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 six 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  * getTransclusionText()
  30  *   Extracts the text of a template with only <includeonly>, etc., parsed
  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  *  $wgMaxArticleSize*
  42  *
  43  *  * only within ParserOptions
  44  * </pre>
  45  *
  46  * @ingroup Parser
  47  */
  48 class Parser
  49 {
  50         /**
  51          * Update this version number when the ParserOutput format
  52          * changes in an incompatible way, so the parser cache
  53          * can automatically discard old data.
  54          */
  55         const VERSION = '1.6.4';
  56
  57         # Flags for Parser::setFunctionHook
  58         # Also available as global constants from Defines.php
  59         const SFH_NO_HASH = 1;
  60         const SFH_OBJECT_ARGS = 2;
  61
  62         # Constants needed for external link processing
  63         # Everything except bracket, space, or control characters
  64         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  65         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  66                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  67
  68         // State constants for the definition list colon extraction
  69         const COLON_STATE_TEXT = 0;
  70         const COLON_STATE_TAG = 1;
  71         const COLON_STATE_TAGSTART = 2;
  72         const COLON_STATE_CLOSETAG = 3;
  73         const COLON_STATE_TAGSLASH = 4;
  74         const COLON_STATE_COMMENT = 5;
  75         const COLON_STATE_COMMENTDASH = 6;
  76         const COLON_STATE_COMMENTDASHDASH = 7;
  77
  78         // Flags for preprocessToDom
  79         const PTD_FOR_INCLUSION = 1;
  80
  81         // Allowed values for $this->mOutputType
  82         // Parameter to startExternalParse().
  83         const OT_HTML = 1;
  84         const OT_WIKI = 2;
  85         const OT_PREPROCESS = 3;
  86         const OT_MSG = 3;
  87         const OT_INCLUDES = 4;
  88
  89         // Marker Suffix needs to be accessible staticly.
  90         const MARKER_SUFFIX = "-QINU\x7f";
  91
  92         /**#@+
  93          * @private
  94          */
  95         # Persistent:
  96         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  97                 $mSubsts, $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex,
  98                 $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList,
  99                 $mVarCache, $mConf, $mFunctionTagHooks;
 100
 101
 102         # Cleared with clearState():
 103         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 104         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 105         var $mLinkHolders, $mLinkID;
 106         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 107         var $mTplExpandCache; // empty-frame expansion cache
 108         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 109         var $mExpensiveFunctionCount; // number of expensive parser function calls
 110
 111         # Temporary
 112         # These are variables reset at least once per parse regardless of $clearState
 113         var $mOptions,      // ParserOptions object
 114                 $mTitle,        // Title context, used for self-link rendering and similar things
 115                 $mOutputType,   // Output type, one of the OT_xxx constants
 116                 $ot,            // Shortcut alias, see setOutputType()
 117                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 118                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 119                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 120
 121         /**#@-*/
 122
 123         /**
 124          * Constructor
 125          *
 126          * @public
 127          */
 128         function __construct( $conf = array() ) {
 129                 $this->mConf = $conf;
 130                 $this->mTagHooks = array();
 131                 $this->mTransparentTagHooks = array();
 132                 $this->mFunctionHooks = array();
 133                 $this->mFunctionTagHooks = array();
 134                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 135                 $this->mDefaultStripList = $this->mStripList = array();
 136                 $this->mUrlProtocols = wfUrlProtocols();
 137                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 138                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 139                 $this->mVarCache = array();
 140                 if ( isset( $conf['preprocessorClass'] ) ) {
 141                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 142                 } elseif ( extension_loaded( 'domxml' ) ) {
 143                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 144                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 145                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 146                 } elseif ( extension_loaded( 'dom' ) ) {
 147                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 148                 } else {
 149                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 150                 }
 151                 $this->mMarkerIndex = 0;
 152                 $this->mFirstCall = true;
 153         }
 154
 155         /**
 156          * Reduce memory usage to reduce the impact of circular references
 157          */
 158         function __destruct() {
 159                 if ( isset( $this->mLinkHolders ) ) {
 160                         $this->mLinkHolders->__destruct();
 161                 }
 162                 foreach ( $this as $name => $value ) {
 163                         unset( $this->$name );
 164                 }
 165         }
 166
 167         /**
 168          * Do various kinds of initialisation on the first call of the parser
 169          */
 170         function firstCallInit() {
 171                 if ( !$this->mFirstCall ) {
 172                         return;
 173                 }
 174                 $this->mFirstCall = false;
 175
 176                 wfProfileIn( __METHOD__ );
 177
 178                 CoreParserFunctions::register( $this );
 179                 CoreTagHooks::register( $this );
 180                 $this->initialiseVariables();
 181
 182                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 183                 wfProfileOut( __METHOD__ );
 184         }
 185
 186         /**
 187          * Clear Parser state
 188          *
 189          * @private
 190          */
 191         function clearState() {
 192                 wfProfileIn( __METHOD__ );
 193                 if ( $this->mFirstCall ) {
 194                         $this->firstCallInit();
 195                 }
 196                 $this->mOutput = new ParserOutput;
 197                 $this->mAutonumber = 0;
 198                 $this->mLastSection = '';
 199                 $this->mDTopen = false;
 200                 $this->mIncludeCount = array();
 201                 $this->mStripState = new StripState;
 202                 $this->mArgStack = false;
 203                 $this->mInPre = false;
 204                 $this->mLinkHolders = new LinkHolderArray( $this );
 205                 $this->mLinkID = 0;
 206                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 207                 $this->mVarCache = array();
 208
 209                 /**
 210                  * Prefix for temporary replacement strings for the multipass parser.
 211                  * \x07 should never appear in input as it's disallowed in XML.
 212                  * Using it at the front also gives us a little extra robustness
 213                  * since it shouldn't match when butted up against identifier-like
 214                  * string constructs.
 215                  *
 216                  * Must not consist of all title characters, or else it will change
 217                  * the behaviour of <nowiki> in a link.
 218                  */
 219                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 220                 # Changed to \x7f to allow XML double-parsing -- TS
 221                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 222
 223
 224                 # Clear these on every parse, bug 4549
 225                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 226
 227                 $this->mShowToc = true;
 228                 $this->mForceTocPosition = false;
 229                 $this->mIncludeSizes = array(
 230                         'post-expand' => 0,
 231                         'arg' => 0,
 232                 );
 233                 $this->mPPNodeCount = 0;
 234                 $this->mDefaultSort = false;
 235                 $this->mHeadings = array();
 236                 $this->mDoubleUnderscores = array();
 237                 $this->mExpensiveFunctionCount = 0;
 238
 239                 # Fix cloning
 240                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 241                         $this->mPreprocessor = null;
 242                 }
 243
 244                 wfRunHooks( 'ParserClearState', array( &$this ) );
 245                 wfProfileOut( __METHOD__ );
 246         }
 247
 248         function setOutputType( $ot ) {
 249                 $this->mOutputType = $ot;
 250                 // Shortcut alias
 251                 $this->ot = array(
 252                         'html' => $ot == self::OT_HTML,
 253                         'wiki' => $ot == self::OT_WIKI,
 254                         'pre' => $ot == self::OT_PREPROCESS,
 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                 } elseif ( !( $wgDisableLangConversion
 369                                           || isset( $this->mDoubleUnderscores['notitleconvert'] ) ) ) {
 370                         $this->mOutput->setTitleText( $wgContLang->convert( $title->getPrefixedText() ) );
 371                 }
 372
 373                 $text = $this->mStripState->unstripNoWiki( $text );
 374
 375                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 376
 377                 if ( $this->mTransparentTagHooks ) {
 378                         //!JF Move to its own function
 379                         $uniq_prefix = $this->mUniqPrefix;
 380                         $matches = array();
 381                         $elements = array_keys( $this->mTransparentTagHooks );
 382                         $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 383
 384                         foreach( $matches as $marker => $data ) {
 385                                 list( $element, $content, $params, $tag ) = $data;
 386                                 $tagName = strtolower( $element );
 387                                 if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 388                                         $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 389                                                 array( $content, $params, $this ) );
 390                                 } else {
 391                                         $output = $tag;
 392                                 }
 393                                 $this->mStripState->general->setPair( $marker, $output );
 394                         }
 395                         $text = $this->mStripState->unstripGeneral( $text );
 396                 }
 397
 398                 $text = Sanitizer::normalizeCharReferences( $text );
 399
 400                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 401                         $text = MWTidy::tidy( $text );
 402                 } else {
 403                         # attempt to sanitize at least some nesting problems
 404                         # (bug #2702 and quite a few others)
 405                         $tidyregs = array(
 406                                 # ''Something [http://www.cool.com cool''] -->
 407                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 408                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 409                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 410                                 # fix up an anchor inside another anchor, only
 411                                 # at least for a single single nested link (bug 3695)
 412                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 413                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 414                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 415                                 # contains a div, so fix it up here; replace
 416                                 # div with escaped text
 417                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 418                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 419                                 # remove empty italic or bold tag pairs, some
 420                                 # introduced by rules above
 421                                 '/<([bi])><\/\\1>/' => '',
 422                         );
 423
 424                         $text = preg_replace(
 425                                 array_keys( $tidyregs ),
 426                                 array_values( $tidyregs ),
 427                                 $text );
 428                 }
 429                 global $wgExpensiveParserFunctionLimit;
 430                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 431                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 432                 }
 433
 434                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 435
 436                 # Information on include size limits, for the benefit of users who try to skirt them
 437                 if ( $this->mOptions->getEnableLimitReport() ) {
 438                         global $wgExpensiveParserFunctionLimit;
 439                         $max = $this->mOptions->getMaxIncludeSize();
 440                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 441                         $limitReport =
 442                                 "NewPP limit report\n" .
 443                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 444                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 445                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 446                                 $PFreport;
 447                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 448                         $text .= "\n<!-- \n$limitReport-->\n";
 449                 }
 450                 $this->mOutput->setText( $text );
 451
 452                 $this->mRevisionId = $oldRevisionId;
 453                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 454                 wfProfileOut( $fname );
 455                 wfProfileOut( __METHOD__ );
 456
 457                 return $this->mOutput;
 458         }
 459
 460         /**
 461          * Recursive parser entry point that can be called from an extension tag
 462          * hook.
 463          *
 464          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 465          *
 466          * @param $text String: text extension wants to have parsed
 467          * @param PPFrame $frame: The frame to use for expanding any template variables
 468          */
 469         function recursiveTagParse( $text, $frame=false ) {
 470                 wfProfileIn( __METHOD__ );
 471                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 472                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 473                 $text = $this->internalParse( $text, false, $frame );
 474                 wfProfileOut( __METHOD__ );
 475                 return $text;
 476         }
 477
 478         /**
 479          * Expand templates and variables in the text, producing valid, static wikitext.
 480          * Also removes comments.
 481          */
 482         function preprocess( $text, $title, $options, $revid = null ) {
 483                 wfProfileIn( __METHOD__ );
 484                 $this->clearState();
 485                 $this->setOutputType( self::OT_PREPROCESS );
 486                 $this->mOptions = $options;
 487                 $this->setTitle( $title );
 488                 if( $revid !== null ) {
 489                         $this->mRevisionId = $revid;
 490                 }
 491                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 492                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 493                 $text = $this->replaceVariables( $text );
 494                 $text = $this->mStripState->unstripBoth( $text );
 495                 wfProfileOut( __METHOD__ );
 496                 return $text;
 497         }
 498
 499         /**
 500          * Get the wikitext of a page as though it was transcluded.
 501          *
 502          * Specifically <includeonly> etc. are parsed, redirects are followed, comments
 503          * are removed, but templates arguments and parser functions are untouched.
 504          *
 505          * This is not called by the parser itself, see braceSubstitution for its transclusion.
 506          */
 507         public function getTransclusionText( $title, $options ) {
 508                 // Must initialize first
 509                 $this->clearState();
 510                 $this->setOutputType( self::OT_INCLUDES );
 511                 $this->mOptions = $options;
 512                 $this->setTitle( new FakeTitle );
 513
 514                 list( $text, $title ) = $this->getTemplateDom( $title );
 515                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 516                 return $this->getPreprocessor()->newFrame()->expand( $text, $flags );
 517         }
 518
 519         /**
 520          * Get a random string
 521          *
 522          * @private
 523          * @static
 524          */
 525         function getRandomString() {
 526                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 527         }
 528
 529         function &getTitle() { return $this->mTitle; }
 530         function getOptions() { return $this->mOptions; }
 531         function getRevisionId() { return $this->mRevisionId; }
 532         function getOutput() { return $this->mOutput; }
 533         function nextLinkID() { return $this->mLinkID++; }
 534
 535         function getFunctionLang() {
 536                 global $wgLang, $wgContLang;
 537
 538                 $target = $this->mOptions->getTargetLanguage();
 539                 if ( $target !== null ) {
 540                         return $target;
 541                 } else {
 542                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 543                 }
 544         }
 545
 546         /**
 547          * Get a preprocessor object
 548          */
 549         function getPreprocessor() {
 550                 if ( !isset( $this->mPreprocessor ) ) {
 551                         $class = $this->mPreprocessorClass;
 552                         $this->mPreprocessor = new $class( $this );
 553                 }
 554                 return $this->mPreprocessor;
 555         }
 556
 557         /**
 558          * Replaces all occurrences of HTML-style comments and the given tags
 559          * in the text with a random marker and returns the next text. The output
 560          * parameter $matches will be an associative array filled with data in
 561          * the form:
 562          *   'UNIQ-xxxxx' => array(
 563          *     'element',
 564          *     'tag content',
 565          *     array( 'param' => 'x' ),
 566          *     '<element param="x">tag content</element>' ) )
 567          *
 568          * @param $elements list of element names. Comments are always extracted.
 569          * @param $text Source text string.
