includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * preprocess()
  24  *   removes HTML comments and expands templates
  25  * cleanSig()
  26  *   Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *   Extracts sections from an article for section editing
  29  * getPreloadText()
  30  *   Removes <noinclude> sections, and <includeonly> tags.
  31  *
  32  * Globals used:
  33  *    objects:   $wgLang, $wgContLang
  34  *
  35  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  36  *
  37  * settings:
  38  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  39  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  40  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  41  *  $wgMaxArticleSize*
  42  *
  43  *  * only within ParserOptions
  44  * </pre>
  45  *
  46  * @ingroup Parser
  47  */
  48 class Parser {
  49         /**
  50          * Update this version number when the ParserOutput format
  51          * changes in an incompatible way, so the parser cache
  52          * can automatically discard old data.
  53          */
  54         const VERSION = '1.6.4';
  55
  56         # Flags for Parser::setFunctionHook
  57         # Also available as global constants from Defines.php
  58         const SFH_NO_HASH = 1;
  59         const SFH_OBJECT_ARGS = 2;
  60
  61         # Constants needed for external link processing
  62         # Everything except bracket, space, or control characters
  63         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  64         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  65                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  66
  67         // State constants for the definition list colon extraction
  68         const COLON_STATE_TEXT = 0;
  69         const COLON_STATE_TAG = 1;
  70         const COLON_STATE_TAGSTART = 2;
  71         const COLON_STATE_CLOSETAG = 3;
  72         const COLON_STATE_TAGSLASH = 4;
  73         const COLON_STATE_COMMENT = 5;
  74         const COLON_STATE_COMMENTDASH = 6;
  75         const COLON_STATE_COMMENTDASHDASH = 7;
  76
  77         // Flags for preprocessToDom
  78         const PTD_FOR_INCLUSION = 1;
  79
  80         // Allowed values for $this->mOutputType
  81         // Parameter to startExternalParse().
  82         const OT_HTML = 1; // like parse()
  83         const OT_WIKI = 2; // like preSaveTransform()
  84         const OT_PREPROCESS = 3; // like preprocess()
  85         const OT_MSG = 3;
  86         const OT_PLAIN = 4; // like extractSections() - portions of the original are returned unchanged.
  87
  88         // Marker Suffix needs to be accessible staticly.
  89         const MARKER_SUFFIX = "-QINU\x7f";
  90
  91         /**#@+
  92          * @private
  93          */
  94         # Persistent:
  95         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  96         var $mSubstWords, $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex;
  97         var $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList;
  98         var $mVarCache, $mConf, $mFunctionTagHooks;
  99
 100
 101         # Cleared with clearState():
 102         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 103         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 104         var $mLinkHolders, $mLinkID;
 105         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 106         var $mTplExpandCache; // empty-frame expansion cache
 107         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 108         var $mExpensiveFunctionCount; // number of expensive parser function calls
 109
 110         # Temporary
 111         # These are variables reset at least once per parse regardless of $clearState
 112         var $mOptions;      // ParserOptions object
 113         var $mTitle;        // Title context, used for self-link rendering and similar things
 114         var $mOutputType;   // Output type, one of the OT_xxx constants
 115         var $ot;            // Shortcut alias, see setOutputType()
 116         var $mRevisionId;   // ID to display in {{REVISIONID}} tags
 117         var $mRevisionTimestamp; // The timestamp of the specified revision ID
 118         var $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 119
 120         /**#@-*/
 121
 122         /**
 123          * Constructor
 124          *
 125          * @public
 126          */
 127         function __construct( $conf = array() ) {
 128                 $this->mConf = $conf;
 129                 $this->mTagHooks = array();
 130                 $this->mTransparentTagHooks = array();
 131                 $this->mFunctionHooks = array();
 132                 $this->mFunctionTagHooks = array();
 133                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 134                 $this->mDefaultStripList = $this->mStripList = array();
 135                 $this->mUrlProtocols = wfUrlProtocols();
 136                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 137                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 138                 $this->mVarCache = array();
 139                 if ( isset( $conf['preprocessorClass'] ) ) {
 140                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 141                 } elseif ( extension_loaded( 'domxml' ) ) {
 142                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 143                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 144                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 145                 } elseif ( extension_loaded( 'dom' ) ) {
 146                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 147                 } else {
 148                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 149                 }
 150                 $this->mMarkerIndex = 0;
 151                 $this->mFirstCall = true;
 152         }
 153
 154         /**
 155          * Reduce memory usage to reduce the impact of circular references
 156          */
 157         function __destruct() {
 158                 if ( isset( $this->mLinkHolders ) ) {
 159                         $this->mLinkHolders->__destruct();
 160                 }
 161                 foreach ( $this as $name => $value ) {
 162                         unset( $this->$name );
 163                 }
 164         }
 165
 166         /**
 167          * Do various kinds of initialisation on the first call of the parser
 168          */
 169         function firstCallInit() {
 170                 if ( !$this->mFirstCall ) {
 171                         return;
 172                 }
 173                 $this->mFirstCall = false;
 174
 175                 wfProfileIn( __METHOD__ );
 176
 177                 CoreParserFunctions::register( $this );
 178                 CoreTagHooks::register( $this );
 179                 $this->initialiseVariables();
 180
 181                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 182                 wfProfileOut( __METHOD__ );
 183         }
 184
 185         /**
 186          * Clear Parser state
 187          *
 188          * @private
 189          */
 190         function clearState() {
 191                 wfProfileIn( __METHOD__ );
 192                 if ( $this->mFirstCall ) {
 193                         $this->firstCallInit();
 194                 }
 195                 $this->mOutput = new ParserOutput;
 196                 $this->mAutonumber = 0;
 197                 $this->mLastSection = '';
 198                 $this->mDTopen = false;
 199                 $this->mIncludeCount = array();
 200                 $this->mStripState = new StripState;
 201                 $this->mArgStack = false;
 202                 $this->mInPre = false;
 203                 $this->mLinkHolders = new LinkHolderArray( $this );
 204                 $this->mLinkID = 0;
 205                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 206                 $this->mVarCache = array();
 207
 208                 /**
 209                  * Prefix for temporary replacement strings for the multipass parser.
 210                  * \x07 should never appear in input as it's disallowed in XML.
 211                  * Using it at the front also gives us a little extra robustness
 212                  * since it shouldn't match when butted up against identifier-like
 213                  * string constructs.
 214                  *
 215                  * Must not consist of all title characters, or else it will change
 216                  * the behaviour of <nowiki> in a link.
 217                  */
 218                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 219                 # Changed to \x7f to allow XML double-parsing -- TS
 220                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 221
 222
 223                 # Clear these on every parse, bug 4549
 224                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 225
 226                 $this->mShowToc = true;
 227                 $this->mForceTocPosition = false;
 228                 $this->mIncludeSizes = array(
 229                         'post-expand' => 0,
 230                         'arg' => 0,
 231                 );
 232                 $this->mPPNodeCount = 0;
 233                 $this->mDefaultSort = false;
 234                 $this->mHeadings = array();
 235                 $this->mDoubleUnderscores = array();
 236                 $this->mExpensiveFunctionCount = 0;
 237
 238                 # Fix cloning
 239                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 240                         $this->mPreprocessor = null;
 241                 }
 242
 243                 wfRunHooks( 'ParserClearState', array( &$this ) );
 244                 wfProfileOut( __METHOD__ );
 245         }
 246
 247         function setOutputType( $ot ) {
 248                 $this->mOutputType = $ot;
 249                 // Shortcut alias
 250                 $this->ot = array(
 251                         'html' => $ot == self::OT_HTML,
 252                         'wiki' => $ot == self::OT_WIKI,
 253                         'pre' => $ot == self::OT_PREPROCESS,
 254                         'plain' => $ot == self::OT_PLAIN,
 255                 );
 256         }
 257
 258         /**
 259          * Set the context title
 260          */
 261         function setTitle( $t ) {
 262                 if ( !$t || $t instanceof FakeTitle ) {
 263                         $t = Title::newFromText( 'NO TITLE' );
 264                 }
 265
 266                 if ( strval( $t->getFragment() ) !== '' ) {
 267                         # Strip the fragment to avoid various odd effects
 268                         $this->mTitle = clone $t;
 269                         $this->mTitle->setFragment( '' );
 270                 } else {
 271                         $this->mTitle = $t;
 272                 }
 273         }
 274
 275         /**
 276          * Accessor for mUniqPrefix.
 277          *
 278          * @public
 279          */
 280         function uniqPrefix() {
 281                 if ( !isset( $this->mUniqPrefix ) ) {
 282                         // @todo Fixme: this is probably *horribly wrong*
 283                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 284                         // if this is called for a parser cache hit, the parser may not
 285                         // have ever been initialized in the first place.
 286                         // Not really sure what the heck is supposed to be going on here.
 287                         return '';
 288                         // throw new MWException( "Accessing uninitialized mUniqPrefix" );
 289                 }
 290                 return $this->mUniqPrefix;
 291         }
 292
 293         /**
 294          * Convert wikitext to HTML
 295          * Do not call this function recursively.
 296          *
 297          * @param $text String: text we want to parse
 298          * @param $title A title object
 299          * @param $options ParserOptions
 300          * @param $linestart boolean
 301          * @param $clearState boolean
 302          * @param $revid Int: number to pass in {{REVISIONID}}
 303          * @return ParserOutput a ParserOutput
 304          */
 305         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 306                 /**
 307                  * First pass--just handle <nowiki> sections, pass the rest off
 308                  * to internalParse() which does all the real work.
 309                  */
 310
 311                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion;
 312                 $fname = __METHOD__.'-' . wfGetCaller();
 313                 wfProfileIn( __METHOD__ );
 314                 wfProfileIn( $fname );
 315
 316                 if ( $clearState ) {
 317                         $this->clearState();
 318                 }
 319
 320                 $this->mOptions = $options;
 321                 $this->setTitle( $title ); // Page title has to be set for the pre-processor
 322
 323                 $oldRevisionId = $this->mRevisionId;
 324                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 325                 if ( $revid !== null ) {
 326                         $this->mRevisionId = $revid;
 327                         $this->mRevisionTimestamp = null;
 328                 }
 329                 $this->setOutputType( self::OT_HTML );
 330                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 331                 # No more strip!
 332                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 333                 $text = $this->internalParse( $text );
 334
 335                 $text = $this->mStripState->unstripGeneral( $text );
 336
 337                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 338                 $fixtags = array(
 339                         # french spaces, last one Guillemet-left
 340                         # only if there is something before the space
 341                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 342                         # french spaces, Guillemet-right
 343                         '/(\\302\\253) /' => '\\1&nbsp;',
 344                         '/&nbsp;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 345                 );
 346                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 347
 348                 $text = $this->doBlockLevels( $text, $linestart );
 349
 350                 $this->replaceLinkHolders( $text );
 351
 352                 // The position of the convert() call should not be changed. it
 353                 // assumes that the links are all replaced and the only thing left
 354                 // is the <nowiki> mark.
 355                 if ( !( $wgDisableLangConversion
 356                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 357                                 || $this->mTitle->isTalkPage()
 358                                 || $this->mTitle->isConversionTable() ) ) {
 359                         $text = $wgContLang->convert( $text );
 360                 }
 361
 362                 // A title may have been set in a conversion rule.
 363                 // Note that if a user tries to set a title in a conversion
 364                 // rule but content conversion was not done, then the parser
 365                 // won't pick it up.  This is probably expected behavior.
 366                 if ( $wgContLang->getConvRuleTitle() ) {
 367                         $this->mOutput->setTitleText( $wgContLang->getConvRuleTitle() );
 368                 }
 369
 370                 $text = $this->mStripState->unstripNoWiki( $text );
 371
 372                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 373
 374 //!JF Move to its own function
 375
 376                 $uniq_prefix = $this->mUniqPrefix;
 377                 $matches = array();
 378                 $elements = array_keys( $this->mTransparentTagHooks );
 379                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 380
 381                 foreach ( $matches as $marker => $data ) {
 382                         list( $element, $content, $params, $tag ) = $data;
 383                         $tagName = strtolower( $element );
 384                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 385                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 386                         } else {
 387                                 $output = $tag;
 388                         }
 389                         $this->mStripState->general->setPair( $marker, $output );
 390                 }
 391                 $text = $this->mStripState->unstripGeneral( $text );
 392
 393                 $text = Sanitizer::normalizeCharReferences( $text );
 394
 395                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 396                         $text = MWTidy::tidy( $text );
 397                 } else {
 398                         # attempt to sanitize at least some nesting problems
 399                         # (bug #2702 and quite a few others)
 400                         $tidyregs = array(
 401                                 # ''Something [http://www.cool.com cool''] -->
 402                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 403                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 404                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 405                                 # fix up an anchor inside another anchor, only
 406                                 # at least for a single single nested link (bug 3695)
 407                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 408                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 409                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 410                                 # contains a div, so fix it up here; replace
 411                                 # div with escaped text
 412                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 413                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 414                                 # remove empty italic or bold tag pairs, some
 415                                 # introduced by rules above
 416                                 '/<([bi])><\/\\1>/' => '',
 417                         );
 418
 419                         $text = preg_replace(
 420                                 array_keys( $tidyregs ),
 421                                 array_values( $tidyregs ),
 422                                 $text );
 423                 }
 424                 global $wgExpensiveParserFunctionLimit;
 425                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 426                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 427                 }
 428
 429                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 430
 431                 # Information on include size limits, for the benefit of users who try to skirt them
 432                 if ( $this->mOptions->getEnableLimitReport() ) {
 433                         $max = $this->mOptions->getMaxIncludeSize();
 434                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 435                         $limitReport =
 436                                 "NewPP limit report\n" .
 437                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 438                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 439                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 440                                 $PFreport;
 441                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 442                         $text .= "\n<!-- \n$limitReport-->\n";
 443                 }
 444                 $this->mOutput->setText( $text );
 445
 446                 $this->mRevisionId = $oldRevisionId;
 447                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 448                 wfProfileOut( $fname );
 449                 wfProfileOut( __METHOD__ );
 450
 451                 return $this->mOutput;
 452         }
 453
 454         /**
 455          * Recursive parser entry point that can be called from an extension tag
 456          * hook.
 457          *
 458          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 459          *
 460          * @param $text String: text extension wants to have parsed
 461          * @param PPFrame $frame: The frame to use for expanding any template variables
 462          */
 463         function recursiveTagParse( $text, $frame=false ) {
 464                 wfProfileIn( __METHOD__ );
 465                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 466                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 467                 $text = $this->internalParse( $text, false, $frame );
 468                 wfProfileOut( __METHOD__ );
 469                 return $text;
 470         }
 471
 472         /**
 473          * Expand templates and variables in the text, producing valid, static wikitext.
 474          * Also removes comments.
 475          */
 476         function preprocess( $text, $title, $options, $revid = null ) {
 477                 wfProfileIn( __METHOD__ );
 478                 $this->clearState();
 479                 $this->setOutputType( self::OT_PREPROCESS );
 480                 $this->mOptions = $options;
 481                 $this->setTitle( $title );
 482                 if ( $revid !== null ) {
 483                         $this->mRevisionId = $revid;
 484                 }
 485                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 486                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 487                 $text = $this->replaceVariables( $text );
 488                 $text = $this->mStripState->unstripBoth( $text );
 489                 wfProfileOut( __METHOD__ );
 490                 return $text;
 491         }
 492
 493         /**
 494          * Process the wikitext for the ?preload= feature. (bug 5210)
 495          *
 496          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 497          * comments, templates, arguments, tags hooks and parser functions are untouched.
 498          */
 499         public function getPreloadText( $text, $title, $options ) {
 500                 // Parser (re)initialisation
 501                 $this->clearState();
 502                 $this->setOutputType( self::OT_PLAIN );
 503                 $this->mOptions = $options;
 504                 $this->setTitle( $title );
 505
 506                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 507                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 508                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 509         }
 510
 511         /**
 512          * Get a random string
 513          *
 514          * @private
 515          * @static
 516          */
 517         function getRandomString() {
 518                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 519         }
 520
 521         function &getTitle() { return $this->mTitle; }
 522         function getOptions() { return $this->mOptions; }
 523         function getRevisionId() { return $this->mRevisionId; }
 524         function getOutput() { return $this->mOutput; }
 525         function nextLinkID() { return $this->mLinkID++; }
 526
 527         function getFunctionLang() {
 528                 global $wgLang, $wgContLang;
 529
 530                 $target = $this->mOptions->getTargetLanguage();
 531                 if ( $target !== null ) {
 532                         return $target;
 533                 } else {
 534                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 535                 }
 536         }
 537
 538         /**
 539          * Get a preprocessor object
 540          */
 541         function getPreprocessor() {
 542                 if ( !isset( $this->mPreprocessor ) ) {
 543                         $class = $this->mPreprocessorClass;
 544                         $this->mPreprocessor = new $class( $this );
 545                 }
 546                 return $this->mPreprocessor;
 547         }
 548
 549         /**
 550          * Replaces all occurrences of HTML-style comments and the given tags
 551          * in the text with a random marker and returns the next text. The output
 552          * parameter $matches will be an associative array filled with data in
 553          * the form:
 554          *   'UNIQ-xxxxx' => array(
 555          *     'element',
 556          *     'tag content',
 557          *     array( 'param' => 'x' ),
 558          *     '<element param="x">tag content</element>' ) )
 559          *
 560          * @param $elements list of element names. Comments are always extracted.
 561          * @param $text Source text string.
 562          * @param $uniq_prefix
 563          *
 564          * @public
 565          * @static
 566          */
 567         function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 568                 static $n = 1;
 569                 $stripped = '';
 570                 $matches = array();
 571
 572                 $taglist = implode( '|', $elements );
 573                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 574
 575                 while ( $text != '' ) {
 576                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 577                         $stripped .= $p[0];
 578                         if ( count( $p ) < 5 ) {
 579                                 break;
 580                         }
 581                         if ( count( $p ) > 5 ) {
 582                                 // comment
 583                                 $element    = $p[4];
 584                                 $attributes = '';
 585                                 $close      = '';
 586                                 $inside     = $p[5];
 587                         } else {
 588                                 // tag
 589                                 $element    = $p[1];
 590                                 $attributes = $p[2];
 591                                 $close      = $p[3];
 592                                 $inside     = $p[4];
 593                         }
 594
 595                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 596                         $stripped .= $marker;
 597
 598                         if ( $close === '/>' ) {
 599                                 // Empty element tag, <tag />
 600                                 $content = null;
 601                                 $text = $inside;
 602                                 $tail = null;
 603                         } else {
 604                                 if ( $element === '!--' ) {
 605                                         $end = '/(-->)/';
 606                                 } else {
 607                                         $end = "/(<\\/$element\\s*>)/i";
 608                                 }
 609                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 610                                 $content = $q[0];
 611                                 if ( count( $q ) < 3 ) {
 612                                         # No end tag -- let it run out to the end of the text.
 613                                         $tail = '';
 614                                         $text = '';
 615                                 } else {
 616                                         $tail = $q[1];
 617                                         $text = $q[2];
 618                                 }
 619                         }
 620
 621                         $matches[$marker] = array( $element,
 622                                 $content,
 623                                 Sanitizer::decodeTagAttributes( $attributes ),
 624                                 "<$element$attributes$close$content$tail" );
 625                 }
 626                 return $stripped;
 627         }
 628
 629         /**
 630          * Get a list of strippable XML-like elements
 631          */
 632         function getStripList() {
 633                 return $this->mStripList;
 634         }
 635
 636         /**
 637          * @deprecated use replaceVariables
 638          */
 639         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 640                 return $text;
 641         }
 642
 643         /**
 644          * Restores pre, math, and other extensions removed by strip()
 645          *
 646          * always call unstripNoWiki() after this one
 647          * @private
 648          * @deprecated use $this->mStripState->unstrip()
 649          */
 650         function unstrip( $text, $state ) {
 651                 return $state->unstripGeneral( $text );
 652         }
 653
 654         /**
 655          * Always call this after unstrip() to preserve the order
 656          *
 657          * @private
 658          * @deprecated use $this->mStripState->unstrip()
 659          */
 660         function unstripNoWiki( $text, $state ) {
 661                 return $state->unstripNoWiki( $text );
 662         }
 663
 664         /**
 665          * @deprecated use $this->mStripState->unstripBoth()
 666          */
 667         function unstripForHTML( $text ) {
 668                 return $this->mStripState->unstripBoth( $text );
 669         }
 670
 671         /**
 672          * Add an item to the strip state
 673          * Returns the unique tag which must be inserted into the stripped text
 674          * The tag will be replaced with the original text in unstrip()
 675          *
 676          * @private
 677          */
 678         function insertStripItem( $text ) {
 679                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 680                 $this->mMarkerIndex++;
 681                 $this->mStripState->general->setPair( $rnd, $text );
 682                 return $rnd;
 683         }
 684
 685         /**
 686          * Interface with html tidy
 687          * @deprecated Use MWTidy::tidy()
 688          */
 689         public static function tidy( $text ) {
 690                 wfDeprecated( __METHOD__ );
 691                 return MWTidy::tidy( $text );
 692         }
 693
 694         /**
 695          * parse the wiki syntax used to render tables
 696          *
 697          * @private
 698          */
 699         function doTableStuff( $text ) {
 700                 wfProfileIn( __METHOD__ );
 701
 702                 $lines = StringUtils::explode( "\n", $text );
 703                 $out = '';
 704                 $td_history = array(); // Is currently a td tag open?
