( $text );
            }

            # Link not escaped by : , create the various objects
            if ( $noforce && !$nt->wasLocalInterwiki() ) {
                # Interwikis
                wfProfileIn( __METHOD__ . "-interwiki" );
                if (
                    $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
                        Language::fetchLanguageName( $iw, null, 'mw' ) ||
                        in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
                    )
                ) {
                    # Bug 24502: filter duplicates
                    if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
                        $this->mLangLinkLanguages[$iw] = true;
                        $this->mOutput->addLanguageLink( $nt->getFullText() );
                    }

                    $s = rtrim( $s . $prefix );
                    $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
                    wfProfileOut( __METHOD__ . "-interwiki" );
                    continue;
                }
                wfProfileOut( __METHOD__ . "-interwiki" );

                if ( $ns == NS_FILE ) {
                    wfProfileIn( __METHOD__ . "-image" );
                    if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
                        if ( $wasblank ) {
                            # if no parameters were passed, $text
                            # becomes something like "File:Foo.png",
                            # which we don't want to pass on to the
                            # image generator
                            $text = '';
                        } else {
                            # recursively parse links inside the image caption
                            # actually, this will parse them in any other parameters, too,
                            # but it might be hard to fix that, and it doesn't matter ATM
                            $text = $this->replaceExternalLinks( $text );
                            $holders->merge( $this->replaceInternalLinks2( $text ) );
                        }
                        # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
                        $s .= $prefix . $this->armorLinks(
                            $this->makeImage( $nt, $text, $holders ) ) . $trail;
                    } else {
                        $s .= $prefix . $trail;
                    }
                    wfProfileOut( __METHOD__ . "-image" );
                    continue;
                }

                if ( $ns == NS_CATEGORY ) {
                    wfProfileIn( __METHOD__ . "-category" );
                    $s = rtrim( $s . "\n" ); # bug 87

                    if ( $wasblank ) {
                        $sortkey = $this->getDefaultSort();
                    } else {
                        $sortkey = $text;
                    }
                    $sortkey = Sanitizer::decodeCharReferences( $sortkey );
                    $sortkey = str_replace( "\n", '', $sortkey );
                    $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
                    $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );

                    $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;

                    wfProfileOut( __METHOD__ . "-category" );
                    continue;
                }
            }

            # Self-link checking. For some languages, variants of the title are checked in
            # LinkHolderArray::doVariants() to allow batching the existence checks necessary
            # for linking to a different variant.
            if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
                $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
                continue;
            }

            # NS_MEDIA is a pseudo-namespace for linking directly to a file
            # @todo FIXME: Should do batch file existence checks, see comment below
            if ( $ns == NS_MEDIA ) {
                wfProfileIn( __METHOD__ . "-media" );
                # Give extensions a chance to select the file revision for us
                $options = array();
                $descQuery = false;
                wfRunHooks( 'BeforeParserFetchFileAndTitle',
                    array( $this, $nt, &$options, &$descQuery ) );
                # Fetch and register the file (file title may be different via hooks)
                list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
                # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
                $s .= $prefix . $this->armorLinks(
                    Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
                wfProfileOut( __METHOD__ . "-media" );
                continue;
            }

            wfProfileIn( __METHOD__ . "-always_known" );
            # Some titles, such as valid special pages or files in foreign repos, should
            # be shown as bluelinks even though they're not included in the page table
            #
            # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
            # batch file existence checks for NS_FILE and NS_MEDIA
            if ( $iw == '' && $nt->isAlwaysKnown() ) {
                $this->mOutput->addLink( $nt );
                $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
            } else {
                # Links will be added to the output link list after checking
                $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
            }
            wfProfileOut( __METHOD__ . "-always_known" );
        }
        wfProfileOut( __METHOD__ );
        return $holders;
    }

    public function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
        list( $inside, $trail ) = Linker::splitTrail( $trail );

        if ( is_string( $query ) ) {
            $query = wfCgiToArray( $query );
        }
        if ( $text == '' ) {
            $text = htmlspecialchars( $nt->getPrefixedText() );
        }

        $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );

        return $this->armorLinks( $link ) . $trail;
    }

    public function armorLinks( $text ) {
        return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
            "{$this->mUniqPrefix}NOPARSE$1", $text );
    }

    public function areSubpagesAllowed() {
        # Some namespaces don't allow subpages
        return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
    }

    public function maybeDoSubpageLink( $target, &$text ) {
        return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
    }

    public function closeParagraph() {
        $result = '';
        if ( $this->mLastSection != '' ) {
            $result = '</' . $this->mLastSection . ">\n";
        }
        $this->mInPre = false;
        $this->mLastSection = '';
        return $result;
    }

    public function getCommon( $st1, $st2 ) {
        $fl = strlen( $st1 );
        $shorter = strlen( $st2 );
        if ( $fl < $shorter ) {
            $shorter = $fl;
        }

        for ( $i = 0; $i < $shorter; ++$i ) {
            if ( $st1[$i] != $st2[$i] ) {
                break;
            }
        }
        return $i;
    }

    public function openList( $char ) {
        $result = $this->closeParagraph();

        if ( '*' === $char ) {
            $result .= "<ul><li>";
        } elseif ( '#' === $char ) {
            $result .= "<ol><li>";
        } elseif ( ':' === $char ) {
            $result .= "<dl><dd>";
        } elseif ( ';' === $char ) {
            $result .= "<dl><dt>";
            $this->mDTopen = true;
        } else {
            $result = '<!-- ERR 1 -->';
        }

        return $result;
    }

    public function nextItem( $char ) {
        if ( '*' === $char || '#' === $char ) {
            return "</li>\n<li>";
        } elseif ( ':' === $char || ';' === $char ) {
            $close = "</dd>\n";
            if ( $this->mDTopen ) {
                $close = "</dt>\n";
            }
            if ( ';' === $char ) {
                $this->mDTopen = true;
                return $close . '<dt>';
            } else {
                $this->mDTopen = false;
                return $close . '<dd>';
            }
        }
        return '<!-- ERR 2 -->';
    }

    public function closeList( $char ) {
        if ( '*' === $char ) {
            $text = "</li></ul>";
        } elseif ( '#' === $char ) {
            $text = "</li></ol>";
        } elseif ( ':' === $char ) {
            if ( $this->mDTopen ) {
                $this->mDTopen = false;
                $text = "</dt></dl>";
            } else {
                $text = "</dd></dl>";
            }
        } else {
            return '<!-- ERR 3 -->';
        }
        return $text;
    }
    public function doBlockLevels( $text, $linestart ) {
        wfProfileIn( __METHOD__ );

        # Parsing through the text line by line.  The main thing
        # happening here is handling of block-level elements p, pre,
        # and making lists from lines starting with * # : etc.
        #
        $textLines = StringUtils::explode( "\n", $text );

        $lastPrefix = $output = '';
        $this->mDTopen = $inBlockElem = false;
        $prefixLength = 0;
        $paragraphStack = false;
        $inBlockquote = false;

        foreach ( $textLines as $oLine ) {
            # Fix up $linestart
            if ( !$linestart ) {
                $output .= $oLine;
                $linestart = true;
                continue;
            }
            # * = ul
            # # = ol
            # ; = dt
            # : = dd

            $lastPrefixLength = strlen( $lastPrefix );
            $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
            $preOpenMatch = preg_match( '/<pre/i', $oLine );
            # If not in a <pre> element, scan for and figure out what prefixes are there.
            if ( !$this->mInPre ) {
                # Multiple prefixes may abut each other for nested lists.
                $prefixLength = strspn( $oLine, '*#:;' );
                $prefix = substr( $oLine, 0, $prefixLength );

                # eh?
                # ; and : are both from definition-lists, so they're equivalent
                #  for the purposes of determining whether or not we need to open/close
                #  elements.
                $prefix2 = str_replace( ';', ':', $prefix );
                $t = substr( $oLine, $prefixLength );
                $this->mInPre = (bool)$preOpenMatch;
            } else {
                # Don't interpret any other prefixes in preformatted text
                $prefixLength = 0;
                $prefix = $prefix2 = '';
                $t = $oLine;
            }

            # List generation
            if ( $prefixLength && $lastPrefix === $prefix2 ) {
                # Same as the last item, so no need to deal with nesting or opening stuff
                $output .= $this->nextItem( substr( $prefix, -1 ) );
                $paragraphStack = false;

                if ( substr( $prefix, -1 ) === ';' ) {
                    # The one nasty exception: definition lists work like this:
                    # ; title : definition text
                    # So we check for : in the remainder text to split up the
                    # title and definition, without b0rking links.
                    $term = $t2 = '';
                    if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
                        $t = $t2;
                        $output .= $term . $this->nextItem( ':' );
                    }
                }
            } elseif ( $prefixLength || $lastPrefixLength ) {
                # We need to open or close prefixes, or both.

                # Either open or close a level...
                $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
                $paragraphStack = false;

                # Close all the prefixes which aren't shared.
                while ( $commonPrefixLength < $lastPrefixLength ) {
                    $output .= $this->closeList( $lastPrefix[$lastPrefixLength - 1] );
                    --$lastPrefixLength;
                }

                # Continue the current prefix if appropriate.
                if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
                    $output .= $this->nextItem( $prefix[$commonPrefixLength - 1] );
                }

                # Open prefixes where appropriate.
                if (  $lastPrefix && $prefixLength > $commonPrefixLength ) {
                    $output .= "\n";
                }
                while ( $prefixLength > $commonPrefixLength ) {
                    $char = substr( $prefix, $commonPrefixLength, 1 );
                    $output .= $this->openList( $char );

                    if ( ';' === $char ) {
                        # @todo FIXME: This is dupe of code above
                        if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
                            $t = $t2;
                            $output .= $term . $this->nextItem( ':' );
                        }
                    }
                    ++$commonPrefixLength;
                }
                if ( !$prefixLength && $lastPrefix ) {
                    $output .= "\n";
                }
                $lastPrefix = $prefix2;
            }

            # If we have no prefixes, go to paragraph mode.
            if ( 0 == $prefixLength ) {
                wfProfileIn( __METHOD__ . "-paragraph" );
                # No prefix (not in list)--go to paragraph mode
                # XXX: use a stack for nestable elements like span, table and div
                $openmatch = preg_match(
                    '/(?:<table|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|'
                        . '<p|<ul|<ol|<dl|<li|<\\/tr|<\\/td|<\\/th)/iS',
                    $t
                );
                $closematch = preg_match(
                    '/(?:<\\/table|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'
                        . '<td|<th|<\\/?blockquote|<\\/?div|<hr|<\\/pre|<\\/p|<\\/mw:|'
                        . $this->mUniqPrefix
                        . '-pre|<\\/li|<\\/ul|<\\/ol|<\\/dl|<\\/?center)/iS',
                    $t
                );

                if ( $openmatch or $closematch ) {
                    $paragraphStack = false;
                    # @todo bug 5718: paragraph closed
                    $output .= $this->closeParagraph();
                    if ( $preOpenMatch and !$preCloseMatch ) {
                        $this->mInPre = true;
                    }
                    $bqOffset = 0;
                    while ( preg_match( '/<(\\/?)blockquote[\s>]/i', $t, $bqMatch, PREG_OFFSET_CAPTURE, $bqOffset ) ) {
                        $inBlockquote = !$bqMatch[1][0]; // is this a close tag?
                        $bqOffset = $bqMatch[0][1] + strlen( $bqMatch[0][0] );
                    }
                    $inBlockElem = !$closematch;
                } elseif ( !$inBlockElem && !$this->mInPre ) {
                    if ( ' ' == substr( $t, 0, 1 )
                        && ( $this->mLastSection === 'pre' || trim( $t ) != '' )
                        && !$inBlockquote
                    ) {
                        # pre
                        if ( $this->mLastSection !== 'pre' ) {
                            $paragraphStack = false;
                            $output .= $this->closeParagraph() . '<pre>';
                            $this->mLastSection = 'pre';
                        }
                        $t = substr( $t, 1 );
                    } else {
                        # paragraph
                        if ( trim( $t ) === '' ) {
                            if ( $paragraphStack ) {
                                $output .= $paragraphStack . '<br />';
                                $paragraphStack = false;
                                $this->mLastSection = 'p';
                            } else {
                                if ( $this->mLastSection !== 'p' ) {
                                    $output .= $this->closeParagraph();
                                    $this->mLastSection = '';
                                    $paragraphStack = '<p>';
                                } else {
                                    $paragraphStack = '</p><p>';
                                }
                            }
                        } else {
                            if ( $paragraphStack ) {
                                $output .= $paragraphStack;
                                $paragraphStack = false;
                                $this->mLastSection = 'p';
                            } elseif ( $this->mLastSection !== 'p' ) {
                                $output .= $this->closeParagraph() . '<p>';
                                $this->mLastSection = 'p';
                            }
                        }
                    }
                }
                wfProfileOut( __METHOD__ . "-paragraph" );
            }
            # somewhere above we forget to get out of pre block (bug 785)
            if ( $preCloseMatch && $this->mInPre ) {
                $this->mInPre = false;
            }
            if ( $paragraphStack === false ) {
                $output .= $t;
                if ( $prefixLength === 0 ) {
                    $output .= "\n";
                }
            }
        }
        while ( $prefixLength ) {
            $output .= $this->closeList( $prefix2[$prefixLength - 1] );
            --$prefixLength;
            if ( !$prefixLength ) {
                $output .= "\n";
            }
        }
        if ( $this->mLastSection != '' ) {
            $output .= '</' . $this->mLastSection . '>';
            $this->mLastSection = '';
        }

        wfProfileOut( __METHOD__ );
        return $output;
    }

    public function findColonNoLinks( $str, &$before, &$after ) {
        wfProfileIn( __METHOD__ );