 570          * @param $uniq_prefix
 571          *
 572          * @public
 573          * @static
 574          */
 575         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 576                 static $n = 1;
 577                 $stripped = '';
 578                 $matches = array();
 579
 580                 $taglist = implode( '|', $elements );
 581                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 582
 583                 while ( $text != '' ) {
 584                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 585                         $stripped .= $p[0];
 586                         if( count( $p ) < 5 ) {
 587                                 break;
 588                         }
 589                         if( count( $p ) > 5 ) {
 590                                 // comment
 591                                 $element    = $p[4];
 592                                 $attributes = '';
 593                                 $close      = '';
 594                                 $inside     = $p[5];
 595                         } else {
 596                                 // tag
 597                                 $element    = $p[1];
 598                                 $attributes = $p[2];
 599                                 $close      = $p[3];
 600                                 $inside     = $p[4];
 601                         }
 602
 603                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 604                         $stripped .= $marker;
 605
 606                         if ( $close === '/>' ) {
 607                                 // Empty element tag, <tag />
 608                                 $content = null;
 609                                 $text = $inside;
 610                                 $tail = null;
 611                         } else {
 612                                 if( $element === '!--' ) {
 613                                         $end = '/(-->)/';
 614                                 } else {
 615                                         $end = "/(<\\/$element\\s*>)/i";
 616                                 }
 617                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 618                                 $content = $q[0];
 619                                 if( count( $q ) < 3 ) {
 620                                         # No end tag -- let it run out to the end of the text.
 621                                         $tail = '';
 622                                         $text = '';
 623                                 } else {
 624                                         $tail = $q[1];
 625                                         $text = $q[2];
 626                                 }
 627                         }
 628
 629                         $matches[$marker] = array( $element,
 630                                 $content,
 631                                 Sanitizer::decodeTagAttributes( $attributes ),
 632                                 "<$element$attributes$close$content$tail" );
 633                 }
 634                 return $stripped;
 635         }
 636
 637         /**
 638          * Get a list of strippable XML-like elements
 639          */
 640         function getStripList() {
 641                 return $this->mStripList;
 642         }
 643
 644         /**
 645          * @deprecated use replaceVariables
 646          */
 647         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 648                 return $text;
 649         }
 650
 651         /**
 652          * Restores pre, math, and other extensions removed by strip()
 653          *
 654          * always call unstripNoWiki() after this one
 655          * @private
 656          * @deprecated use $this->mStripState->unstrip()
 657          */
 658         function unstrip( $text, $state ) {
 659                 return $state->unstripGeneral( $text );
 660         }
 661
 662         /**
 663          * Always call this after unstrip() to preserve the order
 664          *
 665          * @private
 666          * @deprecated use $this->mStripState->unstrip()
 667          */
 668         function unstripNoWiki( $text, $state ) {
 669                 return $state->unstripNoWiki( $text );
 670         }
 671
 672         /**
 673          * @deprecated use $this->mStripState->unstripBoth()
 674          */
 675         function unstripForHTML( $text ) {
 676                 return $this->mStripState->unstripBoth( $text );
 677         }
 678
 679         /**
 680          * Add an item to the strip state
 681          * Returns the unique tag which must be inserted into the stripped text
 682          * The tag will be replaced with the original text in unstrip()
 683          *
 684          * @private
 685          */
 686         function insertStripItem( $text ) {
 687                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 688                 $this->mMarkerIndex++;
 689                 $this->mStripState->general->setPair( $rnd, $text );
 690                 return $rnd;
 691         }
 692
 693         /**
 694          * Interface with html tidy
 695          * @deprecated Use MWTidy::tidy()
 696          */
 697         public static function tidy( $text ) {
 698                 wfDeprecated( __METHOD__ );
 699                 return MWTidy::tidy( $text );
 700         }
 701
 702         /**
 703          * parse the wiki syntax used to render tables
 704          *
 705          * @private
 706          */
 707         function doTableStuff ( $text ) {
 708                 wfProfileIn( __METHOD__ );
 709
 710                 $lines = StringUtils::explode( "\n", $text );
 711                 $out = '';
 712                 $td_history = array (); // Is currently a td tag open?
 713                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 714                 $tr_history = array (); // Is currently a tr tag open?
 715                 $tr_attributes = array (); // history of tr attributes
 716                 $has_opened_tr = array(); // Did this table open a <tr> element?
 717                 $indent_level = 0; // indent level of the table
 718
 719                 foreach ( $lines as $outLine ) {
 720                         $line = trim( $outLine );
 721
 722                         if( $line == '' ) { // empty line, go to next line
 723                                 $out .= $outLine."\n";
 724                                 continue;
 725                         }
 726                         $first_character = $line[0];
 727                         $matches = array();
 728
 729                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 730                                 // First check if we are starting a new table
 731                                 $indent_level = strlen( $matches[1] );
 732
 733                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 734                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 735
 736                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 737                                 array_push( $td_history , false );
 738                                 array_push( $last_tag_history , '' );
 739                                 array_push( $tr_history , false );
 740                                 array_push( $tr_attributes , '' );
 741                                 array_push( $has_opened_tr , false );
 742                         } else if ( count ( $td_history ) == 0 ) {
 743                                 // Don't do any of the following
 744                                 $out .= $outLine."\n";
 745                                 continue;
 746                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 747                                 // We are ending a table
 748                                 $line = '</table>' . substr ( $line , 2 );
 749                                 $last_tag = array_pop ( $last_tag_history );
 750
 751                                 if ( !array_pop ( $has_opened_tr ) ) {
 752                                         $line = "<tr><td></td></tr>{$line}";
 753                                 }
 754
 755                                 if ( array_pop ( $tr_history ) ) {
 756                                         $line = "</tr>{$line}";
 757                                 }
 758
 759                                 if ( array_pop ( $td_history ) ) {
 760                                         $line = "</{$last_tag}>{$line}";
 761                                 }
 762                                 array_pop ( $tr_attributes );
 763                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 764                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 765                                 // Now we have a table row
 766                                 $line = preg_replace( '#^\|-+#', '', $line );
 767
 768                                 // Whats after the tag is now only attributes
 769                                 $attributes = $this->mStripState->unstripBoth( $line );
 770                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 771                                 array_pop( $tr_attributes );
 772                                 array_push( $tr_attributes, $attributes );
 773
 774                                 $line = '';
 775                                 $last_tag = array_pop ( $last_tag_history );
 776                                 array_pop ( $has_opened_tr );
 777                                 array_push ( $has_opened_tr , true );
 778
 779                                 if ( array_pop ( $tr_history ) ) {
 780                                         $line = '</tr>';
 781                                 }
 782
 783                                 if ( array_pop ( $td_history ) ) {
 784                                         $line = "</{$last_tag}>{$line}";
 785                                 }
 786
 787                                 $outLine = $line;
 788                                 array_push ( $tr_history , false );
 789                                 array_push ( $td_history , false );
 790                                 array_push ( $last_tag_history , '' );
 791                         }
 792                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 793                                 // This might be cell elements, td, th or captions
 794                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 795                                         $first_character = '+';
 796                                         $line = substr ( $line , 1 );
 797                                 }
 798
 799                                 $line = substr ( $line , 1 );
 800
 801                                 if ( $first_character === '!' ) {
 802                                         $line = str_replace ( '!!' , '||' , $line );
 803                                 }
 804
 805                                 // Split up multiple cells on the same line.
 806                                 // FIXME : This can result in improper nesting of tags processed
 807                                 // by earlier parser steps, but should avoid splitting up eg
 808                                 // attribute values containing literal "||".
 809                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 810
 811                                 $outLine = '';
 812
 813                                 // Loop through each table cell
 814                                 foreach ( $cells as $cell )
 815                                 {
 816                                         $previous = '';
 817                                         if ( $first_character !== '+' )
 818                                         {
 819                                                 $tr_after = array_pop ( $tr_attributes );
 820                                                 if ( !array_pop ( $tr_history ) ) {
 821                                                         $previous = "<tr{$tr_after}>\n";
 822                                                 }
 823                                                 array_push ( $tr_history , true );
 824                                                 array_push ( $tr_attributes , '' );
 825                                                 array_pop ( $has_opened_tr );
 826                                                 array_push ( $has_opened_tr , true );
 827                                         }
 828
 829                                         $last_tag = array_pop ( $last_tag_history );
 830
 831                                         if ( array_pop ( $td_history ) ) {
 832                                                 $previous = "</{$last_tag}>{$previous}";
 833                                         }
 834
 835                                         if ( $first_character === '|' ) {
 836                                                 $last_tag = 'td';
 837                                         } else if ( $first_character === '!' ) {
 838                                                 $last_tag = 'th';
 839                                         } else if ( $first_character === '+' ) {
 840                                                 $last_tag = 'caption';
 841                                         } else {
 842                                                 $last_tag = '';
 843                                         }
 844
 845                                         array_push ( $last_tag_history , $last_tag );
 846
 847                                         // A cell could contain both parameters and data
 848                                         $cell_data = explode ( '|' , $cell , 2 );
 849
 850                                         // Bug 553: Note that a '|' inside an invalid link should not
 851                                         // be mistaken as delimiting cell parameters
 852                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 853                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 854                                         } else if ( count ( $cell_data ) == 1 )
 855                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 856                                         else {
 857                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 858                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 859                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 860                                         }
 861
 862                                         $outLine .= $cell;
 863                                         array_push ( $td_history , true );
 864                                 }
 865                         }
 866                         $out .= $outLine . "\n";
 867                 }
 868
 869                 // Closing open td, tr && table
 870                 while ( count ( $td_history ) > 0 )
 871                 {
 872                         if ( array_pop ( $td_history ) ) {
 873                                 $out .= "</td>\n";
 874                         }
 875                         if ( array_pop ( $tr_history ) ) {
 876                                 $out .= "</tr>\n";
 877                         }
 878                         if ( !array_pop ( $has_opened_tr ) ) {
 879                                 $out .= "<tr><td></td></tr>\n" ;
 880                         }
 881
 882                         $out .= "</table>\n";
 883                 }
 884
 885                 // Remove trailing line-ending (b/c)
 886                 if ( substr( $out, -1 ) === "\n" ) {
 887                         $out = substr( $out, 0, -1 );
 888                 }
 889
 890                 // special case: don't return empty table
 891                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 892                         $out = '';
 893                 }
 894
 895                 wfProfileOut( __METHOD__ );
 896
 897                 return $out;
 898         }
 899
 900         /**
 901          * Helper function for parse() that transforms wiki markup into
 902          * HTML. Only called for $mOutputType == self::OT_HTML.
 903          *
 904          * @private
 905          */
 906         function internalParse( $text, $isMain = true, $frame=false ) {
 907                 wfProfileIn( __METHOD__ );
 908
 909                 $origText = $text;
 910
 911                 # Hook to suspend the parser in this state
 912                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 913                         wfProfileOut( __METHOD__ );
 914                         return $text ;
 915                 }
 916
 917                 // if $frame is provided, then use $frame for replacing any variables
 918                 if ($frame) {
 919                         // use frame depth to infer how include/noinclude tags should be handled
 920                         // depth=0 means this is the top-level document; otherwise it's an included document
 921                         if( !$frame->depth )
 922                                 $flag = 0;
 923                         else
 924                                 $flag = Parser::PTD_FOR_INCLUSION;
 925                         $dom = $this->preprocessToDom( $text, $flag );
 926                         $text = $frame->expand( $dom );
 927                 }
 928                 // if $frame is not provided, then use old-style replaceVariables
 929                 else {
 930                         $text = $this->replaceVariables( $text );
 931                 }
 932
 933                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 934                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 935
 936                 // Tables need to come after variable replacement for things to work
 937                 // properly; putting them before other transformations should keep
 938                 // exciting things like link expansions from showing up in surprising
 939                 // places.
 940                 $text = $this->doTableStuff( $text );
 941
 942                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 943
 944                 $text = $this->doDoubleUnderscore( $text );
 945
 946                 $text = $this->doHeadings( $text );
 947                 if( $this->mOptions->getUseDynamicDates() ) {
 948                         $df = DateFormatter::getInstance();
 949                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 950                 }
 951                 $text = $this->doAllQuotes( $text );
 952                 $text = $this->replaceInternalLinks( $text );
 953                 $text = $this->replaceExternalLinks( $text );
 954
 955                 # replaceInternalLinks may sometimes leave behind
 956                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 957                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 958
 959                 $text = $this->doMagicLinks( $text );
 960                 $text = $this->formatHeadings( $text, $origText, $isMain );
 961
 962                 wfProfileOut( __METHOD__ );
 963                 return $text;
 964         }
 965
 966         /**
 967          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 968          * magic external links.
 969          *
 970          * DML
 971          * @private
 972          */
 973         function doMagicLinks( $text ) {
 974                 wfProfileIn( __METHOD__ );
 975                 $prots = $this->mUrlProtocols;
 976                 $urlChar = self::EXT_LINK_URL_CLASS;
 977                 $text = preg_replace_callback(
 978                         '!(?:                           # Start cases
 979                                 (<a.*?</a>) |               # m[1]: Skip link text
 980                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 981                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 982                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 983                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 984                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 985                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 986                                     [0-9Xx]                 # check digit
 987                                     \b)
 988                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 989                 wfProfileOut( __METHOD__ );
 990                 return $text;
 991         }
 992
 993         function magicLinkCallback( $m ) {
 994                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 995                         # Skip anchor
 996                         return $m[0];
 997                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 998                         # Skip HTML element
 999                         return $m[0];
1000                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1001                         # Free external link
1002                         return $this->makeFreeExternalLink( $m[0] );
1003                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1004                         # RFC or PMID
1005                         $CssClass = '';
1006                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1007                                 $keyword = 'RFC';
1008                                 $urlmsg = 'rfcurl';
1009                                 $CssClass = 'mw-magiclink-rfc';
1010                                 $id = $m[4];
1011                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1012                                 $keyword = 'PMID';
1013                                 $urlmsg = 'pubmedurl';
1014                                 $CssClass = 'mw-magiclink-pmid';
1015                                 $id = $m[4];
1016                         } else {
1017                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1018                                         substr($m[0], 0, 20 ) . '"' );
1019                         }
1020                         $url = wfMsg( $urlmsg, $id);
1021                         $sk = $this->mOptions->getSkin();
1022                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1023                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1024                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1025                         # ISBN
1026                         $isbn = $m[5];
1027                         $num = strtr( $isbn, array(
1028                                 '-' => '',
1029                                 ' ' => '',
1030                                 'x' => 'X',
1031                         ));
1032                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1033                         return'<a href="' .