 705                 $last_tag_history = array(); // Save history of last lag activated (td, th or caption)
 706                 $tr_history = array(); // Is currently a tr tag open?
 707                 $tr_attributes = array(); // history of tr attributes
 708                 $has_opened_tr = array(); // Did this table open a <tr> element?
 709                 $indent_level = 0; // indent level of the table
 710
 711                 foreach ( $lines as $outLine ) {
 712                         $line = trim( $outLine );
 713
 714                         if ( $line == '' ) { // empty line, go to next line
 715                                 $out .= $outLine."\n";
 716                                 continue;
 717                         }
 718                         $first_character = $line[0];
 719                         $matches = array();
 720
 721                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 722                                 // First check if we are starting a new table
 723                                 $indent_level = strlen( $matches[1] );
 724
 725                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 726                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 727
 728                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 729                                 array_push( $td_history , false );
 730                                 array_push( $last_tag_history , '' );
 731                                 array_push( $tr_history , false );
 732                                 array_push( $tr_attributes , '' );
 733                                 array_push( $has_opened_tr , false );
 734                         } elseif ( count( $td_history ) == 0 ) {
 735                                 // Don't do any of the following
 736                                 $out .= $outLine."\n";
 737                                 continue;
 738                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 739                                 // We are ending a table
 740                                 $line = '</table>' . substr( $line , 2 );
 741                                 $last_tag = array_pop( $last_tag_history );
 742
 743                                 if ( !array_pop( $has_opened_tr ) ) {
 744                                         $line = "<tr><td></td></tr>{$line}";
 745                                 }
 746
 747                                 if ( array_pop( $tr_history ) ) {
 748                                         $line = "</tr>{$line}";
 749                                 }
 750
 751                                 if ( array_pop( $td_history ) ) {
 752                                         $line = "</{$last_tag}>{$line}";
 753                                 }
 754                                 array_pop( $tr_attributes );
 755                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 756                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 757                                 // Now we have a table row
 758                                 $line = preg_replace( '#^\|-+#', '', $line );
 759
 760                                 // Whats after the tag is now only attributes
 761                                 $attributes = $this->mStripState->unstripBoth( $line );
 762                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 763                                 array_pop( $tr_attributes );
 764                                 array_push( $tr_attributes, $attributes );
 765
 766                                 $line = '';
 767                                 $last_tag = array_pop( $last_tag_history );
 768                                 array_pop( $has_opened_tr );
 769                                 array_push( $has_opened_tr , true );
 770
 771                                 if ( array_pop( $tr_history ) ) {
 772                                         $line = '</tr>';
 773                                 }
 774
 775                                 if ( array_pop( $td_history ) ) {
 776                                         $line = "</{$last_tag}>{$line}";
 777                                 }
 778
 779                                 $outLine = $line;
 780                                 array_push( $tr_history , false );
 781                                 array_push( $td_history , false );
 782                                 array_push( $last_tag_history , '' );
 783                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 784                                 // This might be cell elements, td, th or captions
 785                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 786                                         $first_character = '+';
 787                                         $line = substr( $line , 1 );
 788                                 }
 789
 790                                 $line = substr( $line , 1 );
 791
 792                                 if ( $first_character === '!' ) {
 793                                         $line = str_replace( '!!' , '||' , $line );
 794                                 }
 795
 796                                 // Split up multiple cells on the same line.
 797                                 // FIXME : This can result in improper nesting of tags processed
 798                                 // by earlier parser steps, but should avoid splitting up eg
 799                                 // attribute values containing literal "||".
 800                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 801
 802                                 $outLine = '';
 803
 804                                 // Loop through each table cell
 805                                 foreach ( $cells as $cell ) {
 806                                         $previous = '';
 807                                         if ( $first_character !== '+' ) {
 808                                                 $tr_after = array_pop( $tr_attributes );
 809                                                 if ( !array_pop( $tr_history ) ) {
 810                                                         $previous = "<tr{$tr_after}>\n";
 811                                                 }
 812                                                 array_push( $tr_history , true );
 813                                                 array_push( $tr_attributes , '' );
 814                                                 array_pop( $has_opened_tr );
 815                                                 array_push( $has_opened_tr , true );
 816                                         }
 817
 818                                         $last_tag = array_pop( $last_tag_history );
 819
 820                                         if ( array_pop( $td_history ) ) {
 821                                                 $previous = "</{$last_tag}>{$previous}";
 822                                         }
 823
 824                                         if ( $first_character === '|' ) {
 825                                                 $last_tag = 'td';
 826                                         } elseif ( $first_character === '!' ) {
 827                                                 $last_tag = 'th';
 828                                         } elseif ( $first_character === '+' ) {
 829                                                 $last_tag = 'caption';
 830                                         } else {
 831                                                 $last_tag = '';
 832                                         }
 833
 834                                         array_push( $last_tag_history , $last_tag );
 835
 836                                         // A cell could contain both parameters and data
 837                                         $cell_data = explode( '|' , $cell , 2 );
 838
 839                                         // Bug 553: Note that a '|' inside an invalid link should not
 840                                         // be mistaken as delimiting cell parameters
 841                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 842                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 843                                         } elseif ( count( $cell_data ) == 1 ) {
 844                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 845                                         } else {
 846                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 847                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 848                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 849                                         }
 850
 851                                         $outLine .= $cell;
 852                                         array_push( $td_history , true );
 853                                 }
 854                         }
 855                         $out .= $outLine . "\n";
 856                 }
 857
 858                 // Closing open td, tr && table
 859                 while ( count( $td_history ) > 0 ) {
 860                         if ( array_pop( $td_history ) ) {
 861                                 $out .= "</td>\n";
 862                         }
 863                         if ( array_pop( $tr_history ) ) {
 864                                 $out .= "</tr>\n";
 865                         }
 866                         if ( !array_pop( $has_opened_tr ) ) {
 867                                 $out .= "<tr><td></td></tr>\n" ;
 868                         }
 869
 870                         $out .= "</table>\n";
 871                 }
 872
 873                 // Remove trailing line-ending (b/c)
 874                 if ( substr( $out, -1 ) === "\n" ) {
 875                         $out = substr( $out, 0, -1 );
 876                 }
 877
 878                 // special case: don't return empty table
 879                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 880                         $out = '';
 881                 }
 882
 883                 wfProfileOut( __METHOD__ );
 884
 885                 return $out;
 886         }
 887
 888         /**
 889          * Helper function for parse() that transforms wiki markup into
 890          * HTML. Only called for $mOutputType == self::OT_HTML.
 891          *
 892          * @private
 893          */
 894         function internalParse( $text, $isMain = true, $frame=false ) {
 895                 wfProfileIn( __METHOD__ );
 896
 897                 $origText = $text;
 898
 899                 # Hook to suspend the parser in this state
 900                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 901                         wfProfileOut( __METHOD__ );
 902                         return $text ;
 903                 }
 904
 905                 // if $frame is provided, then use $frame for replacing any variables
 906                 if ( $frame ) {
 907                         // use frame depth to infer how include/noinclude tags should be handled
 908                         // depth=0 means this is the top-level document; otherwise it's an included document
 909                         if ( !$frame->depth ) {
 910                                 $flag = 0;
 911                         } else {
 912                                 $flag = Parser::PTD_FOR_INCLUSION;
 913                         $dom = $this->preprocessToDom( $text, $flag );
 914                         $text = $frame->expand( $dom );
 915                         }
 916                 } else {
 917                         // if $frame is not provided, then use old-style replaceVariables
 918                         $text = $this->replaceVariables( $text );
 919                 }
 920
 921                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 922                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 923
 924                 // Tables need to come after variable replacement for things to work
 925                 // properly; putting them before other transformations should keep
 926                 // exciting things like link expansions from showing up in surprising
 927                 // places.
 928                 $text = $this->doTableStuff( $text );
 929
 930                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 931
 932                 $text = $this->doDoubleUnderscore( $text );
 933
 934                 $text = $this->doHeadings( $text );
 935                 if ( $this->mOptions->getUseDynamicDates() ) {
 936                         $df = DateFormatter::getInstance();
 937                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 938                 }
 939                 $text = $this->doAllQuotes( $text );
 940                 $text = $this->replaceInternalLinks( $text );
 941                 $text = $this->replaceExternalLinks( $text );
 942
 943                 # replaceInternalLinks may sometimes leave behind
 944                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 945                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
 946
 947                 $text = $this->doMagicLinks( $text );
 948                 $text = $this->formatHeadings( $text, $origText, $isMain );
 949
 950                 wfProfileOut( __METHOD__ );
 951                 return $text;
 952         }
 953
 954         /**
 955          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 956          * magic external links.
 957          *
 958          * DML
 959          * @private
 960          */
 961         function doMagicLinks( $text ) {
 962                 wfProfileIn( __METHOD__ );
 963                 $prots = $this->mUrlProtocols;
 964                 $urlChar = self::EXT_LINK_URL_CLASS;
 965                 $text = preg_replace_callback(
 966                         '!(?:                           # Start cases
 967                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
 968                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 969                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 970                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 971                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 972                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 973                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 974                                     [0-9Xx]                 # check digit
 975                                     \b)
 976                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 977                 wfProfileOut( __METHOD__ );
 978                 return $text;
 979         }
 980
 981         function magicLinkCallback( $m ) {
 982                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 983                         # Skip anchor
 984                         return $m[0];
 985                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 986                         # Skip HTML element
 987                         return $m[0];
 988                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 989                         # Free external link
 990                         return $this->makeFreeExternalLink( $m[0] );
 991                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 992                         # RFC or PMID
 993                         $CssClass = '';
 994                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 995                                 $keyword = 'RFC';
 996                                 $urlmsg = 'rfcurl';
 997                                 $CssClass = 'mw-magiclink-rfc';
 998                                 $id = $m[4];
 999                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1000                                 $keyword = 'PMID';
1001                                 $urlmsg = 'pubmedurl';
1002                                 $CssClass = 'mw-magiclink-pmid';
1003                                 $id = $m[4];
1004                         } else {
1005                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1006                                         substr( $m[0], 0, 20 ) . '"' );
1007                         }
1008                         $url = wfMsg( $urlmsg, $id);
1009                         $sk = $this->mOptions->getSkin();
1010                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1011                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1012                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1013                         # ISBN
1014                         $isbn = $m[5];
1015                         $num = strtr( $isbn, array(
1016                                 '-' => '',
1017                                 ' ' => '',
1018                                 'x' => 'X',
1019                         ));
1020                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1021                         return'<a href="' .
1022                                 $titleObj->escapeLocalUrl() .
1023                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1024                 } else {
1025                         return $m[0];
1026                 }
1027         }
1028
1029         /**
1030          * Make a free external link, given a user-supplied URL
1031          * @return HTML
1032          * @private
1033          */
1034         function makeFreeExternalLink( $url ) {
1035                 global $wgContLang;
1036                 wfProfileIn( __METHOD__ );
1037
1038                 $sk = $this->mOptions->getSkin();
1039                 $trail = '';
1040
1041                 # The characters '<' and '>' (which were escaped by
1042                 # removeHTMLtags()) should not be included in
1043                 # URLs, per RFC 2396.
1044                 $m2 = array();
1045                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1046                         $trail = substr( $url, $m2[0][1] ) . $trail;
1047                         $url = substr( $url, 0, $m2[0][1] );
1048                 }
1049
1050                 # Move trailing punctuation to $trail
1051                 $sep = ',;\.:!?';
1052                 # If there is no left bracket, then consider right brackets fair game too
1053                 if ( strpos( $url, '(' ) === false ) {
1054                         $sep .= ')';
1055                 }
1056
1057                 $numSepChars = strspn( strrev( $url ), $sep );
1058                 if ( $numSepChars ) {
1059                         $trail = substr( $url, -$numSepChars ) . $trail;
1060                         $url = substr( $url, 0, -$numSepChars );
1061                 }
1062
1063                 $url = Sanitizer::cleanUrl( $url );
1064
1065                 # Is this an external image?
1066                 $text = $this->maybeMakeExternalImage( $url );
1067                 if ( $text === false ) {
1068                         # Not an image, make a link
1069                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1070                                 $this->getExternalLinkAttribs( $url ) );
1071                         # Register it in the output object...
1072                         # Replace unnecessary URL escape codes with their equivalent characters
1073                         $pasteurized = self::replaceUnusualEscapes( $url );
1074                         $this->mOutput->addExternalLink( $pasteurized );
1075                 }
1076                 wfProfileOut( __METHOD__ );
1077                 return $text . $trail;
1078         }
1079
1080
1081         /**
1082          * Parse headers and return html
1083          *
1084          * @private
1085          */
1086         function doHeadings( $text ) {
1087                 wfProfileIn( __METHOD__ );
1088                 for ( $i = 6; $i >= 1; --$i ) {
1089                         $h = str_repeat( '=', $i );
1090                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1091                           "<h$i>\\1</h$i>", $text );
1092                 }
1093                 wfProfileOut( __METHOD__ );
1094                 return $text;
1095         }
1096
1097         /**
1098          * Replace single quotes with HTML markup
1099          * @private
1100          * @return string the altered text
1101          */
1102         function doAllQuotes( $text ) {
1103                 wfProfileIn( __METHOD__ );
1104                 $outtext = '';
1105                 $lines = StringUtils::explode( "\n", $text );
1106                 foreach ( $lines as $line ) {
1107                         $outtext .= $this->doQuotes( $line ) . "\n";
1108                 }
1109                 $outtext = substr( $outtext, 0,-1 );
1110                 wfProfileOut( __METHOD__ );
1111                 return $outtext;
1112         }
1113
1114         /**
1115          * Helper function for doAllQuotes()
1116          */
1117         public function doQuotes( $text ) {
1118                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1119                 if ( count( $arr ) == 1 ) {
1120                         return $text;
1121                 } else {
1122                         # First, do some preliminary work. This may shift some apostrophes from
1123                         # being mark-up to being text. It also counts the number of occurrences
1124                         # of bold and italics mark-ups.
1125                         $i = 0;
1126                         $numbold = 0;
1127                         $numitalics = 0;
1128                         foreach ( $arr as $r ) {
1129                                 if ( ( $i % 2 ) == 1 ) {
1130                                         # If there are ever four apostrophes, assume the first is supposed to
1131                                         # be text, and the remaining three constitute mark-up for bold text.
1132                                         if ( strlen( $arr[$i] ) == 4 ) {
1133                                                 $arr[$i-1] .= "'";
1134                                                 $arr[$i] = "'''";
1135                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1136                                                 # If there are more than 5 apostrophes in a row, assume they're all
1137                                                 # text except for the last 5.
1138                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1139                                                 $arr[$i] = "'''''";
1140                                         }
1141                                         # Count the number of occurrences of bold and italics mark-ups.
1142                                         # We are not counting sequences of five apostrophes.
1143                                         if ( strlen( $arr[$i] ) == 2 ) {
1144                                                 $numitalics++;
1145                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1146                                                 $numbold++;
1147                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1148                                                 $numitalics++;
1149                                                 $numbold++;
1150                                         }
1151                                 }
1152                                 $i++;
1153                         }
1154
1155                         # If there is an odd number of both bold and italics, it is likely
1156                         # that one of the bold ones was meant to be an apostrophe followed
1157                         # by italics. Which one we cannot know for certain, but it is more
1158                         # likely to be one that has a single-letter word before it.
1159                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1160                                 $i = 0;
1161                                 $firstsingleletterword = -1;
1162                                 $firstmultiletterword = -1;
1163                                 $firstspace = -1;
1164                                 foreach ( $arr as $r ) {
1165                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1166                                                 $x1 = substr( $arr[$i-1], -1 );
1167                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1168                                                 if ( $x1 === ' ' ) {
1169                                                         if ( $firstspace == -1 ) {
1170                                                                 $firstspace = $i;
1171                                                         }
1172                                                 } elseif ( $x2 === ' ') {
1173                                                         if ( $firstsingleletterword == -1 ) {
1174                                                                 $firstsingleletterword = $i;
1175                                                         }
1176                                                 } else {
1177                                                         if ( $firstmultiletterword == -1 ) {
1178                                                                 $firstmultiletterword = $i;
1179                                                         }
1180                                                 }
1181                                         }
1182                                         $i++;
1183                                 }
1184
1185                                 # If there is a single-letter word, use it!
1186                                 if ( $firstsingleletterword > -1 ) {
1187                                         $arr[$firstsingleletterword] = "''";
1188                                         $arr[$firstsingleletterword-1] .= "'";
1189                                 } elseif ( $firstmultiletterword > -1 ) {
1190                                         # If not, but there's a multi-letter word, use that one.
1191                                         $arr[$firstmultiletterword] = "''";
1192                                         $arr[$firstmultiletterword-1] .= "'";
1193                                 } elseif ( $firstspace > -1 ) {
1194                                         # ... otherwise use the first one that has neither.
1195                                         # (notice that it is possible for all three to be -1 if, for example,
1196                                         # there is only one pentuple-apostrophe in the line)
1197                                         $arr[$firstspace] = "''";
1198                                         $arr[$firstspace-1] .= "'";
1199                                 }
1200                         }
1201
1202                         # Now let's actually convert our apostrophic mush to HTML!
1203                         $output = '';
1204                         $buffer = '';
1205                         $state = '';
1206                         $i = 0;
1207                         foreach ( $arr as $r ) {
1208                                 if ( ( $i % 2 ) == 0 ) {
1209                                         if ( $state === 'both' ) {
1210                                                 $buffer .= $r;
1211                                         } else {
1212                                                 $output .= $r;
1213                                         }
1214                                 } else {
1215                                         if ( strlen( $r ) == 2 ) {
1216                                                 if ( $state === 'i' ) {
1217                                                         $output .= '</i>'; $state = '';
1218                                                 } elseif ( $state === 'bi' ) {
1219                                                         $output .= '</i>'; $state = 'b';
1220                                                 } elseif ( $state === 'ib' ) {
1221                                                         $output .= '</b></i><b>'; $state = 'b';
1222                                                 } elseif ( $state === 'both' ) {
1223                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1224                                                 } else { # $state can be 'b' or ''
1225                                                         $output .= '<i>'; $state .= 'i';
1226                                                 }
1227                                         } elseif ( strlen( $r ) == 3 ) {
1228                                                 if ( $state === 'b' ) {
1229                                                         $output .= '</b>'; $state = '';
1230                                                 } elseif ( $state === 'bi' ) {
1231                                                         $output .= '</i></b><i>'; $state = 'i';
1232                                                 } elseif ( $state === 'ib' ) {
1233                                                         $output .= '</b>'; $state = 'i';
1234                                                 } elseif ( $state === 'both' ) {
1235                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1236                                                 } else { # $state can be 'i' or ''
1237                                                         $output .= '<b>'; $state .= 'b';
1238                                                 }
1239                                         } elseif ( strlen( $r ) == 5 ) {
1240                                                 if ( $state === 'b' ) {
1241                                                         $output .= '</b><i>'; $state = 'i';
1242                                                 } elseif ( $state === 'i' ) {
1243                                                         $output .= '</i><b>'; $state = 'b';
1244                                                 } elseif ( $state === 'bi' ) {
1245                                                         $output .= '</i></b>'; $state = '';
1246                                                 } elseif ( $state === 'ib' ) {
1247                                                         $output .= '</b></i>'; $state = '';
1248                                                 } elseif ( $state === 'both' ) {
1249                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1250                                                 } else { # ($state == '')
1251                                                         $buffer = ''; $state = 'both';
1252                                                 }
1253                                         }
1254                                 }
1255                                 $i++;
1256                         }
1257                         # Now close all remaining tags.  Notice that the order is important.
1258                         if ( $state === 'b' || $state === 'ib' ) {
1259                                 $output .= '</b>';
1260                         }
1261                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1262                                 $output .= '</i>';
1263                         }
1264                         if ( $state === 'bi' ) {
1265                                 $output .= '</b>';
1266                         }
1267                         # There might be lonely ''''', so make sure we have a buffer
1268                         if ( $state === 'both' && $buffer ) {
1269                                 $output .= '<b><i>'.$buffer.'</i></b>';
1270                         }
1271                         return $output;
1272                 }
1273         }
1274
1275         /**
1276          * Replace external links (REL)
1277          *
1278          * Note: this is all very hackish and the order of execution matters a lot.
1279          * Make sure to run maintenance/parserTests.php if you change this code.
1280          *
1281          * @private
1282          */
1283         function replaceExternalLinks( $text ) {
1284                 global $wgContLang;
1285                 wfProfileIn( __METHOD__ );
1286
1287                 $sk = $this->mOptions->getSkin();
1288
1289                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1290                 $s = array_shift( $bits );
1291
1292                 $i = 0;
1293                 while ( $i<count( $bits ) ) {
1294                         $url = $bits[$i++];
1295                         $protocol = $bits[$i++];
1296                         $text = $bits[$i++];
1297                         $trail = $bits[$i++];
1298
1299                         # The characters '<' and '>' (which were escaped by
1300                         # removeHTMLtags()) should not be included in
1301                         # URLs, per RFC 2396.