        $pos = strpos( $str, ':' );
        if ( $pos === false ) {
            # Nothing to find!
            wfProfileOut( __METHOD__ );
            return false;
        }

        $lt = strpos( $str, '<' );
        if ( $lt === false || $lt > $pos ) {
            # Easy; no tag nesting to worry about
            $before = substr( $str, 0, $pos );
            $after = substr( $str, $pos + 1 );
            wfProfileOut( __METHOD__ );
            return $pos;
        }

        # Ugly state machine to walk through avoiding tags.
        $state = self::COLON_STATE_TEXT;
        $stack = 0;
        $len = strlen( $str );
        for ( $i = 0; $i < $len; $i++ ) {
            $c = $str[$i];

            switch ( $state ) {
            # (Using the number is a performance hack for common cases)
            case 0: # self::COLON_STATE_TEXT:
                switch ( $c ) {
                case "<":
                    # Could be either a <start> tag or an </end> tag
                    $state = self::COLON_STATE_TAGSTART;
                    break;
                case ":":
                    if ( $stack == 0 ) {
                        # We found it!
                        $before = substr( $str, 0, $i );
                        $after = substr( $str, $i + 1 );
                        wfProfileOut( __METHOD__ );
                        return $i;
                    }
                    # Embedded in a tag; don't break it.
                    break;
                default:
                    # Skip ahead looking for something interesting
                    $colon = strpos( $str, ':', $i );
                    if ( $colon === false ) {
                        # Nothing else interesting
                        wfProfileOut( __METHOD__ );
                        return false;
                    }
                    $lt = strpos( $str, '<', $i );
                    if ( $stack === 0 ) {
                        if ( $lt === false || $colon < $lt ) {
                            # We found it!
                            $before = substr( $str, 0, $colon );
                            $after = substr( $str, $colon + 1 );
                            wfProfileOut( __METHOD__ );
                            return $i;
                        }
                    }
                    if ( $lt === false ) {
                        # Nothing else interesting to find; abort!
                        # We're nested, but there's no close tags left. Abort!
                        break 2;
                    }
                    # Skip ahead to next tag start
                    $i = $lt;
                    $state = self::COLON_STATE_TAGSTART;
                }
                break;
            case 1: # self::COLON_STATE_TAG:
                # In a <tag>
                switch ( $c ) {
                case ">":
                    $stack++;
                    $state = self::COLON_STATE_TEXT;
                    break;
                case "/":
                    # Slash may be followed by >?
                    $state = self::COLON_STATE_TAGSLASH;
                    break;
                default:
                    # ignore
                }
                break;
            case 2: # self::COLON_STATE_TAGSTART:
                switch ( $c ) {
                case "/":
                    $state = self::COLON_STATE_CLOSETAG;
                    break;
                case "!":
                    $state = self::COLON_STATE_COMMENT;
                    break;
                case ">":
                    # Illegal early close? This shouldn't happen D:
                    $state = self::COLON_STATE_TEXT;
                    break;
                default:
                    $state = self::COLON_STATE_TAG;
                }
                break;
            case 3: # self::COLON_STATE_CLOSETAG:
                # In a </tag>
                if ( $c === ">" ) {
                    $stack--;
                    if ( $stack < 0 ) {
                        wfDebug( __METHOD__ . ": Invalid input; too many close tags\n" );
                        wfProfileOut( __METHOD__ );
                        return false;
                    }
                    $state = self::COLON_STATE_TEXT;
                }
                break;
            case self::COLON_STATE_TAGSLASH:
                if ( $c === ">" ) {
                    # Yes, a self-closed tag <blah/>
                    $state = self::COLON_STATE_TEXT;
                } else {
                    # Probably we're jumping the gun, and this is an attribute
                    $state = self::COLON_STATE_TAG;
                }
                break;
            case 5: # self::COLON_STATE_COMMENT:
                if ( $c === "-" ) {
                    $state = self::COLON_STATE_COMMENTDASH;
                }
                break;
            case self::COLON_STATE_COMMENTDASH:
                if ( $c === "-" ) {
                    $state = self::COLON_STATE_COMMENTDASHDASH;
                } else {
                    $state = self::COLON_STATE_COMMENT;
                }
                break;
            case self::COLON_STATE_COMMENTDASHDASH:
                if ( $c === ">" ) {
                    $state = self::COLON_STATE_TEXT;
                } else {
                    $state = self::COLON_STATE_COMMENT;
                }
                break;
            default:
                wfProfileOut( __METHOD__ );
                throw new MWException( "State machine error in " . __METHOD__ );
            }
        }
        if ( $stack > 0 ) {
            wfDebug( __METHOD__ . ": Invalid input; not enough close tags (stack $stack, state $state)\n" );
            wfProfileOut( __METHOD__ );
            return false;
        }
        wfProfileOut( __METHOD__ );
        return false;
    }

    public function getVariableValue( $index, $frame = false ) {
        global $wgContLang, $wgSitename, $wgServer, $wgServerName;
        global $wgArticlePath, $wgScriptPath, $wgStylePath;

        if ( is_null( $this->mTitle ) ) {
            // If no title set, bad things are going to happen
            // later. Title should always be set since this
            // should only be called in the middle of a parse
            // operation (but the unit-tests do funky stuff)
            throw new MWException( __METHOD__ . ' Should only be '
                . ' called while parsing (no title set)' );
        }

        if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
            if ( isset( $this->mVarCache[$index] ) ) {
                return $this->mVarCache[$index];
            }
        }

        $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
        wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );

        $pageLang = $this->getFunctionLang();

        switch ( $index ) {
            case '!':
                $value = '|';
                break;
            case 'currentmonth':
                $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
                break;
            case 'currentmonth1':
                $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
                break;
            case 'currentmonthname':
                $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
                break;
            case 'currentmonthnamegen':
                $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
                break;
            case 'currentmonthabbrev':
                $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
                break;
            case 'currentday':
                $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
                break;
            case 'currentday2':
                $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
                break;
            case 'localmonth':
                $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
                break;
            case 'localmonth1':
                $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
                break;
            case 'localmonthname':
                $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
                break;
            case 'localmonthnamegen':
                $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
                break;
            case 'localmonthabbrev':
                $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
                break;
            case 'localday':
                $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
                break;
            case 'localday2':
                $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
                break;
            case 'pagename':
                $value = wfEscapeWikiText( $this->mTitle->getText() );
                break;
            case 'pagenamee':
                $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
                break;
            case 'fullpagename':
                $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
                break;
            case 'fullpagenamee':
                $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
                break;
            case 'subpagename':
                $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
                break;
            case 'subpagenamee':
                $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
                break;
            case 'rootpagename':
                $value = wfEscapeWikiText( $this->mTitle->getRootText() );
                break;
            case 'rootpagenamee':
                $value = wfEscapeWikiText( wfUrlEncode( str_replace(
                    ' ',
                    '_',
                    $this->mTitle->getRootText()
                ) ) );
                break;
            case 'basepagename':
                $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
                break;
            case 'basepagenamee':
                $value = wfEscapeWikiText( wfUrlEncode( str_replace(
                    ' ',
                    '_',
                    $this->mTitle->getBaseText()
                ) ) );
                break;
            case 'talkpagename':
                if ( $this->mTitle->canTalk() ) {
                    $talkPage = $this->mTitle->getTalkPage();
                    $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
                } else {
                    $value = '';
                }
                break;
            case 'talkpagenamee':
                if ( $this->mTitle->canTalk() ) {
                    $talkPage = $this->mTitle->getTalkPage();
                    $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
                } else {
                    $value = '';
                }
                break;
            case 'subjectpagename':
                $subjPage = $this->mTitle->getSubjectPage();
                $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
                break;
            case 'subjectpagenamee':
                $subjPage = $this->mTitle->getSubjectPage();
                $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
                break;
            case 'pageid': // requested in bug 23427
                $pageid = $this->getTitle()->getArticleID();
                if ( $pageid == 0 ) {
                    # 0 means the page doesn't exist in the database,
                    # which means the user is previewing a new page.
                    # The vary-revision flag must be set, because the magic word
                    # will have a different value once the page is saved.
                    $this->mOutput->setFlag( 'vary-revision' );
                    wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
                }
                $value = $pageid ? $pageid : null;
                break;
            case 'revisionid':
                # Let the edit saving system know we should parse the page
                # *after* a revision ID has been assigned.
                $this->mOutput->setFlag( 'vary-revision' );
                wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
                $value = $this->mRevisionId;
                break;
            case 'revisionday':
                # Let the edit saving system know we should parse the page
                # *after* a revision ID has been assigned. This is for null edits.
                $this->mOutput->setFlag( 'vary-revision' );
                wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
                $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
                break;
            case 'revisionday2':
                # Let the edit saving system know we should parse the page
                # *after* a revision ID has been assigned. This is for null edits.
                $this->mOutput->setFlag( 'vary-revision' );
                wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
                $value = substr( $this->getRevisionTimestamp(), 6, 2 );
                break;
            case 'revisionmonth':
                # Let the edit saving system know we should parse the page
                # *after* a revision ID has been assigned. This is for null edits.
                $this->mOutput->setFlag( 'vary-revision' );
                wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
                $value = substr( $this->getRevisionTimestamp(), 4, 2 );
                break;
            case 'revisionmonth1':
                # Let the edit saving system know we should parse the page
                # *after* a revision ID has been assigned. This is for null edits.
                $this->mOutput->setFlag( 'vary-revision' );
                wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
                $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
                break;
            case 'revisionyear':
                # Let the edit saving system know we should parse the page
                # *after* a revision ID has been assigned. This is for null edits.
                $this->mOutput->setFlag( 'vary-revision' );
                wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
                $value = substr( $this->getRevisionTimestamp(), 0, 4 );
                break;
            case 'revisiontimestamp':
                # Let the edit saving system know we should parse the page
                # *after* a revision ID has been assigned. This is for null edits.
                $this->mOutput->setFlag( 'vary-revision' );
                wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
                $value = $this->getRevisionTimestamp();
                break;
            case 'revisionuser':
                # Let the edit saving system know we should parse the page
                # *after* a revision ID has been assigned. This is for null edits.
                $this->mOutput->setFlag( 'vary-revision' );
                wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
                $value = $this->getRevisionUser();
                break;
            case 'revisionsize':
                # Let the edit saving system know we should parse the page
                # *after* a revision ID has been assigned. This is for null edits.
                $this->mOutput->setFlag( 'vary-revision' );
                wfDebug( __METHOD__ . ": {{REVISIONSIZE}} used, setting vary-revision...\n" );
                $value = $this->getRevisionSize();
                break;
            case 'namespace':
                $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
                break;
            case 'namespacee':
                $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
                break;
            case 'namespacenumber':
                $value = $this->mTitle->getNamespace();
                break;
            case 'talkspace':
                $value = $this->mTitle->canTalk()
                    ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
                    : '';
                break;
            case 'talkspacee':
                $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
                break;
            case 'subjectspace':
                $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
                break;
            case 'subjectspacee':
                $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
                break;
            case 'currentdayname':
                $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
                break;
            case 'currentyear':
                $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
                break;
            case 'currenttime':
                $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
                break;
            case 'currenthour':
                $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
                break;
            case 'currentweek':
                # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
                # int to remove the padding
                $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
                break;
            case 'currentdow':
                $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
                break;
            case 'localdayname':
                $value = $pageLang->getWeekdayName(
                    (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
                );
                break;
            case 'localyear':
                $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
                break;
            case 'localtime':
                $value = $pageLang->time(
                    MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
                    false,
                    false
                );
                break;
            case 'localhour':
                $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
                break;
            case 'localweek':
                # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
                # int to remove the padding
                $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
                break;
            case 'localdow':
                $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
                break;
            case 'numberofarticles':
                $value = $pageLang->formatNum( SiteStats::articles() );
                break;
            case 'numberoffiles':
                $value = $pageLang->formatNum( SiteStats::images() );
                break;
            case 'numberofusers':
                $value = $pageLang->formatNum( SiteStats::users() );
                break;
            case 'numberofactiveusers':
                $value = $pageLang->formatNum( SiteStats::activeUsers() );
                break;
            case 'numberofpages':
                $value = $pageLang->formatNum( SiteStats::pages() );
                break;
            case 'numberofadmins':
                $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
                break;
            case 'numberofedits':
                $value = $pageLang->formatNum( SiteStats::edits() );
                break;
            case 'numberofviews':
                global $wgDisableCounters;
                $value = !$wgDisableCounters ? $pageLang->formatNum( SiteStats::views() ) : '';
                break;
            case 'currenttimestamp':
                $value = wfTimestamp( TS_MW, $ts );
                break;
            case 'localtimestamp':
                $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
                break;
            case 'currentversion':
                $value = SpecialVersion::getVersion();
                break;
            case 'articlepath':
                return $wgArticlePath;
            case 'sitename':
                return $wgSitename;
            case 'server':
                return $wgServer;
            case 'servername':
                return $wgServerName;
            case 'scriptpath':
                return $wgScriptPath;
            case 'stylepath':
                return $wgStylePath;
            case 'directionmark':
                return $pageLang->getDirMark();
            case 'contentlanguage':
                global $wgLanguageCode;
                return $wgLanguageCode;
            case 'cascadingsources':
                $value = CoreParserFunctions::cascadingsources( $this );
                break;
            default:
                $ret = null;
                wfRunHooks(
                    'ParserGetVariableValueSwitch',
                    array( &$this, &$this->mVarCache, &$index, &$ret, &$frame )
                );