1034                                 $titleObj->escapeLocalUrl() .
1035                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1036                 } else {
1037                         return $m[0];
1038                 }
1039         }
1040
1041         /**
1042          * Make a free external link, given a user-supplied URL
1043          * @return HTML
1044          * @private
1045          */
1046         function makeFreeExternalLink( $url ) {
1047                 global $wgContLang;
1048                 wfProfileIn( __METHOD__ );
1049
1050                 $sk = $this->mOptions->getSkin();
1051                 $trail = '';
1052
1053                 # The characters '<' and '>' (which were escaped by
1054                 # removeHTMLtags()) should not be included in
1055                 # URLs, per RFC 2396.
1056                 $m2 = array();
1057                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1058                         $trail = substr($url, $m2[0][1]) . $trail;
1059                         $url = substr($url, 0, $m2[0][1]);
1060                 }
1061
1062                 # Move trailing punctuation to $trail
1063                 $sep = ',;\.:!?';
1064                 # If there is no left bracket, then consider right brackets fair game too
1065                 if ( strpos( $url, '(' ) === false ) {
1066                         $sep .= ')';
1067                 }
1068
1069                 $numSepChars = strspn( strrev( $url ), $sep );
1070                 if ( $numSepChars ) {
1071                         $trail = substr( $url, -$numSepChars ) . $trail;
1072                         $url = substr( $url, 0, -$numSepChars );
1073                 }
1074
1075                 $url = Sanitizer::cleanUrl( $url );
1076
1077                 # Is this an external image?
1078                 $text = $this->maybeMakeExternalImage( $url );
1079                 if ( $text === false ) {
1080                         # Not an image, make a link
1081                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1082                                 $this->getExternalLinkAttribs( $url ) );
1083                         # Register it in the output object...
1084                         # Replace unnecessary URL escape codes with their equivalent characters
1085                         $pasteurized = self::replaceUnusualEscapes( $url );
1086                         $this->mOutput->addExternalLink( $pasteurized );
1087                 }
1088                 wfProfileOut( __METHOD__ );
1089                 return $text . $trail;
1090         }
1091
1092
1093         /**
1094          * Parse headers and return html
1095          *
1096          * @private
1097          */
1098         function doHeadings( $text ) {
1099                 wfProfileIn( __METHOD__ );
1100                 for ( $i = 6; $i >= 1; --$i ) {
1101                         $h = str_repeat( '=', $i );
1102                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1103                           "<h$i>\\1</h$i>", $text );
1104                 }
1105                 wfProfileOut( __METHOD__ );
1106                 return $text;
1107         }
1108
1109         /**
1110          * Replace single quotes with HTML markup
1111          * @private
1112          * @return string the altered text
1113          */
1114         function doAllQuotes( $text ) {
1115                 wfProfileIn( __METHOD__ );
1116                 $outtext = '';
1117                 $lines = StringUtils::explode( "\n", $text );
1118                 foreach ( $lines as $line ) {
1119                         $outtext .= $this->doQuotes( $line ) . "\n";
1120                 }
1121                 $outtext = substr($outtext, 0,-1);
1122                 wfProfileOut( __METHOD__ );
1123                 return $outtext;
1124         }
1125
1126         /**
1127          * Helper function for doAllQuotes()
1128          */
1129         public function doQuotes( $text ) {
1130                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1131                 if ( count( $arr ) == 1 )
1132                         return $text;
1133                 else
1134                 {
1135                         # First, do some preliminary work. This may shift some apostrophes from
1136                         # being mark-up to being text. It also counts the number of occurrences
1137                         # of bold and italics mark-ups.
1138                         $i = 0;
1139                         $numbold = 0;
1140                         $numitalics = 0;
1141                         foreach ( $arr as $r )
1142                         {
1143                                 if ( ( $i % 2 ) == 1 )
1144                                 {
1145                                         # If there are ever four apostrophes, assume the first is supposed to
1146                                         # be text, and the remaining three constitute mark-up for bold text.
1147                                         if ( strlen( $arr[$i] ) == 4 )
1148                                         {
1149                                                 $arr[$i-1] .= "'";
1150                                                 $arr[$i] = "'''";
1151                                         }
1152                                         # If there are more than 5 apostrophes in a row, assume they're all
1153                                         # text except for the last 5.
1154                                         else if ( strlen( $arr[$i] ) > 5 )
1155                                         {
1156                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1157                                                 $arr[$i] = "'''''";
1158                                         }
1159                                         # Count the number of occurrences of bold and italics mark-ups.
1160                                         # We are not counting sequences of five apostrophes.
1161                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1162                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1163                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1164                                 }
1165                                 $i++;
1166                         }
1167
1168                         # If there is an odd number of both bold and italics, it is likely
1169                         # that one of the bold ones was meant to be an apostrophe followed
1170                         # by italics. Which one we cannot know for certain, but it is more
1171                         # likely to be one that has a single-letter word before it.
1172                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1173                         {
1174                                 $i = 0;
1175                                 $firstsingleletterword = -1;
1176                                 $firstmultiletterword = -1;
1177                                 $firstspace = -1;
1178                                 foreach ( $arr as $r )
1179                                 {
1180                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1181                                         {
1182                                                 $x1 = substr ($arr[$i-1], -1);
1183                                                 $x2 = substr ($arr[$i-1], -2, 1);
1184                                                 if ($x1 === ' ') {
1185                                                         if ($firstspace == -1) $firstspace = $i;
1186                                                 } else if ($x2 === ' ') {
1187                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1188                                                 } else {
1189                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1190                                                 }
1191                                         }
1192                                         $i++;
1193                                 }
1194
1195                                 # If there is a single-letter word, use it!
1196                                 if ($firstsingleletterword > -1)
1197                                 {
1198                                         $arr [ $firstsingleletterword ] = "''";
1199                                         $arr [ $firstsingleletterword-1 ] .= "'";
1200                                 }
1201                                 # If not, but there's a multi-letter word, use that one.
1202                                 else if ($firstmultiletterword > -1)
1203                                 {
1204                                         $arr [ $firstmultiletterword ] = "''";
1205                                         $arr [ $firstmultiletterword-1 ] .= "'";
1206                                 }
1207                                 # ... otherwise use the first one that has neither.
1208                                 # (notice that it is possible for all three to be -1 if, for example,
1209                                 # there is only one pentuple-apostrophe in the line)
1210                                 else if ($firstspace > -1)
1211                                 {
1212                                         $arr [ $firstspace ] = "''";
1213                                         $arr [ $firstspace-1 ] .= "'";
1214                                 }
1215                         }
1216
1217                         # Now let's actually convert our apostrophic mush to HTML!
1218                         $output = '';
1219                         $buffer = '';
1220                         $state = '';
1221                         $i = 0;
1222                         foreach ($arr as $r)
1223                         {
1224                                 if (($i % 2) == 0)
1225                                 {
1226                                         if ($state === 'both')
1227                                                 $buffer .= $r;
1228                                         else
1229                                                 $output .= $r;
1230                                 }
1231                                 else
1232                                 {
1233                                         if (strlen ($r) == 2)
1234                                         {
1235                                                 if ($state === 'i')
1236                                                 { $output .= '</i>'; $state = ''; }
1237                                                 else if ($state === 'bi')
1238                                                 { $output .= '</i>'; $state = 'b'; }
1239                                                 else if ($state === 'ib')
1240                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1241                                                 else if ($state === 'both')
1242                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1243                                                 else # $state can be 'b' or ''
1244                                                 { $output .= '<i>'; $state .= 'i'; }
1245                                         }
1246                                         else if (strlen ($r) == 3)
1247                                         {
1248                                                 if ($state === 'b')
1249                                                 { $output .= '</b>'; $state = ''; }
1250                                                 else if ($state === 'bi')
1251                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1252                                                 else if ($state === 'ib')
1253                                                 { $output .= '</b>'; $state = 'i'; }
1254                                                 else if ($state === 'both')
1255                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1256                                                 else # $state can be 'i' or ''
1257                                                 { $output .= '<b>'; $state .= 'b'; }
1258                                         }
1259                                         else if (strlen ($r) == 5)
1260                                         {
1261                                                 if ($state === 'b')
1262                                                 { $output .= '</b><i>'; $state = 'i'; }
1263                                                 else if ($state === 'i')
1264                                                 { $output .= '</i><b>'; $state = 'b'; }
1265                                                 else if ($state === 'bi')
1266                                                 { $output .= '</i></b>'; $state = ''; }
1267                                                 else if ($state === 'ib')
1268                                                 { $output .= '</b></i>'; $state = ''; }
1269                                                 else if ($state === 'both')
1270                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1271                                                 else # ($state == '')
1272                                                 { $buffer = ''; $state = 'both'; }
1273                                         }
1274                                 }
1275                                 $i++;
1276                         }
1277                         # Now close all remaining tags.  Notice that the order is important.
1278                         if ($state === 'b' || $state === 'ib')
1279                                 $output .= '</b>';
1280                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1281                                 $output .= '</i>';
1282                         if ($state === 'bi')
1283                                 $output .= '</b>';
1284                         # There might be lonely ''''', so make sure we have a buffer
1285                         if ($state === 'both' && $buffer)
1286                                 $output .= '<b><i>'.$buffer.'</i></b>';
1287                         return $output;
1288                 }
1289         }
1290
1291         /**
1292          * Replace external links (REL)
1293          *
1294          * Note: this is all very hackish and the order of execution matters a lot.
1295          * Make sure to run maintenance/parserTests.php if you change this code.
1296          *
1297          * @private
1298          */
1299         function replaceExternalLinks( $text ) {
1300                 global $wgContLang;
1301                 wfProfileIn( __METHOD__ );
1302
1303                 $sk = $this->mOptions->getSkin();
1304
1305                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1306                 $s = array_shift( $bits );
1307
1308                 $i = 0;
1309                 while ( $i<count( $bits ) ) {
1310                         $url = $bits[$i++];
1311                         $protocol = $bits[$i++];
1312                         $text = $bits[$i++];
1313                         $trail = $bits[$i++];
1314
1315                         # The characters '<' and '>' (which were escaped by
1316                         # removeHTMLtags()) should not be included in
1317                         # URLs, per RFC 2396.
1318                         $m2 = array();
1319                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1320                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1321                                 $url = substr($url, 0, $m2[0][1]);
1322                         }
1323
1324                         # If the link text is an image URL, replace it with an <img> tag
1325                         # This happened by accident in the original parser, but some people used it extensively
1326                         $img = $this->maybeMakeExternalImage( $text );
1327                         if ( $img !== false ) {
1328                                 $text = $img;
1329                         }
1330
1331                         $dtrail = '';
1332
1333                         # Set linktype for CSS - if URL==text, link is essentially free
1334                         $linktype = ($text === $url) ? 'free' : 'text';
1335
1336                         # No link text, e.g. [http://domain.tld/some.link]
1337                         if ( $text == '' ) {
1338                                 # Autonumber if allowed. See bug #5918
1339                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1340                                         $langObj = $this->getFunctionLang();
1341                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1342                                         $linktype = 'autonumber';
1343                                 } else {
1344                                         # Otherwise just use the URL
1345                                         $text = htmlspecialchars( $url );
1346                                         $linktype = 'free';
1347                                 }
1348                         } else {
1349                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1350                                 # Check for trail
1351                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1352                         }
1353
1354                         $text = $wgContLang->markNoConversion($text);
1355
1356                         $url = Sanitizer::cleanUrl( $url );
1357
1358                         # Use the encoded URL
1359                         # This means that users can paste URLs directly into the text
1360                         # Funny characters like &ouml; aren't valid in URLs anyway
1361                         # This was changed in August 2004
1362                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1363                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1364
1365                         # Register link in the output object.
1366                         # Replace unnecessary URL escape codes with the referenced character
1367                         # This prevents spammers from hiding links from the filters
1368                         $pasteurized = self::replaceUnusualEscapes( $url );
1369                         $this->mOutput->addExternalLink( $pasteurized );
1370                 }
1371
1372                 wfProfileOut( __METHOD__ );
1373                 return $s;
1374         }
1375
1376         /**
1377          * Get an associative array of additional HTML attributes appropriate for a
1378          * particular external link.  This currently may include rel => nofollow
1379          * (depending on configuration, namespace, and the URL's domain) and/or a
1380          * target attribute (depending on configuration).
1381          *
1382          * @param string $url Optional URL, to extract the domain from for rel =>
1383          *   nofollow if appropriate
1384          * @return array Associative array of HTML attributes
1385          */
1386         function getExternalLinkAttribs( $url = false ) {
1387                 $attribs = array();
1388                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1389                 $ns = $this->mTitle->getNamespace();
1390                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1391                         $attribs['rel'] = 'nofollow';
1392
1393                         global $wgNoFollowDomainExceptions;
1394                         if ( $wgNoFollowDomainExceptions ) {
1395                                 $bits = wfParseUrl( $url );
1396                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1397                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1398                                                 if( substr( $bits['host'], -strlen( $domain ) )
1399                                                 == $domain ) {
1400                                                         unset( $attribs['rel'] );
1401                                                         break;
1402                                                 }
1403                                         }
1404                                 }
1405                         }
1406                 }
1407                 if ( $this->mOptions->getExternalLinkTarget() ) {
1408                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1409                 }
1410                 return $attribs;
1411         }
1412
1413
1414         /**
1415          * Replace unusual URL escape codes with their equivalent characters
1416          * @param string
1417          * @return string
1418          * @static
1419          * @todo  This can merge genuinely required bits in the path or query string,
1420          *        breaking legit URLs. A proper fix would treat the various parts of
1421          *        the URL differently; as a workaround, just use the output for
1422          *        statistical records, not for actual linking/output.