1302                         $m2 = array();
1303                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1304                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1305                                 $url = substr( $url, 0, $m2[0][1] );
1306                         }
1307
1308                         # If the link text is an image URL, replace it with an <img> tag
1309                         # This happened by accident in the original parser, but some people used it extensively
1310                         $img = $this->maybeMakeExternalImage( $text );
1311                         if ( $img !== false ) {
1312                                 $text = $img;
1313                         }
1314
1315                         $dtrail = '';
1316
1317                         # Set linktype for CSS - if URL==text, link is essentially free
1318                         $linktype = ( $text === $url ) ? 'free' : 'text';
1319
1320                         # No link text, e.g. [http://domain.tld/some.link]
1321                         if ( $text == '' ) {
1322                                 # Autonumber if allowed. See bug #5918
1323                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1324                                         $langObj = $this->getFunctionLang();
1325                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1326                                         $linktype = 'autonumber';
1327                                 } else {
1328                                         # Otherwise just use the URL
1329                                         $text = htmlspecialchars( $url );
1330                                         $linktype = 'free';
1331                                 }
1332                         } else {
1333                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1334                                 # Check for trail
1335                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1336                         }
1337
1338                         $text = $wgContLang->markNoConversion( $text );
1339
1340                         $url = Sanitizer::cleanUrl( $url );
1341
1342                         # Use the encoded URL
1343                         # This means that users can paste URLs directly into the text
1344                         # Funny characters like &ouml; aren't valid in URLs anyway
1345                         # This was changed in August 2004
1346                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1347                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1348
1349                         # Register link in the output object.
1350                         # Replace unnecessary URL escape codes with the referenced character
1351                         # This prevents spammers from hiding links from the filters
1352                         $pasteurized = self::replaceUnusualEscapes( $url );
1353                         $this->mOutput->addExternalLink( $pasteurized );
1354                 }
1355
1356                 wfProfileOut( __METHOD__ );
1357                 return $s;
1358         }
1359
1360         /**
1361          * Get an associative array of additional HTML attributes appropriate for a
1362          * particular external link.  This currently may include rel => nofollow
1363          * (depending on configuration, namespace, and the URL's domain) and/or a
1364          * target attribute (depending on configuration).
1365          *
1366          * @param string $url Optional URL, to extract the domain from for rel =>
1367          *   nofollow if appropriate
1368          * @return array Associative array of HTML attributes
1369          */
1370         function getExternalLinkAttribs( $url = false ) {
1371                 $attribs = array();
1372                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1373                 $ns = $this->mTitle->getNamespace();
1374                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1375                         $attribs['rel'] = 'nofollow';
1376
1377                         global $wgNoFollowDomainExceptions;
1378                         if ( $wgNoFollowDomainExceptions ) {
1379                                 $bits = wfParseUrl( $url );
1380                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1381                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1382                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1383                                                         unset( $attribs['rel'] );
1384                                                         break;
1385                                                 }
1386                                         }
1387                                 }
1388                         }
1389                 }
1390                 if ( $this->mOptions->getExternalLinkTarget() ) {
1391                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1392                 }
1393                 return $attribs;
1394         }
1395
1396
1397         /**
1398          * Replace unusual URL escape codes with their equivalent characters
1399          * @param string
1400          * @return string
1401          * @static
1402          * @todo  This can merge genuinely required bits in the path or query string,
1403          *        breaking legit URLs. A proper fix would treat the various parts of
1404          *        the URL differently; as a workaround, just use the output for
1405          *        statistical records, not for actual linking/output.
1406          */
1407         static function replaceUnusualEscapes( $url ) {
1408                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1409                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1410         }
1411
1412         /**
1413          * Callback function used in replaceUnusualEscapes().
1414          * Replaces unusual URL escape codes with their equivalent character
1415          * @static
1416          * @private
1417          */
1418         private static function replaceUnusualEscapesCallback( $matches ) {
1419                 $char = urldecode( $matches[0] );
1420                 $ord = ord( $char );
1421                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1422                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1423                         // No, shouldn't be escaped
1424                         return $char;
1425                 } else {
1426                         // Yes, leave it escaped
1427                         return $matches[0];
1428                 }
1429         }
1430
1431         /**
1432          * make an image if it's allowed, either through the global
1433          * option, through the exception, or through the on-wiki whitelist
1434          * @private
1435          */
1436         function maybeMakeExternalImage( $url ) {
1437                 $sk = $this->mOptions->getSkin();
1438                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1439                 $imagesexception = !empty( $imagesfrom );
1440                 $text = false;
1441                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1442                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1443                         $imagematch = false;
1444                         foreach ( $imagesfrom as $match ) {
1445                                 if ( strpos( $url, $match ) === 0 ) {
1446                                         $imagematch = true;
1447                                         break;
1448                                 }
1449                         }
1450                 } elseif ( $imagesexception ) {
1451                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1452                 } else {
1453                         $imagematch = false;
1454                 }
1455                 if ( $this->mOptions->getAllowExternalImages()
1456                      || ( $imagesexception && $imagematch ) ) {
1457                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1458                                 # Image found
1459                                 $text = $sk->makeExternalImage( $url );
1460                         }
1461                 }
1462                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1463                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1464                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1465                         foreach ( $whitelist as $entry ) {
1466                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1467                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1468                                         continue;
1469                                 }
1470                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1471                                         # Image matches a whitelist entry
1472                                         $text = $sk->makeExternalImage( $url );
1473                                         break;
1474                                 }
1475                         }
1476                 }
1477                 return $text;
1478         }
1479
1480         /**
1481          * Process [[ ]] wikilinks
1482          * @return processed text
1483          *
1484          * @private
1485          */
1486         function replaceInternalLinks( $s ) {
1487                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1488                 return $s;
1489         }
1490
1491         /**
1492          * Process [[ ]] wikilinks (RIL)
1493          * @return LinkHolderArray
1494          *
1495          * @private
1496          */
1497         function replaceInternalLinks2( &$s ) {
1498                 global $wgContLang;
1499
1500                 wfProfileIn( __METHOD__ );
1501
1502                 wfProfileIn( __METHOD__.'-setup' );
1503                 static $tc = FALSE, $e1, $e1_img;
1504                 # the % is needed to support urlencoded titles as well
1505                 if ( !$tc ) {
1506                         $tc = Title::legalChars() . '#%';
1507                         # Match a link having the form [[namespace:link|alternate]]trail
1508                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1509                         # Match cases where there is no "]]", which might still be images
1510                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1511                 }
1512
1513                 $sk = $this->mOptions->getSkin();
1514                 $holders = new LinkHolderArray( $this );
1515
1516                 # split the entire text string on occurences of [[
1517                 $a = StringUtils::explode( '[[', ' ' . $s );
1518                 # get the first element (all text up to first [[), and remove the space we added
1519                 $s = $a->current();
1520                 $a->next();
1521                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1522                 $s = substr( $s, 1 );
1523
1524                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1525                 $e2 = null;
1526                 if ( $useLinkPrefixExtension ) {
1527                         # Match the end of a line for a word that's not followed by whitespace,
1528                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1529                         $e2 = wfMsgForContent( 'linkprefix' );
1530                 }
1531
1532                 if ( is_null( $this->mTitle ) ) {
1533                         wfProfileOut( __METHOD__.'-setup' );
1534                         wfProfileOut( __METHOD__ );
1535                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1536                 }
1537                 $nottalk = !$this->mTitle->isTalkPage();
1538
1539                 if ( $useLinkPrefixExtension ) {
1540                         $m = array();
1541                         if ( preg_match( $e2, $s, $m ) ) {
1542                                 $first_prefix = $m[2];
1543                         } else {
1544                                 $first_prefix = false;
1545                         }
1546                 } else {
1547                         $prefix = '';
1548                 }
1549
1550                 if ( $wgContLang->hasVariants() ) {
1551                         $selflink = $wgContLang->convertLinkToAllVariants( $this->mTitle->getPrefixedText() );
1552                 } else {
1553                         $selflink = array( $this->mTitle->getPrefixedText() );
1554                 }
1555                 $useSubpages = $this->areSubpagesAllowed();
1556                 wfProfileOut( __METHOD__.'-setup' );
1557
1558                 # Loop for each link
1559                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1560                         # Check for excessive memory usage
1561                         if ( $holders->isBig() ) {
1562                                 # Too big
1563                                 # Do the existence check, replace the link holders and clear the array
1564                                 $holders->replace( $s );
1565                                 $holders->clear();
1566                         }
1567
1568                         if ( $useLinkPrefixExtension ) {
1569                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1570                                 if ( preg_match( $e2, $s, $m ) ) {
1571                                         $prefix = $m[2];
1572                                         $s = $m[1];
1573                                 } else {
1574                                         $prefix='';
1575                                 }
1576                                 # first link
1577                                 if ( $first_prefix ) {
1578                                         $prefix = $first_prefix;
1579                                         $first_prefix = false;
1580                                 }
1581                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1582                         }
1583
1584                         $might_be_img = false;
1585
1586                         wfProfileIn( __METHOD__."-e1" );
1587                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1588                                 $text = $m[2];
1589                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1590                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1591                                 # the real problem is with the $e1 regex
1592                                 # See bug 1300.
1593                                 #
1594                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1595                                 # and no image is in sight. See bug 2095.
1596                                 #
1597                                 if ( $text !== '' &&
1598                                         substr( $m[3], 0, 1 ) === ']' &&
1599                                         strpos( $text, '[' ) !== false
1600                                 )
1601                                 {
1602                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1603                                         $m[3] = substr( $m[3], 1 );
1604                                 }
1605                                 # fix up urlencoded title texts
1606                                 if ( strpos( $m[1], '%' ) !== false ) {
1607                                         # Should anchors '#' also be rejected?
1608                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode( $m[1] ) );
1609                                 }
1610                                 $trail = $m[3];
1611                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1612                                 $might_be_img = true;
1613                                 $text = $m[2];
1614                                 if ( strpos( $m[1], '%' ) !== false ) {
1615                                         $m[1] = urldecode( $m[1] );
1616                                 }
1617                                 $trail = "";
1618                         } else { # Invalid form; output directly
1619                                 $s .= $prefix . '[[' . $line ;
1620                                 wfProfileOut( __METHOD__."-e1" );
1621                                 continue;
1622                         }
1623                         wfProfileOut( __METHOD__."-e1" );
1624                         wfProfileIn( __METHOD__."-misc" );
1625
1626                         # Don't allow internal links to pages containing
1627                         # PROTO: where PROTO is a valid URL protocol; these
1628                         # should be external links.
1629                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1630                                 $s .= $prefix . '[[' . $line ;
1631                                 wfProfileOut( __METHOD__."-misc" );
1632                                 continue;
1633                         }
1634
1635                         # Make subpage if necessary
1636                         if ( $useSubpages ) {
1637                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1638                         } else {
1639                                 $link = $m[1];
1640                         }
1641
1642                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1643                         if ( !$noforce ) {
1644                                 # Strip off leading ':'
1645                                 $link = substr( $link, 1 );
1646                         }
1647
1648                         wfProfileOut( __METHOD__."-misc" );
1649                         wfProfileIn( __METHOD__."-title" );
1650                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1651                         if ( $nt === null ) {
1652                                 $s .= $prefix . '[[' . $line;
1653                                 wfProfileOut( __METHOD__."-title" );
1654                                 continue;
1655                         }
1656
1657                         $ns = $nt->getNamespace();
1658                         $iw = $nt->getInterWiki();
1659                         wfProfileOut( __METHOD__."-title" );
1660
1661                         if ( $might_be_img ) { # if this is actually an invalid link
1662                                 wfProfileIn( __METHOD__."-might_be_img" );
1663                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1664                                         $found = false;
1665                                         while ( true ) {
1666                                                 # look at the next 'line' to see if we can close it there
1667                                                 $a->next();
1668                                                 $next_line = $a->current();
1669                                                 if ( $next_line === false || $next_line === null ) {
1670                                                         break;
1671                                                 }
1672                                                 $m = explode( ']]', $next_line, 3 );
1673                                                 if ( count( $m ) == 3 ) {
1674                                                         # the first ]] closes the inner link, the second the image
1675                                                         $found = true;
1676                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1677                                                         $trail = $m[2];
1678                                                         break;
1679                                                 } elseif ( count( $m ) == 2 ) {
1680                                                         # if there's exactly one ]] that's fine, we'll keep looking
1681                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1682                                                 } else {
1683                                                         # if $next_line is invalid too, we need look no further
1684                                                         $text .= '[[' . $next_line;
1685                                                         break;
1686                                                 }
1687                                         }
1688                                         if ( !$found ) {
1689                                                 # we couldn't find the end of this imageLink, so output it raw
1690                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1691                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1692                                                 $s .= "{$prefix}[[$link|$text";
1693                                                 # note: no $trail, because without an end, there *is* no trail
1694                                                 wfProfileOut( __METHOD__."-might_be_img" );
1695                                                 continue;
1696                                         }
1697                                 } else { # it's not an image, so output it raw
1698                                         $s .= "{$prefix}[[$link|$text";
1699                                         # note: no $trail, because without an end, there *is* no trail
1700                                         wfProfileOut( __METHOD__."-might_be_img" );
1701                                         continue;
1702                                 }
1703                                 wfProfileOut( __METHOD__."-might_be_img" );
1704                         }
1705
1706                         $wasblank = ( $text  == '' );
1707                         if ( $wasblank ) {
1708                                 $text = $link;
1709                         }
1710
1711                         # Link not escaped by : , create the various objects
1712                         if ( $noforce ) {
1713
1714                                 # Interwikis
1715                                 wfProfileIn( __METHOD__."-interwiki" );
1716                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1717                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1718                                         $s = rtrim( $s . $prefix );
1719                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1720                                         wfProfileOut( __METHOD__."-interwiki" );
1721                                         continue;
1722                                 }
1723                                 wfProfileOut( __METHOD__."-interwiki" );
1724
1725                                 if ( $ns == NS_FILE ) {
1726                                         wfProfileIn( __METHOD__."-image" );
1727                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1728                                                 if ( $wasblank ) {
1729                                                         # if no parameters were passed, $text
1730                                                         # becomes something like "File:Foo.png",
1731                                                         # which we don't want to pass on to the
1732                                                         # image generator
1733                                                         $text = '';
1734                                                 } else {
1735                                                         # recursively parse links inside the image caption
1736                                                         # actually, this will parse them in any other parameters, too,
1737                                                         # but it might be hard to fix that, and it doesn't matter ATM
1738                                                         $text = $this->replaceExternalLinks( $text );
1739                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1740                                                 }
1741                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1742                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1743                                         } else {
1744                                                 $s .= $prefix . $trail;
1745                                         }
1746                                         $this->mOutput->addImage( $nt->getDBkey() );
1747                                         wfProfileOut( __METHOD__."-image" );
1748                                         continue;
1749
1750                                 }
1751
1752                                 if ( $ns == NS_CATEGORY ) {
1753                                         wfProfileIn( __METHOD__."-category" );
1754                                         $s = rtrim( $s . "\n" ); # bug 87
1755
1756                                         if ( $wasblank ) {
1757                                                 $sortkey = $this->getDefaultSort();
1758                                         } else {
1759                                                 $sortkey = $text;
1760                                         }
1761                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1762                                         $sortkey = str_replace( "\n", '', $sortkey );
1763                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1764                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1765
1766                                         /**
1767                                          * Strip the whitespace Category links produce, see bug 87
1768                                          * @todo We might want to use trim($tmp, "\n") here.
1769                                          */
1770                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '': $prefix . $trail;
1771
1772                                         wfProfileOut( __METHOD__."-category" );
1773                                         continue;
1774                                 }
1775                         }
1776
1777                         # Self-link checking
1778                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1779                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1780                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1781                                         continue;
1782                                 }
1783                         }
1784
1785                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1786                         # FIXME: Should do batch file existence checks, see comment below
1787                         if ( $ns == NS_MEDIA ) {
1788                                 wfProfileIn( __METHOD__."-media" );
1789                                 # Give extensions a chance to select the file revision for us
1790                                 $skip = $time = false;
1791                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1792                                 if ( $skip ) {
1793                                         $link = $sk->link( $nt );
1794                                 } else {
1795                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1796                                 }
1797                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1798                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1799                                 $this->mOutput->addImage( $nt->getDBkey() );
1800                                 wfProfileOut( __METHOD__."-media" );
1801                                 continue;
1802                         }
1803
1804                         wfProfileIn( __METHOD__."-always_known" );
1805                         # Some titles, such as valid special pages or files in foreign repos, should
1806                         # be shown as bluelinks even though they're not included in the page table
1807                         #
1808                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1809                         # batch file existence checks for NS_FILE and NS_MEDIA
1810                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1811                                 $this->mOutput->addLink( $nt );
1812                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1813                         } else {
1814                                 # Links will be added to the output link list after checking
1815                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1816                         }
1817                         wfProfileOut( __METHOD__."-always_known" );
1818                 }
1819                 wfProfileOut( __METHOD__ );
1820                 return $holders;
1821         }
1822
1823         /**
1824          * Make a link placeholder. The text returned can be later resolved to a real link with
1825          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1826          * parsing of interwiki links, and secondly to allow all existence checks and
1827          * article length checks (for stub links) to be bundled into a single query.
1828          *
1829          * @deprecated
1830          */
1831         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1832                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1833         }
1834
1835         /**
1836          * Render a forced-blue link inline; protect against double expansion of
1837          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1838          * Since this little disaster has to split off the trail text to avoid
1839          * breaking URLs in the following text without breaking trails on the
1840          * wiki links, it's been made into a horrible function.
1841          *
1842          * @param Title $nt
1843          * @param string $text
1844          * @param string $query
1845          * @param string $trail
1846          * @param string $prefix
1847          * @return string HTML-wikitext mix oh yuck
1848          */
1849         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1850                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1851                 $sk = $this->mOptions->getSkin();
1852                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1853                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1854                 return $this->armorLinks( $link ) . $trail;
1855         }
1856
1857         /**
1858          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1859          * going to go through further parsing steps before inline URL expansion.
1860          *
1861          * Not needed quite as much as it used to be since free links are a bit
1862          * more sensible these days. But bracketed links are still an issue.
1863          *
1864          * @param string more-or-less HTML
1865          * @return string less-or-more HTML with NOPARSE bits
1866          */
1867         function armorLinks( $text ) {
1868                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1869                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1870         }
1871
1872         /**
1873          * Return true if subpage links should be expanded on this page.
1874          * @return bool
1875          */
1876         function areSubpagesAllowed() {
1877                 # Some namespaces don't allow subpages
1878                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1879         }
1880
1881         /**
1882          * Handle link to subpage if necessary
1883          * @param string $target the source of the link
1884          * @param string &$text the link text, modified as necessary
1885          * @return string the full name of the link
1886          * @private
1887          */
1888         function maybeDoSubpageLink( $target, &$text ) {
1889                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1890         }
1891
1892         /**#@+
1893          * Used by doBlockLevels()
1894          * @private
1895          */
1896         /* private */ function closeParagraph() {
1897                 $result = '';
1898                 if ( $this->mLastSection != '' ) {
1899                         $result = '</' . $this->mLastSection  . ">\n";
1900                 }
1901                 $this->mInPre = false;
1902                 $this->mLastSection = '';
1903                 return $result;
1904         }
1905         /**
1906          * getCommon() returns the length of the longest common substring
1907          * of both arguments, starting at the beginning of both.
1908          * @private
1909          */
1910         function getCommon( $st1, $st2 ) {
1911                 $fl = strlen( $st1 );
1912                 $shorter = strlen( $st2 );
1913                 if ( $fl < $shorter ) {
1914                         $shorter = $fl;
1915                 }
1916
1917                 for ( $i = 0; $i < $shorter; ++$i ) {
1918                         if ( $st1{$i} != $st2{$i} ) {
1919                                 break;
1920                         }
1921                 }
1922                 return $i;
1923         }
1924         /**
1925          * These next three functions open, continue, and close the list
1926          * element appropriate to the prefix character passed into them.
1927          * @private
1928          */
1929         function openList( $char ) {
1930                 $result = $this->closeParagraph();
1931
1932                 if ( '*' === $char ) {
1933                         $result .= '<ul><li>';
1934                 } elseif ( '#' === $char ) {
1935                         $result .= '<ol><li>';
1936                 } elseif ( ':' === $char ) {
1937                         $result .= '<dl><dd>';
1938                 } elseif ( ';' === $char ) {
1939                         $result .= '<dl><dt>';
1940                         $this->mDTopen = true;
1941                 } else {
1942                         $result = '<!-- ERR 1 -->';
1943                 }
1944
1945                 return $result;
1946         }
1947
1948         /* private */ function nextItem( $char ) {
1949                 if ( '*' === $char || '#' === $char ) {
1950                         return '</li><li>';
1951                 } elseif ( ':' === $char || ';' === $char ) {
1952                         $close = '</dd>';
1953                         if ( $this->mDTopen ) {
1954                                 $close = '</dt>';
1955                         }
1956                         if ( ';' === $char ) {
1957                                 $this->mDTopen = true;
1958                                 return $close . '<dt>';
1959                         } else {
1960                                 $this->mDTopen = false;
1961                                 return $close . '<dd>';
1962                         }
1963                 }
1964                 return '<!-- ERR 2 -->';
1965         }
1966
1967         /* private */ function closeList( $char ) {
1968                 if ( '*' === $char ) {
1969                         $text = '</li></ul>';
1970                 } elseif ( '#' === $char ) {
1971                         $text = '</li></ol>';
1972                 } elseif ( ':' === $char ) {
1973                         if ( $this->mDTopen ) {
1974                                 $this->mDTopen = false;
1975                                 $text = '</dt></dl>';
1976                         } else {
1977                                 $text = '</dd></dl>';
1978                         }
1979                 } else {
1980                         return '<!-- ERR 3 -->';
1981                 }
1982                 return $text."\n";
1983         }
1984         /**#@-*/
1985
1986         /**
1987          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1988          *
1989          * @param $linestart bool whether or not this is at the start of a line.