                return $ret;
        }

        if ( $index ) {
            $this->mVarCache[$index] = $value;
        }

        return $value;
    }

    public function initialiseVariables() {
        wfProfileIn( __METHOD__ );
        $variableIDs = MagicWord::getVariableIDs();
        $substIDs = MagicWord::getSubstIDs();

        $this->mVariables = new MagicWordArray( $variableIDs );
        $this->mSubstWords = new MagicWordArray( $substIDs );
        wfProfileOut( __METHOD__ );
    }

    public function preprocessToDom( $text, $flags = 0 ) {
        $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
        return $dom;
    }

    public static function splitWhitespace( $s ) {
        $ltrimmed = ltrim( $s );
        $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
        $trimmed = rtrim( $ltrimmed );
        $diff = strlen( $ltrimmed ) - strlen( $trimmed );
        if ( $diff > 0 ) {
            $w2 = substr( $ltrimmed, -$diff );
        } else {
            $w2 = '';
        }
        return array( $w1, $trimmed, $w2 );
    }

    public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
        # Is there any text? Also, Prevent too big inclusions!
        if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
            return $text;
        }
        wfProfileIn( __METHOD__ );

        if ( $frame === false ) {
            $frame = $this->getPreprocessor()->newFrame();
        } elseif ( !( $frame instanceof PPFrame ) ) {
            wfDebug( __METHOD__ . " called using plain parameters instead of "
                . "a PPFrame instance. Creating custom frame.\n" );
            $frame = $this->getPreprocessor()->newCustomFrame( $frame );
        }

        $dom = $this->preprocessToDom( $text );
        $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
        $text = $frame->expand( $dom, $flags );

        wfProfileOut( __METHOD__ );
        return $text;
    }

    public static function createAssocArgs( $args ) {
        $assocArgs = array();
        $index = 1;
        foreach ( $args as $arg ) {
            $eqpos = strpos( $arg, '=' );
            if ( $eqpos === false ) {
                $assocArgs[$index++] = $arg;
            } else {
                $name = trim( substr( $arg, 0, $eqpos ) );
                $value = trim( substr( $arg, $eqpos + 1 ) );
                if ( $value === false ) {
                    $value = '';
                }
                if ( $name !== false ) {
                    $assocArgs[$name] = $value;
                }
            }
        }

        return $assocArgs;
    }

    public function limitationWarn( $limitationType, $current = '', $max = '' ) {
        # does no harm if $current and $max are present but are unnecessary for the message
        $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
            ->inLanguage( $this->mOptions->getUserLangObj() )->text();
        $this->mOutput->addWarning( $warning );
        $this->addTrackingCategory( "$limitationType-category" );
    }

    public function braceSubstitution( $piece, $frame ) {
        wfProfileIn( __METHOD__ );
        wfProfileIn( __METHOD__ . '-setup' );

        // Flags

        // $text has been filled
        $found = false;
        // wiki markup in $text should be escaped
        $nowiki = false;
        // $text is HTML, armour it against wikitext transformation
        $isHTML = false;
        // Force interwiki transclusion to be done in raw mode not rendered
        $forceRawInterwiki = false;
        // $text is a DOM node needing expansion in a child frame
        $isChildObj = false;
        // $text is a DOM node needing expansion in the current frame
        $isLocalObj = false;

        # Title object, where $text came from
        $title = false;

        # $part1 is the bit before the first |, and must contain only title characters.
        # Various prefixes will be stripped from it later.
        $titleWithSpaces = $frame->expand( $piece['title'] );
        $part1 = trim( $titleWithSpaces );
        $titleText = false;

        # Original title text preserved for various purposes
        $originalTitle = $part1;

        # $args is a list of argument nodes, starting from index 0, not including $part1
        # @todo FIXME: If piece['parts'] is null then the call to getLength()
        # below won't work b/c this $args isn't an object
        $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
        wfProfileOut( __METHOD__ . '-setup' );

        $titleProfileIn = null; // profile templates

        # SUBST
        wfProfileIn( __METHOD__ . '-modifiers' );
        if ( !$found ) {

            $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );

            # Possibilities for substMatch: "subst", "safesubst" or FALSE
            # Decide whether to expand template or keep wikitext as-is.
            if ( $this->ot['wiki'] ) {
                if ( $substMatch === false ) {
                    $literal = true;  # literal when in PST with no prefix
                } else {
                    $literal = false; # expand when in PST with subst: or safesubst:
                }
            } else {
                if ( $substMatch == 'subst' ) {
                    $literal = true;  # literal when not in PST with plain subst:
                } else {
                    $literal = false; # expand when not in PST with safesubst: or no prefix
                }
            }
            if ( $literal ) {
                $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
                $isLocalObj = true;
                $found = true;
            }
        }

        # Variables
        if ( !$found && $args->getLength() == 0 ) {
            $id = $this->mVariables->matchStartToEnd( $part1 );
            if ( $id !== false ) {
                $text = $this->getVariableValue( $id, $frame );
                if ( MagicWord::getCacheTTL( $id ) > -1 ) {
                    $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
                }
                $found = true;
            }
        }

        # MSG, MSGNW and RAW
        if ( !$found ) {
            # Check for MSGNW:
            $mwMsgnw = MagicWord::get( 'msgnw' );
            if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
                $nowiki = true;
            } else {
                # Remove obsolete MSG:
                $mwMsg = MagicWord::get( 'msg' );
                $mwMsg->matchStartAndRemove( $part1 );
            }

            # Check for RAW:
            $mwRaw = MagicWord::get( 'raw' );
            if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
                $forceRawInterwiki = true;
            }
        }
        wfProfileOut( __METHOD__ . '-modifiers' );

        # Parser functions
        if ( !$found ) {
            wfProfileIn( __METHOD__ . '-pfunc' );

            $colonPos = strpos( $part1, ':' );
            if ( $colonPos !== false ) {
                $func = substr( $part1, 0, $colonPos );
                $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
                for ( $i = 0; $i < $args->getLength(); $i++ ) {
                    $funcArgs[] = $args->item( $i );
                }
                try {
                    $result = $this->callParserFunction( $frame, $func, $funcArgs );
                } catch ( Exception $ex ) {
                    wfProfileOut( __METHOD__ . '-pfunc' );
                    wfProfileOut( __METHOD__ );
                    throw $ex;
                }

                # The interface for parser functions allows for extracting
                # flags into the local scope. Extract any forwarded flags
                # here.
                extract( $result );
            }
            wfProfileOut( __METHOD__ . '-pfunc' );
        }

        # Finish mangling title and then check for loops.
        # Set $title to a Title object and $titleText to the PDBK
        if ( !$found ) {
            $ns = NS_TEMPLATE;
            # Split the title into page and subpage
            $subpage = '';
            $relative = $this->maybeDoSubpageLink( $part1, $subpage );
            if ( $part1 !== $relative ) {
                $part1 = $relative;
                $ns = $this->mTitle->getNamespace();
            }
            $title = Title::newFromText( $part1, $ns );
            if ( $title ) {
                $titleText = $title->getPrefixedText();
                # Check for language variants if the template is not found
                if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
                    $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
                }
                # Do recursion depth check
                $limit = $this->mOptions->getMaxTemplateDepth();
                if ( $frame->depth >= $limit ) {
                    $found = true;
                    $text = '<span class="error">'
                        . wfMessage( 'parser-template-recursion-depth-warning' )
                            ->numParams( $limit )->inContentLanguage()->text()
                        . '</span>';
                }
            }
        }

        # Load from database
        if ( !$found && $title ) {
            if ( !Profiler::instance()->isPersistent() ) {
                # Too many unique items can kill profiling DBs/collectors
                $titleProfileIn = __METHOD__ . "-title-" . $title->getPrefixedDBkey();
                wfProfileIn( $titleProfileIn ); // template in
            }
            wfProfileIn( __METHOD__ . '-loadtpl' );
            if ( !$title->isExternal() ) {
                if ( $title->isSpecialPage()
                    && $this->mOptions->getAllowSpecialInclusion()
                    && $this->ot['html']
                ) {
                    // Pass the template arguments as URL parameters.
                    // "uselang" will have no effect since the Language object
                    // is forced to the one defined in ParserOptions.
                    $pageArgs = array();
                    $argsLength = $args->getLength();
                    for ( $i = 0; $i < $argsLength; $i++ ) {
                        $bits = $args->item( $i )->splitArg();
                        if ( strval( $bits['index'] ) === '' ) {
                            $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
                            $value = trim( $frame->expand( $bits['value'] ) );
                            $pageArgs[$name] = $value;
                        }
                    }

                    // Create a new context to execute the special page
                    $context = new RequestContext;
                    $context->setTitle( $title );
                    $context->setRequest( new FauxRequest( $pageArgs ) );
                    $context->setUser( $this->getUser() );
                    $context->setLanguage( $this->mOptions->getUserLangObj() );
                    $ret = SpecialPageFactory::capturePath( $title, $context );
                    if ( $ret ) {
                        $text = $context->getOutput()->getHTML();
                        $this->mOutput->addOutputPageMetadata( $context->getOutput() );
                        $found = true;
                        $isHTML = true;
                        $this->disableCache();
                    }
                } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
                    $found = false; # access denied
                    wfDebug( __METHOD__ . ": template inclusion denied for " .
                        $title->getPrefixedDBkey() . "\n" );
                } else {
                    list( $text, $title ) = $this->getTemplateDom( $title );
                    if ( $text !== false ) {
                        $found = true;
                        $isChildObj = true;
                    }
                }

                # If the title is valid but undisplayable, make a link to it
                if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
                    $text = "[[:$titleText]]";
                    $found = true;
                }
            } elseif ( $title->isTrans() ) {
                # Interwiki transclusion
                if ( $this->ot['html'] && !$forceRawInterwiki ) {
                    $text = $this->interwikiTransclude( $title, 'render' );
                    $isHTML = true;
                } else {
                    $text = $this->interwikiTransclude( $title, 'raw' );
                    # Preprocess it like a template
                    $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
                    $isChildObj = true;
                }
                $found = true;
            }