1423          */
1424         static function replaceUnusualEscapes( $url ) {
1425                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1426                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1427         }
1428
1429         /**
1430          * Callback function used in replaceUnusualEscapes().
1431          * Replaces unusual URL escape codes with their equivalent character
1432          * @static
1433          * @private
1434          */
1435         private static function replaceUnusualEscapesCallback( $matches ) {
1436                 $char = urldecode( $matches[0] );
1437                 $ord = ord( $char );
1438                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1439                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1440                         // No, shouldn't be escaped
1441                         return $char;
1442                 } else {
1443                         // Yes, leave it escaped
1444                         return $matches[0];
1445                 }
1446         }
1447
1448         /**
1449          * make an image if it's allowed, either through the global
1450          * option, through the exception, or through the on-wiki whitelist
1451          * @private
1452          */
1453         function maybeMakeExternalImage( $url ) {
1454                 $sk = $this->mOptions->getSkin();
1455                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1456                 $imagesexception = !empty($imagesfrom);
1457                 $text = false;
1458                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1459                 if( $imagesexception && is_array( $imagesfrom ) ) {
1460                         $imagematch = false;
1461                         foreach( $imagesfrom as $match ) {
1462                                 if( strpos( $url, $match ) === 0 ) {
1463                                         $imagematch = true;
1464                                         break;
1465                                 }
1466                         }
1467                 } elseif( $imagesexception ) {
1468                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1469                 } else {
1470                         $imagematch = false;
1471                 }
1472                 if ( $this->mOptions->getAllowExternalImages()
1473                      || ( $imagesexception && $imagematch ) ) {
1474                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1475                                 # Image found
1476                                 $text = $sk->makeExternalImage( $url );
1477                         }
1478                 }
1479                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1480                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1481                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1482                         foreach( $whitelist as $entry ) {
1483                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1484                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1485                                         continue;
1486                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1487                                         # Image matches a whitelist entry
1488                                         $text = $sk->makeExternalImage( $url );
1489                                         break;
1490                                 }
1491                         }
1492                 }
1493                 return $text;
1494         }
1495
1496         /**
1497          * Process [[ ]] wikilinks
1498          * @return processed text
1499          *
1500          * @private
1501          */
1502         function replaceInternalLinks( $s ) {
1503                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1504                 return $s;
1505         }
1506
1507         /**
1508          * Process [[ ]] wikilinks (RIL)
1509          * @return LinkHolderArray
1510          *
1511          * @private
1512          */
1513         function replaceInternalLinks2( &$s ) {
1514                 global $wgContLang;
1515
1516                 wfProfileIn( __METHOD__ );
1517
1518                 wfProfileIn( __METHOD__.'-setup' );
1519                 static $tc = FALSE, $e1, $e1_img;
1520                 # the % is needed to support urlencoded titles as well
1521                 if ( !$tc ) {
1522                         $tc = Title::legalChars() . '#%';
1523                         # Match a link having the form [[namespace:link|alternate]]trail
1524                         $e1 = "/^([{$tc}]*)(\\|.*?)?]](.*)\$/sD";
1525                         # Match cases where there is no "]]", which might still be images
1526                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1527                 }
1528
1529                 $sk = $this->mOptions->getSkin();
1530                 $holders = new LinkHolderArray( $this );
1531
1532                 #split the entire text string on occurences of [[
1533                 $a = StringUtils::explode( '[[', ' ' . $s );
1534                 #get the first element (all text up to first [[), and remove the space we added
1535                 $s = $a->current();
1536                 $a->next();
1537                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1538                 $s = substr( $s, 1 );
1539
1540                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1541                 $e2 = null;
1542                 if ( $useLinkPrefixExtension ) {
1543                         # Match the end of a line for a word that's not followed by whitespace,
1544                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1545                         $e2 = wfMsgForContent( 'linkprefix' );
1546                 }
1547
1548                 if( is_null( $this->mTitle ) ) {
1549                         wfProfileOut( __METHOD__.'-setup' );
1550                         wfProfileOut( __METHOD__ );
1551                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1552                 }
1553                 $nottalk = !$this->mTitle->isTalkPage();
1554
1555                 if ( $useLinkPrefixExtension ) {
1556                         $m = array();
1557                         if ( preg_match( $e2, $s, $m ) ) {
1558                                 $first_prefix = $m[2];
1559                         } else {
1560                                 $first_prefix = false;
1561                         }
1562                 } else {
1563                         $prefix = '';
1564                 }
1565
1566                 if($wgContLang->hasVariants()) {
1567                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1568                 } else {
1569                         $selflink = array($this->mTitle->getPrefixedText());
1570                 }
1571                 $useSubpages = $this->areSubpagesAllowed();
1572                 wfProfileOut( __METHOD__.'-setup' );
1573
1574                 # Loop for each link
1575                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1576                         # Check for excessive memory usage
1577                         if ( $holders->isBig() ) {
1578                                 # Too big
1579                                 # Do the existence check, replace the link holders and clear the array
1580                                 $holders->replace( $s );
1581                                 $holders->clear();
1582                         }
1583
1584                         if ( $useLinkPrefixExtension ) {
1585                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1586                                 if ( preg_match( $e2, $s, $m ) ) {
1587                                         $prefix = $m[2];
1588                                         $s = $m[1];
1589                                 } else {
1590                                         $prefix='';
1591                                 }
1592                                 # first link
1593                                 if($first_prefix) {
1594                                         $prefix = $first_prefix;
1595                                         $first_prefix = false;
1596                                 }
1597                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1598                         }
1599
1600                         $might_be_img = false;
1601
1602                         wfProfileIn( __METHOD__."-e1" );
1603                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1604
1605                                 if( $m[2] === '' ) {
1606                                         $text = '';
1607                                 } elseif( $m[2] === '|' ) {
1608                                         $text = $this->getPipeTrickText( $m[1] );
1609                                 } else {
1610                                         $text = substr( $m[2], 1 );
1611                                 }
1612
1613                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1614                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1615                                 # the real problem is with the $e1 regex
1616                                 # See bug 1300.
1617                                 #
1618                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1619                                 # and no image is in sight. See bug 2095.
1620                                 #
1621                                 if( $text !== '' &&
1622                                         substr( $m[3], 0, 1 ) === ']' &&
1623                                         strpos($text, '[') !== false
1624                                 )
1625                                 {
1626                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1627                                         $m[3] = substr( $m[3], 1 );
1628                                 }
1629
1630                                 # Handle pipe-trick for [[|<blah>]]
1631                                 $lnk = $m[1] === '' ? $this->getPipeTrickLink( $text ) : $m[1];
1632                                 # fix up urlencoded title texts
1633                                 if( strpos( $lnk, '%' ) !== false ) {
1634                                         # Should anchors '#' also be rejected?
1635                                         $lnk = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($lnk) );
1636                                 }
1637
1638                                 $trail = $m[3];
1639                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1640                                 $might_be_img = true;
1641                                 $text = $m[2];
1642                                 $lnk = strpos( $m[1], '%' ) === false ? $m[1] : urldecode( $m[1] );
1643                                 $trail = "";
1644                         } else { # Invalid form; output directly
1645                                 $s .= $prefix . '[[' . $line ;
1646                                 wfProfileOut( __METHOD__."-e1" );
1647                                 continue;
1648                         }
1649                         wfProfileOut( __METHOD__."-e1" );
1650                         wfProfileIn( __METHOD__."-misc" );
1651
1652                         # Don't allow internal links to pages containing
1653                         # PROTO: where PROTO is a valid URL protocol; these
1654                         # should be external links.
1655                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $lnk ) ) {
1656                                 $s .= $prefix . '[[' . $line ;
1657                                 wfProfileOut( __METHOD__."-misc" );
1658                                 continue;
1659                         }
1660
1661                         # Make subpage if necessary
1662                         if ( $useSubpages ) {
1663                                 $link = $this->maybeDoSubpageLink( $lnk, $text );
1664                         } else {
1665                                 $link = $lnk;
1666                         }
1667
1668                         $noforce = (substr( $lnk, 0, 1 ) !== ':');
1669                         if (!$noforce) {
1670                                 # Strip off leading ':'
1671                                 $link = substr( $link, 1 );
1672                         }
1673
1674                         wfProfileOut( __METHOD__."-misc" );
1675                         wfProfileIn( __METHOD__."-title" );
1676                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1677                         if ( $nt === null ) {
1678                                 $s .= $prefix . '[[' . $line;
1679                                 wfProfileOut( __METHOD__."-title" );
1680                                 continue;
1681                         }
1682
1683                         $ns = $nt->getNamespace();
1684                         $iw = $nt->getInterWiki();
1685                         wfProfileOut( __METHOD__."-title" );
1686
1687                         if ( $might_be_img ) { # if this is actually an invalid link
1688                                 wfProfileIn( __METHOD__."-might_be_img" );
1689                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1690                                         $found = false;
1691                                         while ( true ) {
1692                                                 #look at the next 'line' to see if we can close it there
1693                                                 $a->next();
1694                                                 $next_line = $a->current();
1695                                                 if ( $next_line === false || $next_line === null ) {
1696                                                         break;
1697                                                 }
1698                                                 $m = explode( ']]', $next_line, 3 );
1699                                                 if ( count( $m ) == 3 ) {
1700                                                         # the first ]] closes the inner link, the second the image
1701                                                         $found = true;
1702                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1703                                                         $trail = $m[2];
1704                                                         break;
1705                                                 } elseif ( count( $m ) == 2 ) {
1706                                                         #if there's exactly one ]] that's fine, we'll keep looking
1707                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1708                                                 } else {
1709                                                         #if $next_line is invalid too, we need look no further
1710                                                         $text .= '[[' . $next_line;
1711                                                         break;
1712                                                 }
1713                                         }
1714                                         if ( !$found ) {
1715                                                 # we couldn't find the end of this imageLink, so output it raw
1716                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1717                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1718                                                 $s .= "{$prefix}[[$link|$text";
1719                                                 # note: no $trail, because without an end, there *is* no trail
1720                                                 wfProfileOut( __METHOD__."-might_be_img" );
1721                                                 continue;
1722                                         }
1723                                 } else { #it's not an image, so output it raw
1724                                         $s .= "{$prefix}[[$link|$text";
1725                                         # note: no $trail, because without an end, there *is* no trail
1726                                         wfProfileOut( __METHOD__."-might_be_img" );
1727                                         continue;
1728                                 }
1729                                 wfProfileOut( __METHOD__."-might_be_img" );
1730                         }
1731
1732                         $wasblank = ( $text  == '' );
1733                         if ( $wasblank ) $text = $link;
1734
1735                         # Link not escaped by : , create the various objects
1736                         if ( $noforce ) {
1737
1738                                 # Interwikis
1739                                 wfProfileIn( __METHOD__."-interwiki" );
1740                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1741                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1742                                         $s = rtrim($s . $prefix);
1743                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1744                                         wfProfileOut( __METHOD__."-interwiki" );
1745                                         continue;
1746                                 }
1747                                 wfProfileOut( __METHOD__."-interwiki" );
1748
1749                                 if ( $ns == NS_FILE ) {
1750                                         wfProfileIn( __METHOD__."-image" );
1751                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1752                                                 if ( $wasblank ) {
1753                                                         # if no parameters were passed, $text
1754                                                         # becomes something like "File:Foo.png",
1755                                                         # which we don't want to pass on to the
1756                                                         # image generator
1757                                                         $text = '';
1758                                                 } else {
1759                                                         # recursively parse links inside the image caption
1760                                                         # actually, this will parse them in any other parameters, too,
1761                                                         # but it might be hard to fix that, and it doesn't matter ATM
1762                                                         $text = $this->replaceExternalLinks($text);
1763                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1764                                                 }
1765                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1766                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1767                                         } else {
1768                                                 $s .= $prefix . $trail;
1769                                         }
1770                                         $this->mOutput->addImage( $nt->getDBkey() );
1771                                         wfProfileOut( __METHOD__."-image" );
1772                                         continue;
1773
1774                                 }
1775
1776                                 if ( $ns == NS_CATEGORY ) {
1777                                         wfProfileIn( __METHOD__."-category" );
1778                                         $s = rtrim($s . "\n"); # bug 87
1779
1780                                         if ( $wasblank ) {
1781                                                 $sortkey = $this->getDefaultSort();
1782                                         } else {
1783                                                 $sortkey = $text;
1784                                         }
1785                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1786                                         $sortkey = str_replace( "\n", '', $sortkey );
1787                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1788                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1789
1790                                         /**
1791                                          * Strip the whitespace Category links produce, see bug 87
1792                                          * @todo We might want to use trim($tmp, "\n") here.