1990          * @private
1991          * @return string the lists rendered as HTML
1992          */
1993         function doBlockLevels( $text, $linestart ) {
1994                 wfProfileIn( __METHOD__ );
1995
1996                 # Parsing through the text line by line.  The main thing
1997                 # happening here is handling of block-level elements p, pre,
1998                 # and making lists from lines starting with * # : etc.
1999                 #
2000                 $textLines = StringUtils::explode( "\n", $text );
2001
2002                 $lastPrefix = $output = '';
2003                 $this->mDTopen = $inBlockElem = false;
2004                 $prefixLength = 0;
2005                 $paragraphStack = false;
2006
2007                 foreach ( $textLines as $oLine ) {
2008                         # Fix up $linestart
2009                         if ( !$linestart ) {
2010                                 $output .= $oLine;
2011                                 $linestart = true;
2012                                 continue;
2013                         }
2014                         // * = ul
2015                         // # = ol
2016                         // ; = dt
2017                         // : = dd
2018
2019                         $lastPrefixLength = strlen( $lastPrefix );
2020                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2021                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2022                         // If not in a <pre> element, scan for and figure out what prefixes are there.
2023                         if ( !$this->mInPre ) {
2024                                 # Multiple prefixes may abut each other for nested lists.
2025                                 $prefixLength = strspn( $oLine, '*#:;' );
2026                                 $prefix = substr( $oLine, 0, $prefixLength );
2027
2028                                 # eh?
2029                                 // ; and : are both from definition-lists, so they're equivalent
2030                                 //  for the purposes of determining whether or not we need to open/close
2031                                 //  elements.
2032                                 $prefix2 = str_replace( ';', ':', $prefix );
2033                                 $t = substr( $oLine, $prefixLength );
2034                                 $this->mInPre = (bool)$preOpenMatch;
2035                         } else {
2036                                 # Don't interpret any other prefixes in preformatted text
2037                                 $prefixLength = 0;
2038                                 $prefix = $prefix2 = '';
2039                                 $t = $oLine;
2040                         }
2041
2042                         # List generation
2043                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2044                                 # Same as the last item, so no need to deal with nesting or opening stuff
2045                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2046                                 $paragraphStack = false;
2047
2048                                 if ( substr( $prefix, -1 ) === ';') {
2049                                         # The one nasty exception: definition lists work like this:
2050                                         # ; title : definition text
2051                                         # So we check for : in the remainder text to split up the
2052                                         # title and definition, without b0rking links.
2053                                         $term = $t2 = '';
2054                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2055                                                 $t = $t2;
2056                                                 $output .= $term . $this->nextItem( ':' );
2057                                         }
2058                                 }
2059                         } elseif ( $prefixLength || $lastPrefixLength ) {
2060                                 // We need to open or close prefixes, or both.
2061
2062                                 # Either open or close a level...
2063                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2064                                 $paragraphStack = false;
2065
2066                                 // Close all the prefixes which aren't shared.
2067                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2068                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2069                                         --$lastPrefixLength;
2070                                 }
2071
2072                                 // Continue the current prefix if appropriate.
2073                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2074                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2075                                 }
2076
2077                                 // Open prefixes where appropriate.
2078                                 while ( $prefixLength > $commonPrefixLength ) {
2079                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2080                                         $output .= $this->openList( $char );
2081
2082                                         if ( ';' === $char ) {
2083                                                 # FIXME: This is dupe of code above
2084                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2085                                                         $t = $t2;
2086                                                         $output .= $term . $this->nextItem( ':' );
2087                                                 }
2088                                         }
2089                                         ++$commonPrefixLength;
2090                                 }
2091                                 $lastPrefix = $prefix2;
2092                         }
2093
2094                         // If we have no prefixes, go to paragraph mode.
2095                         if ( 0 == $prefixLength ) {
2096                                 wfProfileIn( __METHOD__."-paragraph" );
2097                                 # No prefix (not in list)--go to paragraph mode
2098                                 // XXX: use a stack for nestable elements like span, table and div
2099                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2100                                 $closematch = preg_match(
2101                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2102                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2103                                 if ( $openmatch or $closematch ) {
2104                                         $paragraphStack = false;
2105                                         # TODO bug 5718: paragraph closed
2106                                         $output .= $this->closeParagraph();
2107                                         if ( $preOpenMatch and !$preCloseMatch ) {
2108                                                 $this->mInPre = true;
2109                                         }
2110                                         if ( $closematch ) {
2111                                                 $inBlockElem = false;
2112                                         } else {
2113                                                 $inBlockElem = true;
2114                                         }
2115                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2116                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2117                                                 // pre
2118                                                 if ( $this->mLastSection !== 'pre' ) {
2119                                                         $paragraphStack = false;
2120                                                         $output .= $this->closeParagraph().'<pre>';
2121                                                         $this->mLastSection = 'pre';
2122                                                 }
2123                                                 $t = substr( $t, 1 );
2124                                         } else {
2125                                                 // paragraph
2126                                                 if ( trim( $t ) == '' ) {
2127                                                         if ( $paragraphStack ) {
2128                                                                 $output .= $paragraphStack.'<br />';
2129                                                                 $paragraphStack = false;
2130                                                                 $this->mLastSection = 'p';
2131                                                         } else {
2132                                                                 if ( $this->mLastSection !== 'p' ) {
2133                                                                         $output .= $this->closeParagraph();
2134                                                                         $this->mLastSection = '';
2135                                                                         $paragraphStack = '<p>';
2136                                                                 } else {
2137                                                                         $paragraphStack = '</p><p>';
2138                                                                 }
2139                                                         }
2140                                                 } else {
2141                                                         if ( $paragraphStack ) {
2142                                                                 $output .= $paragraphStack;
2143                                                                 $paragraphStack = false;
2144                                                                 $this->mLastSection = 'p';
2145                                                         } elseif ( $this->mLastSection !== 'p' ) {
2146                                                                 $output .= $this->closeParagraph().'<p>';
2147                                                                 $this->mLastSection = 'p';
2148                                                         }
2149                                                 }
2150                                         }
2151                                 }
2152                                 wfProfileOut( __METHOD__."-paragraph" );
2153                         }
2154                         // somewhere above we forget to get out of pre block (bug 785)
2155                         if ( $preCloseMatch && $this->mInPre ) {
2156                                 $this->mInPre = false;
2157                         }
2158                         if ( $paragraphStack === false ) {
2159                                 $output .= $t."\n";
2160                         }
2161                 }
2162                 while ( $prefixLength ) {
2163                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2164                         --$prefixLength;
2165                 }
2166                 if ( $this->mLastSection != '' ) {
2167                         $output .= '</' . $this->mLastSection . '>';
2168                         $this->mLastSection = '';
2169                 }
2170
2171                 wfProfileOut( __METHOD__ );
2172                 return $output;
2173         }
2174
2175         /**
2176          * Split up a string on ':', ignoring any occurences inside tags
2177          * to prevent illegal overlapping.
2178          * @param string $str the string to split
2179          * @param string &$before set to everything before the ':'
2180          * @param string &$after set to everything after the ':'
2181          * return string the position of the ':', or false if none found
2182          */
2183         function findColonNoLinks( $str, &$before, &$after ) {
2184                 wfProfileIn( __METHOD__ );
2185
2186                 $pos = strpos( $str, ':' );
2187                 if ( $pos === false ) {
2188                         // Nothing to find!
2189                         wfProfileOut( __METHOD__ );
2190                         return false;
2191                 }
2192
2193                 $lt = strpos( $str, '<' );
2194                 if ( $lt === false || $lt > $pos ) {
2195                         // Easy; no tag nesting to worry about
2196                         $before = substr( $str, 0, $pos );
2197                         $after = substr( $str, $pos+1 );
2198                         wfProfileOut( __METHOD__ );
2199                         return $pos;
2200                 }
2201
2202                 // Ugly state machine to walk through avoiding tags.
2203                 $state = self::COLON_STATE_TEXT;
2204                 $stack = 0;
2205                 $len = strlen( $str );
2206                 for( $i = 0; $i < $len; $i++ ) {
2207                         $c = $str{$i};
2208
2209                         switch( $state ) {
2210                         // (Using the number is a performance hack for common cases)
2211                         case 0: // self::COLON_STATE_TEXT:
2212                                 switch( $c ) {
2213                                 case "<":
2214                                         // Could be either a <start> tag or an </end> tag
2215                                         $state = self::COLON_STATE_TAGSTART;
2216                                         break;
2217                                 case ":":
2218                                         if ( $stack == 0 ) {
2219                                                 // We found it!
2220                                                 $before = substr( $str, 0, $i );
2221                                                 $after = substr( $str, $i + 1 );
2222                                                 wfProfileOut( __METHOD__ );
2223                                                 return $i;
2224                                         }
2225                                         // Embedded in a tag; don't break it.
2226                                         break;
2227                                 default:
2228                                         // Skip ahead looking for something interesting
2229                                         $colon = strpos( $str, ':', $i );
2230                                         if ( $colon === false ) {
2231                                                 // Nothing else interesting
2232                                                 wfProfileOut( __METHOD__ );
2233                                                 return false;
2234                                         }
2235                                         $lt = strpos( $str, '<', $i );
2236                                         if ( $stack === 0 ) {
2237                                                 if ( $lt === false || $colon < $lt ) {
2238                                                         // We found it!
2239                                                         $before = substr( $str, 0, $colon );
2240                                                         $after = substr( $str, $colon + 1 );
2241                                                         wfProfileOut( __METHOD__ );
2242                                                         return $i;
2243                                                 }
2244                                         }
2245                                         if ( $lt === false ) {
2246                                                 // Nothing else interesting to find; abort!
2247                                                 // We're nested, but there's no close tags left. Abort!
2248                                                 break 2;
2249                                         }
2250                                         // Skip ahead to next tag start
2251                                         $i = $lt;
2252                                         $state = self::COLON_STATE_TAGSTART;
2253                                 }
2254                                 break;
2255                         case 1: // self::COLON_STATE_TAG:
2256                                 // In a <tag>
2257                                 switch( $c ) {
2258                                 case ">":
2259                                         $stack++;
2260                                         $state = self::COLON_STATE_TEXT;
2261                                         break;
2262                                 case "/":
2263                                         // Slash may be followed by >?
2264                                         $state = self::COLON_STATE_TAGSLASH;
2265                                         break;
2266                                 default:
2267                                         // ignore
2268                                 }
2269                                 break;
2270                         case 2: // self::COLON_STATE_TAGSTART:
2271                                 switch( $c ) {
2272                                 case "/":
2273                                         $state = self::COLON_STATE_CLOSETAG;
2274                                         break;
2275                                 case "!":
2276                                         $state = self::COLON_STATE_COMMENT;
2277                                         break;
2278                                 case ">":
2279                                         // Illegal early close? This shouldn't happen D:
2280                                         $state = self::COLON_STATE_TEXT;
2281                                         break;
2282                                 default:
2283                                         $state = self::COLON_STATE_TAG;
2284                                 }
2285                                 break;
2286                         case 3: // self::COLON_STATE_CLOSETAG:
2287                                 // In a </tag>
2288                                 if ( $c === ">" ) {
2289                                         $stack--;
2290                                         if ( $stack < 0 ) {
2291                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2292                                                 wfProfileOut( __METHOD__ );
2293                                                 return false;
2294                                         }
2295                                         $state = self::COLON_STATE_TEXT;
2296                                 }
2297                                 break;
2298                         case self::COLON_STATE_TAGSLASH:
2299                                 if ( $c === ">" ) {
2300                                         // Yes, a self-closed tag <blah/>
2301                                         $state = self::COLON_STATE_TEXT;
2302                                 } else {
2303                                         // Probably we're jumping the gun, and this is an attribute
2304                                         $state = self::COLON_STATE_TAG;
2305                                 }
2306                                 break;
2307                         case 5: // self::COLON_STATE_COMMENT:
2308                                 if ( $c === "-" ) {
2309                                         $state = self::COLON_STATE_COMMENTDASH;
2310                                 }
2311                                 break;
2312                         case self::COLON_STATE_COMMENTDASH:
2313                                 if ( $c === "-" ) {
2314                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2315                                 } else {
2316                                         $state = self::COLON_STATE_COMMENT;
2317                                 }
2318                                 break;
2319                         case self::COLON_STATE_COMMENTDASHDASH:
2320                                 if ( $c === ">" ) {
2321                                         $state = self::COLON_STATE_TEXT;
2322                                 } else {
2323                                         $state = self::COLON_STATE_COMMENT;
2324                                 }
2325                                 break;
2326                         default:
2327                                 throw new MWException( "State machine error in " . __METHOD__ );
2328                         }
2329                 }
2330                 if ( $stack > 0 ) {
2331                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2332                         return false;
2333                 }
2334                 wfProfileOut( __METHOD__ );
2335                 return false;
2336         }
2337
2338         /**
2339          * Return value of a magic variable (like PAGENAME)
2340          *
2341          * @private
2342          */
2343         function getVariableValue( $index, $frame=false ) {
2344                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2345                 global $wgScriptPath, $wgStylePath;
2346
2347                 /**
2348                  * Some of these require message or data lookups and can be
2349                  * expensive to check many times.
2350                  */
2351                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2352                         if ( isset( $this->mVarCache[$index] ) ) {
2353                                 return $this->mVarCache[$index];
2354                         }
2355                 }
2356
2357                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2358                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2359
2360                 # Use the time zone
2361                 global $wgLocaltimezone;
2362                 if ( isset( $wgLocaltimezone ) ) {
2363                         $oldtz = date_default_timezone_get();
2364                         date_default_timezone_set( $wgLocaltimezone );
2365                 }
2366
2367                 $localTimestamp = date( 'YmdHis', $ts );
2368                 $localMonth = date( 'm', $ts );
2369                 $localMonth1 = date( 'n', $ts );
2370                 $localMonthName = date( 'n', $ts );
2371                 $localDay = date( 'j', $ts );
2372                 $localDay2 = date( 'd', $ts );
2373                 $localDayOfWeek = date( 'w', $ts );
2374                 $localWeek = date( 'W', $ts );
2375                 $localYear = date( 'Y', $ts );
2376                 $localHour = date( 'H', $ts );
2377                 if ( isset( $wgLocaltimezone ) ) {
2378                         date_default_timezone_set( $oldtz );
2379                 }
2380
2381                 switch ( $index ) {
2382                         case 'currentmonth':
2383                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2384                                 break;
2385                         case 'currentmonth1':
2386                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2387                                 break;
2388                         case 'currentmonthname':
2389                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2390                                 break;
2391                         case 'currentmonthnamegen':
2392                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2393                                 break;
2394                         case 'currentmonthabbrev':
2395                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2396                                 break;
2397                         case 'currentday':
2398                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2399                                 break;
2400                         case 'currentday2':
2401                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2402                                 break;
2403                         case 'localmonth':
2404                                 $value = $wgContLang->formatNum( $localMonth );
2405                                 break;
2406                         case 'localmonth1':
2407                                 $value = $wgContLang->formatNum( $localMonth1 );
2408                                 break;
2409                         case 'localmonthname':
2410                                 $value = $wgContLang->getMonthName( $localMonthName );
2411                                 break;
2412                         case 'localmonthnamegen':
2413                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2414                                 break;
2415                         case 'localmonthabbrev':
2416                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2417                                 break;
2418                         case 'localday':
2419                                 $value = $wgContLang->formatNum( $localDay );
2420                                 break;
2421                         case 'localday2':
2422                                 $value = $wgContLang->formatNum( $localDay2 );
2423                                 break;
2424                         case 'pagename':
2425                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2426                                 break;
2427                         case 'pagenamee':
2428                                 $value = $this->mTitle->getPartialURL();
2429                                 break;
2430                         case 'fullpagename':
2431                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2432                                 break;
2433                         case 'fullpagenamee':
2434                                 $value = $this->mTitle->getPrefixedURL();
2435                                 break;
2436                         case 'subpagename':
2437                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2438                                 break;
2439                         case 'subpagenamee':
2440                                 $value = $this->mTitle->getSubpageUrlForm();
2441                                 break;
2442                         case 'basepagename':
2443                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2444                                 break;
2445                         case 'basepagenamee':
2446                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2447                                 break;
2448                         case 'talkpagename':
2449                                 if ( $this->mTitle->canTalk() ) {
2450                                         $talkPage = $this->mTitle->getTalkPage();
2451                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2452                                 } else {
2453                                         $value = '';
2454                                 }
2455                                 break;
2456                         case 'talkpagenamee':
2457                                 if ( $this->mTitle->canTalk() ) {
2458                                         $talkPage = $this->mTitle->getTalkPage();
2459                                         $value = $talkPage->getPrefixedUrl();
2460                                 } else {
2461                                         $value = '';
2462                                 }
2463                                 break;
2464                         case 'subjectpagename':
2465                                 $subjPage = $this->mTitle->getSubjectPage();
2466                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2467                                 break;
2468                         case 'subjectpagenamee':
2469                                 $subjPage = $this->mTitle->getSubjectPage();
2470                                 $value = $subjPage->getPrefixedUrl();
2471                                 break;
2472                         case 'revisionid':
2473                                 // Let the edit saving system know we should parse the page
2474                                 // *after* a revision ID has been assigned.
2475                                 $this->mOutput->setFlag( 'vary-revision' );
2476                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2477                                 $value = $this->mRevisionId;
2478                                 break;
2479                         case 'revisionday':
2480                                 // Let the edit saving system know we should parse the page
2481                                 // *after* a revision ID has been assigned. This is for null edits.
2482                                 $this->mOutput->setFlag( 'vary-revision' );
2483                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2484                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2485                                 break;
2486                         case 'revisionday2':
2487                                 // Let the edit saving system know we should parse the page
2488                                 // *after* a revision ID has been assigned. This is for null edits.
2489                                 $this->mOutput->setFlag( 'vary-revision' );
2490                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2491                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2492                                 break;
2493                         case 'revisionmonth':
2494                                 // Let the edit saving system know we should parse the page
2495                                 // *after* a revision ID has been assigned. This is for null edits.
2496                                 $this->mOutput->setFlag( 'vary-revision' );
2497                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2498                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2499                                 break;
2500                         case 'revisionyear':
2501                                 // Let the edit saving system know we should parse the page
2502                                 // *after* a revision ID has been assigned. This is for null edits.
2503                                 $this->mOutput->setFlag( 'vary-revision' );
2504                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2505                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2506                                 break;
2507                         case 'revisiontimestamp':
2508                                 // Let the edit saving system know we should parse the page
2509                                 // *after* a revision ID has been assigned. This is for null edits.
2510                                 $this->mOutput->setFlag( 'vary-revision' );
2511                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2512                                 $value = $this->getRevisionTimestamp();
2513                                 break;
2514                         case 'revisionuser':
2515                                 // Let the edit saving system know we should parse the page
2516                                 // *after* a revision ID has been assigned. This is for null edits.
2517                                 $this->mOutput->setFlag( 'vary-revision' );
2518                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2519                                 $value = $this->getRevisionUser();
2520                                 break;
2521                         case 'namespace':
2522                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2523                                 break;
2524                         case 'namespacee':
2525                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2526                                 break;
2527                         case 'talkspace':
2528                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2529                                 break;
2530                         case 'talkspacee':
2531                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2532                                 break;
2533                         case 'subjectspace':
2534                                 $value = $this->mTitle->getSubjectNsText();
2535                                 break;
2536                         case 'subjectspacee':
2537                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2538                                 break;
2539                         case 'currentdayname':
2540                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2541                                 break;
2542                         case 'currentyear':
2543                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2544                                 break;
2545                         case 'currenttime':
2546                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2547                                 break;
2548                         case 'currenthour':
2549                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2550                                 break;
2551                         case 'currentweek':
2552                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2553                                 // int to remove the padding
2554                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2555                                 break;
2556                         case 'currentdow':
2557                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2558                                 break;
2559                         case 'localdayname':
2560                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2561                                 break;
2562                         case 'localyear':
2563                                 $value = $wgContLang->formatNum( $localYear, true );
2564                                 break;
2565                         case 'localtime':
2566                                 $value = $wgContLang->time( $localTimestamp, false, false );
2567                                 break;
2568                         case 'localhour':
2569                                 $value = $wgContLang->formatNum( $localHour, true );
2570                                 break;
2571                         case 'localweek':
2572                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2573                                 // int to remove the padding
2574                                 $value = $wgContLang->formatNum( (int)$localWeek );
2575                                 break;
2576                         case 'localdow':
2577                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2578                                 break;
2579                         case 'numberofarticles':
2580                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2581                                 break;
2582                         case 'numberoffiles':
2583                                 $value = $wgContLang->formatNum( SiteStats::images() );
2584                                 break;
2585                         case 'numberofusers':
2586                                 $value = $wgContLang->formatNum( SiteStats::users() );
2587                                 break;
2588                         case 'numberofactiveusers':
2589                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2590                                 break;
2591                         case 'numberofpages':
2592                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2593                                 break;
2594                         case 'numberofadmins':
2595                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2596                                 break;
2597                         case 'numberofedits':
2598                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2599                                 break;
2600                         case 'numberofviews':
2601                                 $value = $wgContLang->formatNum( SiteStats::views() );
2602                                 break;
2603                         case 'currenttimestamp':
2604                                 $value = wfTimestamp( TS_MW, $ts );
2605                                 break;
2606                         case 'localtimestamp':
2607                                 $value = $localTimestamp;
2608                                 break;
2609                         case 'currentversion':
2610                                 $value = SpecialVersion::getVersion();
2611                                 break;
2612                         case 'sitename':
2613                                 return $wgSitename;
2614                         case 'server':
2615                                 return $wgServer;
2616                         case 'servername':
2617                                 return $wgServerName;
2618                         case 'scriptpath':
2619                                 return $wgScriptPath;
2620                         case 'stylepath':
2621                                 return $wgStylePath;
2622                         case 'directionmark':
2623                                 return $wgContLang->getDirMark();
2624                         case 'contentlanguage':
2625                                 global $wgContLanguageCode;
2626                                 return $wgContLanguageCode;
2627                         default:
2628                                 $ret = null;
2629                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2630                                         return $ret;
2631                                 } else {
2632                                         return null;
2633                                 }
2634                 }
2635
2636                 if ( $index )
2637                         $this->mVarCache[$index] = $value;
2638
2639                 return $value;
2640         }
2641
2642         /**
2643          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2644          *
2645          * @private
2646          */
2647         function initialiseVariables() {
2648                 wfProfileIn( __METHOD__ );
2649                 $variableIDs = MagicWord::getVariableIDs();
2650                 $substIDs = MagicWord::getSubstIDs();
2651
2652                 $this->mVariables = new MagicWordArray( $variableIDs );
2653                 $this->mSubstWords = new MagicWordArray( $substIDs );
2654                 wfProfileOut( __METHOD__ );
2655         }
2656
2657         /**
2658          * Preprocess some wikitext and return the document tree.