            # Do infinite loop check
            # This has to be done after redirect resolution to avoid infinite loops via redirects
            if ( !$frame->loopCheck( $title ) ) {
                $found = true;
                $text = '<span class="error">'
                    . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
                    . '</span>';
                wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
            }
            wfProfileOut( __METHOD__ . '-loadtpl' );
        }

        # If we haven't found text to substitute by now, we're done
        # Recover the source wikitext and return it
        if ( !$found ) {
            $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
            if ( $titleProfileIn ) {
                wfProfileOut( $titleProfileIn ); // template out
            }
            wfProfileOut( __METHOD__ );
            return array( 'object' => $text );
        }

        # Expand DOM-style return values in a child frame
        if ( $isChildObj ) {
            # Clean up argument array
            $newFrame = $frame->newChild( $args, $title );

            if ( $nowiki ) {
                $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
            } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
                # Expansion is eligible for the empty-frame cache
                $text = $newFrame->cachedExpand( $titleText, $text );
            } else {
                # Uncached expansion
                $text = $newFrame->expand( $text );
            }
        }
        if ( $isLocalObj && $nowiki ) {
            $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
            $isLocalObj = false;
        }

        if ( $titleProfileIn ) {
            wfProfileOut( $titleProfileIn ); // template out
        }

        # Replace raw HTML by a placeholder
        if ( $isHTML ) {
            $text = $this->insertStripItem( $text );
        } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
            # Escape nowiki-style return values
            $text = wfEscapeWikiText( $text );
        } elseif ( is_string( $text )
            && !$piece['lineStart']
            && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
        ) {
            # Bug 529: if the template begins with a table or block-level
            # element, it should be treated as beginning a new line.
            # This behavior is somewhat controversial.
            $text = "\n" . $text;
        }

        if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
            # Error, oversize inclusion
            if ( $titleText !== false ) {
                # Make a working, properly escaped link if possible (bug 23588)
                $text = "[[:$titleText]]";
            } else {
                # This will probably not be a working link, but at least it may
                # provide some hint of where the problem is
                preg_replace( '/^:/', '', $originalTitle );
                $text = "[[:$originalTitle]]";
            }
            $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
                . 'post-expand include size too large -->' );
            $this->limitationWarn( 'post-expand-template-inclusion' );
        }

        if ( $isLocalObj ) {
            $ret = array( 'object' => $text );
        } else {
            $ret = array( 'text' => $text );
        }

        wfProfileOut( __METHOD__ );
        return $ret;
    }

    public function callParserFunction( $frame, $function, array $args = array() ) {
        global $wgContLang;

        wfProfileIn( __METHOD__ );

        # Case sensitive functions
        if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
            $function = $this->mFunctionSynonyms[1][$function];
        } else {
            # Case insensitive functions
            $function = $wgContLang->lc( $function );
            if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
                $function = $this->mFunctionSynonyms[0][$function];
            } else {
                wfProfileOut( __METHOD__ );
                return array( 'found' => false );
            }
        }

        wfProfileIn( __METHOD__ . '-pfunc-' . $function );
        list( $callback, $flags ) = $this->mFunctionHooks[$function];

        # Workaround for PHP bug 35229 and similar
        if ( !is_callable( $callback ) ) {
            wfProfileOut( __METHOD__ . '-pfunc-' . $function );
            wfProfileOut( __METHOD__ );
            throw new MWException( "Tag hook for $function is not callable\n" );
        }

        $allArgs = array( &$this );
        if ( $flags & SFH_OBJECT_ARGS ) {
            # Convert arguments to PPNodes and collect for appending to $allArgs
            $funcArgs = array();
            foreach ( $args as $k => $v ) {
                if ( $v instanceof PPNode || $k === 0 ) {
                    $funcArgs[] = $v;
                } else {
                    $funcArgs[] = $this->mPreprocessor->newPartNodeArray( array( $k => $v ) )->item( 0 );
                }
            }

            # Add a frame parameter, and pass the arguments as an array
            $allArgs[] = $frame;
            $allArgs[] = $funcArgs;
        } else {
            # Convert arguments to plain text and append to $allArgs
            foreach ( $args as $k => $v ) {
                if ( $v instanceof PPNode ) {
                    $allArgs[] = trim( $frame->expand( $v ) );
                } elseif ( is_int( $k ) && $k >= 0 ) {
                    $allArgs[] = trim( $v );
                } else {
                    $allArgs[] = trim( "$k=$v" );
                }
            }
        }

        $result = call_user_func_array( $callback, $allArgs );

        # The interface for function hooks allows them to return a wikitext
        # string or an array containing the string and any flags. This mungs
        # things around to match what this method should return.
        if ( !is_array( $result ) ) {
            $result = array(
                'found' => true,
                'text' => $result,
            );
        } else {
            if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
                $result['text'] = $result[0];
            }
            unset( $result[0] );
            $result += array(
                'found' => true,
            );
        }

        $noparse = true;
        $preprocessFlags = 0;
        if ( isset( $result['noparse'] ) ) {
            $noparse = $result['noparse'];
        }
        if ( isset( $result['preprocessFlags'] ) ) {
            $preprocessFlags = $result['preprocessFlags'];
        }

        if ( !$noparse ) {
            $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
            $result['isChildObj'] = true;
        }
        wfProfileOut( __METHOD__ . '-pfunc-' . $function );
        wfProfileOut( __METHOD__ );

        return $result;
    }

    public function getTemplateDom( $title ) {
        $cacheTitle = $title;
        $titleText = $title->getPrefixedDBkey();

        if ( isset( $this->mTplRedirCache[$titleText] ) ) {
            list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
            $title = Title::makeTitle( $ns, $dbk );
            $titleText = $title->getPrefixedDBkey();
        }
        if ( isset( $this->mTplDomCache[$titleText] ) ) {
            return array( $this->mTplDomCache[$titleText], $title );
        }

        # Cache miss, go to the database
        list( $text, $title ) = $this->fetchTemplateAndTitle( $title );

        if ( $text === false ) {
            $this->mTplDomCache[$titleText] = false;
            return array( false, $title );
        }

        $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
        $this->mTplDomCache[$titleText] = $dom;

        if ( !$title->equals( $cacheTitle ) ) {
            $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
                array( $title->getNamespace(), $cdb = $title->getDBkey() );
        }

        return array( $dom, $title );
    }

    public function fetchTemplateAndTitle( $title ) {
        // Defaults to Parser::statelessFetchTemplate()
        $templateCb = $this->mOptions->getTemplateCallback();
        $stuff = call_user_func( $templateCb, $title, $this );
        $text = $stuff['text'];
        $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
        if ( isset( $stuff['deps'] ) ) {
            foreach ( $stuff['deps'] as $dep ) {
                $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
                if ( $dep['title']->equals( $this->getTitle() ) ) {
                    // If we transclude ourselves, the final result
                    // will change based on the new version of the page
                    $this->mOutput->setFlag( 'vary-revision' );
                }
            }
        }
        return array( $text, $finalTitle );
    }

    public function fetchTemplate( $title ) {
        $rv = $this->fetchTemplateAndTitle( $title );
        return $rv[0];
    }

    public static function statelessFetchTemplate( $title, $parser = false ) {
        $text = $skip = false;
        $finalTitle = $title;
        $deps = array();

        # Loop to fetch the article, with up to 1 redirect
        for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
            # Give extensions a chance to select the revision instead
            $id = false; # Assume current
            wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
                array( $parser, $title, &$skip, &$id ) );

            if ( $skip ) {
                $text = false;
                $deps[] = array(
                    'title' => $title,
                    'page_id' => $title->getArticleID(),
                    'rev_id' => null
                );
                break;
            }
            # Get the revision
            $rev = $id
                ? Revision::newFromId( $id )
                : Revision::newFromTitle( $title, false, Revision::READ_NORMAL );
            $rev_id = $rev ? $rev->getId() : 0;
            # If there is no current revision, there is no page
            if ( $id === false && !$rev ) {
                $linkCache = LinkCache::singleton();
                $linkCache->addBadLinkObj( $title );
            }

            $deps[] = array(
                'title' => $title,
                'page_id' => $title->getArticleID(),
                'rev_id' => $rev_id );
            if ( $rev && !$title->equals( $rev->getTitle() ) ) {
                # We fetched a rev from a different title; register it too...
                $deps[] = array(
                    'title' => $rev->getTitle(),
                    'page_id' => $rev->getPage(),
                    'rev_id' => $rev_id );
            }

            if ( $rev ) {
                $content = $rev->getContent();
                $text = $content ? $content->getWikitextForTransclusion() : null;

                if ( $text === false || $text === null ) {
                    $text = false;
                    break;
                }
            } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
                global $wgContLang;
                $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
                if ( !$message->exists() ) {
                    $text = false;
                    break;
                }
                $content = $message->content();
                $text = $message->plain();
            } else {
                break;
            }
            if ( !$content ) {
                break;
            }
            # Redirect?
            $finalTitle = $title;
            $title = $content->getRedirectTarget();
        }
        return array(
            'text' => $text,
            'finalTitle' => $finalTitle,
            'deps' => $deps );
    }

    public function fetchFile( $title, $options = array() ) {
        $res = $this->fetchFileAndTitle( $title, $options );
        return $res[0];
    }

    public function fetchFileAndTitle( $title, $options = array() ) {
        $file = $this->fetchFileNoRegister( $title, $options );

        $time = $file ? $file->getTimestamp() : false;
        $sha1 = $file ? $file->getSha1() : false;
        # Register the file as a dependency...
        $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
        if ( $file && !$title->equals( $file->getTitle() ) ) {
            # Update fetched file title
            $title = $file->getTitle();
            $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
        }
        return array( $file, $title );
    }

    protected function fetchFileNoRegister( $title, $options = array() ) {
        if ( isset( $options['broken'] ) ) {
            $file = false; // broken thumbnail forced by hook
        } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
            $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
        } else { // get by (name,timestamp)
            $file = wfFindFile( $title, $options );
        }
        return $file;
    }

    public function interwikiTransclude( $title, $action ) {
        global $wgEnableScaryTranscluding;

        if ( !$wgEnableScaryTranscluding ) {
            return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
        }

        $url = $title->getFullURL( array( 'action' => $action ) );

        if ( strlen( $url ) > 255 ) {
            return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
        }
        return $this->fetchScaryTemplateMaybeFromCache( $url );
    }

    public function fetchScaryTemplateMaybeFromCache( $url ) {
        global $wgTranscludeCacheExpiry;
        $dbr = wfGetDB( DB_SLAVE );
        $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
        $obj = $dbr->selectRow( 'transcache', array( 'tc_time', 'tc_contents' ),
                array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
        if ( $obj ) {
            return $obj->tc_contents;
        }

        $req = MWHttpRequest::factory( $url );
        $status = $req->execute(); // Status object
        if ( $status->isOK() ) {
            $text = $req->getContent();
        } elseif ( $req->getStatus() != 200 ) {
            // Though we failed to fetch the content, this status is useless.
            return wfMessage( 'scarytranscludefailed-httpstatus' )
                ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
        } else {
            return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
        }

        $dbw = wfGetDB( DB_MASTER );
        $dbw->replace( 'transcache', array( 'tc_url' ), array(
            'tc_url' => $url,
            'tc_time' => $dbw->timestamp( time() ),
            'tc_contents' => $text
        ) );
        return $text;
    }

    public function argSubstitution( $piece, $frame ) {
        wfProfileIn( __METHOD__ );

        $error = false;
        $parts = $piece['parts'];
        $nameWithSpaces = $frame->expand( $piece['title'] );
        $argName = trim( $nameWithSpaces );
        $object = false;
        $text = $frame->getArgument( $argName );
        if ( $text === false && $parts->getLength() > 0
            && ( $this->ot['html']
                || $this->ot['pre']
                || ( $this->ot['wiki'] && $frame->isTemplate() )
            )
        ) {
            # No match in frame, use the supplied default
            $object = $parts->item( 0 )->getChildren();
        }
        if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
            $error = '<!-- WARNING: argument omitted, expansion size too large -->';
            $this->limitationWarn( 'post-expand-template-argument' );
        }

        if ( $text === false && $object === false ) {
            # No match anywhere
            $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
        }
        if ( $error !== false ) {
            $text .= $error;
        }
        if ( $object !== false ) {
            $ret = array( 'object' => $object );
        } else {
            $ret = array( 'text' => $text );
        }

        wfProfileOut( __METHOD__ );
        return $ret;
    }

    public function extensionSubstitution( $params, $frame ) {
        $name = $frame->expand( $params['name'] );
        $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
        $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
        $marker = "{$this->mUniqPrefix}-$name-"
            . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;