1793                                          */
1794                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1795
1796                                         wfProfileOut( __METHOD__."-category" );
1797                                         continue;
1798                                 }
1799                         }
1800
1801                         # Self-link checking
1802                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1803                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1804                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1805                                         continue;
1806                                 }
1807                         }
1808
1809                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1810                         # FIXME: Should do batch file existence checks, see comment below
1811                         if( $ns == NS_MEDIA ) {
1812                                 wfProfileIn( __METHOD__."-media" );
1813                                 # Give extensions a chance to select the file revision for us
1814                                 $skip = $time = false;
1815                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1816                                 if ( $skip ) {
1817                                         $link = $sk->link( $nt );
1818                                 } else {
1819                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1820                                 }
1821                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1822                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1823                                 $this->mOutput->addImage( $nt->getDBkey() );
1824                                 wfProfileOut( __METHOD__."-media" );
1825                                 continue;
1826                         }
1827
1828                         wfProfileIn( __METHOD__."-always_known" );
1829                         # Some titles, such as valid special pages or files in foreign repos, should
1830                         # be shown as bluelinks even though they're not included in the page table
1831                         #
1832                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1833                         # batch file existence checks for NS_FILE and NS_MEDIA
1834                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1835                                 $this->mOutput->addLink( $nt );
1836                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1837                         } else {
1838                                 # Links will be added to the output link list after checking
1839                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1840                         }
1841                         wfProfileOut( __METHOD__."-always_known" );
1842                 }
1843                 wfProfileOut( __METHOD__ );
1844                 return $holders;
1845         }
1846
1847         /**
1848          * Make a link placeholder. The text returned can be later resolved to a real link with
1849          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1850          * parsing of interwiki links, and secondly to allow all existence checks and
1851          * article length checks (for stub links) to be bundled into a single query.
1852          *
1853          * @deprecated
1854          */
1855         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1856                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1857         }
1858
1859         /**
1860          * Render a forced-blue link inline; protect against double expansion of
1861          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1862          * Since this little disaster has to split off the trail text to avoid
1863          * breaking URLs in the following text without breaking trails on the
1864          * wiki links, it's been made into a horrible function.
1865          *
1866          * @param Title $nt
1867          * @param string $text
1868          * @param string $query
1869          * @param string $trail
1870          * @param string $prefix
1871          * @return string HTML-wikitext mix oh yuck
1872          */
1873         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1874                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1875                 $sk = $this->mOptions->getSkin();
1876                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1877                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1878                 return $this->armorLinks( $link ) . $trail;
1879         }
1880
1881         /**
1882          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1883          * going to go through further parsing steps before inline URL expansion.
1884          *
1885          * Not needed quite as much as it used to be since free links are a bit
1886          * more sensible these days. But bracketed links are still an issue.
1887          *
1888          * @param string more-or-less HTML
1889          * @return string less-or-more HTML with NOPARSE bits
1890          */
1891         function armorLinks( $text ) {
1892                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1893                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1894         }
1895
1896         /**
1897          * Return true if subpage links should be expanded on this page.
1898          * @return bool
1899          */
1900         function areSubpagesAllowed() {
1901                 # Some namespaces don't allow subpages
1902                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1903         }
1904
1905         /**
1906          * Handle link to subpage if necessary
1907          * @param string $target the source of the link
1908          * @param string &$text the link text, modified as necessary
1909          * @return string the full name of the link
1910          * @private
1911          */
1912         function maybeDoSubpageLink($target, &$text) {
1913                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1914         }
1915
1916         /**
1917          * From the [[title|]] return link-text as though the used typed [[title|link-text]]
1918          * @param string $link from [[$link|]]
1919          * @return string $text for [[$link|$text]]
1920          */
1921         function getPipeTrickText( $link ) {
1922                 return Linker::getPipeTrickText( $link );
1923         }
1924
1925         /**
1926          * From the [[|link-text]] return the title as though the user typed [[title|link-text]]
1927          * @param string $text from [[|$text]]
1928          * @param Title $title to resolve the link against
1929          * @return string $link for [[$link|$text]]
1930          */
1931         function getPipeTrickLink( $text ) {
1932                 return Linker::getPipeTrickLink( $text, $this->mTitle );
1933         }
1934
1935         /**#@+
1936          * Used by doBlockLevels()
1937          * @private
1938          */
1939         /* private */ function closeParagraph() {
1940                 $result = '';
1941                 if ( $this->mLastSection != '' ) {
1942                         $result = '</' . $this->mLastSection  . ">\n";
1943                 }
1944                 $this->mInPre = false;
1945                 $this->mLastSection = '';
1946                 return $result;
1947         }
1948         # getCommon() returns the length of the longest common substring
1949         # of both arguments, starting at the beginning of both.
1950         #
1951         /* private */ function getCommon( $st1, $st2 ) {
1952                 $fl = strlen( $st1 );
1953                 $shorter = strlen( $st2 );
1954                 if ( $fl < $shorter ) { $shorter = $fl; }
1955
1956                 for ( $i = 0; $i < $shorter; ++$i ) {
1957                         if ( $st1{$i} != $st2{$i} ) { break; }
1958                 }
1959                 return $i;
1960         }
1961         # These next three functions open, continue, and close the list
1962         # element appropriate to the prefix character passed into them.
1963         #
1964         /* private */ function openList( $char ) {
1965                 $result = $this->closeParagraph();
1966
1967                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1968                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1969                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1970                 elseif ( ';' === $char ) {
1971                         $result .= '<dl><dt>';
1972                         $this->mDTopen = true;
1973                 }
1974                 else { $result = '<!-- ERR 1 -->'; }
1975
1976                 return $result;
1977         }
1978
1979         /* private */ function nextItem( $char ) {
1980                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1981                 elseif ( ':' === $char || ';' === $char ) {
1982                         $close = '</dd>';
1983                         if ( $this->mDTopen ) { $close = '</dt>'; }
1984                         if ( ';' === $char ) {
1985                                 $this->mDTopen = true;
1986                                 return $close . '<dt>';
1987                         } else {
1988                                 $this->mDTopen = false;
1989                                 return $close . '<dd>';
1990                         }
1991                 }
1992                 return '<!-- ERR 2 -->';
1993         }
1994
1995         /* private */ function closeList( $char ) {
1996                 if ( '*' === $char ) { $text = '</li></ul>'; }
1997                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
1998                 elseif ( ':' === $char ) {
1999                         if ( $this->mDTopen ) {
2000                                 $this->mDTopen = false;
2001                                 $text = '</dt></dl>';
2002                         } else {
2003                                 $text = '</dd></dl>';
2004                         }
2005                 }
2006                 else {  return '<!-- ERR 3 -->'; }
2007                 return $text."\n";
2008         }
2009         /**#@-*/
2010
2011         /**
2012          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2013          *
2014          * @param $linestart bool whether or not this is at the start of a line.
2015          * @private
2016          * @return string the lists rendered as HTML
2017          */
2018         function doBlockLevels( $text, $linestart ) {
2019                 wfProfileIn( __METHOD__ );
2020
2021                 # Parsing through the text line by line.  The main thing
2022                 # happening here is handling of block-level elements p, pre,
2023                 # and making lists from lines starting with * # : etc.
2024                 #
2025                 $textLines = StringUtils::explode( "\n", $text );
2026
2027                 $lastPrefix = $output = '';
2028                 $this->mDTopen = $inBlockElem = false;
2029                 $prefixLength = 0;
2030                 $paragraphStack = false;
2031
2032                 foreach ( $textLines as $oLine ) {
2033                         # Fix up $linestart
2034                         if ( !$linestart ) {
2035                                 $output .= $oLine;
2036                                 $linestart = true;
2037                                 continue;
2038                         }
2039                         // * = ul
2040                         // # = ol
2041                         // ; = dt
2042                         // : = dd
2043
2044                         $lastPrefixLength = strlen( $lastPrefix );
2045                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2046                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2047                         // If not in a <pre> element, scan for and figure out what prefixes are there.
2048                         if ( !$this->mInPre ) {
2049                                 # Multiple prefixes may abut each other for nested lists.
2050                                 $prefixLength = strspn( $oLine, '*#:;' );
2051                                 $prefix = substr( $oLine, 0, $prefixLength );
2052
2053                                 # eh?
2054                                 // ; and : are both from definition-lists, so they're equivalent
2055                                 //  for the purposes of determining whether or not we need to open/close
2056                                 //  elements.
2057                                 $prefix2 = str_replace( ';', ':', $prefix );
2058                                 $t = substr( $oLine, $prefixLength );
2059                                 $this->mInPre = (bool)$preOpenMatch;
2060                         } else {
2061                                 # Don't interpret any other prefixes in preformatted text
2062                                 $prefixLength = 0;
2063                                 $prefix = $prefix2 = '';
2064                                 $t = $oLine;
2065                         }
2066
2067                         # List generation
2068                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2069                                 # Same as the last item, so no need to deal with nesting or opening stuff
2070                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2071                                 $paragraphStack = false;
2072
2073                                 if ( substr( $prefix, -1 ) === ';') {
2074                                         # The one nasty exception: definition lists work like this:
2075                                         # ; title : definition text
2076                                         # So we check for : in the remainder text to split up the
2077                                         # title and definition, without b0rking links.
2078                                         $term = $t2 = '';
2079                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2080                                                 $t = $t2;
2081                                                 $output .= $term . $this->nextItem( ':' );
2082                                         }
2083                                 }
2084                         } elseif( $prefixLength || $lastPrefixLength ) {
2085                                 // We need to open or close prefixes, or both.
2086
2087                                 # Either open or close a level...
2088                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2089                                 $paragraphStack = false;
2090
2091                                 // Close all the prefixes which aren't shared.
2092                                 while( $commonPrefixLength < $lastPrefixLength ) {
2093                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2094                                         --$lastPrefixLength;
2095                                 }
2096
2097                                 // Continue the current prefix if appropriate.
2098                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2099                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2100                                 }
2101
2102                                 // Open prefixes where appropriate.
2103                                 while ( $prefixLength > $commonPrefixLength ) {
2104                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2105                                         $output .= $this->openList( $char );
2106
2107                                         if ( ';' === $char ) {
2108                                                 # FIXME: This is dupe of code above
2109                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2110                                                         $t = $t2;
2111                                                         $output .= $term . $this->nextItem( ':' );
2112                                                 }
2113                                         }
2114                                         ++$commonPrefixLength;
2115                                 }
2116                                 $lastPrefix = $prefix2;
2117                         }
2118
2119                         // If we have no prefixes, go to paragraph mode.
2120                         if( 0 == $prefixLength ) {
2121                                 wfProfileIn( __METHOD__."-paragraph" );
2122                                 # No prefix (not in list)--go to paragraph mode
2123                                 // XXX: use a stack for nestable elements like span, table and div
2124                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2125                                 $closematch = preg_match(
2126                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2127                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2128                                 if ( $openmatch or $closematch ) {
2129                                         $paragraphStack = false;
2130                                         # TODO bug 5718: paragraph closed
2131                                         $output .= $this->closeParagraph();
2132                                         if ( $preOpenMatch and !$preCloseMatch ) {
2133                                                 $this->mInPre = true;
2134                                         }
2135                                         if ( $closematch ) {
2136                                                 $inBlockElem = false;
2137                                         } else {
2138                                                 $inBlockElem = true;
2139                                         }
2140                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2141                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2142                                                 // pre
2143                                                 if ($this->mLastSection !== 'pre') {
2144                                                         $paragraphStack = false;
2145                                                         $output .= $this->closeParagraph().'<pre>';
2146                                                         $this->mLastSection = 'pre';
2147                                                 }
2148                                                 $t = substr( $t, 1 );
2149                                         } else {
2150                                                 // paragraph
2151                                                 if ( trim($t) == '' ) {
2152                                                         if ( $paragraphStack ) {
2153                                                                 $output .= $paragraphStack.'<br />';
2154                                                                 $paragraphStack = false;
2155                                                                 $this->mLastSection = 'p';
2156                                                         } else {
2157                                                                 if ($this->mLastSection !== 'p' ) {
2158                                                                         $output .= $this->closeParagraph();
2159                                                                         $this->mLastSection = '';
2160                                                                         $paragraphStack = '<p>';
2161                                                                 } else {
2162                                                                         $paragraphStack = '</p><p>';
2163                                                                 }
2164                                                         }
2165                                                 } else {
2166                                                         if ( $paragraphStack ) {
2167                                                                 $output .= $paragraphStack;
2168                                                                 $paragraphStack = false;
2169                                                                 $this->mLastSection = 'p';
2170                                                         } else if ($this->mLastSection !== 'p') {
2171                                                                 $output .= $this->closeParagraph().'<p>';
2172                                                                 $this->mLastSection = 'p';
2173                                                         }
2174                                                 }
2175                                         }
2176                                 }
2177                                 wfProfileOut( __METHOD__."-paragraph" );
2178                         }
2179                         // somewhere above we forget to get out of pre block (bug 785)
2180                         if($preCloseMatch && $this->mInPre) {
2181                                 $this->mInPre = false;
2182                         }
2183                         if ($paragraphStack === false) {
2184                                 $output .= $t."\n";
2185                         }
2186                 }
2187                 while ( $prefixLength ) {
2188                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2189                         --$prefixLength;
2190                 }
2191                 if ( $this->mLastSection != '' ) {
2192                         $output .= '</' . $this->mLastSection . '>';
2193                         $this->mLastSection = '';
2194                 }
2195
2196                 wfProfileOut( __METHOD__ );
2197                 return $output;
2198         }
2199
2200         /**
2201          * Split up a string on ':', ignoring any occurences inside tags
2202          * to prevent illegal overlapping.
2203          * @param string $str the string to split
2204          * @param string &$before set to everything before the ':'
2205          * @param string &$after set to everything after the ':'
2206          * return string the position of the ':', or false if none found
2207          */
2208         function findColonNoLinks($str, &$before, &$after) {
2209                 wfProfileIn( __METHOD__ );
2210
2211                 $pos = strpos( $str, ':' );
2212                 if( $pos === false ) {
2213                         // Nothing to find!
2214                         wfProfileOut( __METHOD__ );
2215                         return false;
2216                 }
2217
2218                 $lt = strpos( $str, '<' );
2219                 if( $lt === false || $lt > $pos ) {
2220                         // Easy; no tag nesting to worry about
2221                         $before = substr( $str, 0, $pos );
2222                         $after = substr( $str, $pos+1 );
2223                         wfProfileOut( __METHOD__ );
2224                         return $pos;
2225                 }
2226
2227                 // Ugly state machine to walk through avoiding tags.