2659          * This is the ghost of replace_variables().
2660          *
2661          * @param string $text The text to parse
2662          * @param integer flags Bitwise combination of:
2663          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2664          *                                     included. Default is to assume a direct page view.
2665          *
2666          * The generated DOM tree must depend only on the input text and the flags.
2667          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2668          *
2669          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2670          * change in the DOM tree for a given text, must be passed through the section identifier
2671          * in the section edit link and thus back to extractSections().
2672          *
2673          * The output of this function is currently only cached in process memory, but a persistent
2674          * cache may be implemented at a later date which takes further advantage of these strict
2675          * dependency requirements.
2676          *
2677          * @private
2678          */
2679         function preprocessToDom( $text, $flags = 0 ) {
2680                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2681                 return $dom;
2682         }
2683
2684         /*
2685          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2686          */
2687         public static function splitWhitespace( $s ) {
2688                 $ltrimmed = ltrim( $s );
2689                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2690                 $trimmed = rtrim( $ltrimmed );
2691                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2692                 if ( $diff > 0 ) {
2693                         $w2 = substr( $ltrimmed, -$diff );
2694                 } else {
2695                         $w2 = '';
2696                 }
2697                 return array( $w1, $trimmed, $w2 );
2698         }
2699
2700         /**
2701          * Replace magic variables, templates, and template arguments
2702          * with the appropriate text. Templates are substituted recursively,
2703          * taking care to avoid infinite loops.
2704          *
2705          * Note that the substitution depends on value of $mOutputType:
2706          *  self::OT_WIKI: only {{subst:}} templates
2707          *  self::OT_PREPROCESS: templates but not extension tags
2708          *  self::OT_HTML: all templates and extension tags
2709          *
2710          * @param string $tex The text to transform
2711          * @param PPFrame $frame Object describing the arguments passed to the template.
2712          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2713          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2714          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2715          * @private
2716          */
2717         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2718                 # Is there any text? Also, Prevent too big inclusions!
2719                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2720                         return $text;
2721                 }
2722                 wfProfileIn( __METHOD__ );
2723
2724                 if ( $frame === false ) {
2725                         $frame = $this->getPreprocessor()->newFrame();
2726                 } elseif ( !( $frame instanceof PPFrame ) ) {
2727                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2728                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2729                 }
2730
2731                 $dom = $this->preprocessToDom( $text );
2732                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2733                 $text = $frame->expand( $dom, $flags );
2734
2735                 wfProfileOut( __METHOD__ );
2736                 return $text;
2737         }
2738
2739         // Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2740         static function createAssocArgs( $args ) {
2741                 $assocArgs = array();
2742                 $index = 1;
2743                 foreach ( $args as $arg ) {
2744                         $eqpos = strpos( $arg, '=' );
2745                         if ( $eqpos === false ) {
2746                                 $assocArgs[$index++] = $arg;
2747                         } else {
2748                                 $name = trim( substr( $arg, 0, $eqpos ) );
2749                                 $value = trim( substr( $arg, $eqpos+1 ) );
2750                                 if ( $value === false ) {
2751                                         $value = '';
2752                                 }
2753                                 if ( $name !== false ) {
2754                                         $assocArgs[$name] = $value;
2755                                 }
2756                         }
2757                 }
2758
2759                 return $assocArgs;
2760         }
2761
2762         /**
2763          * Warn the user when a parser limitation is reached
2764          * Will warn at most once the user per limitation type
2765          *
2766          * @param string $limitationType, should be one of:
2767          *   'expensive-parserfunction' (corresponding messages:
2768          *       'expensive-parserfunction-warning',
2769          *       'expensive-parserfunction-category')
2770          *   'post-expand-template-argument' (corresponding messages:
2771          *       'post-expand-template-argument-warning',
2772          *       'post-expand-template-argument-category')
2773          *   'post-expand-template-inclusion' (corresponding messages:
2774          *       'post-expand-template-inclusion-warning',
2775          *       'post-expand-template-inclusion-category')
2776          * @params int $current, $max When an explicit limit has been
2777          *       exceeded, provide the values (optional)
2778          */
2779         function limitationWarn( $limitationType, $current=null, $max=null) {
2780                 // does no harm if $current and $max are present but are unnecessary for the message
2781                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2782                 $this->mOutput->addWarning( $warning );
2783                 $this->addTrackingCategory( "$limitationType-category" );
2784         }
2785
2786         /**
2787          * Return the text of a template, after recursively
2788          * replacing any variables or templates within the template.
2789          *
2790          * @param array $piece The parts of the template
2791          *  $piece['title']: the title, i.e. the part before the |
2792          *  $piece['parts']: the parameter array
2793          *  $piece['lineStart']: whether the brace was at the start of a line
2794          * @param PPFrame The current frame, contains template arguments
2795          * @return string the text of the template
2796          * @private
2797          */
2798         function braceSubstitution( $piece, $frame ) {
2799                 global $wgContLang, $wgNonincludableNamespaces;
2800                 wfProfileIn( __METHOD__ );
2801                 wfProfileIn( __METHOD__.'-setup' );
2802
2803                 # Flags
2804                 $found = false;             # $text has been filled
2805                 $nowiki = false;            # wiki markup in $text should be escaped
2806                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2807                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2808                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2809                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2810
2811                 # Title object, where $text came from
2812                 $title = null;
2813
2814                 # $part1 is the bit before the first |, and must contain only title characters.
2815                 # Various prefixes will be stripped from it later.
2816                 $titleWithSpaces = $frame->expand( $piece['title'] );
2817                 $part1 = trim( $titleWithSpaces );
2818                 $titleText = false;
2819
2820                 # Original title text preserved for various purposes
2821                 $originalTitle = $part1;
2822
2823                 # $args is a list of argument nodes, starting from index 0, not including $part1
2824                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2825                 wfProfileOut( __METHOD__.'-setup' );
2826
2827                 # SUBST
2828                 wfProfileIn( __METHOD__.'-modifiers' );
2829                 if ( !$found ) {
2830
2831                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2832
2833                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2834                         # Decide whether to expand template or keep wikitext as-is.
2835                         if ( $this->ot['wiki'] ) {
2836                                 if ( $substMatch === false ) {
2837                                         $literal = true;  # literal when in PST with no prefix
2838                                 } else {
2839                                         $literal = false; # expand when in PST with subst: or safesubst:
2840                                 }
2841                         } else {
2842                                 if ( $substMatch == 'subst' ) {
2843                                         $literal = true;  # literal when not in PST with plain subst:
2844                                 } else {
2845                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2846                                 }
2847                         }
2848                         if ( $literal ) {
2849                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2850                                 $isLocalObj = true;
2851                                 $found = true;
2852                         }
2853                 }
2854
2855                 # Variables
2856                 if ( !$found && $args->getLength() == 0 ) {
2857                         $id = $this->mVariables->matchStartToEnd( $part1 );
2858                         if ( $id !== false ) {
2859                                 $text = $this->getVariableValue( $id, $frame );
2860                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
2861                                         $this->mOutput->mContainsOldMagic = true;
2862                                 }
2863                                 $found = true;
2864                         }
2865                 }
2866
2867                 # MSG, MSGNW and RAW
2868                 if ( !$found ) {
2869                         # Check for MSGNW:
2870                         $mwMsgnw = MagicWord::get( 'msgnw' );
2871                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2872                                 $nowiki = true;
2873                         } else {
2874                                 # Remove obsolete MSG:
2875                                 $mwMsg = MagicWord::get( 'msg' );
2876                                 $mwMsg->matchStartAndRemove( $part1 );
2877                         }
2878
2879                         # Check for RAW:
2880                         $mwRaw = MagicWord::get( 'raw' );
2881                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2882                                 $forceRawInterwiki = true;
2883                         }
2884                 }
2885                 wfProfileOut( __METHOD__.'-modifiers' );
2886
2887                 # Parser functions
2888                 if ( !$found ) {
2889                         wfProfileIn( __METHOD__ . '-pfunc' );
2890
2891                         $colonPos = strpos( $part1, ':' );
2892                         if ( $colonPos !== false ) {
2893                                 # Case sensitive functions
2894                                 $function = substr( $part1, 0, $colonPos );
2895                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2896                                         $function = $this->mFunctionSynonyms[1][$function];
2897                                 } else {
2898                                         # Case insensitive functions
2899                                         $function = $wgContLang->lc( $function );
2900                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2901                                                 $function = $this->mFunctionSynonyms[0][$function];
2902                                         } else {
2903                                                 $function = false;
2904                                         }
2905                                 }
2906                                 if ( $function ) {
2907                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2908                                         $initialArgs = array( &$this );
2909                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2910                                         if ( $flags & SFH_OBJECT_ARGS ) {
2911                                                 # Add a frame parameter, and pass the arguments as an array
2912                                                 $allArgs = $initialArgs;
2913                                                 $allArgs[] = $frame;
2914                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2915                                                         $funcArgs[] = $args->item( $i );
2916                                                 }
2917                                                 $allArgs[] = $funcArgs;
2918                                         } else {
2919                                                 # Convert arguments to plain text
2920                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2921                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2922                                                 }
2923                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2924                                         }
2925
2926                                         # Workaround for PHP bug 35229 and similar
2927                                         if ( !is_callable( $callback ) ) {
2928                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2929                                                 wfProfileOut( __METHOD__ );
2930                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2931                                         }
2932                                         $result = call_user_func_array( $callback, $allArgs );
2933                                         $found = true;
2934                                         $noparse = true;
2935                                         $preprocessFlags = 0;
2936
2937                                         if ( is_array( $result ) ) {
2938                                                 if ( isset( $result[0] ) ) {
2939                                                         $text = $result[0];
2940                                                         unset( $result[0] );
2941                                                 }
2942
2943                                                 // Extract flags into the local scope
2944                                                 // This allows callers to set flags such as nowiki, found, etc.
2945                                                 extract( $result );
2946                                         } else {
2947                                                 $text = $result;
2948                                         }
2949                                         if ( !$noparse ) {
2950                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2951                                                 $isChildObj = true;
2952                                         }
2953                                 }
2954                         }
2955                         wfProfileOut( __METHOD__ . '-pfunc' );
2956                 }
2957
2958                 # Finish mangling title and then check for loops.
2959                 # Set $title to a Title object and $titleText to the PDBK
2960                 if ( !$found ) {
2961                         $ns = NS_TEMPLATE;
2962                         # Split the title into page and subpage
2963                         $subpage = '';
2964                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2965                         if ( $subpage !== '' ) {
2966                                 $ns = $this->mTitle->getNamespace();
2967                         }
2968                         $title = Title::newFromText( $part1, $ns );
2969                         if ( $title ) {
2970                                 $titleText = $title->getPrefixedText();
2971                                 # Check for language variants if the template is not found
2972                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
2973                                         $wgContLang->findVariantLink( $part1, $title, true );
2974                                 }
2975                                 # Do recursion depth check
2976                                 $limit = $this->mOptions->getMaxTemplateDepth();
2977                                 if ( $frame->depth >= $limit ) {
2978                                         $found = true;
2979                                         $text = '<span class="error">'
2980                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
2981                                                 . '</span>';
2982                                 }
2983                         }
2984                 }
2985
2986                 # Load from database
2987                 if ( !$found && $title ) {
2988                         wfProfileIn( __METHOD__ . '-loadtpl' );
2989                         if ( !$title->isExternal() ) {
2990                                 if ( $title->getNamespace() == NS_SPECIAL
2991                                         && $this->mOptions->getAllowSpecialInclusion()
2992                                         && $this->ot['html'] )
2993                                 {
2994                                         $text = SpecialPage::capturePath( $title );
2995                                         if ( is_string( $text ) ) {
2996                                                 $found = true;
2997                                                 $isHTML = true;
2998                                                 $this->disableCache();
2999                                         }
3000                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3001                                         $found = false; // access denied
3002                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3003                                 } else {
3004                                         list( $text, $title ) = $this->getTemplateDom( $title );
3005                                         if ( $text !== false ) {
3006                                                 $found = true;
3007                                                 $isChildObj = true;
3008                                         }
3009                                 }
3010
3011                                 # If the title is valid but undisplayable, make a link to it
3012                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3013                                         $text = "[[:$titleText]]";
3014                                         $found = true;
3015                                 }
3016                         } elseif ( $title->isTrans() ) {
3017                                 // Interwiki transclusion
3018                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3019                                         $text = $this->interwikiTransclude( $title, 'render' );
3020                                         $isHTML = true;
3021                                 } else {
3022                                         $text = $this->interwikiTransclude( $title, 'raw' );
3023                                         // Preprocess it like a template
3024                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3025                                         $isChildObj = true;
3026                                 }
3027                                 $found = true;
3028                         }
3029
3030                         # Do infinite loop check
3031                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3032                         if ( !$frame->loopCheck( $title ) ) {
3033                                 $found = true;
3034                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3035                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3036                         }
3037                         wfProfileOut( __METHOD__ . '-loadtpl' );
3038                 }
3039
3040                 # If we haven't found text to substitute by now, we're done
3041                 # Recover the source wikitext and return it
3042                 if ( !$found ) {
3043                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3044                         wfProfileOut( __METHOD__ );
3045                         return array( 'object' => $text );
3046                 }
3047
3048                 # Expand DOM-style return values in a child frame
3049                 if ( $isChildObj ) {
3050                         # Clean up argument array
3051                         $newFrame = $frame->newChild( $args, $title );
3052
3053                         if ( $nowiki ) {
3054                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3055                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3056                                 # Expansion is eligible for the empty-frame cache
3057                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3058                                         $text = $this->mTplExpandCache[$titleText];
3059                                 } else {
3060                                         $text = $newFrame->expand( $text );
3061                                         $this->mTplExpandCache[$titleText] = $text;
3062                                 }
3063                         } else {
3064                                 # Uncached expansion
3065                                 $text = $newFrame->expand( $text );
3066                         }
3067                 }
3068                 if ( $isLocalObj && $nowiki ) {
3069                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3070                         $isLocalObj = false;
3071                 }
3072
3073                 # Replace raw HTML by a placeholder
3074                 # Add a blank line preceding, to prevent it from mucking up
3075                 # immediately preceding headings
3076                 if ( $isHTML ) {
3077                         $text = "\n\n" . $this->insertStripItem( $text );
3078                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3079                         # Escape nowiki-style return values
3080                         $text = wfEscapeWikiText( $text );
3081                 } elseif ( is_string( $text )
3082                         && !$piece['lineStart']
3083                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3084                 {
3085                         # Bug 529: if the template begins with a table or block-level
3086                         # element, it should be treated as beginning a new line.
3087                         # This behaviour is somewhat controversial.
3088                         $text = "\n" . $text;
3089                 }
3090
3091                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3092                         # Error, oversize inclusion
3093                         $text = "[[$originalTitle]]" .
3094                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3095                         $this->limitationWarn( 'post-expand-template-inclusion' );
3096                 }
3097
3098                 if ( $isLocalObj ) {
3099                         $ret = array( 'object' => $text );
3100                 } else {
3101                         $ret = array( 'text' => $text );
3102                 }
3103
3104                 wfProfileOut( __METHOD__ );
3105                 return $ret;
3106         }
3107
3108         /**
3109          * Get the semi-parsed DOM representation of a template with a given title,
3110          * and its redirect destination title. Cached.
3111          */
3112         function getTemplateDom( $title ) {
3113                 $cacheTitle = $title;
3114                 $titleText = $title->getPrefixedDBkey();
3115
3116                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3117                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3118                         $title = Title::makeTitle( $ns, $dbk );
3119                         $titleText = $title->getPrefixedDBkey();
3120                 }
3121                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3122                         return array( $this->mTplDomCache[$titleText], $title );
3123                 }
3124
3125                 // Cache miss, go to the database
3126                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3127
3128                 if ( $text === false ) {
3129                         $this->mTplDomCache[$titleText] = false;
3130                         return array( false, $title );
3131                 }
3132
3133                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3134                 $this->mTplDomCache[ $titleText ] = $dom;
3135
3136                 if ( !$title->equals( $cacheTitle ) ) {
3137                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3138                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3139                 }
3140
3141                 return array( $dom, $title );
3142         }
3143
3144         /**
3145          * Fetch the unparsed text of a template and register a reference to it.
3146          */
3147         function fetchTemplateAndTitle( $title ) {
3148                 $templateCb = $this->mOptions->getTemplateCallback();
3149                 $stuff = call_user_func( $templateCb, $title, $this );
3150                 $text = $stuff['text'];
3151                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3152                 if ( isset( $stuff['deps'] ) ) {
3153                         foreach ( $stuff['deps'] as $dep ) {
3154                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3155                         }
3156                 }
3157                 return array( $text, $finalTitle );
3158         }
3159
3160         function fetchTemplate( $title ) {
3161                 $rv = $this->fetchTemplateAndTitle( $title );
3162                 return $rv[0];
3163         }
3164
3165         /**
3166          * Static function to get a template
3167          * Can be overridden via ParserOptions::setTemplateCallback().
3168          */
3169         static function statelessFetchTemplate( $title, $parser=false ) {
3170                 $text = $skip = false;
3171                 $finalTitle = $title;
3172                 $deps = array();
3173
3174                 // Loop to fetch the article, with up to 1 redirect
3175                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3176                         # Give extensions a chance to select the revision instead
3177                         $id = false; // Assume current
3178                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3179
3180                         if ( $skip ) {
3181                                 $text = false;
3182                                 $deps[] = array(
3183                                         'title' => $title,
3184                                         'page_id' => $title->getArticleID(),
3185                                         'rev_id' => null );
3186                                 break;
3187                         }
3188                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3189                         $rev_id = $rev ? $rev->getId() : 0;
3190                         // If there is no current revision, there is no page
3191                         if ( $id === false && !$rev ) {
3192                                 $linkCache = LinkCache::singleton();
3193                                 $linkCache->addBadLinkObj( $title );
3194                         }
3195
3196                         $deps[] = array(
3197                                 'title' => $title,
3198                                 'page_id' => $title->getArticleID(),
3199                                 'rev_id' => $rev_id );
3200
3201                         if ( $rev ) {
3202                                 $text = $rev->getText();
3203                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3204                                 global $wgContLang;
3205                                 $message = $wgContLang->lcfirst( $title->getText() );
3206                                 $text = wfMsgForContentNoTrans( $message );
3207                                 if ( wfEmptyMsg( $message, $text ) ) {
3208                                         $text = false;
3209                                         break;
3210                                 }
3211                         } else {
3212                                 break;
3213                         }
3214                         if ( $text === false ) {
3215                                 break;
3216                         }
3217                         // Redirect?
3218                         $finalTitle = $title;
3219                         $title = Title::newFromRedirect( $text );
3220                 }
3221                 return array(
3222                         'text' => $text,
3223                         'finalTitle' => $finalTitle,
3224                         'deps' => $deps );
3225         }
3226
3227         /**
3228          * Transclude an interwiki link.