        $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
            ( $this->ot['html'] || $this->ot['pre'] );
        if ( $isFunctionTag ) {
            $markerType = 'none';
        } else {
            $markerType = 'general';
        }
        if ( $this->ot['html'] || $isFunctionTag ) {
            $name = strtolower( $name );
            $attributes = Sanitizer::decodeTagAttributes( $attrText );
            if ( isset( $params['attributes'] ) ) {
                $attributes = $attributes + $params['attributes'];
            }

            if ( isset( $this->mTagHooks[$name] ) ) {
                # Workaround for PHP bug 35229 and similar
                if ( !is_callable( $this->mTagHooks[$name] ) ) {
                    throw new MWException( "Tag hook for $name is not callable\n" );
                }
                $output = call_user_func_array( $this->mTagHooks[$name],
                    array( $content, $attributes, $this, $frame ) );
            } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
                list( $callback, ) = $this->mFunctionTagHooks[$name];
                if ( !is_callable( $callback ) ) {
                    throw new MWException( "Tag hook for $name is not callable\n" );
                }

                $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
            } else {
                $output = '<span class="error">Invalid tag extension name: ' .
                    htmlspecialchars( $name ) . '</span>';
            }

            if ( is_array( $output ) ) {
                # Extract flags to local scope (to override $markerType)
                $flags = $output;
                $output = $flags[0];
                unset( $flags[0] );
                extract( $flags );
            }
        } else {
            if ( is_null( $attrText ) ) {
                $attrText = '';
            }
            if ( isset( $params['attributes'] ) ) {
                foreach ( $params['attributes'] as $attrName => $attrValue ) {
                    $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
                        htmlspecialchars( $attrValue ) . '"';
                }
            }
            if ( $content === null ) {
                $output = "<$name$attrText/>";
            } else {
                $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
                $output = "<$name$attrText>$content$close";
            }
        }

        if ( $markerType === 'none' ) {
            return $output;
        } elseif ( $markerType === 'nowiki' ) {
            $this->mStripState->addNoWiki( $marker, $output );
        } elseif ( $markerType === 'general' ) {
            $this->mStripState->addGeneral( $marker, $output );
        } else {
            throw new MWException( __METHOD__ . ': invalid marker type' );
        }
        return $marker;
    }

    public function incrementIncludeSize( $type, $size ) {
        if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
            return false;
        } else {
            $this->mIncludeSizes[$type] += $size;
            return true;
        }
    }

    public function incrementExpensiveFunctionCount() {
        $this->mExpensiveFunctionCount++;
        return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
    }

    public function doDoubleUnderscore( $text ) {
        wfProfileIn( __METHOD__ );

        # The position of __TOC__ needs to be recorded
        $mw = MagicWord::get( 'toc' );
        if ( $mw->match( $text ) ) {
            $this->mShowToc = true;
            $this->mForceTocPosition = true;

            # Set a placeholder. At the end we'll fill it in with the TOC.
            $text = $mw->replace( '<!--MWTOC-->', $text, 1 );

            # Only keep the first one.
            $text = $mw->replace( '', $text );
        }

        # Now match and remove the rest of them
        $mwa = MagicWord::getDoubleUnderscoreArray();
        $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );

        if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
            $this->mOutput->mNoGallery = true;
        }
        if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
            $this->mShowToc = false;
        }
        if ( isset( $this->mDoubleUnderscores['hiddencat'] )
            && $this->mTitle->getNamespace() == NS_CATEGORY
        ) {
            $this->addTrackingCategory( 'hidden-category-category' );
        }
        # (bug 8068) Allow control over whether robots index a page.
        #
        # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here!  This
        # is not desirable, the last one on the page should win.
        if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
            $this->mOutput->setIndexPolicy( 'noindex' );
            $this->addTrackingCategory( 'noindex-category' );
        }
        if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
            $this->mOutput->setIndexPolicy( 'index' );
            $this->addTrackingCategory( 'index-category' );
        }

        # Cache all double underscores in the database
        foreach ( $this->mDoubleUnderscores as $key => $val ) {
            $this->mOutput->setProperty( $key, '' );
        }

        wfProfileOut( __METHOD__ );
        return $text;
    }

    public function addTrackingCategory( $msg ) {
        if ( $this->mTitle->getNamespace() === NS_SPECIAL ) {
            wfDebug( __METHOD__ . ": Not adding tracking category $msg to special page!\n" );
            return false;
        }
        // Important to parse with correct title (bug 31469)
        $cat = wfMessage( $msg )
            ->title( $this->getTitle() )
            ->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->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
            return true;
        } else {
            wfDebug( __METHOD__ . ": [[MediaWiki:$msg]] is not a valid title!\n" );
            return false;
        }
    }

    public function formatHeadings( $text, $origText, $isMain = true ) {
        global $wgMaxTocLevel, $wgExperimentalHtmlIds;

        # Inhibit editsection links if requested in the page
        if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
            $maybeShowEditLink = $showEditLink = false;
        } else {
            $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
            $showEditLink = $this->mOptions->getEditSection();
        }
        if ( $showEditLink ) {
            $this->mOutput->setEditSectionTokens( true );
        }

        # Get all headlines for numbering them and adding funky stuff like [edit]
        # links - this is for later, but we need the number of headlines right now
        $matches = array();
        $numMatches = preg_match_all(
            '/<H(?P<level>[1-6])(?P<attrib>.*?' . '>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
            $text,
            $matches
        );

        # if there are fewer than 4 headlines in the article, do not show TOC
        # unless it's been explicitly enabled.
        $enoughToc = $this->mShowToc &&
            ( ( $numMatches >= 4 ) || $this->mForceTocPosition );

        # Allow user to stipulate that a page should have a "new section"
        # link added via __NEWSECTIONLINK__
        if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
            $this->mOutput->setNewSection( true );
        }

        # Allow user to remove the "new section"
        # link via __NONEWSECTIONLINK__
        if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
            $this->mOutput->hideNewSection( true );
        }

        # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
        # override above conditions and always show TOC above first header
        if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
            $this->mShowToc = true;
            $enoughToc = true;
        }

        # headline counter
        $headlineCount = 0;
        $numVisible = 0;

        # Ugh .. the TOC should have neat indentation levels which can be
        # passed to the skin functions. These are determined here
        $toc = '';
        $full = '';
        $head = array();
        $sublevelCount = array();
        $levelCount = array();
        $level = 0;
        $prevlevel = 0;
        $toclevel = 0;
        $prevtoclevel = 0;
        $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
        $baseTitleText = $this->mTitle->getPrefixedDBkey();
        $oldType = $this->mOutputType;
        $this->setOutputType( self::OT_WIKI );
        $frame = $this->getPreprocessor()->newFrame();
        $root = $this->preprocessToDom( $origText );
        $node = $root->getFirstChild();
        $byteOffset = 0;
        $tocraw = array();
        $refers = array();

        foreach ( $matches[3] as $headline ) {
            $isTemplate = false;
            $titleText = false;
            $sectionIndex = false;
            $numbering = '';
            $markerMatches = array();
            if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
                $serial = $markerMatches[1];
                list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
                $isTemplate = ( $titleText != $baseTitleText );
                $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
            }

            if ( $toclevel ) {
                $prevlevel = $level;
            }
            $level = $matches[1][$headlineCount];

            if ( $level > $prevlevel ) {
                # Increase TOC level
                $toclevel++;
                $sublevelCount[$toclevel] = 0;
                if ( $toclevel < $wgMaxTocLevel ) {
                    $prevtoclevel = $toclevel;
                    $toc .= Linker::tocIndent();
                    $numVisible++;
                }
            } elseif ( $level < $prevlevel && $toclevel > 1 ) {
                # Decrease TOC level, find level to jump to

                for ( $i = $toclevel; $i > 0; $i-- ) {
                    if ( $levelCount[$i] == $level ) {
                        # Found last matching level
                        $toclevel = $i;
                        break;
                    } elseif ( $levelCount[$i] < $level ) {
                        # Found first matching level below current level
                        $toclevel = $i + 1;
                        break;
                    }
                }
                if ( $i == 0 ) {
                    $toclevel = 1;
                }
                if ( $toclevel < $wgMaxTocLevel ) {
                    if ( $prevtoclevel < $wgMaxTocLevel ) {
                        # Unindent only if the previous toc level was shown :p
                        $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
                        $prevtoclevel = $toclevel;
                    } else {
                        $toc .= Linker::tocLineEnd();
                    }
                }
            } else {
                # No change in level, end TOC line
                if ( $toclevel < $wgMaxTocLevel ) {
                    $toc .= Linker::tocLineEnd();
                }
            }

            $levelCount[$toclevel] = $level;

            # count number of headlines for each level
            $sublevelCount[$toclevel]++;
            $dot = 0;
            for ( $i = 1; $i <= $toclevel; $i++ ) {
                if ( !empty( $sublevelCount[$i] ) ) {
                    if ( $dot ) {
                        $numbering .= '.';
                    }
                    $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
                    $dot = 1;
                }
            }

            # The safe header is a version of the header text safe to use for links

            # Remove link placeholders by the link text.
            #     <!--LINK number-->
            # turns into
            #     link text with suffix
            # Do this before unstrip since link text can contain strip markers
            $safeHeadline = $this->replaceLinkHoldersText( $headline );

            # Avoid insertion of weird stuff like <math> by expanding the relevant sections
            $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );

            # Strip out HTML (first regex removes any tag not allowed)
            # Allowed tags are:
            # * <sup> and <sub> (bug 8393)
            # * <i> (bug 26375)
            # * <b> (r105284)
            # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
            #
            # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
            # to allow setting directionality in toc items.
            $tocline = preg_replace(
                array(
                    '#<(?!/?(span|sup|sub|i|b)(?: [^>]*)?>).*?' . '>#',
                    '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|i|b))(?: .*?)?' . '>#'
                ),
                array( '', '<$1>' ),
                $safeHeadline
            );
            $tocline = trim( $tocline );

            # For the anchor, strip out HTML-y stuff period
            $safeHeadline = preg_replace( '/<.*?' . '>/', '', $safeHeadline );
            $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );

            # Save headline for section edit hint before it's escaped
            $headlineHint = $safeHeadline;

            if ( $wgExperimentalHtmlIds ) {
                # For reverse compatibility, provide an id that's
                # HTML4-compatible, like we used to.
                #
                # It may be worth noting, academically, that it's possible for
                # the legacy anchor to conflict with a non-legacy headline
                # anchor on the page.  In this case likely the "correct" thing
                # would be to either drop the legacy anchors or make sure
                # they're numbered first.  However, this would require people
                # to type in section names like "abc_.D7.93.D7.90.D7.A4"
                # manually, so let's not bother worrying about it.
                $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
                    array( 'noninitial', 'legacy' ) );
                $safeHeadline = Sanitizer::escapeId( $safeHeadline );

                if ( $legacyHeadline == $safeHeadline ) {
                    # No reason to have both (in fact, we can't)
                    $legacyHeadline = false;
                }
            } else {
                $legacyHeadline = false;
                $safeHeadline = Sanitizer::escapeId( $safeHeadline,
                    'noninitial' );
            }

            # HTML names must be case-insensitively unique (bug 10721).
            # This does not apply to Unicode characters per
            # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
            # @todo FIXME: We may be changing them depending on the current locale.
            $arrayKey = strtolower( $safeHeadline );
            if ( $legacyHeadline === false ) {
                $legacyArrayKey = false;
            } else {
                $legacyArrayKey = strtolower( $legacyHeadline );
            }

            # count how many in assoc. array so we can track dupes in anchors
            if ( isset( $refers[$arrayKey] ) ) {
                $refers[$arrayKey]++;
            } else {
                $refers[$arrayKey] = 1;
            }
            if ( isset( $refers[$legacyArrayKey] ) ) {
                $refers[$legacyArrayKey]++;
            } else {
                $refers[$legacyArrayKey] = 1;
            }

            # Don't number the heading if it is the only one (looks silly)
            if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
                # the two are different if the line contains a link
                $headline = Html::element(
                    'span',
                    array( 'class' => 'mw-headline-number' ),
                    $numbering
                ) . ' ' . $headline;
            }