2228                 $state = self::COLON_STATE_TEXT;
2229                 $stack = 0;
2230                 $len = strlen( $str );
2231                 for( $i = 0; $i < $len; $i++ ) {
2232                         $c = $str{$i};
2233
2234                         switch( $state ) {
2235                         // (Using the number is a performance hack for common cases)
2236                         case 0: // self::COLON_STATE_TEXT:
2237                                 switch( $c ) {
2238                                 case "<":
2239                                         // Could be either a <start> tag or an </end> tag
2240                                         $state = self::COLON_STATE_TAGSTART;
2241                                         break;
2242                                 case ":":
2243                                         if( $stack == 0 ) {
2244                                                 // We found it!
2245                                                 $before = substr( $str, 0, $i );
2246                                                 $after = substr( $str, $i + 1 );
2247                                                 wfProfileOut( __METHOD__ );
2248                                                 return $i;
2249                                         }
2250                                         // Embedded in a tag; don't break it.
2251                                         break;
2252                                 default:
2253                                         // Skip ahead looking for something interesting
2254                                         $colon = strpos( $str, ':', $i );
2255                                         if( $colon === false ) {
2256                                                 // Nothing else interesting
2257                                                 wfProfileOut( __METHOD__ );
2258                                                 return false;
2259                                         }
2260                                         $lt = strpos( $str, '<', $i );
2261                                         if( $stack === 0 ) {
2262                                                 if( $lt === false || $colon < $lt ) {
2263                                                         // We found it!
2264                                                         $before = substr( $str, 0, $colon );
2265                                                         $after = substr( $str, $colon + 1 );
2266                                                         wfProfileOut( __METHOD__ );
2267                                                         return $i;
2268                                                 }
2269                                         }
2270                                         if( $lt === false ) {
2271                                                 // Nothing else interesting to find; abort!
2272                                                 // We're nested, but there's no close tags left. Abort!
2273                                                 break 2;
2274                                         }
2275                                         // Skip ahead to next tag start
2276                                         $i = $lt;
2277                                         $state = self::COLON_STATE_TAGSTART;
2278                                 }
2279                                 break;
2280                         case 1: // self::COLON_STATE_TAG:
2281                                 // In a <tag>
2282                                 switch( $c ) {
2283                                 case ">":
2284                                         $stack++;
2285                                         $state = self::COLON_STATE_TEXT;
2286                                         break;
2287                                 case "/":
2288                                         // Slash may be followed by >?
2289                                         $state = self::COLON_STATE_TAGSLASH;
2290                                         break;
2291                                 default:
2292                                         // ignore
2293                                 }
2294                                 break;
2295                         case 2: // self::COLON_STATE_TAGSTART:
2296                                 switch( $c ) {
2297                                 case "/":
2298                                         $state = self::COLON_STATE_CLOSETAG;
2299                                         break;
2300                                 case "!":
2301                                         $state = self::COLON_STATE_COMMENT;
2302                                         break;
2303                                 case ">":
2304                                         // Illegal early close? This shouldn't happen D:
2305                                         $state = self::COLON_STATE_TEXT;
2306                                         break;
2307                                 default:
2308                                         $state = self::COLON_STATE_TAG;
2309                                 }
2310                                 break;
2311                         case 3: // self::COLON_STATE_CLOSETAG:
2312                                 // In a </tag>
2313                                 if( $c === ">" ) {
2314                                         $stack--;
2315                                         if( $stack < 0 ) {
2316                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2317                                                 wfProfileOut( __METHOD__ );
2318                                                 return false;
2319                                         }
2320                                         $state = self::COLON_STATE_TEXT;
2321                                 }
2322                                 break;
2323                         case self::COLON_STATE_TAGSLASH:
2324                                 if( $c === ">" ) {
2325                                         // Yes, a self-closed tag <blah/>
2326                                         $state = self::COLON_STATE_TEXT;
2327                                 } else {
2328                                         // Probably we're jumping the gun, and this is an attribute
2329                                         $state = self::COLON_STATE_TAG;
2330                                 }
2331                                 break;
2332                         case 5: // self::COLON_STATE_COMMENT:
2333                                 if( $c === "-" ) {
2334                                         $state = self::COLON_STATE_COMMENTDASH;
2335                                 }
2336                                 break;
2337                         case self::COLON_STATE_COMMENTDASH:
2338                                 if( $c === "-" ) {
2339                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2340                                 } else {
2341                                         $state = self::COLON_STATE_COMMENT;
2342                                 }
2343                                 break;
2344                         case self::COLON_STATE_COMMENTDASHDASH:
2345                                 if( $c === ">" ) {
2346                                         $state = self::COLON_STATE_TEXT;
2347                                 } else {
2348                                         $state = self::COLON_STATE_COMMENT;
2349                                 }
2350                                 break;
2351                         default:
2352                                 throw new MWException( "State machine error in " . __METHOD__ );
2353                         }
2354                 }
2355                 if( $stack > 0 ) {
2356                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2357                         return false;
2358                 }
2359                 wfProfileOut( __METHOD__ );
2360                 return false;
2361         }
2362
2363         /**
2364          * Return value of a magic variable (like PAGENAME)
2365          *
2366          * @private
2367          */
2368         function getVariableValue( $index, $frame=false ) {
2369                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2370                 global $wgScriptPath, $wgStylePath;
2371
2372                 /**
2373                  * Some of these require message or data lookups and can be
2374                  * expensive to check many times.
2375                  */
2376                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2377                         if ( isset( $this->mVarCache[$index] ) ) {
2378                                 return $this->mVarCache[$index];
2379                         }
2380                 }
2381
2382                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2383                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2384
2385                 # Use the time zone
2386                 global $wgLocaltimezone;
2387                 if ( isset( $wgLocaltimezone ) ) {
2388                         $oldtz = date_default_timezone_get();
2389                         date_default_timezone_set( $wgLocaltimezone );
2390                 }
2391
2392                 $localTimestamp = date( 'YmdHis', $ts );
2393                 $localMonth = date( 'm', $ts );
2394                 $localMonth1 = date( 'n', $ts );
2395                 $localMonthName = date( 'n', $ts );
2396                 $localDay = date( 'j', $ts );
2397                 $localDay2 = date( 'd', $ts );
2398                 $localDayOfWeek = date( 'w', $ts );
2399                 $localWeek = date( 'W', $ts );
2400                 $localYear = date( 'Y', $ts );
2401                 $localHour = date( 'H', $ts );
2402                 if ( isset( $wgLocaltimezone ) ) {
2403                         date_default_timezone_set( $oldtz );
2404                 }
2405
2406                 switch ( $index ) {
2407                         case 'currentmonth':
2408                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2409                                 break;
2410                         case 'currentmonth1':
2411                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2412                                 break;
2413                         case 'currentmonthname':
2414                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2415                                 break;
2416                         case 'currentmonthnamegen':
2417                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2418                                 break;
2419                         case 'currentmonthabbrev':
2420                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2421                                 break;
2422                         case 'currentday':
2423                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2424                                 break;
2425                         case 'currentday2':
2426                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2427                                 break;
2428                         case 'localmonth':
2429                                 $value = $wgContLang->formatNum( $localMonth );
2430                                 break;
2431                         case 'localmonth1':
2432                                 $value = $wgContLang->formatNum( $localMonth1 );
2433                                 break;
2434                         case 'localmonthname':
2435                                 $value = $wgContLang->getMonthName( $localMonthName );
2436                                 break;
2437                         case 'localmonthnamegen':
2438                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2439                                 break;
2440                         case 'localmonthabbrev':
2441                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2442                                 break;
2443                         case 'localday':
2444                                 $value = $wgContLang->formatNum( $localDay );
2445                                 break;
2446                         case 'localday2':
2447                                 $value = $wgContLang->formatNum( $localDay2 );
2448                                 break;
2449                         case 'pagename':
2450                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2451                                 break;
2452                         case 'pagenamee':
2453                                 $value = $this->mTitle->getPartialURL();
2454                                 break;
2455                         case 'fullpagename':
2456                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2457                                 break;
2458                         case 'fullpagenamee':
2459                                 $value = $this->mTitle->getPrefixedURL();
2460                                 break;
2461                         case 'subpagename':
2462                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2463                                 break;
2464                         case 'subpagenamee':
2465                                 $value = $this->mTitle->getSubpageUrlForm();
2466                                 break;
2467                         case 'basepagename':
2468                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2469                                 break;
2470                         case 'basepagenamee':
2471                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2472                                 break;
2473                         case 'talkpagename':
2474                                 if( $this->mTitle->canTalk() ) {
2475                                         $talkPage = $this->mTitle->getTalkPage();
2476                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2477                                 } else {
2478                                         $value = '';
2479                                 }
2480                                 break;
2481                         case 'talkpagenamee':
2482                                 if( $this->mTitle->canTalk() ) {
2483                                         $talkPage = $this->mTitle->getTalkPage();
2484                                         $value = $talkPage->getPrefixedUrl();
2485                                 } else {
2486                                         $value = '';
2487                                 }
2488                                 break;
2489                         case 'subjectpagename':
2490                                 $subjPage = $this->mTitle->getSubjectPage();
2491                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2492                                 break;
2493                         case 'subjectpagenamee':
2494                                 $subjPage = $this->mTitle->getSubjectPage();
2495                                 $value = $subjPage->getPrefixedUrl();
2496                                 break;
2497                         case 'pipetrick':
2498                                 $text = $this->mTitle->getText();
2499                                 $value = $this->getPipeTrickText( $text );
2500                                 break;
2501                         case 'pipetricke':
2502                                 $text = $this->mTitle->getText();
2503                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->getPipeTrickText( $text ) ) );
2504                                 break;
2505                         case 'revisionid':
2506                                 // Let the edit saving system know we should parse the page
2507                                 // *after* a revision ID has been assigned.
2508                                 $this->mOutput->setFlag( 'vary-revision' );
2509                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2510                                 $value = $this->mRevisionId;
2511                                 break;
2512                         case 'revisionday':
2513                                 // Let the edit saving system know we should parse the page
2514                                 // *after* a revision ID has been assigned. This is for null edits.
2515                                 $this->mOutput->setFlag( 'vary-revision' );
2516                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2517                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2518                                 break;
2519                         case 'revisionday2':
2520                                 // Let the edit saving system know we should parse the page
2521                                 // *after* a revision ID has been assigned. This is for null edits.
2522                                 $this->mOutput->setFlag( 'vary-revision' );
2523                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2524                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2525                                 break;
2526                         case 'revisionmonth':
2527                                 // Let the edit saving system know we should parse the page
2528                                 // *after* a revision ID has been assigned. This is for null edits.
2529                                 $this->mOutput->setFlag( 'vary-revision' );
2530                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2531                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2532                                 break;
2533                         case 'revisionyear':
2534                                 // Let the edit saving system know we should parse the page
2535                                 // *after* a revision ID has been assigned. This is for null edits.
2536                                 $this->mOutput->setFlag( 'vary-revision' );
2537                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2538                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2539                                 break;
2540                         case 'revisiontimestamp':
2541                                 // Let the edit saving system know we should parse the page
2542                                 // *after* a revision ID has been assigned. This is for null edits.
2543                                 $this->mOutput->setFlag( 'vary-revision' );
2544                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2545                                 $value = $this->getRevisionTimestamp();
2546                                 break;
2547                         case 'revisionuser':
2548                                 // Let the edit saving system know we should parse the page
2549                                 // *after* a revision ID has been assigned. This is for null edits.
2550                                 $this->mOutput->setFlag( 'vary-revision' );
2551                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2552                                 $value = $this->getRevisionUser();
2553                                 break;
2554                         case 'namespace':
2555                                 $value = str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2556                                 break;
2557                         case 'namespacee':
2558                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2559                                 break;
2560                         case 'talkspace':
2561                                 $value = $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2562                                 break;
2563                         case 'talkspacee':
2564                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2565                                 break;
2566                         case 'subjectspace':
2567                                 $value = $this->mTitle->getSubjectNsText();
2568                                 break;
2569                         case 'subjectspacee':
2570                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2571                                 break;
2572                         case 'currentdayname':
2573                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2574                                 break;
2575                         case 'currentyear':
2576                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2577                                 break;
2578                         case 'currenttime':
2579                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2580                                 break;
2581                         case 'currenthour':
2582                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2583                                 break;
2584                         case 'currentweek':
2585                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2586                                 // int to remove the padding
2587                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2588                                 break;
2589                         case 'currentdow':
2590                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2591                                 break;
2592                         case 'localdayname':
2593                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2594                                 break;
2595                         case 'localyear':
2596                                 $value = $wgContLang->formatNum( $localYear, true );
2597                                 break;
2598                         case 'localtime':
2599                                 $value = $wgContLang->time( $localTimestamp, false, false );
2600                                 break;
2601                         case 'localhour':
2602                                 $value = $wgContLang->formatNum( $localHour, true );
2603                                 break;
2604                         case 'localweek':
2605                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2606                                 // int to remove the padding
2607                                 $value = $wgContLang->formatNum( (int)$localWeek );
2608                                 break;
2609                         case 'localdow':
2610                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2611                                 break;
2612                         case 'numberofarticles':
2613                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2614                                 break;
2615                         case 'numberoffiles':
2616                                 $value = $wgContLang->formatNum( SiteStats::images() );
2617                                 break;
2618                         case 'numberofusers':
2619                                 $value = $wgContLang->formatNum( SiteStats::users() );
2620                                 break;
2621                         case 'numberofactiveusers':
2622                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2623                                 break;
2624                         case 'numberofpages':
2625                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2626                                 break;
2627                         case 'numberofadmins':
2628                                 $value = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2629                                 break;
2630                         case 'numberofedits':
2631                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2632                                 break;
2633                         case 'numberofviews':
2634                                 $value = $wgContLang->formatNum( SiteStats::views() );
2635                                 break;
2636                         case 'currenttimestamp':
2637                                 $value = wfTimestamp( TS_MW, $ts );
2638                                 break;
2639                         case 'localtimestamp':
2640                                 $value = $localTimestamp;
2641                                 break;
2642                         case 'currentversion':
2643                                 $value = SpecialVersion::getVersion();
2644                                 break;
2645                         case 'sitename':
2646                                 return $wgSitename;
2647                         case 'server':
2648                                 return $wgServer;
2649                         case 'servername':
2650                                 return $wgServerName;
2651                         case 'scriptpath':
2652                                 return $wgScriptPath;
2653                         case 'stylepath':
2654                                 return $wgStylePath;
2655                         case 'directionmark':
2656                                 return $wgContLang->getDirMark();
2657                         case 'contentlanguage':
2658                                 global $wgContLanguageCode;
2659                                 return $wgContLanguageCode;
2660                         default:
2661                                 $ret = null;
2662                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) )
2663                                         return $ret;
2664                                 else
2665                                         return null;
2666                 }
2667
2668                 if ( $index )
2669                         $this->mVarCache[$index] = $value;
2670
2671                 return $value;
2672         }
2673
2674         /**
2675          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2676          *
2677          * @private
2678          */
2679         function initialiseVariables() {
2680                 wfProfileIn( __METHOD__ );
2681                 $variableIDs = MagicWord::getVariableIDs();
2682                 $substIDs = MagicWord::getSubstIDs();
2683
2684                 $this->mVariables = new MagicWordArray( $variableIDs );
2685                 $this->mSubsts = new MagicWordArray( $substIDs );
2686                 wfProfileOut( __METHOD__ );
2687         }
2688
2689         /**
2690          * Preprocess some wikitext and return the document tree.