3229          */
3230         function interwikiTransclude( $title, $action ) {
3231                 global $wgEnableScaryTranscluding;
3232
3233                 if ( !$wgEnableScaryTranscluding ) {
3234                         return wfMsg('scarytranscludedisabled');
3235                 }
3236
3237                 $url = $title->getFullUrl( "action=$action" );
3238
3239                 if ( strlen( $url ) > 255 ) {
3240                         return wfMsg( 'scarytranscludetoolong' );
3241                 }
3242                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3243         }
3244
3245         function fetchScaryTemplateMaybeFromCache( $url ) {
3246                 global $wgTranscludeCacheExpiry;
3247                 $dbr = wfGetDB( DB_SLAVE );
3248                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3249                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3250                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3251                 if ( $obj ) {
3252                         return $obj->tc_contents;
3253                 }
3254
3255                 $text = Http::get( $url );
3256                 if ( !$text ) {
3257                         return wfMsg( 'scarytranscludefailed', $url );
3258                 }
3259
3260                 $dbw = wfGetDB( DB_MASTER );
3261                 $dbw->replace( 'transcache', array('tc_url'), array(
3262                         'tc_url' => $url,
3263                         'tc_time' => $dbw->timestamp( time() ),
3264                         'tc_contents' => $text)
3265                 );
3266                 return $text;
3267         }
3268
3269
3270         /**
3271          * Triple brace replacement -- used for template arguments
3272          * @private
3273          */
3274         function argSubstitution( $piece, $frame ) {
3275                 wfProfileIn( __METHOD__ );
3276
3277                 $error = false;
3278                 $parts = $piece['parts'];
3279                 $nameWithSpaces = $frame->expand( $piece['title'] );
3280                 $argName = trim( $nameWithSpaces );
3281                 $object = false;
3282                 $text = $frame->getArgument( $argName );
3283                 if (  $text === false && $parts->getLength() > 0
3284                   && (
3285                     $this->ot['html']
3286                     || $this->ot['pre']
3287                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3288                   )
3289                 ) {
3290                         # No match in frame, use the supplied default
3291                         $object = $parts->item( 0 )->getChildren();
3292                 }
3293                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3294                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3295                         $this->limitationWarn( 'post-expand-template-argument' );
3296                 }
3297
3298                 if ( $text === false && $object === false ) {
3299                         # No match anywhere
3300                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3301                 }
3302                 if ( $error !== false ) {
3303                         $text .= $error;
3304                 }
3305                 if ( $object !== false ) {
3306                         $ret = array( 'object' => $object );
3307                 } else {
3308                         $ret = array( 'text' => $text );
3309                 }
3310
3311                 wfProfileOut( __METHOD__ );
3312                 return $ret;
3313         }
3314
3315         /**
3316          * Return the text to be used for a given extension tag.
3317          * This is the ghost of strip().
3318          *
3319          * @param array $params Associative array of parameters:
3320          *     name       PPNode for the tag name
3321          *     attr       PPNode for unparsed text where tag attributes are thought to be
3322          *     attributes Optional associative array of parsed attributes
3323          *     inner      Contents of extension element
3324          *     noClose    Original text did not have a close tag
3325          * @param PPFrame $frame
3326          */
3327         function extensionSubstitution( $params, $frame ) {
3328                 global $wgRawHtml, $wgContLang;
3329
3330                 $name = $frame->expand( $params['name'] );
3331                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3332                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3333                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3334
3335                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3336                         ( $this->ot['html'] || $this->ot['pre'] );
3337                 if ( $isFunctionTag ) {
3338                         $markerType = 'none';
3339                 } else {
3340                         $markerType = 'general';
3341                 }
3342                 if ( $this->ot['html'] || $isFunctionTag ) {
3343                         $name = strtolower( $name );
3344                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3345                         if ( isset( $params['attributes'] ) ) {
3346                                 $attributes = $attributes + $params['attributes'];
3347                         }
3348
3349                         if ( isset( $this->mTagHooks[$name] ) ) {
3350                                 # Workaround for PHP bug 35229 and similar
3351                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3352                                         throw new MWException( "Tag hook for $name is not callable\n" );
3353                                 }
3354                                 $output = call_user_func_array( $this->mTagHooks[$name],
3355                                         array( $content, $attributes, $this, $frame ) );
3356                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3357                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3358                                 if ( !is_callable( $callback ) ) {
3359                                         throw new MWException( "Tag hook for $name is not callable\n" );
3360                                 }
3361
3362                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3363                         } else {
3364                                 $output = '<span class="error">Invalid tag extension name: ' .
3365                                         htmlspecialchars( $name ) . '</span>';
3366                         }
3367
3368                         if ( is_array( $output ) ) {
3369                                 // Extract flags to local scope (to override $markerType)
3370                                 $flags = $output;
3371                                 $output = $flags[0];
3372                                 unset( $flags[0] );
3373                                 extract( $flags );
3374                         }
3375                 } else {
3376                         if ( is_null( $attrText ) ) {
3377                                 $attrText = '';
3378                         }
3379                         if ( isset( $params['attributes'] ) ) {
3380                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3381                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3382                                                 htmlspecialchars( $attrValue ) . '"';
3383                                 }
3384                         }
3385                         if ( $content === null ) {
3386                                 $output = "<$name$attrText/>";
3387                         } else {
3388                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3389                                 $output = "<$name$attrText>$content$close";
3390                         }
3391                 }
3392
3393                 if ( $markerType === 'none' ) {
3394                         return $output;
3395                 } elseif ( $markerType === 'nowiki' ) {
3396                         $this->mStripState->nowiki->setPair( $marker, $output );
3397                 } elseif ( $markerType === 'general' ) {
3398                         $this->mStripState->general->setPair( $marker, $output );
3399                 } else {
3400                         throw new MWException( __METHOD__.': invalid marker type' );
3401                 }
3402                 return $marker;
3403         }
3404
3405         /**
3406          * Increment an include size counter
3407          *
3408          * @param string $type The type of expansion
3409          * @param integer $size The size of the text
3410          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3411          */
3412         function incrementIncludeSize( $type, $size ) {
3413                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3414                         return false;
3415                 } else {
3416                         $this->mIncludeSizes[$type] += $size;
3417                         return true;
3418                 }
3419         }
3420
3421         /**
3422          * Increment the expensive function count
3423          *
3424          * @return boolean False if the limit has been exceeded
3425          */
3426         function incrementExpensiveFunctionCount() {
3427                 global $wgExpensiveParserFunctionLimit;
3428                 $this->mExpensiveFunctionCount++;
3429                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3430                         return true;
3431                 }
3432                 return false;
3433         }
3434
3435         /**
3436          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3437          * Fills $this->mDoubleUnderscores, returns the modified text
3438          */
3439         function doDoubleUnderscore( $text ) {
3440                 wfProfileIn( __METHOD__ );
3441
3442                 // The position of __TOC__ needs to be recorded
3443                 $mw = MagicWord::get( 'toc' );
3444                 if ( $mw->match( $text ) ) {
3445                         $this->mShowToc = true;
3446                         $this->mForceTocPosition = true;
3447
3448                         // Set a placeholder. At the end we'll fill it in with the TOC.
3449                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3450
3451                         // Only keep the first one.
3452                         $text = $mw->replace( '', $text );
3453                 }
3454
3455                 // Now match and remove the rest of them
3456                 $mwa = MagicWord::getDoubleUnderscoreArray();
3457                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3458
3459                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3460                         $this->mOutput->mNoGallery = true;
3461                 }
3462                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3463                         $this->mShowToc = false;
3464                 }
3465                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3466                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3467                         $this->addTrackingCategory( 'hidden-category-category' );
3468                 }
3469                 # (bug 8068) Allow control over whether robots index a page.
3470                 #
3471                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3472                 # is not desirable, the last one on the page should win.
3473                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3474                         $this->mOutput->setIndexPolicy( 'noindex' );
3475                         $this->addTrackingCategory( 'noindex-category' );
3476                 }
3477                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3478                         $this->mOutput->setIndexPolicy( 'index' );
3479                         $this->addTrackingCategory( 'index-category' );
3480                 }
3481
3482                 wfProfileOut( __METHOD__ );
3483                 return $text;
3484         }
3485
3486         /**
3487          * Add a tracking category, getting the title from a system message,
3488          * or print a debug message if the title is invalid.
3489          * @param $msg String message key
3490          * @return Bool whether the addition was successful
3491          */
3492         protected function addTrackingCategory( $msg ) {
3493                 $cat = wfMsgForContent( $msg );
3494
3495                 # Allow tracking categories to be disabled by setting them to "-"
3496                 if ( $cat === '-' ) {
3497                         return false;
3498                 }
3499
3500                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3501                 if ( $containerCategory ) {
3502                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3503                         return true;
3504                 } else {
3505                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3506                         return false;
3507                 }
3508         }
3509
3510         /**
3511          * This function accomplishes several tasks:
3512          * 1) Auto-number headings if that option is enabled
3513          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3514          * 3) Add a Table of contents on the top for users who have enabled the option
3515          * 4) Auto-anchor headings
3516          *
3517          * It loops through all headlines, collects the necessary data, then splits up the
3518          * string and re-inserts the newly formatted headlines.
3519          *
3520          * @param string $text
3521          * @param string $origText Original, untouched wikitext
3522          * @param boolean $isMain
3523          * @private
3524          */
3525         function formatHeadings( $text, $origText, $isMain=true ) {
3526                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3527
3528                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3529                 $showEditLink = $this->mOptions->getEditSection();
3530
3531                 // Do not call quickUserCan unless necessary
3532                 if ( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3533                         $showEditLink = 0;
3534                 }
3535
3536                 # Inhibit editsection links if requested in the page
3537                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3538                         $showEditLink = 0;
3539                 }
3540
3541                 # Get all headlines for numbering them and adding funky stuff like [edit]
3542                 # links - this is for later, but we need the number of headlines right now
3543                 $matches = array();
3544                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3545
3546                 # if there are fewer than 4 headlines in the article, do not show TOC
3547                 # unless it's been explicitly enabled.
3548                 $enoughToc = $this->mShowToc &&
3549                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3550
3551                 # Allow user to stipulate that a page should have a "new section"
3552                 # link added via __NEWSECTIONLINK__
3553                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3554                         $this->mOutput->setNewSection( true );
3555                 }
3556
3557                 # Allow user to remove the "new section"
3558                 # link via __NONEWSECTIONLINK__
3559                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3560                         $this->mOutput->hideNewSection( true );
3561                 }
3562
3563                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3564                 # override above conditions and always show TOC above first header
3565                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3566                         $this->mShowToc = true;
3567                         $enoughToc = true;
3568                 }
3569
3570                 # We need this to perform operations on the HTML
3571                 $sk = $this->mOptions->getSkin();
3572
3573                 # headline counter
3574                 $headlineCount = 0;
3575                 $numVisible = 0;
3576
3577                 # Ugh .. the TOC should have neat indentation levels which can be
3578                 # passed to the skin functions. These are determined here
3579                 $toc = '';
3580                 $full = '';
3581                 $head = array();
3582                 $sublevelCount = array();
3583                 $levelCount = array();
3584                 $toclevel = 0;
3585                 $level = 0;
3586                 $prevlevel = 0;
3587                 $toclevel = 0;
3588                 $prevtoclevel = 0;
3589                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3590                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3591                 $oldType = $this->mOutputType;
3592                 $this->setOutputType( self::OT_WIKI );
3593                 $frame = $this->getPreprocessor()->newFrame();
3594                 $root = $this->preprocessToDom( $origText );
3595                 $node = $root->getFirstChild();
3596                 $byteOffset = 0;
3597                 $tocraw = array();
3598
3599                 foreach ( $matches[3] as $headline ) {
3600                         $isTemplate = false;
3601                         $titleText = false;
3602                         $sectionIndex = false;
3603                         $numbering = '';
3604                         $markerMatches = array();
3605                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3606                                 $serial = $markerMatches[1];
3607                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3608                                 $isTemplate = ( $titleText != $baseTitleText );
3609                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3610                         }
3611
3612                         if ( $toclevel ) {
3613                                 $prevlevel = $level;
3614                                 $prevtoclevel = $toclevel;
3615                         }
3616                         $level = $matches[1][$headlineCount];
3617
3618                         if ( $level > $prevlevel ) {
3619                                 # Increase TOC level
3620                                 $toclevel++;
3621                                 $sublevelCount[$toclevel] = 0;
3622                                 if ( $toclevel<$wgMaxTocLevel ) {
3623                                         $prevtoclevel = $toclevel;
3624                                         $toc .= $sk->tocIndent();
3625                                         $numVisible++;
3626                                 }
3627                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3628                                 # Decrease TOC level, find level to jump to
3629
3630                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3631                                         if ( $levelCount[$i] == $level ) {
3632                                                 # Found last matching level
3633                                                 $toclevel = $i;
3634                                                 break;
3635                                         } elseif ( $levelCount[$i] < $level ) {
3636                                                 # Found first matching level below current level
3637                                                 $toclevel = $i + 1;
3638                                                 break;
3639                                         }
3640                                 }
3641                                 if ( $i == 0 ) {
3642                                         $toclevel = 1;
3643                                 }
3644                                 if ( $toclevel<$wgMaxTocLevel ) {
3645                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3646                                                 # Unindent only if the previous toc level was shown :p
3647                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3648                                                 $prevtoclevel = $toclevel;
3649                                         } else {
3650                                                 $toc .= $sk->tocLineEnd();
3651                                         }
3652                                 }
3653                         } else {
3654                                 # No change in level, end TOC line
3655                                 if ( $toclevel<$wgMaxTocLevel ) {
3656                                         $toc .= $sk->tocLineEnd();
3657                                 }
3658                         }
3659
3660                         $levelCount[$toclevel] = $level;
3661
3662                         # count number of headlines for each level
3663                         @$sublevelCount[$toclevel]++;
3664                         $dot = 0;
3665                         for( $i = 1; $i <= $toclevel; $i++ ) {
3666                                 if ( !empty( $sublevelCount[$i] ) ) {
3667                                         if ( $dot ) {
3668                                                 $numbering .= '.';
3669                                         }
3670                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3671                                         $dot = 1;
3672                                 }
3673                         }
3674
3675                         # The safe header is a version of the header text safe to use for links
3676                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3677                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3678
3679                         # Remove link placeholders by the link text.
3680                         #     <!--LINK number-->
3681                         # turns into
3682                         #     link text with suffix
3683                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3684
3685                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3686                         $tocline = preg_replace(
3687                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3688                                 array( '',                          '<$1>' ),
3689                                 $safeHeadline
3690                         );
3691                         $tocline = trim( $tocline );
3692
3693                         # For the anchor, strip out HTML-y stuff period
3694                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3695                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3696                         $safeHeadline = trim( $safeHeadline );
3697
3698                         # Save headline for section edit hint before it's escaped
3699                         $headlineHint = $safeHeadline;
3700
3701                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3702                                 # For reverse compatibility, provide an id that's
3703                                 # HTML4-compatible, like we used to.
3704                                 #
3705                                 # It may be worth noting, academically, that it's possible for
3706                                 # the legacy anchor to conflict with a non-legacy headline
3707                                 # anchor on the page.  In this case likely the "correct" thing
3708                                 # would be to either drop the legacy anchors or make sure
3709                                 # they're numbered first.  However, this would require people
3710                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3711                                 # manually, so let's not bother worrying about it.
3712                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3713                                         array( 'noninitial', 'legacy' ) );
3714                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3715
3716                                 if ( $legacyHeadline == $safeHeadline ) {
3717                                         # No reason to have both (in fact, we can't)
3718                                         $legacyHeadline = false;
3719                                 }
3720                         } else {
3721                                 $legacyHeadline = false;
3722                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3723                                         'noninitial' );
3724                         }
3725
3726                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3727                         # Does this apply to Unicode characters?  Because we aren't
3728                         # handling those here.
3729                         $arrayKey = strtolower( $safeHeadline );
3730                         if ( $legacyHeadline === false ) {
3731                                 $legacyArrayKey = false;
3732                         } else {
3733                                 $legacyArrayKey = strtolower( $legacyHeadline );
3734                         }
3735
3736                         # count how many in assoc. array so we can track dupes in anchors
3737                         if ( isset( $refers[$arrayKey] ) ) {
3738                                 $refers[$arrayKey]++;
3739                         } else {
3740                                 $refers[$arrayKey] = 1;
3741                         }
3742                         if ( isset( $refers[$legacyArrayKey] ) ) {
3743                                 $refers[$legacyArrayKey]++;
3744                         } else {
3745                                 $refers[$legacyArrayKey] = 1;
3746                         }
3747
3748                         # Don't number the heading if it is the only one (looks silly)
3749                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3750                                 # the two are different if the line contains a link
3751                                 $headline = $numbering . ' ' . $headline;
3752                         }
3753
3754                         # Create the anchor for linking from the TOC to the section
3755                         $anchor = $safeHeadline;
3756                         $legacyAnchor = $legacyHeadline;
3757                         if ( $refers[$arrayKey] > 1 ) {
3758                                 $anchor .= '_' . $refers[$arrayKey];
3759                         }
3760                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3761                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3762                         }
3763                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3764                                 $toc .= $sk->tocLine( $anchor, $tocline,
3765                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3766                         }
3767
3768                         # Add the section to the section tree
3769                         # Find the DOM node for this header
3770                         while ( $node && !$isTemplate ) {
3771                                 if ( $node->getName() === 'h' ) {
3772                                         $bits = $node->splitHeading();
3773                                         if ( $bits['i'] == $sectionIndex )
3774                                                 break;
3775                                 }
3776                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3777                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3778                                 $node = $node->getNextSibling();
3779                         }
3780                         $tocraw[] = array(
3781                                 'toclevel' => $toclevel,
3782                                 'level' => $level,
3783                                 'line' => $tocline,
3784                                 'number' => $numbering,
3785                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3786                                 'fromtitle' => $titleText,
3787                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3788                                 'anchor' => $anchor,
3789                         );
3790
3791                         # give headline the correct <h#> tag
3792                         if ( $showEditLink && $sectionIndex !== false ) {
3793                                 if ( $isTemplate ) {
3794                                         # Put a T flag in the section identifier, to indicate to extractSections()
3795                                         # that sections inside <includeonly> should be counted.
3796                                         $editlink = $sk->doEditSectionLink( Title::newFromText( $titleText ), "T-$sectionIndex" );
3797                                 } else {
3798                                         $editlink = $sk->doEditSectionLink( $this->mTitle, $sectionIndex, $headlineHint );
3799                                 }
3800                         } else {
3801                                 $editlink = '';
3802                         }
3803                         $head[$headlineCount] = $sk->makeHeadline( $level,
3804                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3805                                 $editlink, $legacyAnchor );
3806
3807                         $headlineCount++;
3808                 }
3809
3810                 $this->setOutputType( $oldType );
3811
3812                 # Never ever show TOC if no headers
3813                 if ( $numVisible < 1 ) {
3814                         $enoughToc = false;
3815                 }
3816
3817                 if ( $enoughToc ) {
3818                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3819                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3820                         }
3821                         $toc = $sk->tocList( $toc );
3822                         $this->mOutput->setTOCHTML( $toc );
3823                 }
3824
3825                 if ( $isMain ) {
3826                         $this->mOutput->setSections( $tocraw );
3827                 }
3828
3829                 # split up and insert constructed headlines
3830
3831                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3832                 $i = 0;
3833
3834                 foreach ( $blocks as $block ) {
3835                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3836                                 # This is the [edit] link that appears for the top block of text when
3837                                 # section editing is enabled
3838
3839                                 # Disabled because it broke block formatting
3840                                 # For example, a bullet point in the top line
3841                                 # $full .= $sk->editSectionLink(0);
3842                         }
3843                         $full .= $block;
3844                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3845                                 # Top anchor now in skin
3846                                 $full = $full.$toc;
3847                         }
3848
3849                         if ( !empty( $head[$i] ) ) {
3850                                 $full .= $head[$i];
3851                         }
3852                         $i++;
3853                 }
3854                 if ( $this->mForceTocPosition ) {
3855                         return str_replace( '<!--MWTOC-->', $toc, $full );
3856                 } else {
3857                         return $full;
3858                 }
3859         }
3860
3861         /**
3862          * Merge $tree2 into $tree1 by replacing the section with index
3863          * $section in $tree1 and its descendants with the sections in $tree2.
3864          * Note that in the returned section tree, only the 'index' and
3865          * 'byteoffset' fields are guaranteed to be correct.
3866          * @param $tree1 array Section tree from ParserOutput::getSectons()
3867         *  @param $tree2 array Section tree
3868          * @param $section int Section index
3869          * @param $title Title Title both section trees come from
3870          * @param $len2 int Length of the original wikitext for $tree2
3871          * @return array Merged section tree
3872          */
3873         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3874                 global $wgContLang;
3875                 $newTree = array();
3876                 $targetLevel = false;
3877                 $merged = false;
3878                 $lastLevel = 1;
3879                 $nextIndex = 1;
3880                 $numbering = array( 0 );
3881                 $titletext = $title->getPrefixedDBkey();
3882                 foreach ( $tree1 as $s ) {
3883                         if ( $targetLevel !== false ) {
3884                                 if ( $s['level'] <= $targetLevel ) {
3885                                         // We've skipped enough
3886                                         $targetLevel = false;
3887                                 } else {
3888                                         continue;
3889                                 }
3890                         }
3891                         if ( $s['index'] != $section ||
3892                                         $s['fromtitle'] != $titletext ) {
3893                                 self::incrementNumbering( $numbering,
3894                                         $s['toclevel'], $lastLevel );
3895
3896                                 // Rewrite index, byteoffset and number
3897                                 if ( $s['fromtitle'] == $titletext ) {
3898                                         $s['index'] = $nextIndex++;
3899                                         if ( $merged ) {
3900                                                 $s['byteoffset'] += $len2;
3901                                         }
3902                                 }
3903                                 $s['number']  = implode( '.', array_map(
3904                                         array( $wgContLang, 'formatnum' ),
3905                                         $numbering ) );
3906                                 $lastLevel = $s['toclevel'];
3907                                 $newTree[] = $s;
3908                         } else {
3909                                 // We're at $section
3910                                 // Insert sections from $tree2 here
3911                                 foreach ( $tree2 as $s2 ) {
3912                                         // Rewrite the fields in $s2
3913                                         // before inserting it
3914                                         $s2['toclevel'] += $s['toclevel'] - 1;
3915                                         $s2['level'] += $s['level'] - 1;
3916                                         $s2['index'] = $nextIndex++;
3917                                         $s2['byteoffset'] += $s['byteoffset'];
3918
3919                                         self::incrementNumbering( $numbering,
3920                                                 $s2['toclevel'], $lastLevel );
3921                                         $s2['number']  = implode( '.', array_map(
3922                                                 array( $wgContLang, 'formatnum' ),
3923                                                 $numbering ) );
3924                                         $lastLevel = $s2['toclevel'];
3925                                         $newTree[] = $s2;
3926                                 }
3927                                 // Skip all descendants of $section in $tree1
3928                                 $targetLevel = $s['level'];
3929                                 $merged = true;
3930                         }
3931                 }
3932                 return $newTree;
3933         }
3934
3935         /**
3936          * Increment a section number. Helper function for mergeSectionTrees()
3937          * @param $number array Array representing a section number
3938          * @param $level int Current TOC level (depth)
3939          * @param $lastLevel int Level of previous TOC entry
3940          */
3941         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3942                 if ( $level > $lastLevel ) {
3943                         $number[$level - 1] = 1;
3944                 } elseif ( $level < $lastLevel ) {
3945                         foreach ( $number as $key => $unused )
3946                                 if ( $key >= $level ) {
3947                                         unset( $number[$key] );
3948                                 }
3949                         $number[$level - 1]++;
3950                 } else {
3951                         $number[$level - 1]++;
3952                 }
3953         }
3954
3955         /**
3956          * Transform wiki markup when saving a page by doing \r\n -> \n
3957          * conversion, substitting signatures, {{subst:}} templates, etc.