            # Create the anchor for linking from the TOC to the section
            $anchor = $safeHeadline;
            $legacyAnchor = $legacyHeadline;
            if ( $refers[$arrayKey] > 1 ) {
                $anchor .= '_' . $refers[$arrayKey];
            }
            if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
                $legacyAnchor .= '_' . $refers[$legacyArrayKey];
            }
            if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
                $toc .= Linker::tocLine( $anchor, $tocline,
                    $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
            }

            # Add the section to the section tree
            # Find the DOM node for this header
            $noOffset = ( $isTemplate || $sectionIndex === false );
            while ( $node && !$noOffset ) {
                if ( $node->getName() === 'h' ) {
                    $bits = $node->splitHeading();
                    if ( $bits['i'] == $sectionIndex ) {
                        break;
                    }
                }
                $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
                    $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
                $node = $node->getNextSibling();
            }
            $tocraw[] = array(
                'toclevel' => $toclevel,
                'level' => $level,
                'line' => $tocline,
                'number' => $numbering,
                'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
                'fromtitle' => $titleText,
                'byteoffset' => ( $noOffset ? null : $byteOffset ),
                'anchor' => $anchor,
            );

            # give headline the correct <h#> tag
            if ( $maybeShowEditLink && $sectionIndex !== false ) {
                // Output edit section links as markers with styles that can be customized by skins
                if ( $isTemplate ) {
                    # Put a T flag in the section identifier, to indicate to extractSections()
                    # that sections inside <includeonly> should be counted.
                    $editsectionPage = $titleText;
                    $editsectionSection = "T-$sectionIndex";
                    $editsectionContent = null;
                } else {
                    $editsectionPage = $this->mTitle->getPrefixedText();
                    $editsectionSection = $sectionIndex;
                    $editsectionContent = $headlineHint;
                }
                // We use a bit of pesudo-xml for editsection markers. The
                // language converter is run later on. Using a UNIQ style marker
                // leads to the converter screwing up the tokens when it
                // converts stuff. And trying to insert strip tags fails too. At
                // this point all real inputted tags have already been escaped,
                // so we don't have to worry about a user trying to input one of
                // these markers directly. We use a page and section attribute
                // to stop the language converter from converting these
                // important bits of data, but put the headline hint inside a
                // content block because the language converter is supposed to
                // be able to convert that piece of data.
                // Gets replaced with html in ParserOutput::getText
                $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
                $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
                if ( $editsectionContent !== null ) {
                    $editlink .= '>' . $editsectionContent . '</mw:editsection>';
                } else {
                    $editlink .= '/>';
                }
            } else {
                $editlink = '';
            }
            $head[$headlineCount] = Linker::makeHeadline( $level,
                $matches['attrib'][$headlineCount], $anchor, $headline,
                $editlink, $legacyAnchor );

            $headlineCount++;
        }

        $this->setOutputType( $oldType );

        # Never ever show TOC if no headers
        if ( $numVisible < 1 ) {
            $enoughToc = false;
        }

        if ( $enoughToc ) {
            if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
                $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
            }
            $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
            $this->mOutput->setTOCHTML( $toc );
            $toc = self::TOC_START . $toc . self::TOC_END;
            $this->mOutput->addModules( 'mediawiki.toc' );
        }

        if ( $isMain ) {
            $this->mOutput->setSections( $tocraw );
        }

        # split up and insert constructed headlines
        $blocks = preg_split( '/<H[1-6].*?' . '>[\s\S]*?<\/H[1-6]>/i', $text );
        $i = 0;

        // build an array of document sections
        $sections = array();
        foreach ( $blocks as $block ) {
            // $head is zero-based, sections aren't.
            if ( empty( $head[$i - 1] ) ) {
                $sections[$i] = $block;
            } else {
                $sections[$i] = $head[$i - 1] . $block;
            }

            wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );

            $i++;
        }

        if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
            // append the TOC at the beginning
            // Top anchor now in skin
            $sections[0] = $sections[0] . $toc . "\n";
        }

        $full .= join( '', $sections );

        if ( $this->mForceTocPosition ) {
            return str_replace( '<!--MWTOC-->', $toc, $full );
        } else {
            return $full;
        }
    }

    public function preSaveTransform( $text, Title $title, User $user,
        ParserOptions $options, $clearState = true
    ) {
        if ( $clearState ) {
            $magicScopeVariable = $this->lock();
        }
        $this->startParse( $title, $options, self::OT_WIKI, $clearState );
        $this->setUser( $user );

        $pairs = array(
            "\r\n" => "\n",
        );
        $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
        if ( $options->getPreSaveTransform() ) {
            $text = $this->pstPass2( $text, $user );
        }
        $text = $this->mStripState->unstripBoth( $text );

        $this->setUser( null ); #Reset

        return $text;
    }

    private function pstPass2( $text, $user ) {
        global $wgContLang;

        # Note: This is the timestamp saved as hardcoded wikitext to
        # the database, we use $wgContLang here in order to give
        # everyone the same signature and use the default one rather
        # than the one selected in each user's preferences.
        # (see also bug 12815)
        $ts = $this->mOptions->getTimestamp();
        $timestamp = MWTimestamp::getLocalInstance( $ts );
        $ts = $timestamp->format( 'YmdHis' );
        $tzMsg = $timestamp->format( 'T' );  # might vary on DST changeover!

        # Allow translation of timezones through wiki. format() can return
        # whatever crap the system uses, localised or not, so we cannot
        # ship premade translations.
        $key = 'timezone-' . strtolower( trim( $tzMsg ) );
        $msg = wfMessage( $key )->inContentLanguage();
        if ( $msg->exists() ) {
            $tzMsg = $msg->text();
        }

        $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";

        # Variable replacement
        # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
        $text = $this->replaceVariables( $text );

        # This works almost by chance, as the replaceVariables are done before the getUserSig(),
        # which may corrupt this parser instance via its wfMessage()->text() call-

        # Signatures
        $sigText = $this->getUserSig( $user );
        $text = strtr( $text, array(
            '~~~~~' => $d,
            '~~~~' => "$sigText $d",
            '~~~' => $sigText
        ) );

        # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
        $tc = '[' . Title::legalChars() . ']';
        $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!

        // [[ns:page (context)|]]
        $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
        // [[ns:page（context）|]] (double-width brackets, added in r40257)
        $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
        // [[ns:page (context), context|]] (using either single or double-width comma)
        $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
        // [[|page]] (reverse pipe trick: add context from page title)
        $p2 = "/\[\[\\|($tc+)]]/";

        # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
        $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
        $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
        $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );

        $t = $this->mTitle->getText();
        $m = array();
        if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
            $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
        } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
            $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
        } else {
            # if there's no context, don't bother duplicating the title
            $text = preg_replace( $p2, '[[\\1]]', $text );
        }

        # Trim trailing whitespace
        $text = rtrim( $text );

        return $text;
    }

    public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
        global $wgMaxSigChars;

        $username = $user->getName();

        # If not given, retrieve from the user object.
        if ( $nickname === false ) {
            $nickname = $user->getOption( 'nickname' );
        }

        if ( is_null( $fancySig ) ) {
            $fancySig = $user->getBoolOption( 'fancysig' );
        }

        $nickname = $nickname == null ? $username : $nickname;

        if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
            $nickname = $username;
            wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
        } elseif ( $fancySig !== false ) {
            # Sig. might contain markup; validate this
            if ( $this->validateSig( $nickname ) !== false ) {
                # Validated; clean up (if needed) and return it
                return $this->cleanSig( $nickname, true );
            } else {
                # Failed to validate; fall back to the default
                $nickname = $username;
                wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
            }
        }

        # Make sure nickname doesnt get a sig in a sig
        $nickname = self::cleanSigInSig( $nickname );

        # If we're still here, make it a link to the user page
        $userText = wfEscapeWikiText( $username );
        $nickText = wfEscapeWikiText( $nickname );
        $msgName = $user->isAnon() ? 'signature-anon' : 'signature';

        return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
            ->title( $this->getTitle() )->text();
    }

    public function validateSig( $text ) {
        return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
    }

    public function cleanSig( $text, $parsing = false ) {
        if ( !$parsing ) {
            global $wgTitle;
            $magicScopeVariable = $this->lock();
            $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
        }

        # Option to disable this feature
        if ( !$this->mOptions->getCleanSignatures() ) {
            return $text;
        }

        # @todo FIXME: Regex doesn't respect extension tags or nowiki
        #  => Move this logic to braceSubstitution()
        $substWord = MagicWord::get( 'subst' );
        $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
        $substText = '{{' . $substWord->getSynonym( 0 );

        $text = preg_replace( $substRegex, $substText, $text );
        $text = self::cleanSigInSig( $text );
        $dom = $this->preprocessToDom( $text );
        $frame = $this->getPreprocessor()->newFrame();
        $text = $frame->expand( $dom );

        if ( !$parsing ) {
            $text = $this->mStripState->unstripBoth( $text );
        }

        return $text;
    }

    public static function cleanSigInSig( $text ) {
        $text = preg_replace( '/~{3,5}/', '', $text );
        return $text;
    }

    public function startExternalParse( Title $title = null, ParserOptions $options,
        $outputType, $clearState = true
    ) {
        $this->startParse( $title, $options, $outputType, $clearState );
    }

    private function startParse( Title $title = null, ParserOptions $options,
        $outputType, $clearState = true
    ) {
        $this->setTitle( $title );
        $this->mOptions = $options;
        $this->setOutputType( $outputType );
        if ( $clearState ) {
            $this->clearState();
        }
    }

    public function transformMsg( $text, $options, $title = null ) {
        static $executing = false;

        # Guard against infinite recursion
        if ( $executing ) {
            return $text;
        }
        $executing = true;

        wfProfileIn( __METHOD__ );
        if ( !$title ) {
            global $wgTitle;
            $title = $wgTitle;
        }

        $text = $this->preprocess( $text, $title, $options );

        $executing = false;
        wfProfileOut( __METHOD__ );
        return $text;
    }

    public function setHook( $tag, $callback ) {
        $tag = strtolower( $tag );
        if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
            throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
        }
        $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
        $this->mTagHooks[$tag] = $callback;
        if ( !in_array( $tag, $this->mStripList ) ) {
            $this->mStripList[] = $tag;
        }

        return $oldVal;
    }

    public function setTransparentTagHook( $tag, $callback ) {
        $tag = strtolower( $tag );
        if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
            throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
        }
        $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
        $this->mTransparentTagHooks[$tag] = $callback;

        return $oldVal;
    }

    public function clearTagHooks() {
        $this->mTagHooks = array();
        $this->mFunctionTagHooks = array();
        $this->mStripList = $this->mDefaultStripList;
    }

    public function setFunctionHook( $id, $callback, $flags = 0 ) {
        global $wgContLang;

        $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
        $this->mFunctionHooks[$id] = array( $callback, $flags );

        # Add to function cache
        $mw = MagicWord::get( $id );
        if ( !$mw ) {
            throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
        }

        $synonyms = $mw->getSynonyms();
        $sensitive = intval( $mw->isCaseSensitive() );

        foreach ( $synonyms as $syn ) {
            # Case
            if ( !$sensitive ) {
                $syn = $wgContLang->lc( $syn );
            }
            # Add leading hash
            if ( !( $flags & SFH_NO_HASH ) ) {
                $syn = '#' . $syn;
            }
            # Remove trailing colon
            if ( substr( $syn, -1, 1 ) === ':' ) {
                $syn = substr( $syn, 0, -1 );
            }
            $this->mFunctionSynonyms[$sensitive][$syn] = $id;
        }
        return $oldVal;
    }

    public function getFunctionHooks() {
        return array_keys( $this->mFunctionHooks );
    }

    public function setFunctionTagHook( $tag, $callback, $flags ) {
        $tag = strtolower( $tag );
        if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
            throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
        }
        $old = isset( $this->mFunctionTagHooks[$tag] ) ?
            $this->mFunctionTagHooks[$tag] : null;
        $this->mFunctionTagHooks[$tag] = array( $callback, $flags );

        if ( !in_array( $tag, $this->mStripList ) ) {
            $this->mStripList[] = $tag;
        }

        return $old;
    }

    public function replaceLinkHolders( &$text, $options = 0 ) {
        return $this->mLinkHolders->replace( $text );
    }

    public function replaceLinkHoldersText( $text ) {
        return $this->mLinkHolders->replaceText( $text );
    }

    public function renderImageGallery( $text, $params ) {
        wfProfileIn( __METHOD__ );