2691          * This is the ghost of replace_variables().
2692          *
2693          * @param string $text The text to parse
2694          * @param integer flags Bitwise combination of:
2695          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2696          *                                     included. Default is to assume a direct page view.
2697          *
2698          * The generated DOM tree must depend only on the input text and the flags.
2699          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2700          *
2701          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2702          * change in the DOM tree for a given text, must be passed through the section identifier
2703          * in the section edit link and thus back to extractSections().
2704          *
2705          * The output of this function is currently only cached in process memory, but a persistent
2706          * cache may be implemented at a later date which takes further advantage of these strict
2707          * dependency requirements.
2708          *
2709          * @private
2710          */
2711         function preprocessToDom ( $text, $flags = 0 ) {
2712                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2713                 return $dom;
2714         }
2715
2716         /*
2717          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2718          */
2719         public static function splitWhitespace( $s ) {
2720                 $ltrimmed = ltrim( $s );
2721                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2722                 $trimmed = rtrim( $ltrimmed );
2723                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2724                 if ( $diff > 0 ) {
2725                         $w2 = substr( $ltrimmed, -$diff );
2726                 } else {
2727                         $w2 = '';
2728                 }
2729                 return array( $w1, $trimmed, $w2 );
2730         }
2731
2732         /**
2733          * Replace magic variables, templates, and template arguments
2734          * with the appropriate text. Templates are substituted recursively,
2735          * taking care to avoid infinite loops.
2736          *
2737          * Note that the substitution depends on value of $mOutputType:
2738          *  self::OT_WIKI: only {{subst:}} templates
2739          *  self::OT_PREPROCESS: templates but not extension tags
2740          *  self::OT_HTML: all templates and extension tags
2741          *
2742          * @param string $tex The text to transform
2743          * @param PPFrame $frame Object describing the arguments passed to the template.
2744          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2745          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2746          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2747          * @private
2748          */
2749         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2750                 # Is there any text? Also, Prevent too big inclusions!
2751                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2752                         return $text;
2753                 }
2754                 wfProfileIn( __METHOD__ );
2755
2756                 if ( $frame === false ) {
2757                         $frame = $this->getPreprocessor()->newFrame();
2758                 } elseif ( !( $frame instanceof PPFrame ) ) {
2759                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2760                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2761                 }
2762
2763                 $dom = $this->preprocessToDom( $text );
2764                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2765                 $text = $frame->expand( $dom, $flags );
2766
2767                 wfProfileOut( __METHOD__ );
2768                 return $text;
2769         }
2770
2771         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2772         static function createAssocArgs( $args ) {
2773                 $assocArgs = array();
2774                 $index = 1;
2775                 foreach( $args as $arg ) {
2776                         $eqpos = strpos( $arg, '=' );
2777                         if ( $eqpos === false ) {
2778                                 $assocArgs[$index++] = $arg;
2779                         } else {
2780                                 $name = trim( substr( $arg, 0, $eqpos ) );
2781                                 $value = trim( substr( $arg, $eqpos+1 ) );
2782                                 if ( $value === false ) {
2783                                         $value = '';
2784                                 }
2785                                 if ( $name !== false ) {
2786                                         $assocArgs[$name] = $value;
2787                                 }
2788                         }
2789                 }
2790
2791                 return $assocArgs;
2792         }
2793
2794         /**
2795          * Warn the user when a parser limitation is reached
2796          * Will warn at most once the user per limitation type
2797          *
2798          * @param string $limitationType, should be one of:
2799          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2800          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2801          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2802          * @params int $current, $max When an explicit limit has been
2803          *       exceeded, provide the values (optional)
2804          */
2805         function limitationWarn( $limitationType, $current=null, $max=null) {
2806                 //does no harm if $current and $max are present but are unnecessary for the message
2807                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2808                 $this->mOutput->addWarning( $warning );
2809                 $this->addTrackingCategory( "$limitationType-category" );
2810         }
2811
2812         /**
2813          * Return the text of a template, after recursively
2814          * replacing any variables or templates within the template.
2815          *
2816          * @param array $piece The parts of the template
2817          *  $piece['title']: the title, i.e. the part before the |
2818          *  $piece['parts']: the parameter array
2819          *  $piece['lineStart']: whether the brace was at the start of a line
2820          * @param PPFrame The current frame, contains template arguments
2821          * @return string the text of the template
2822          * @private
2823          */
2824         function braceSubstitution( $piece, $frame ) {
2825                 global $wgContLang, $wgNonincludableNamespaces;
2826                 wfProfileIn( __METHOD__ );
2827                 wfProfileIn( __METHOD__.'-setup' );
2828
2829                 # Flags
2830                 $found = false;             # $text has been filled
2831                 $nowiki = false;            # wiki markup in $text should be escaped
2832                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2833                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2834                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2835                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2836
2837                 # Title object, where $text came from
2838                 $title = null;
2839
2840                 # $part1 is the bit before the first |, and must contain only title characters.
2841                 # Various prefixes will be stripped from it later.
2842                 $titleWithSpaces = $frame->expand( $piece['title'] );
2843                 $part1 = trim( $titleWithSpaces );
2844                 $titleText = false;
2845
2846                 # Original title text preserved for various purposes
2847                 $originalTitle = $part1;
2848
2849                 # $args is a list of argument nodes, starting from index 0, not including $part1
2850                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2851                 wfProfileOut( __METHOD__.'-setup' );
2852
2853                 # SUBST
2854                 wfProfileIn( __METHOD__.'-modifiers' );
2855                 if ( !$found ) {
2856
2857                         $substMatch = $this->mSubsts->matchStartAndRemove( $part1 );
2858
2859                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2860                         # Whether to include depends also on whether we are in the pre-save-transform
2861                         #
2862                         # safesubst || (subst && PST) || (false && !PST) => transclude (skip the if)
2863                         # (false && PST) || (subst && !PST)  => return input (handled by if)
2864                         if ( $substMatch != 'safesubst' && ($substMatch == 'subst' xor $this->ot['wiki']) ) {
2865                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2866                                 $isLocalObj = true;
2867                                 $found = true;
2868                         }
2869                 }
2870
2871                 # Variables
2872                 if ( !$found && $args->getLength() == 0 ) {
2873                         $id = $this->mVariables->matchStartToEnd( $part1 );
2874                         if ( $id !== false ) {
2875                                 $text = $this->getVariableValue( $id, $frame );
2876                                 if (MagicWord::getCacheTTL($id)>-1)
2877                                         $this->mOutput->mContainsOldMagic = true;
2878                                 $found = true;
2879                         }
2880                 }
2881
2882                 # MSG, MSGNW and RAW
2883                 if ( !$found ) {
2884                         # Check for MSGNW:
2885                         $mwMsgnw = MagicWord::get( 'msgnw' );
2886                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2887                                 $nowiki = true;
2888                         } else {
2889                                 # Remove obsolete MSG:
2890                                 $mwMsg = MagicWord::get( 'msg' );
2891                                 $mwMsg->matchStartAndRemove( $part1 );
2892                         }
2893
2894                         # Check for RAW:
2895                         $mwRaw = MagicWord::get( 'raw' );
2896                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2897                                 $forceRawInterwiki = true;
2898                         }
2899                 }
2900                 wfProfileOut( __METHOD__.'-modifiers' );
2901
2902                 # Parser functions
2903                 if ( !$found ) {
2904                         wfProfileIn( __METHOD__ . '-pfunc' );
2905
2906                         $colonPos = strpos( $part1, ':' );
2907                         if ( $colonPos !== false ) {
2908                                 # Case sensitive functions
2909                                 $function = substr( $part1, 0, $colonPos );
2910                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2911                                         $function = $this->mFunctionSynonyms[1][$function];
2912                                 } else {
2913                                         # Case insensitive functions
2914                                         $function = $wgContLang->lc( $function );
2915                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2916                                                 $function = $this->mFunctionSynonyms[0][$function];
2917                                         } else {
2918                                                 $function = false;
2919                                         }
2920                                 }
2921                                 if ( $function ) {
2922                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2923                                         $initialArgs = array( &$this );
2924                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2925                                         if ( $flags & SFH_OBJECT_ARGS ) {
2926                                                 # Add a frame parameter, and pass the arguments as an array
2927                                                 $allArgs = $initialArgs;
2928                                                 $allArgs[] = $frame;
2929                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2930                                                         $funcArgs[] = $args->item( $i );
2931                                                 }
2932                                                 $allArgs[] = $funcArgs;
2933                                         } else {
2934                                                 # Convert arguments to plain text
2935                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2936                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2937                                                 }
2938                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2939                                         }
2940
2941                                         # Workaround for PHP bug 35229 and similar
2942                                         if ( !is_callable( $callback ) ) {
2943                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2944                                                 wfProfileOut( __METHOD__ );
2945                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2946                                         }
2947                                         $result = call_user_func_array( $callback, $allArgs );
2948                                         $found = true;
2949                                         $noparse = true;
2950                                         $preprocessFlags = 0;
2951
2952                                         if ( is_array( $result ) ) {
2953                                                 if ( isset( $result[0] ) ) {
2954                                                         $text = $result[0];
2955                                                         unset( $result[0] );
2956                                                 }
2957
2958                                                 // Extract flags into the local scope
2959                                                 // This allows callers to set flags such as nowiki, found, etc.
2960                                                 extract( $result );
2961                                         } else {
2962                                                 $text = $result;
2963                                         }
2964                                         if ( !$noparse ) {
2965                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2966                                                 $isChildObj = true;
2967                                         }
2968                                 }
2969                         }
2970                         wfProfileOut( __METHOD__ . '-pfunc' );
2971                 }
2972
2973                 # Finish mangling title and then check for loops.
2974                 # Set $title to a Title object and $titleText to the PDBK
2975                 if ( !$found ) {
2976                         $ns = NS_TEMPLATE;
2977                         # Split the title into page and subpage
2978                         $subpage = '';
2979                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2980                         if ($subpage !== '') {
2981                                 $ns = $this->mTitle->getNamespace();
2982                         }
2983                         $title = Title::newFromText( $part1, $ns );
2984                         if ( $title ) {
2985                                 $titleText = $title->getPrefixedText();
2986                                 # Check for language variants if the template is not found
2987                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2988                                         $wgContLang->findVariantLink( $part1, $title, true );
2989                                 }
2990                                 # Do recursion depth check
2991                                 $limit = $this->mOptions->getMaxTemplateDepth();
2992                                 if ( $frame->depth >= $limit ) {
2993                                         $found = true;
2994                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2995                                 }
2996                         }
2997                 }
2998
2999                 # Load from database
3000                 if ( !$found && $title ) {
3001                         wfProfileIn( __METHOD__ . '-loadtpl' );
3002                         if ( !$title->isExternal() ) {
3003                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3004                                         $text = SpecialPage::capturePath( $title );
3005                                         if ( is_string( $text ) ) {
3006                                                 $found = true;
3007                                                 $isHTML = true;
3008                                                 $this->disableCache();
3009                                         }
3010                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3011                                         $found = false; //access denied
3012                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3013                                 } else {
3014                                         list( $text, $title ) = $this->getTemplateDom( $title );
3015                                         if ( $text !== false ) {
3016                                                 $found = true;
3017                                                 $isChildObj = true;
3018                                         }
3019                                 }
3020
3021                                 # If the title is valid but undisplayable, make a link to it
3022                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3023                                         $text = "[[:$titleText]]";
3024                                         $found = true;
3025                                 }
3026                         } elseif ( $title->isTrans() ) {
3027                                 // Interwiki transclusion
3028                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3029                                         $text = $this->interwikiTransclude( $title, 'render' );
3030                                         $isHTML = true;
3031                                 } else {
3032                                         $text = $this->interwikiTransclude( $title, 'raw' );
3033                                         // Preprocess it like a template
3034                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3035                                         $isChildObj = true;
3036                                 }
3037                                 $found = true;
3038                         }
3039
3040                         # Do infinite loop check
3041                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3042                         if ( !$frame->loopCheck( $title ) ) {
3043                                 $found = true;
3044                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3045                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3046                         }
3047                         wfProfileOut( __METHOD__ . '-loadtpl' );
3048                 }
3049
3050                 # If we haven't found text to substitute by now, we're done
3051                 # Recover the source wikitext and return it
3052                 if ( !$found ) {
3053                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3054                         wfProfileOut( __METHOD__ );
3055                         return array( 'object' => $text );
3056                 }
3057
3058                 # Expand DOM-style return values in a child frame
3059                 if ( $isChildObj ) {
3060                         # Clean up argument array
3061                         $newFrame = $frame->newChild( $args, $title );
3062
3063                         if ( $nowiki ) {
3064                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3065                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3066                                 # Expansion is eligible for the empty-frame cache
3067                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3068                                         $text = $this->mTplExpandCache[$titleText];
3069                                 } else {
3070                                         $text = $newFrame->expand( $text );
3071                                         $this->mTplExpandCache[$titleText] = $text;
3072                                 }
3073                         } else {
3074                                 # Uncached expansion
3075                                 $text = $newFrame->expand( $text );
3076                         }
3077                 }
3078                 if ( $isLocalObj && $nowiki ) {
3079                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3080                         $isLocalObj = false;
3081                 }
3082
3083                 # Replace raw HTML by a placeholder
3084                 # Add a blank line preceding, to prevent it from mucking up
3085                 # immediately preceding headings
3086                 if ( $isHTML ) {
3087                         $text = "\n\n" . $this->insertStripItem( $text );
3088                 }
3089                 # Escape nowiki-style return values
3090                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3091                         $text = wfEscapeWikiText( $text );
3092                 }
3093                 # Bug 529: if the template begins with a table or block-level
3094                 # element, it should be treated as beginning a new line.