3958          *
3959          * @param string $text the text to transform
3960          * @param Title &$title the Title object for the current article
3961          * @param User $user the User object describing the current user
3962          * @param ParserOptions $options parsing options
3963          * @param bool $clearState whether to clear the parser state first
3964          * @return string the altered wiki markup
3965          * @public
3966          */
3967         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3968                 $this->mOptions = $options;
3969                 $this->setTitle( $title );
3970                 $this->setOutputType( self::OT_WIKI );
3971
3972                 if ( $clearState ) {
3973                         $this->clearState();
3974                 }
3975
3976                 $pairs = array(
3977                         "\r\n" => "\n",
3978                 );
3979                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3980                 $text = $this->pstPass2( $text, $user );
3981                 $text = $this->mStripState->unstripBoth( $text );
3982                 return $text;
3983         }
3984
3985         /**
3986          * Pre-save transform helper function
3987          * @private
3988          */
3989         function pstPass2( $text, $user ) {
3990                 global $wgContLang, $wgLocaltimezone;
3991
3992                 # Note: This is the timestamp saved as hardcoded wikitext to
3993                 # the database, we use $wgContLang here in order to give
3994                 # everyone the same signature and use the default one rather
3995                 # than the one selected in each user's preferences.
3996                 # (see also bug 12815)
3997                 $ts = $this->mOptions->getTimestamp();
3998                 if ( isset( $wgLocaltimezone ) ) {
3999                         $tz = $wgLocaltimezone;
4000                 } else {
4001                         $tz = date_default_timezone_get();
4002                 }
4003
4004                 $unixts = wfTimestamp( TS_UNIX, $ts );
4005                 $oldtz = date_default_timezone_get();
4006                 date_default_timezone_set( $tz );
4007                 $ts = date( 'YmdHis', $unixts );
4008                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4009
4010                 # Allow translation of timezones trough wiki. date() can return
4011                 # whatever crap the system uses, localised or not, so we cannot
4012                 # ship premade translations.
4013                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4014                 $value = wfMsgForContent( $key );
4015                 if ( !wfEmptyMsg( $key, $value ) ) {
4016                         $tzMsg = $value;
4017                 }
4018
4019                 date_default_timezone_set( $oldtz );
4020
4021                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4022
4023                 # Variable replacement
4024                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4025                 $text = $this->replaceVariables( $text );
4026
4027                 # Signatures
4028                 $sigText = $this->getUserSig( $user );
4029                 $text = strtr( $text, array(
4030                         '~~~~~' => $d,
4031                         '~~~~' => "$sigText $d",
4032                         '~~~' => $sigText
4033                 ) );
4034
4035                 # Context links: [[|name]] and [[name (context)|]]
4036                 global $wgLegalTitleChars;
4037                 $tc = "[$wgLegalTitleChars]";
4038                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4039
4040                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4041                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4042                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4043                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4044
4045                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4046                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4047                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4048                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4049
4050                 $t = $this->mTitle->getText();
4051                 $m = array();
4052                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4053                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4054                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4055                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4056                 } else {
4057                         # if there's no context, don't bother duplicating the title
4058                         $text = preg_replace( $p2, '[[\\1]]', $text );
4059                 }
4060
4061                 # Trim trailing whitespace
4062                 $text = rtrim( $text );
4063
4064                 return $text;
4065         }
4066
4067         /**
4068          * Fetch the user's signature text, if any, and normalize to
4069          * validated, ready-to-insert wikitext.
4070          * If you have pre-fetched the nickname or the fancySig option, you can
4071          * specify them here to save a database query.
4072          *
4073          * @param User $user
4074          * @return string
4075          */
4076         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4077                 global $wgMaxSigChars;
4078
4079                 $username = $user->getName();
4080
4081                 // If not given, retrieve from the user object.
4082                 if ( $nickname === false )
4083                         $nickname = $user->getOption( 'nickname' );
4084
4085                 if ( is_null( $fancySig ) ) {
4086                         $fancySig = $user->getBoolOption( 'fancysig' );
4087                 }
4088
4089                 $nickname = $nickname == null ? $username : $nickname;
4090
4091                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4092                         $nickname = $username;
4093                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4094                 } elseif ( $fancySig !== false ) {
4095                         # Sig. might contain markup; validate this
4096                         if ( $this->validateSig( $nickname ) !== false ) {
4097                                 # Validated; clean up (if needed) and return it
4098                                 return $this->cleanSig( $nickname, true );
4099                         } else {
4100                                 # Failed to validate; fall back to the default
4101                                 $nickname = $username;
4102                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4103                         }
4104                 }
4105
4106                 // Make sure nickname doesnt get a sig in a sig
4107                 $nickname = $this->cleanSigInSig( $nickname );
4108
4109                 # If we're still here, make it a link to the user page
4110                 $userText = wfEscapeWikiText( $username );
4111                 $nickText = wfEscapeWikiText( $nickname );
4112                 if ( $user->isAnon() )  {
4113                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4114                 } else {
4115                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4116                 }
4117         }
4118
4119         /**
4120          * Check that the user's signature contains no bad XML
4121          *
4122          * @param string $text
4123          * @return mixed An expanded string, or false if invalid.
4124          */
4125         function validateSig( $text ) {
4126                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4127         }
4128
4129         /**
4130          * Clean up signature text
4131          *
4132          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4133          * 2) Substitute all transclusions
4134          *
4135          * @param string $text
4136          * @param $parsing Whether we're cleaning (preferences save) or parsing
4137          * @return string Signature text
4138          */
4139         function cleanSig( $text, $parsing = false ) {
4140                 if ( !$parsing ) {
4141                         global $wgTitle;
4142                         $this->clearState();
4143                         $this->setTitle( $wgTitle );
4144                         $this->mOptions = new ParserOptions;
4145                         $this->setOutputType = self::OT_PREPROCESS;
4146                 }
4147
4148                 # Option to disable this feature
4149                 if ( !$this->mOptions->getCleanSignatures() ) {
4150                         return $text;
4151                 }
4152
4153                 # FIXME: regex doesn't respect extension tags or nowiki
4154                 #  => Move this logic to braceSubstitution()
4155                 $substWord = MagicWord::get( 'subst' );
4156                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4157                 $substText = '{{' . $substWord->getSynonym( 0 );
4158
4159                 $text = preg_replace( $substRegex, $substText, $text );
4160                 $text = $this->cleanSigInSig( $text );
4161                 $dom = $this->preprocessToDom( $text );
4162                 $frame = $this->getPreprocessor()->newFrame();
4163                 $text = $frame->expand( $dom );
4164
4165                 if ( !$parsing ) {
4166                         $text = $this->mStripState->unstripBoth( $text );
4167                 }
4168
4169                 return $text;
4170         }
4171
4172         /**
4173          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4174          * @param string $text
4175          * @return string Signature text with /~{3,5}/ removed
4176          */
4177         function cleanSigInSig( $text ) {
4178                 $text = preg_replace( '/~{3,5}/', '', $text );
4179                 return $text;
4180         }
4181
4182         /**
4183          * Set up some variables which are usually set up in parse()
4184          * so that an external function can call some class members with confidence
4185          * @public
4186          */
4187         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4188                 $this->setTitle( $title );
4189                 $this->mOptions = $options;
4190                 $this->setOutputType( $outputType );
4191                 if ( $clearState ) {
4192                         $this->clearState();
4193                 }
4194         }
4195
4196         /**
4197          * Wrapper for preprocess()
4198          *
4199          * @param string $text the text to preprocess
4200          * @param ParserOptions $options  options
4201          * @return string
4202          * @public
4203          */
4204         function transformMsg( $text, $options ) {
4205                 global $wgTitle;
4206                 static $executing = false;
4207
4208                 # Guard against infinite recursion
4209                 if ( $executing ) {
4210                         return $text;
4211                 }
4212                 $executing = true;
4213
4214                 wfProfileIn( __METHOD__ );
4215                 $text = $this->preprocess( $text, $wgTitle, $options );
4216
4217                 $executing = false;
4218                 wfProfileOut( __METHOD__ );
4219                 return $text;
4220         }
4221
4222         /**
4223          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4224          * The callback should have the following form:
4225          *    function myParserHook( $text, $params, &$parser ) { ... }
4226          *
4227          * Transform and return $text. Use $parser for any required context, e.g. use
4228          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4229          *
4230          * @public
4231          *
4232          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4233          * @param mixed $callback The callback function (and object) to use for the tag
4234          *
4235          * @return The old value of the mTagHooks array associated with the hook
4236          */
4237         function setHook( $tag, $callback ) {
4238                 $tag = strtolower( $tag );
4239                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4240                 $this->mTagHooks[$tag] = $callback;
4241                 if ( !in_array( $tag, $this->mStripList ) ) {
4242                         $this->mStripList[] = $tag;
4243                 }
4244
4245                 return $oldVal;
4246         }
4247
4248         function setTransparentTagHook( $tag, $callback ) {
4249                 $tag = strtolower( $tag );
4250                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4251                 $this->mTransparentTagHooks[$tag] = $callback;
4252
4253                 return $oldVal;
4254         }
4255
4256         /**
4257          * Remove all tag hooks
4258          */
4259         function clearTagHooks() {
4260                 $this->mTagHooks = array();
4261                 $this->mStripList = $this->mDefaultStripList;
4262         }
4263
4264         /**
4265          * Create a function, e.g. {{sum:1|2|3}}
4266          * The callback function should have the form:
4267          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4268          *
4269          * Or with SFH_OBJECT_ARGS:
4270          *    function myParserFunction( $parser, $frame, $args ) { ... }
4271          *
4272          * The callback may either return the text result of the function, or an array with the text
4273          * in element 0, and a number of flags in the other elements. The names of the flags are
4274          * specified in the keys. Valid flags are:
4275          *   found                     The text returned is valid, stop processing the template. This
4276          *                             is on by default.
4277          *   nowiki                    Wiki markup in the return value should be escaped
4278          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4279          *
4280          * @public
4281          *
4282          * @param string $id The magic word ID
4283          * @param mixed $callback The callback function (and object) to use
4284          * @param integer $flags a combination of the following flags:
4285          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4286          *
4287          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4288          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4289          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4290          *     the arguments, and to control the way they are expanded.
4291          *
4292          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4293          *     arguments, for instance:
4294          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4295          *
4296          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4297          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4298          *     working if/when this is changed.
4299          *
4300          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4301          *     expansion.
4302          *
4303          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4304          *     about the methods available in PPFrame and PPNode.
4305          *
4306          * @return The old callback function for this name, if any
4307          */
4308         function setFunctionHook( $id, $callback, $flags = 0 ) {
4309                 global $wgContLang;
4310
4311                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4312                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4313
4314                 # Add to function cache
4315                 $mw = MagicWord::get( $id );
4316                 if ( !$mw )
4317                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4318
4319                 $synonyms = $mw->getSynonyms();
4320                 $sensitive = intval( $mw->isCaseSensitive() );
4321
4322                 foreach ( $synonyms as $syn ) {
4323                         # Case
4324                         if ( !$sensitive ) {
4325                                 $syn = $wgContLang->lc( $syn );
4326                         }
4327                         # Add leading hash
4328                         if ( !( $flags & SFH_NO_HASH ) ) {
4329                                 $syn = '#' . $syn;
4330                         }
4331                         # Remove trailing colon
4332                         if ( substr( $syn, -1, 1 ) === ':' ) {
4333                                 $syn = substr( $syn, 0, -1 );
4334                         }
4335                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4336                 }
4337                 return $oldVal;
4338         }
4339
4340         /**
4341          * Get all registered function hook identifiers
4342          *
4343          * @return array
4344          */
4345         function getFunctionHooks() {
4346                 return array_keys( $this->mFunctionHooks );
4347         }
4348
4349         /**
4350          * Create a tag function, e.g. <test>some stuff</test>.
4351          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4352          * Unlike parser functions, their content is not preprocessed.
4353          */
4354         function setFunctionTagHook( $tag, $callback, $flags ) {
4355                 $tag = strtolower( $tag );
4356                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4357                         $this->mFunctionTagHooks[$tag] : null;
4358                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4359
4360                 if ( !in_array( $tag, $this->mStripList ) ) {
4361                         $this->mStripList[] = $tag;
4362                 }
4363
4364                 return $old;
4365         }
4366
4367         /**
4368          * FIXME: update documentation. makeLinkObj() is deprecated.
4369          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4370          * Placeholders created in Skin::makeLinkObj()
4371          * Returns an array of link CSS classes, indexed by PDBK.
4372          */
4373         function replaceLinkHolders( &$text, $options = 0 ) {
4374                 return $this->mLinkHolders->replace( $text );
4375         }
4376
4377         /**
4378          * Replace <!--LINK--> link placeholders with plain text of links
4379          * (not HTML-formatted).
4380          * @param string $text
4381          * @return string
4382          */
4383         function replaceLinkHoldersText( $text ) {
4384                 return $this->mLinkHolders->replaceText( $text );
4385         }
4386
4387         /**
4388          * Renders an image gallery from a text with one line per image.
4389          * text labels may be given by using |-style alternative text. E.g.
4390          *   Image:one.jpg|The number "1"
4391          *   Image:tree.jpg|A tree
4392          * given as text will return the HTML of a gallery with two images,
4393          * labeled 'The number "1"' and
4394          * 'A tree'.
4395          */
4396         function renderImageGallery( $text, $params ) {
4397                 $ig = new ImageGallery();
4398                 $ig->setContextTitle( $this->mTitle );
4399                 $ig->setShowBytes( false );
4400                 $ig->setShowFilename( false );
4401                 $ig->setParser( $this );
4402                 $ig->setHideBadImages();
4403                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4404                 $ig->useSkin( $this->mOptions->getSkin() );
4405                 $ig->mRevisionId = $this->mRevisionId;
4406
4407                 if ( isset( $params['showfilename'] ) ) {
4408                         $ig->setShowFilename( true );
4409                 } else {
4410                         $ig->setShowFilename( false );
4411                 }
4412                 if ( isset( $params['caption'] ) ) {
4413                         $caption = $params['caption'];
4414                         $caption = htmlspecialchars( $caption );
4415                         $caption = $this->replaceInternalLinks( $caption );
4416                         $ig->setCaptionHtml( $caption );
4417                 }
4418                 if ( isset( $params['perrow'] ) ) {
4419                         $ig->setPerRow( $params['perrow'] );
4420                 }
4421                 if ( isset( $params['widths'] ) ) {
4422                         $ig->setWidths( $params['widths'] );
4423                 }
4424                 if ( isset( $params['heights'] ) ) {
4425                         $ig->setHeights( $params['heights'] );
4426                 }
4427
4428                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4429
4430                 $lines = StringUtils::explode( "\n", $text );
4431                 foreach ( $lines as $line ) {
4432                         # match lines like these:
4433                         # Image:someimage.jpg|This is some image
4434                         $matches = array();
4435                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4436                         # Skip empty lines
4437                         if ( count( $matches ) == 0 ) {
4438                                 continue;
4439                         }
4440
4441                         if ( strpos( $matches[0], '%' ) !== false ) {
4442                                 $matches[1] = urldecode( $matches[1] );
4443                         }
4444                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4445                         $nt =& $tp;
4446                         if ( is_null( $nt ) ) {
4447                                 # Bogus title. Ignore these so we don't bomb out later.
4448                                 continue;
4449                         }
4450                         if ( isset( $matches[3] ) ) {
4451                                 $label = $matches[3];
4452                         } else {
4453                                 $label = '';
4454                         }
4455
4456                         $html = $this->recursiveTagParse( trim( $label ) );
4457
4458                         $ig->add( $nt, $html );
4459
4460                         # Only add real images (bug #5586)
4461                         if ( $nt->getNamespace() == NS_FILE ) {
4462                                 $this->mOutput->addImage( $nt->getDBkey() );
4463                         }
4464                 }
4465                 return $ig->toHTML();
4466         }
4467
4468         function getImageParams( $handler ) {
4469                 if ( $handler ) {
4470                         $handlerClass = get_class( $handler );
4471                 } else {
4472                         $handlerClass = '';
4473                 }
4474                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4475                         // Initialise static lists
4476                         static $internalParamNames = array(
4477                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4478                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4479                                         'bottom', 'text-bottom' ),
4480                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4481                                         'upright', 'border', 'link', 'alt' ),
4482                         );
4483                         static $internalParamMap;
4484                         if ( !$internalParamMap ) {
4485                                 $internalParamMap = array();
4486                                 foreach ( $internalParamNames as $type => $names ) {
4487                                         foreach ( $names as $name ) {
4488                                                 $magicName = str_replace( '-', '_', "img_$name" );
4489                                                 $internalParamMap[$magicName] = array( $type, $name );
4490                                         }
4491                                 }
4492                         }
4493
4494                         // Add handler params
4495                         $paramMap = $internalParamMap;
4496                         if ( $handler ) {
4497                                 $handlerParamMap = $handler->getParamMap();
4498                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4499                                         $paramMap[$magic] = array( 'handler', $paramName );
4500                                 }
4501                         }
4502                         $this->mImageParams[$handlerClass] = $paramMap;
4503                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4504                 }
4505                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4506         }
4507
4508         /**
4509          * Parse image options text and use it to make an image
4510          * @param Title $title
4511          * @param string $options
4512          * @param LinkHolderArray $holders
4513          */
4514         function makeImage( $title, $options, $holders = false ) {
4515                 # Check if the options text is of the form "options|alt text"
4516                 # Options are:
4517                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4518                 #  * left       no resizing, just left align. label is used for alt= only
4519                 #  * right      same, but right aligned
4520                 #  * none       same, but not aligned
4521                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4522                 #  * center     center the image
4523                 #  * frame      Keep original image size, no magnify-button.
4524                 #  * framed     Same as "frame"
4525                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4526                 #  * upright    reduce width for upright images, rounded to full __0 px
4527                 #  * border     draw a 1px border around the image
4528                 #  * alt        Text for HTML alt attribute (defaults to empty)
4529                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4530                 # vertical-align values (no % or length right now):
4531                 #  * baseline
4532                 #  * sub
4533                 #  * super
4534                 #  * top
4535                 #  * text-top
4536                 #  * middle
4537                 #  * bottom
4538                 #  * text-bottom
4539
4540                 $parts = StringUtils::explode( "|", $options );
4541                 $sk = $this->mOptions->getSkin();
4542
4543                 # Give extensions a chance to select the file revision for us
4544                 $skip = $time = $descQuery = false;
4545                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4546
4547                 if ( $skip ) {
4548                         return $sk->link( $title );
4549                 }
4550
4551                 # Get the file
4552                 $imagename = $title->getDBkey();
4553                 $file = wfFindFile( $title, array( 'time' => $time ) );
4554                 # Get parameter map
4555                 $handler = $file ? $file->getHandler() : false;
4556
4557                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4558
4559                 # Process the input parameters
4560                 $caption = '';
4561                 $params = array( 'frame' => array(), 'handler' => array(),
4562                         'horizAlign' => array(), 'vertAlign' => array() );
4563                 foreach ( $parts as $part ) {
4564                         $part = trim( $part );
4565                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4566                         $validated = false;
4567                         if ( isset( $paramMap[$magicName] ) ) {
4568                                 list( $type, $paramName ) = $paramMap[$magicName];
4569
4570                                 // Special case; width and height come in one variable together
4571                                 if ( $type === 'handler' && $paramName === 'width' ) {
4572                                         $m = array();
4573                                         # (bug 13500) In both cases (width/height and width only),
4574                                         # permit trailing "px" for backward compatibility.