        $mode = false;
        if ( isset( $params['mode'] ) ) {
            $mode = $params['mode'];
        }

        try {
            $ig = ImageGalleryBase::factory( $mode );
        } catch ( MWException $e ) {
            // If invalid type set, fallback to default.
            $ig = ImageGalleryBase::factory( false );
        }

        $ig->setContextTitle( $this->mTitle );
        $ig->setShowBytes( false );
        $ig->setShowFilename( false );
        $ig->setParser( $this );
        $ig->setHideBadImages();
        $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );

        if ( isset( $params['showfilename'] ) ) {
            $ig->setShowFilename( true );
        } else {
            $ig->setShowFilename( false );
        }
        if ( isset( $params['caption'] ) ) {
            $caption = $params['caption'];
            $caption = htmlspecialchars( $caption );
            $caption = $this->replaceInternalLinks( $caption );
            $ig->setCaptionHtml( $caption );
        }
        if ( isset( $params['perrow'] ) ) {
            $ig->setPerRow( $params['perrow'] );
        }
        if ( isset( $params['widths'] ) ) {
            $ig->setWidths( $params['widths'] );
        }
        if ( isset( $params['heights'] ) ) {
            $ig->setHeights( $params['heights'] );
        }
        $ig->setAdditionalOptions( $params );

        wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );

        $lines = StringUtils::explode( "\n", $text );
        foreach ( $lines as $line ) {
            # match lines like these:
            # Image:someimage.jpg|This is some image
            $matches = array();
            preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
            # Skip empty lines
            if ( count( $matches ) == 0 ) {
                continue;
            }

            if ( strpos( $matches[0], '%' ) !== false ) {
                $matches[1] = rawurldecode( $matches[1] );
            }
            $title = Title::newFromText( $matches[1], NS_FILE );
            if ( is_null( $title ) ) {
                # Bogus title. Ignore these so we don't bomb out later.
                continue;
            }

            # We need to get what handler the file uses, to figure out parameters.
            # Note, a hook can overide the file name, and chose an entirely different
            # file (which potentially could be of a different type and have different handler).
            $options = array();
            $descQuery = false;
            wfRunHooks( 'BeforeParserFetchFileAndTitle',
                array( $this, $title, &$options, &$descQuery ) );
            # Don't register it now, as ImageGallery does that later.
            $file = $this->fetchFileNoRegister( $title, $options );
            $handler = $file ? $file->getHandler() : false;

            wfProfileIn( __METHOD__ . '-getMagicWord' );
            $paramMap = array(
                'img_alt' => 'gallery-internal-alt',
                'img_link' => 'gallery-internal-link',
            );
            if ( $handler ) {
                $paramMap = $paramMap + $handler->getParamMap();
                // We don't want people to specify per-image widths.
                // Additionally the width parameter would need special casing anyhow.
                unset( $paramMap['img_width'] );
            }

            $mwArray = new MagicWordArray( array_keys( $paramMap ) );
            wfProfileOut( __METHOD__ . '-getMagicWord' );

            $label = '';
            $alt = '';
            $link = '';
            $handlerOptions = array();
            if ( isset( $matches[3] ) ) {
                // look for an |alt= definition while trying not to break existing
                // captions with multiple pipes (|) in it, until a more sensible grammar
                // is defined for images in galleries

                // FIXME: Doing recursiveTagParse at this stage, and the trim before
                // splitting on '|' is a bit odd, and different from makeImage.
                $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
                $parameterMatches = StringUtils::explode( '|', $matches[3] );

                foreach ( $parameterMatches as $parameterMatch ) {
                    list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
                    if ( $magicName ) {
                        $paramName = $paramMap[$magicName];

                        switch ( $paramName ) {
                        case 'gallery-internal-alt':
                            $alt = $this->stripAltText( $match, false );
                            break;
                        case 'gallery-internal-link':
                            $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
                            $chars = self::EXT_LINK_URL_CLASS;
                            $prots = $this->mUrlProtocols;
                            //check to see if link matches an absolute url, if not then it must be a wiki link.
                            if ( preg_match( "/^($prots)$chars+$/u", $linkValue ) ) {
                                $link = $linkValue;
                            } else {
                                $localLinkTitle = Title::newFromText( $linkValue );
                                if ( $localLinkTitle !== null ) {
                                    $link = $localLinkTitle->getLinkURL();
                                }
                            }
                            break;
                        default:
                            // Must be a handler specific parameter.
                            if ( $handler->validateParam( $paramName, $match ) ) {
                                $handlerOptions[$paramName] = $match;
                            } else {
                                // Guess not. Append it to the caption.
                                wfDebug( "$parameterMatch failed parameter validation\n" );
                                $label .= '|' . $parameterMatch;
                            }
                        }

                    } else {
                        // concatenate all other pipes
                        $label .= '|' . $parameterMatch;
                    }
                }
                // remove the first pipe
                $label = substr( $label, 1 );
            }

            $ig->add( $title, $label, $alt, $link, $handlerOptions );
        }
        $html = $ig->toHTML();
        wfRunHooks( 'AfterParserFetchFileAndTitle', array( $this, $ig, &$html ) );
        wfProfileOut( __METHOD__ );
        return $html;
    }

    public function getImageParams( $handler ) {
        if ( $handler ) {
            $handlerClass = get_class( $handler );
        } else {
            $handlerClass = '';
        }
        if ( !isset( $this->mImageParams[$handlerClass] ) ) {
            # Initialise static lists
            static $internalParamNames = array(
                'horizAlign' => array( 'left', 'right', 'center', 'none' ),
                'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
                    'bottom', 'text-bottom' ),
                'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
                    'upright', 'border', 'link', 'alt', 'class' ),
            );
            static $internalParamMap;
            if ( !$internalParamMap ) {
                $internalParamMap = array();
                foreach ( $internalParamNames as $type => $names ) {
                    foreach ( $names as $name ) {
                        $magicName = str_replace( '-', '_', "img_$name" );
                        $internalParamMap[$magicName] = array( $type, $name );
                    }
                }
            }

            # Add handler params
            $paramMap = $internalParamMap;
            if ( $handler ) {
                $handlerParamMap = $handler->getParamMap();
                foreach ( $handlerParamMap as $magic => $paramName ) {
                    $paramMap[$magic] = array( 'handler', $paramName );
                }
            }
            $this->mImageParams[$handlerClass] = $paramMap;
            $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
        }
        return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
    }

    public function makeImage( $title, $options, $holders = false ) {
        # Check if the options text is of the form "options|alt text"
        # Options are:
        #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
        #  * left       no resizing, just left align. label is used for alt= only
        #  * right      same, but right aligned
        #  * none       same, but not aligned
        #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
        #  * center     center the image
        #  * frame      Keep original image size, no magnify-button.
        #  * framed     Same as "frame"
        #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
        #  * upright    reduce width for upright images, rounded to full __0 px
        #  * border     draw a 1px border around the image
        #  * alt        Text for HTML alt attribute (defaults to empty)
        #  * class      Set a class for img node
        #  * link       Set the target of the image link. Can be external, interwiki, or local
        # vertical-align values (no % or length right now):
        #  * baseline
        #  * sub
        #  * super
        #  * top
        #  * text-top
        #  * middle
        #  * bottom
        #  * text-bottom

        $parts = StringUtils::explode( "|", $options );

        # Give extensions a chance to select the file revision for us
        $options = array();
        $descQuery = false;
        wfRunHooks( 'BeforeParserFetchFileAndTitle',
            array( $this, $title, &$options, &$descQuery ) );
        # Fetch and register the file (file title may be different via hooks)
        list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );

        # Get parameter map
        $handler = $file ? $file->getHandler() : false;

        list( $paramMap, $mwArray ) = $this->getImageParams( $handler );

        if ( !$file ) {
            $this->addTrackingCategory( 'broken-file-category' );
        }

        # Process the input parameters
        $caption = '';
        $params = array( 'frame' => array(), 'handler' => array(),
            'horizAlign' => array(), 'vertAlign' => array() );
        $seenformat = false;
        foreach ( $parts as $part ) {
            $part = trim( $part );
            list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
            $validated = false;
            if ( isset( $paramMap[$magicName] ) ) {
                list( $type, $paramName ) = $paramMap[$magicName];

                # Special case; width and height come in one variable together
                if ( $type === 'handler' && $paramName === 'width' ) {
                    $parsedWidthParam = $this->parseWidthParam( $value );
                    if ( isset( $parsedWidthParam['width'] ) ) {
                        $width = $parsedWidthParam['width'];
                        if ( $handler->validateParam( 'width', $width ) ) {
                            $params[$type]['width'] = $width;
                            $validated = true;
                        }
                    }
                    if ( isset( $parsedWidthParam['height'] ) ) {
                        $height = $parsedWidthParam['height'];
                        if ( $handler->validateParam( 'height', $height ) ) {
                            $params[$type]['height'] = $height;
                            $validated = true;
                        }
                    }
                    # else no validation -- bug 13436
                } else {
                    if ( $type === 'handler' ) {
                        # Validate handler parameter
                        $validated = $handler->validateParam( $paramName, $value );
                    } else {
                        # Validate internal parameters
                        switch ( $paramName ) {
                        case 'manualthumb':
                        case 'alt':
                        case 'class':
                            # @todo FIXME: Possibly check validity here for
                            # manualthumb? downstream behavior seems odd with
                            # missing manual thumbs.
                            $validated = true;
                            $value = $this->stripAltText( $value, $holders );
                            break;
                        case 'link':
                            $chars = self::EXT_LINK_URL_CLASS;
                            $prots = $this->mUrlProtocols;
                            if ( $value === '' ) {
                                $paramName = 'no-link';
                                $value = true;
                                $validated = true;
                            } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
                                if ( preg_match( "/^((?i)$prots)$chars+$/u", $value, $m ) ) {
                                    $paramName = 'link-url';
                                    $this->mOutput->addExternalLink( $value );
                                    if ( $this->mOptions->getExternalLinkTarget() ) {
                                        $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
                                    }
                                    $validated = true;
                                }
                            } else {
                                $linkTitle = Title::newFromText( $value );
                                if ( $linkTitle ) {
                                    $paramName = 'link-title';
                                    $value = $linkTitle;
                                    $this->mOutput->addLink( $linkTitle );
                                    $validated = true;
                                }
                            }
                            break;
                        case 'frameless':
                        case 'framed':
                        case 'thumbnail':
                            // use first appearing option, discard others.
                            $validated = ! $seenformat;
                            $seenformat = true;
                            break;
                        default:
                            # Most other things appear to be empty or numeric...
                            $validated = ( $value === false || is_numeric( trim( $value ) ) );
                        }
                    }

                    if ( $validated ) {
                        $params[$type][$paramName] = $value;
                    }
                }
            }
            if ( !$validated ) {
                $caption = $part;
            }
        }

        # Process alignment parameters
        if ( $params['horizAlign'] ) {
            $params['frame']['align'] = key( $params['horizAlign'] );
        }
        if ( $params['vertAlign'] ) {
            $params['frame']['valign'] = key( $params['vertAlign'] );
        }

        $params['frame']['caption'] = $caption;

        # Will the image be presented in a frame, with the caption below?
        $imageIsFramed = isset( $params['frame']['frame'] )
            || isset( $params['frame']['framed'] )
            || isset( $params['frame']['thumbnail'] )
            || isset( $params['frame']['manualthumb'] );

        # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
        # came to also set the caption, ordinary text after the image -- which
        # makes no sense, because that just repeats the text multiple times in
        # screen readers.  It *also* came to set the title attribute.
        #
        # Now that we have an alt attribute, we should not set the alt text to
        # equal the caption: that's worse than useless, it just repeats the
        # text.  This is the framed/thumbnail case.  If there's no caption, we
        # use the unnamed parameter for alt text as well, just for the time be-
        # ing, if the unnamed param is set and the alt param is not.
        #
        # For the future, we need to figure out if we want to tweak this more,
        # e.g., introducing a title= parameter for the title; ignoring the un-
        # named parameter entirely for images without a caption; adding an ex-
        # plicit caption= parameter and preserving the old magic unnamed para-
        # meter for BC; ...
        if ( $imageIsFramed ) { # Framed image
            if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
                # No caption or alt text, add the filename as the alt text so
                # that screen readers at least get some description of the image
                $params['frame']['alt'] = $title->getText();
            }
            # Do not set $params['frame']['title'] because tooltips don't make sense
            # for framed images
        } else { # Inline image
            if ( !isset( $params['frame']['alt'] ) ) {
                # No alt text, use the "caption" for the alt text
                if ( $caption !== '' ) {
                    $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
                } else {
                    # No caption, fall back to using the filename for the
                    # alt text
                    $params['frame']['alt'] = $title->getText();
                }
            }
            # Use the "caption" for the tooltip text
            $params['frame']['title'] = $this->stripAltText( $caption, $holders );
        }

        wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params, $this ) );

        # Linker does the rest
        $time = isset( $options['time'] ) ? $options['time'] : false;
        $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
            $time, $descQuery, $this->mOptions->getThumbSize() );

        # Give the handler a chance to modify the parser object
        if ( $handler ) {
            $handler->parserTransformHook( $this, $file );
        }

        return $ret;
    }

    protected function stripAltText( $caption, $holders ) {
        # Strip bad stuff out of the title (tooltip).  We can't just use
        # replaceLinkHoldersText() here, because if this function is called
        # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
        if ( $holders ) {
            $tooltip = $holders->replaceText( $caption );
        } else {
            $tooltip = $this->replaceLinkHoldersText( $caption );
        }

        # make sure there are no placeholders in thumbnail attributes
        # that are later expanded to html- so expand them now and
        # remove the tags
        $tooltip = $this->mStripState->unstripBoth( $tooltip );
        $tooltip = Sanitizer::stripAllTags( $tooltip );

        return $tooltip;
    }

    public function disableCache() {
        wfDebug( "Parser output marked as uncacheable.\n" );
        if ( !$this->mOutput ) {
            throw new MWException( __METHOD__ .
                " can only be called when actually parsing something" );
        }
        $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
        $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
    }

    public function attributeStripCallback( &$text, $frame = false ) {
        $text = $this->replaceVariables( $text, $frame );
        $text = $this->mStripState->unstripBoth( $text );
        return $text;
    }

    public function getTags() {
        return array_merge(
            array_keys( $this->mTransparentTagHooks ),
            array_keys( $this->mTagHooks ),
            array_keys( $this->mFunctionTagHooks )
        );
    }

    public function replaceTransparentTags( $text ) {
        $matches = array();
        $elements = array_keys( $this->mTransparentTagHooks );
        $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
        $replacements = array();

        foreach ( $matches as $marker => $data ) {
            list( $element, $content, $params, $tag ) = $data;
            $tagName = strtolower( $element );
            if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
                $output = call_user_func_array(
                    $this->mTransparentTagHooks[$tagName],
                    array( $content, $params, $this )
                );
            } else {
                $output = $tag;
            }
            $replacements[$marker] = $output;
        }
        return strtr( $text, $replacements );
    }

    private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
        global $wgTitle; # not generally used but removes an ugly failure mode

        $magicScopeVariable = $this->lock();
        $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
        $outText = '';
        $frame = $this->getPreprocessor()->newFrame();

        # Process section extraction flags
        $flags = 0;
        $sectionParts = explode( '-', $sectionId );
        $sectionIndex = array_pop( $sectionParts );
        foreach ( $sectionParts as $part ) {
            if ( $part === 'T' ) {
                $flags |= self::PTD_FOR_INCLUSION;
            }
        }

        # Check for empty input
        if ( strval( $text ) === '' ) {
            # Only sections 0 and T-0 exist in an empty document
            if ( $sectionIndex == 0 ) {
                if ( $mode === 'get' ) {
                    return '';
                } else {
                    return $newText;
                }
            } else {
                if ( $mode === 'get' ) {
                    return $newText;
                } else {
                    return $text;
                }
            }
        }

        # Preprocess the text
        $root = $this->preprocessToDom( $text, $flags );

        # <h> nodes indicate section breaks
        # They can only occur at the top level, so we can find them by iterating the root's children
        $node = $root->getFirstChild();

        # Find the target section
        if ( $sectionIndex == 0 ) {
            # Section zero doesn't nest, level=big
            $targetLevel = 1000;
        } else {
            while ( $node ) {
                if ( $node->getName() === 'h' ) {
                    $bits = $node->splitHeading();
                    if ( $bits['i'] == $sectionIndex ) {
                        $targetLevel = $bits['level'];
                        break;
                    }
                }
                if ( $mode === 'replace' ) {
                    $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
                }
                $node = $node->getNextSibling();
            }
        }

        if ( !$node ) {
            # Not found
            if ( $mode === 'get' ) {
                return $newText;
            } else {
                return $text;
            }
        }

        # Find the end of the section, including nested sections
        do {
            if ( $node->getName() === 'h' ) {
                $bits = $node->splitHeading();
                $curLevel = $bits['level'];
                if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
                    break;
                }
            }
            if ( $mode === 'get' ) {
                $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
            }
            $node = $node->getNextSibling();
        } while ( $node );

        # Write out the remainder (in replace mode only)
        if ( $mode === 'replace' ) {
            # Output the replacement text
            # Add two newlines on -- trailing whitespace in $newText is conventionally
            # stripped by the editor, so we need both newlines to restore the paragraph gap
            # Only add trailing whitespace if there is newText
            if ( $newText != "" ) {
                $outText .= $newText . "\n\n";
            }

            while ( $node ) {
                $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
                $node = $node->getNextSibling();
            }
        }

        if ( is_string( $outText ) ) {
            # Re-insert stripped tags
            $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
        }

        return $outText;
    }

    public function getSection( $text, $sectionId, $defaultText = '' ) {
        return $this->extractSections( $text, $sectionId, 'get', $defaultText );
    }

    public function replaceSection( $oldText, $sectionId, $newText ) {
        return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
    }

    public function getRevisionId() {
        return $this->mRevisionId;
    }

    public function getRevisionObject() {
        if ( !is_null( $this->mRevisionObject ) ) {
            return $this->mRevisionObject;
        }
        if ( is_null( $this->mRevisionId ) ) {
            return null;
        }

        $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
        return $this->mRevisionObject;
    }

    public function getRevisionTimestamp() {
        if ( is_null( $this->mRevisionTimestamp ) ) {
            wfProfileIn( __METHOD__ );

            global $wgContLang;

            $revObject = $this->getRevisionObject();
            $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();

            # The cryptic '' timezone parameter tells to use the site-default
            # timezone offset instead of the user settings.
            #
            # Since this value will be saved into the parser cache, served
            # to other users, and potentially even used inside links and such,
            # it needs to be consistent for all visitors.
            $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );

            wfProfileOut( __METHOD__ );
        }
        return $this->mRevisionTimestamp;
    }

    public function getRevisionUser() {
        if ( is_null( $this->mRevisionUser ) ) {
            $revObject = $this->getRevisionObject();

            # if this template is subst: the revision id will be blank,
            # so just use the current user's name
            if ( $revObject ) {
                $this->mRevisionUser = $revObject->getUserText();
            } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
                $this->mRevisionUser = $this->getUser()->getName();
            }
        }
        return $this->mRevisionUser;
    }

    public function getRevisionSize() {
        if ( is_null( $this->mRevisionSize ) ) {
            $revObject = $this->getRevisionObject();

            # if this variable is subst: the revision id will be blank,
            # so just use the parser input size, because the own substituation
            # will change the size.
            if ( $revObject ) {
                $this->mRevisionSize = $revObject->getSize();
            } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
                $this->mRevisionSize = $this->mInputSize;
            }
        }
        return $this->mRevisionSize;
    }

    public function setDefaultSort( $sort ) {
        $this->mDefaultSort = $sort;
        $this->mOutput->setProperty( 'defaultsort', $sort );
    }

    public function getDefaultSort() {
        if ( $this->mDefaultSort !== false ) {
            return $this->mDefaultSort;
        } else {
            return '';
        }
    }

    public function getCustomDefaultSort() {
        return $this->mDefaultSort;
    }

    public function guessSectionNameFromWikiText( $text ) {
        # Strip out wikitext links(they break the anchor)
        $text = $this->stripSectionName( $text );
        $text = Sanitizer::normalizeSectionNameWhitespace( $text );
        return '#' . Sanitizer::escapeId( $text, 'noninitial' );
    }

    public function guessLegacySectionNameFromWikiText( $text ) {
        # Strip out wikitext links(they break the anchor)
        $text = $this->stripSectionName( $text );
        $text = Sanitizer::normalizeSectionNameWhitespace( $text );
        return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
    }

    public function stripSectionName( $text ) {
        # Strip internal link markup
        $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
        $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );

        # Strip external link markup
        # @todo FIXME: Not tolerant to blank link text
        # I.E. [https://www.mediawiki.org] will render as [1] or something depending
        # on how many empty links there are on the page - need to figure that out.
        $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );

        # Parse wikitext quotes (italics & bold)
        $text = $this->doQuotes( $text );

        # Strip HTML tags
        $text = StringUtils::delimiterReplace( '<', '>', '', $text );
        return $text;
    }

    public function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
        $magicScopeVariable = $this->lock();
        $this->startParse( $title, $options, $outputType, true );

        $text = $this->replaceVariables( $text );
        $text = $this->mStripState->unstripBoth( $text );
        $text = Sanitizer::removeHTMLtags( $text );
        return $text;
    }

    public function testPst( $text, Title $title, ParserOptions $options ) {
        return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
    }

    public function testPreprocess( $text, Title $title, ParserOptions $options ) {
        return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
    }

    public function markerSkipCallback( $s, $callback ) {
        $i = 0;
        $out = '';
        while ( $i < strlen( $s ) ) {
            $markerStart = strpos( $s, $this->mUniqPrefix, $i );
            if ( $markerStart === false ) {
                $out .= call_user_func( $callback, substr( $s, $i ) );
                break;
            } else {
                $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
                $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
                if ( $markerEnd === false ) {
                    $out .= substr( $s, $markerStart );
                    break;
                } else {
                    $markerEnd += strlen( self::MARKER_SUFFIX );
                    $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
                    $i = $markerEnd;
                }
            }
        }
        return $out;
    }

    public function killMarkers( $text ) {
        return $this->mStripState->killMarkers( $text );
    }

    public function serializeHalfParsedText( $text ) {
        wfProfileIn( __METHOD__ );
        $data = array(
            'text' => $text,
            'version' => self::HALF_PARSED_VERSION,
            'stripState' => $this->mStripState->getSubState( $text ),
            'linkHolders' => $this->mLinkHolders->getSubArray( $text )
        );
        wfProfileOut( __METHOD__ );
        return $data;
    }

    public function unserializeHalfParsedText( $data ) {
        if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
            throw new MWException( __METHOD__ . ': invalid version' );
        }

        # First, extract the strip state.
        $texts = array( $data['text'] );
        $texts = $this->mStripState->merge( $data['stripState'], $texts );

        # Now renumber links
        $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );

        # Should be good to go.
        return $texts[0];
    }

    public function isValidHalfParsedText( $data ) {
        return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
    }

    public function parseWidthParam( $value ) {
        $parsedWidthParam = array();
        if ( $value === '' ) {
            return $parsedWidthParam;
        }
        $m = array();
        # (bug 13500) In both cases (width/height and width only),
        # permit trailing "px" for backward compatibility.
        if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
            $width = intval( $m[1] );
            $height = intval( $m[2] );
            $parsedWidthParam['width'] = $width;
            $parsedWidthParam['height'] = $height;
        } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
            $width = intval( $value );
            $parsedWidthParam['width'] = $width;
        }
        return $parsedWidthParam;
    }

    protected function lock() {
        if ( $this->mInParse ) {
            throw new MWException( "Parser state cleared while parsing. "
                . "Did you call Parser::parse recursively?" );
        }
        $this->mInParse = true;

        $that = $this;
        $recursiveCheck = new ScopedCallback( function() use ( $that ) {
            $that->mInParse = false;
        } );

        return $recursiveCheck;
    }

    public static function stripOuterParagraph( $html ) {
        $m = array();
        if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
            if ( strpos( $m[1], '</p>' ) === false ) {
                $html = $m[1];
            }
        }

        return $html;
    }

    public function getFreshParser() {
        global $wgParserConf;
        if ( $this->mInParse ) {
            return new $wgParserConf['class']( $wgParserConf );
        } else {
            return $this;
        }
    }
}