4575                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4576                                                 $width = intval( $m[1] );
4577                                                 $height = intval( $m[2] );
4578                                                 if ( $handler->validateParam( 'width', $width ) ) {
4579                                                         $params[$type]['width'] = $width;
4580                                                         $validated = true;
4581                                                 }
4582                                                 if ( $handler->validateParam( 'height', $height ) ) {
4583                                                         $params[$type]['height'] = $height;
4584                                                         $validated = true;
4585                                                 }
4586                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4587                                                 $width = intval( $value );
4588                                                 if ( $handler->validateParam( 'width', $width ) ) {
4589                                                         $params[$type]['width'] = $width;
4590                                                         $validated = true;
4591                                                 }
4592                                         } // else no validation -- bug 13436
4593                                 } else {
4594                                         if ( $type === 'handler' ) {
4595                                                 # Validate handler parameter
4596                                                 $validated = $handler->validateParam( $paramName, $value );
4597                                         } else {
4598                                                 # Validate internal parameters
4599                                                 switch( $paramName ) {
4600                                                 case 'manualthumb':
4601                                                 case 'alt':
4602                                                         // @todo Fixme: possibly check validity here for
4603                                                         // manualthumb? downstream behavior seems odd with
4604                                                         // missing manual thumbs.
4605                                                         $validated = true;
4606                                                         $value = $this->stripAltText( $value, $holders );
4607                                                         break;
4608                                                 case 'link':
4609                                                         $chars = self::EXT_LINK_URL_CLASS;
4610                                                         $prots = $this->mUrlProtocols;
4611                                                         if ( $value === '' ) {
4612                                                                 $paramName = 'no-link';
4613                                                                 $value = true;
4614                                                                 $validated = true;
4615                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4616                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4617                                                                         $paramName = 'link-url';
4618                                                                         $this->mOutput->addExternalLink( $value );
4619                                                                         $validated = true;
4620                                                                 }
4621                                                         } else {
4622                                                                 $linkTitle = Title::newFromText( $value );
4623                                                                 if ( $linkTitle ) {
4624                                                                         $paramName = 'link-title';
4625                                                                         $value = $linkTitle;
4626                                                                         $this->mOutput->addLink( $linkTitle );
4627                                                                         $validated = true;
4628                                                                 }
4629                                                         }
4630                                                         break;
4631                                                 default:
4632                                                         // Most other things appear to be empty or numeric...
4633                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4634                                                 }
4635                                         }
4636
4637                                         if ( $validated ) {
4638                                                 $params[$type][$paramName] = $value;
4639                                         }
4640                                 }
4641                         }
4642                         if ( !$validated ) {
4643                                 $caption = $part;
4644                         }
4645                 }
4646
4647                 # Process alignment parameters
4648                 if ( $params['horizAlign'] ) {
4649                         $params['frame']['align'] = key( $params['horizAlign'] );
4650                 }
4651                 if ( $params['vertAlign'] ) {
4652                         $params['frame']['valign'] = key( $params['vertAlign'] );
4653                 }
4654
4655                 $params['frame']['caption'] = $caption;
4656
4657                 # Will the image be presented in a frame, with the caption below?
4658                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4659                                  isset( $params['frame']['framed'] ) ||
4660                                  isset( $params['frame']['thumbnail'] ) ||
4661                                  isset( $params['frame']['manualthumb'] );
4662
4663                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4664                 # came to also set the caption, ordinary text after the image -- which
4665                 # makes no sense, because that just repeats the text multiple times in
4666                 # screen readers.  It *also* came to set the title attribute.
4667                 #
4668                 # Now that we have an alt attribute, we should not set the alt text to
4669                 # equal the caption: that's worse than useless, it just repeats the
4670                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4671                 # use the unnamed parameter for alt text as well, just for the time be-
4672                 # ing, if the unnamed param is set and the alt param is not.
4673                 #
4674                 # For the future, we need to figure out if we want to tweak this more,
4675                 # e.g., introducing a title= parameter for the title; ignoring the un-
4676                 # named parameter entirely for images without a caption; adding an ex-
4677                 # plicit caption= parameter and preserving the old magic unnamed para-
4678                 # meter for BC; ...
4679                 if ( $imageIsFramed ) { # Framed image
4680                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4681                                 # No caption or alt text, add the filename as the alt text so
4682                                 # that screen readers at least get some description of the image
4683                                 $params['frame']['alt'] = $title->getText();
4684                         }
4685                         # Do not set $params['frame']['title'] because tooltips don't make sense
4686                         # for framed images
4687                 } else { # Inline image
4688                         if ( !isset( $params['frame']['alt'] ) ) {
4689                                 # No alt text, use the "caption" for the alt text
4690                                 if ( $caption !== '') {
4691                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4692                                 } else {
4693                                         # No caption, fall back to using the filename for the
4694                                         # alt text
4695                                         $params['frame']['alt'] = $title->getText();
4696                                 }
4697                         }
4698                         # Use the "caption" for the tooltip text
4699                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4700                 }
4701
4702                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4703
4704                 # Linker does the rest
4705                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4706
4707                 # Give the handler a chance to modify the parser object
4708                 if ( $handler ) {
4709                         $handler->parserTransformHook( $this, $file );
4710                 }
4711
4712                 return $ret;
4713         }
4714
4715         protected function stripAltText( $caption, $holders ) {
4716                 # Strip bad stuff out of the title (tooltip).  We can't just use
4717                 # replaceLinkHoldersText() here, because if this function is called
4718                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4719                 if ( $holders ) {
4720                         $tooltip = $holders->replaceText( $caption );
4721                 } else {
4722                         $tooltip = $this->replaceLinkHoldersText( $caption );
4723                 }
4724
4725                 # make sure there are no placeholders in thumbnail attributes
4726                 # that are later expanded to html- so expand them now and
4727                 # remove the tags
4728                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4729                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4730
4731                 return $tooltip;
4732         }
4733
4734         /**
4735          * Set a flag in the output object indicating that the content is dynamic and
4736          * shouldn't be cached.
4737          */
4738         function disableCache() {
4739                 wfDebug( "Parser output marked as uncacheable.\n" );
4740                 $this->mOutput->mCacheTime = -1;
4741         }
4742
4743         /**#@+
4744          * Callback from the Sanitizer for expanding items found in HTML attribute
4745          * values, so they can be safely tested and escaped.
4746          * @param string $text
4747          * @param PPFrame $frame
4748          * @return string
4749          * @private
4750          */
4751         function attributeStripCallback( &$text, $frame = false ) {
4752                 $text = $this->replaceVariables( $text, $frame );
4753                 $text = $this->mStripState->unstripBoth( $text );
4754                 return $text;
4755         }
4756
4757         /**#@-*/
4758
4759         /**#@+
4760          * Accessor/mutator
4761          */
4762         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4763         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4764         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4765         /**#@-*/
4766
4767         /**#@+
4768          * Accessor
4769          */
4770         function getTags() {
4771                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4772         }
4773         /**#@-*/
4774
4775
4776         /**
4777          * Break wikitext input into sections, and either pull or replace
4778          * some particular section's text.
4779          *
4780          * External callers should use the getSection and replaceSection methods.
4781          *
4782          * @param string $text Page wikitext
4783          * @param string $section A section identifier string of the form:
4784          *   <flag1> - <flag2> - ... - <section number>
4785          *
4786          * Currently the only recognised flag is "T", which means the target section number
4787          * was derived during a template inclusion parse, in other words this is a template
4788          * section edit link. If no flags are given, it was an ordinary section edit link.
4789          * This flag is required to avoid a section numbering mismatch when a section is
4790          * enclosed by <includeonly> (bug 6563).
4791          *
4792          * The section number 0 pulls the text before the first heading; other numbers will
4793          * pull the given section along with its lower-level subsections. If the section is
4794          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4795          *
4796          * @param string $mode One of "get" or "replace"
4797          * @param string $newText Replacement text for section data.
4798          * @return string for "get", the extracted section text.
4799          *                for "replace", the whole page with the section replaced.
4800          */
4801         private function extractSections( $text, $section, $mode, $newText='' ) {
4802                 global $wgTitle;
4803                 $this->clearState();
4804                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4805                 $this->mOptions = new ParserOptions;
4806                 $this->setOutputType( self::OT_PLAIN );
4807                 $outText = '';
4808                 $frame = $this->getPreprocessor()->newFrame();
4809
4810                 // Process section extraction flags
4811                 $flags = 0;
4812                 $sectionParts = explode( '-', $section );
4813                 $sectionIndex = array_pop( $sectionParts );
4814                 foreach ( $sectionParts as $part ) {
4815                         if ( $part === 'T' ) {
4816                                 $flags |= self::PTD_FOR_INCLUSION;
4817                         }
4818                 }
4819                 // Preprocess the text
4820                 $root = $this->preprocessToDom( $text, $flags );
4821
4822                 // <h> nodes indicate section breaks
4823                 // They can only occur at the top level, so we can find them by iterating the root's children
4824                 $node = $root->getFirstChild();
4825
4826                 // Find the target section
4827                 if ( $sectionIndex == 0 ) {
4828                         // Section zero doesn't nest, level=big
4829                         $targetLevel = 1000;
4830                 } else {
4831             while ( $node ) {
4832                 if ( $node->getName() === 'h' ) {
4833                     $bits = $node->splitHeading();
4834                                         if ( $bits['i'] == $sectionIndex ) {
4835                                         $targetLevel = $bits['level'];
4836                                                 break;
4837                                         }
4838                                 }
4839                                 if ( $mode === 'replace' ) {
4840                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4841                                 }
4842                                 $node = $node->getNextSibling();
4843                         }
4844                 }
4845
4846                 if ( !$node ) {
4847                         // Not found
4848                         if ( $mode === 'get' ) {
4849                                 return $newText;
4850                         } else {
4851                                 return $text;
4852                         }
4853                 }
4854
4855                 // Find the end of the section, including nested sections
4856                 do {
4857                         if ( $node->getName() === 'h' ) {
4858                                 $bits = $node->splitHeading();
4859                                 $curLevel = $bits['level'];
4860                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4861                                         break;
4862                                 }
4863                         }
4864                         if ( $mode === 'get' ) {
4865                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4866                         }
4867                         $node = $node->getNextSibling();
4868                 } while ( $node );
4869
4870                 // Write out the remainder (in replace mode only)
4871                 if ( $mode === 'replace' ) {
4872                         // Output the replacement text
4873                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4874                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4875                         // Only add trailing whitespace if there is newText
4876                         if ( $newText != "" ) {
4877                                 $outText .= $newText . "\n\n";
4878                         }
4879
4880                         while ( $node ) {
4881                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4882                                 $node = $node->getNextSibling();
4883                         }
4884                 }
4885
4886                 if ( is_string( $outText ) ) {
4887                         // Re-insert stripped tags
4888                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4889                 }
4890
4891                 return $outText;
4892         }
4893
4894         /**
4895          * This function returns the text of a section, specified by a number ($section).
4896          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4897          * the first section before any such heading (section 0).
4898          *
4899          * If a section contains subsections, these are also returned.
4900          *
4901          * @param string $text text to look in
4902          * @param string $section section identifier
4903          * @param string $deftext default to return if section is not found
4904          * @return string text of the requested section
4905          */
4906         public function getSection( $text, $section, $deftext='' ) {
4907                 return $this->extractSections( $text, $section, "get", $deftext );
4908         }
4909
4910         public function replaceSection( $oldtext, $section, $text ) {
4911                 return $this->extractSections( $oldtext, $section, "replace", $text );
4912         }
4913
4914         /**
4915          * Get the timestamp associated with the current revision, adjusted for
4916          * the default server-local timestamp
4917          */
4918         function getRevisionTimestamp() {
4919                 if ( is_null( $this->mRevisionTimestamp ) ) {
4920                         wfProfileIn( __METHOD__ );
4921                         global $wgContLang;
4922                         $dbr = wfGetDB( DB_SLAVE );
4923                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4924                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4925
4926                         // Normalize timestamp to internal MW format for timezone processing.
4927                         // This has the added side-effect of replacing a null value with
4928                         // the current time, which gives us more sensible behavior for
4929                         // previews.
4930                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4931
4932                         // The cryptic '' timezone parameter tells to use the site-default
4933                         // timezone offset instead of the user settings.
4934                         //
4935                         // Since this value will be saved into the parser cache, served
4936                         // to other users, and potentially even used inside links and such,
4937                         // it needs to be consistent for all visitors.
4938                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4939
4940                         wfProfileOut( __METHOD__ );
4941                 }
4942                 return $this->mRevisionTimestamp;
4943         }
4944
4945         /**
4946          * Get the name of the user that edited the last revision
4947          */
4948         function getRevisionUser() {
4949                 // if this template is subst: the revision id will be blank,
4950                 // so just use the current user's name
4951                 if ( $this->mRevisionId ) {
4952                         $revision = Revision::newFromId( $this->mRevisionId );
4953                         $revuser = $revision->getUserText();
4954                 } else {
4955                         global $wgUser;
4956                         $revuser = $wgUser->getName();
4957                 }
4958                 return $revuser;
4959         }
4960
4961         /**
4962          * Mutator for $mDefaultSort
4963          *
4964          * @param $sort New value
4965          */
4966         public function setDefaultSort( $sort ) {
4967                 $this->mDefaultSort = $sort;
4968         }
4969
4970         /**
4971          * Accessor for $mDefaultSort
4972          * Will use the title/prefixed title if none is set
4973          *
4974          * @return string
4975          */
4976         public function getDefaultSort() {
4977                 global $wgCategoryPrefixedDefaultSortkey;
4978                 if ( $this->mDefaultSort !== false ) {
4979                         return $this->mDefaultSort;
4980                 } elseif ( $this->mTitle->getNamespace() == NS_CATEGORY ||
4981                         !$wgCategoryPrefixedDefaultSortkey )
4982                 {
4983                         return $this->mTitle->getText();
4984                 } else {
4985                         return $this->mTitle->getPrefixedText();
4986                 }
4987         }
4988
4989         /**
4990          * Accessor for $mDefaultSort
4991          * Unlike getDefaultSort(), will return false if none is set
4992          *
4993          * @return string or false
4994          */
4995         public function getCustomDefaultSort() {
4996                 return $this->mDefaultSort;
4997         }
4998
4999         /**
5000          * Try to guess the section anchor name based on a wikitext fragment
5001          * presumably extracted from a heading, for example "Header" from
5002          * "== Header ==".
5003          */
5004         public function guessSectionNameFromWikiText( $text ) {
5005                 # Strip out wikitext links(they break the anchor)
5006                 $text = $this->stripSectionName( $text );
5007                 $headline = Sanitizer::decodeCharReferences( $text );
5008                 # strip out HTML
5009                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5010                 $headline = trim( $headline );
5011                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5012                 $replacearray = array(
5013                         '%3A' => ':',
5014                         '%' => '.'
5015                 );
5016                 return str_replace(
5017                         array_keys( $replacearray ),
5018                         array_values( $replacearray ),
5019                         $sectionanchor );
5020         }
5021
5022         /**
5023          * Strips a text string of wikitext for use in a section anchor
5024          *
5025          * Accepts a text string and then removes all wikitext from the
5026          * string and leaves only the resultant text (i.e. the result of
5027          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5028          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5029          * to create valid section anchors by mimicing the output of the
5030          * parser when headings are parsed.
5031          *
5032          * @param $text string Text string to be stripped of wikitext
5033          * for use in a Section anchor
5034          * @return Filtered text string
5035          */
5036         public function stripSectionName( $text ) {
5037                 # Strip internal link markup
5038                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5039                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5040
5041                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5042                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5043                 # on how many empty links there are on the page - need to figure that out.
5044                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5045
5046                 # Parse wikitext quotes (italics & bold)
5047                 $text = $this->doQuotes( $text );
5048
5049                 # Strip HTML tags
5050                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5051                 return $text;
5052         }
5053
5054         function srvus( $text ) {
5055                 return $this->testSrvus( $text, $this->mOutputType );
5056         }
5057
5058         /**
5059          * strip/replaceVariables/unstrip for preprocessor regression testing
5060          */
5061         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5062                 $this->clearState();
5063                 if ( !$title instanceof Title ) {
5064                         $title = Title::newFromText( $title );
5065                 }
5066                 $this->mTitle = $title;
5067                 $this->mOptions = $options;
5068                 $this->setOutputType( $outputType );
5069                 $text = $this->replaceVariables( $text );
5070                 $text = $this->mStripState->unstripBoth( $text );
5071                 $text = Sanitizer::removeHTMLtags( $text );
5072                 return $text;
5073         }
5074
5075         function testPst( $text, $title, $options ) {
5076                 global $wgUser;
5077                 if ( !$title instanceof Title ) {
5078                         $title = Title::newFromText( $title );
5079                 }
5080                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5081         }
5082
5083         function testPreprocess( $text, $title, $options ) {
5084                 if ( !$title instanceof Title ) {
5085                         $title = Title::newFromText( $title );
5086                 }
5087                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5088         }
5089
5090         function markerSkipCallback( $s, $callback ) {
5091                 $i = 0;
5092                 $out = '';
5093                 while ( $i < strlen( $s ) ) {
5094                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5095                         if ( $markerStart === false ) {
5096                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5097                                 break;
5098                         } else {
5099                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5100                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5101                                 if ( $markerEnd === false ) {
5102                                         $out .= substr( $s, $markerStart );
5103                                         break;
5104                                 } else {
5105                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5106                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5107                                         $i = $markerEnd;
5108                                 }
5109                         }
5110                 }
5111                 return $out;
5112         }
5113
5114         function serialiseHalfParsedText( $text ) {
5115                 $data = array();
5116                 $data['text'] = $text;
5117
5118                 // First, find all strip markers, and store their
5119                 //  data in an array.
5120                 $stripState = new StripState;
5121                 $pos = 0;
5122                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5123                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5124                 {
5125                         $end_pos += strlen( self::MARKER_SUFFIX );
5126                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5127
5128                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5129                                 $replaceArray = $stripState->general;
5130                                 $stripText = $this->mStripState->general->data[$marker];
5131                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5132                                 $replaceArray = $stripState->nowiki;
5133                                 $stripText = $this->mStripState->nowiki->data[$marker];
5134                         } else {
5135                                 throw new MWException( "Hanging strip marker: '$marker'." );
5136                         }
5137
5138                         $replaceArray->setPair( $marker, $stripText );
5139                         $pos = $end_pos;
5140                 }
5141                 $data['stripstate'] = $stripState;
5142
5143                 // Now, find all of our links, and store THEIR
5144                 //  data in an array! :)
5145                 $links = array( 'internal' => array(), 'interwiki' => array() );
5146                 $pos = 0;
5147
5148                 // Internal links
5149                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5150                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5151
5152                         $ns = trim( $ns );
5153                         if ( empty( $links['internal'][$ns] ) ) {
5154                                 $links['internal'][$ns] = array();
5155                         }
5156
5157                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5158                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5159                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5160                 }
5161
5162                 $pos = 0;
5163
5164                 // Interwiki links
5165                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5166                         $data = substr( $text, $start_pos );
5167                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5168                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5169                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5170                 }
5171
5172                 $data['linkholder'] = $links;
5173
5174                 return $data;
5175         }
5176
5177         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5178                 if ( !$intPrefix ) {
5179                         $intPrefix = $this->getRandomString();
5180                 }
5181
5182                 // First, extract the strip state.
5183                 $stripState = $data['stripstate'];
5184                 $this->mStripState->general->merge( $stripState->general );
5185                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5186
5187                 // Now, extract the text, and renumber links
5188                 $text = $data['text'];
5189                 $links = $data['linkholder'];
5190
5191                 // Internal...
5192                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5193                         foreach ( $nsLinks as $key => $entry ) {
5194                                 $newKey = $intPrefix . '-' . $key;
5195                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5196
5197                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5198                         }
5199                 }
5200
5201                 // Interwiki...
5202                 foreach ( $links['interwiki'] as $key => $entry ) {
5203                         $newKey = "$intPrefix-$key";
5204                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5205
5206                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5207                 }
5208
5209                 // Should be good to go.
5210                 return $text;
5211         }
5212 }
5213
5214 /**
5215  * @todo document, briefly.
5216  * @ingroup Parser
5217  */
5218 class StripState {
5219         var $general, $nowiki;
5220
5221         function __construct() {
5222                 $this->general = new ReplacementArray;
5223                 $this->nowiki = new ReplacementArray;
5224         }
5225
5226         function unstripGeneral( $text ) {
5227                 wfProfileIn( __METHOD__ );
5228                 do {
5229                         $oldText = $text;
5230                         $text = $this->general->replace( $text );
5231                 } while ( $text !== $oldText );
5232                 wfProfileOut( __METHOD__ );
5233                 return $text;
5234         }
5235
5236         function unstripNoWiki( $text ) {
5237                 wfProfileIn( __METHOD__ );
5238                 do {
5239                         $oldText = $text;
5240                         $text = $this->nowiki->replace( $text );
5241                 } while ( $text !== $oldText );
5242                 wfProfileOut( __METHOD__ );
5243                 return $text;
5244         }
5245
5246         function unstripBoth( $text ) {
5247                 wfProfileIn( __METHOD__ );
5248                 do {
5249                         $oldText = $text;
5250                         $text = $this->general->replace( $text );
5251                         $text = $this->nowiki->replace( $text );
5252                 } while ( $text !== $oldText );
5253                 wfProfileOut( __METHOD__ );
5254                 return $text;
5255         }
5256 }
5257
5258 /**
5259  * @todo document, briefly.
5260  * @ingroup Parser
5261  */
5262 class OnlyIncludeReplacer {
5263         var $output = '';
5264
5265         function replace( $matches ) {
5266                 if ( substr( $matches[1], -1 ) === "\n" ) {
5267                         $this->output .= substr( $matches[1], 0, -1 );
5268                 } else {
5269                         $this->output .= $matches[1];
5270                 }
5271         }
5272 }