3095                 # This behaviour is somewhat controversial.
3096                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3097                         $text = "\n" . $text;
3098                 }
3099
3100                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3101                         # Error, oversize inclusion
3102                         $text = "[[$originalTitle]]" .
3103                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3104                         $this->limitationWarn( 'post-expand-template-inclusion' );
3105                 }
3106
3107                 if ( $isLocalObj ) {
3108                         $ret = array( 'object' => $text );
3109                 } else {
3110                         $ret = array( 'text' => $text );
3111                 }
3112
3113                 wfProfileOut( __METHOD__ );
3114                 return $ret;
3115         }
3116
3117         /**
3118          * Get the semi-parsed DOM representation of a template with a given title,
3119          * and its redirect destination title. Cached.
3120          */
3121         function getTemplateDom( $title ) {
3122                 $cacheTitle = $title;
3123                 $titleText = $title->getPrefixedDBkey();
3124
3125                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3126                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3127                         $title = Title::makeTitle( $ns, $dbk );
3128                         $titleText = $title->getPrefixedDBkey();
3129                 }
3130                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3131                         return array( $this->mTplDomCache[$titleText], $title );
3132                 }
3133
3134                 // Cache miss, go to the database
3135                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3136
3137                 if ( $text === false ) {
3138                         $this->mTplDomCache[$titleText] = false;
3139                         return array( false, $title );
3140                 }
3141
3142                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3143                 $this->mTplDomCache[ $titleText ] = $dom;
3144
3145                 if (! $title->equals($cacheTitle)) {
3146                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3147                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3148                 }
3149
3150                 return array( $dom, $title );
3151         }
3152
3153         /**
3154          * Fetch the unparsed text of a template and register a reference to it.
3155          */
3156         function fetchTemplateAndTitle( $title ) {
3157                 $templateCb = $this->mOptions->getTemplateCallback();
3158                 $stuff = call_user_func( $templateCb, $title, $this );
3159                 $text = $stuff['text'];
3160                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3161                 if ( isset( $stuff['deps'] ) ) {
3162                         foreach ( $stuff['deps'] as $dep ) {
3163                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3164                         }
3165                 }
3166                 return array($text,$finalTitle);
3167         }
3168
3169         function fetchTemplate( $title ) {
3170                 $rv = $this->fetchTemplateAndTitle($title);
3171                 return $rv[0];
3172         }
3173
3174         /**
3175          * Static function to get a template
3176          * Can be overridden via ParserOptions::setTemplateCallback().
3177          */
3178         static function statelessFetchTemplate( $title, $parser=false ) {
3179                 $text = $skip = false;
3180                 $finalTitle = $title;
3181                 $deps = array();
3182
3183                 // Loop to fetch the article, with up to 1 redirect
3184                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3185                         # Give extensions a chance to select the revision instead
3186                         $id = false; // Assume current
3187                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3188
3189                         if( $skip ) {
3190                                 $text = false;
3191                                 $deps[] = array(
3192                                         'title' => $title,
3193                                         'page_id' => $title->getArticleID(),
3194                                         'rev_id' => null );
3195                                 break;
3196                         }
3197                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3198                         $rev_id = $rev ? $rev->getId() : 0;
3199                         // If there is no current revision, there is no page
3200                         if( $id === false && !$rev ) {
3201                                 $linkCache = LinkCache::singleton();
3202                                 $linkCache->addBadLinkObj( $title );
3203                         }
3204
3205                         $deps[] = array(
3206                                 'title' => $title,
3207                                 'page_id' => $title->getArticleID(),
3208                                 'rev_id' => $rev_id );
3209
3210                         if( $rev ) {
3211                                 $text = $rev->getText();
3212                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3213                                 global $wgContLang;
3214                                 $message = $wgContLang->lcfirst( $title->getText() );
3215                                 $text = wfMsgForContentNoTrans( $message );
3216                                 if( wfEmptyMsg( $message, $text ) ) {
3217                                         $text = false;
3218                                         break;
3219                                 }
3220                         } else {
3221                                 break;
3222                         }
3223                         if ( $text === false ) {
3224                                 break;
3225                         }
3226                         // Redirect?
3227                         $finalTitle = $title;
3228                         $title = Title::newFromRedirect( $text );
3229                 }
3230                 return array(
3231                         'text' => $text,
3232                         'finalTitle' => $finalTitle,
3233                         'deps' => $deps );
3234         }
3235
3236         /**
3237          * Transclude an interwiki link.
3238          */
3239         function interwikiTransclude( $title, $action ) {
3240                 global $wgEnableScaryTranscluding;
3241
3242                 if (!$wgEnableScaryTranscluding)
3243                         return wfMsg('scarytranscludedisabled');
3244
3245                 $url = $title->getFullUrl( "action=$action" );
3246
3247                 if (strlen($url) > 255)
3248                         return wfMsg('scarytranscludetoolong');
3249                 return $this->fetchScaryTemplateMaybeFromCache($url);
3250         }
3251
3252         function fetchScaryTemplateMaybeFromCache($url) {
3253                 global $wgTranscludeCacheExpiry;
3254                 $dbr = wfGetDB(DB_SLAVE);
3255                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3256                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3257                                 array('tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3258                 if ($obj) {
3259                         return $obj->tc_contents;
3260                 }
3261
3262                 $text = Http::get($url);
3263                 if (!$text)
3264                         return wfMsg('scarytranscludefailed', $url);
3265
3266                 $dbw = wfGetDB(DB_MASTER);
3267                 $dbw->replace('transcache', array('tc_url'), array(
3268                         'tc_url' => $url,
3269                         'tc_time' => $dbw->timestamp( time() ),
3270                         'tc_contents' => $text));
3271                 return $text;
3272         }
3273
3274
3275         /**
3276          * Triple brace replacement -- used for template arguments
3277          * @private
3278          */
3279         function argSubstitution( $piece, $frame ) {
3280                 wfProfileIn( __METHOD__ );
3281
3282                 $error = false;
3283                 $parts = $piece['parts'];
3284                 $nameWithSpaces = $frame->expand( $piece['title'] );
3285                 $argName = trim( $nameWithSpaces );
3286                 $object = false;
3287                 $text = $frame->getArgument( $argName );
3288                 if (  $text === false && $parts->getLength() > 0
3289                   && (
3290                     $this->ot['html']
3291                     || $this->ot['pre']
3292                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3293                   )
3294                 ) {
3295                         # No match in frame, use the supplied default
3296                         $object = $parts->item( 0 )->getChildren();
3297                 }
3298                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3299                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3300                         $this->limitationWarn( 'post-expand-template-argument' );
3301                 }
3302
3303                 if ( $text === false && $object === false ) {
3304                         # No match anywhere
3305                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3306                 }
3307                 if ( $error !== false ) {
3308                         $text .= $error;
3309                 }
3310                 if ( $object !== false ) {
3311                         $ret = array( 'object' => $object );
3312                 } else {
3313                         $ret = array( 'text' => $text );
3314                 }
3315
3316                 wfProfileOut( __METHOD__ );
3317                 return $ret;
3318         }
3319
3320         /**
3321          * Return the text to be used for a given extension tag.
3322          * This is the ghost of strip().
3323          *
3324          * @param array $params Associative array of parameters:
3325          *     name       PPNode for the tag name
3326          *     attr       PPNode for unparsed text where tag attributes are thought to be
3327          *     attributes Optional associative array of parsed attributes
3328          *     inner      Contents of extension element
3329          *     noClose    Original text did not have a close tag
3330          * @param PPFrame $frame
3331          */
3332         function extensionSubstitution( $params, $frame ) {
3333                 global $wgRawHtml, $wgContLang;
3334
3335                 $name = $frame->expand( $params['name'] );
3336                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3337                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3338                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3339
3340                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3341                         ( $this->ot['html'] || $this->ot['pre'] );
3342                 if ( $isFunctionTag ) {
3343                         $markerType = 'none';
3344                 } else {
3345                         $markerType = 'general';
3346                 }
3347                 if ( $this->ot['html'] || $isFunctionTag ) {
3348                         $name = strtolower( $name );
3349                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3350                         if ( isset( $params['attributes'] ) ) {
3351                                 $attributes = $attributes + $params['attributes'];
3352                         }
3353
3354                         if( isset( $this->mTagHooks[$name] ) ) {
3355                                 # Workaround for PHP bug 35229 and similar
3356                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3357                                         throw new MWException( "Tag hook for $name is not callable\n" );
3358                                 }
3359                                 $output = call_user_func_array( $this->mTagHooks[$name],
3360                                         array( $content, $attributes, $this, $frame ) );
3361                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3362                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3363                                 if( !is_callable( $callback ) )
3364                                         throw new MWException( "Tag hook for $name is not callable\n" );
3365
3366                                 $output = call_user_func_array( $callback,
3367                                         array( &$this, $frame, $content, $attributes ) );
3368                         } else {
3369                                 $output = '<span class="error">Invalid tag extension name: ' .
3370                                         htmlspecialchars( $name ) . '</span>';
3371                         }
3372
3373                         if ( is_array( $output ) ) {
3374                                 // Extract flags to local scope (to override $markerType)
3375                                 $flags = $output;
3376                                 $output = $flags[0];
3377                                 unset( $flags[0] );
3378                                 extract( $flags );
3379                         }
3380                 } else {
3381                         if ( is_null( $attrText ) ) {
3382                                 $attrText = '';
3383                         }
3384                         if ( isset( $params['attributes'] ) ) {
3385                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3386                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3387                                                 htmlspecialchars( $attrValue ) . '"';
3388                                 }
3389                         }
3390                         if ( $content === null ) {
3391                                 $output = "<$name$attrText/>";
3392                         } else {
3393                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3394                                 $output = "<$name$attrText>$content$close";
3395                         }
3396                 }
3397
3398                 if( $markerType === 'none' ) {
3399                         return $output;
3400                 } elseif ( $markerType === 'nowiki' ) {
3401                         $this->mStripState->nowiki->setPair( $marker, $output );
3402                 } elseif ( $markerType === 'general' ) {
3403                         $this->mStripState->general->setPair( $marker, $output );
3404                 } else {
3405                         throw new MWException( __METHOD__.': invalid marker type' );
3406                 }
3407                 return $marker;
3408         }
3409
3410         /**
3411          * Increment an include size counter
3412          *
3413          * @param string $type The type of expansion
3414          * @param integer $size The size of the text
3415          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3416          */
3417         function incrementIncludeSize( $type, $size ) {
3418                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3419                         return false;
3420                 } else {
3421                         $this->mIncludeSizes[$type] += $size;
3422                         return true;
3423                 }
3424         }
3425
3426         /**
3427          * Increment the expensive function count
3428          *
3429          * @return boolean False if the limit has been exceeded
3430          */
3431         function incrementExpensiveFunctionCount() {
3432                 global $wgExpensiveParserFunctionLimit;
3433                 $this->mExpensiveFunctionCount++;
3434                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3435                         return true;
3436                 }
3437                 return false;
3438         }
3439
3440         /**
3441          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3442          * Fills $this->mDoubleUnderscores, returns the modified text
3443          */
3444         function doDoubleUnderscore( $text ) {
3445                 wfProfileIn( __METHOD__ );
3446
3447                 // The position of __TOC__ needs to be recorded
3448                 $mw = MagicWord::get( 'toc' );
3449                 if( $mw->match( $text ) ) {
3450                         $this->mShowToc = true;
3451                         $this->mForceTocPosition = true;
3452
3453                         // Set a placeholder. At the end we'll fill it in with the TOC.
3454                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3455
3456                         // Only keep the first one.
3457                         $text = $mw->replace( '', $text );
3458                 }
3459
3460                 // Now match and remove the rest of them
3461                 $mwa = MagicWord::getDoubleUnderscoreArray();
3462                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3463
3464                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3465                         $this->mOutput->mNoGallery = true;
3466                 }
3467                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3468                         $this->mShowToc = false;
3469                 }
3470                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3471                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3472                         $this->addTrackingCategory( 'hidden-category-category' );
3473                 }
3474                 # (bug 8068) Allow control over whether robots index a page.
3475                 #
3476                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3477                 # is not desirable, the last one on the page should win.
3478                 if( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3479                         $this->mOutput->setIndexPolicy( 'noindex' );
3480                         $this->addTrackingCategory( 'noindex-category' );
3481                 }
3482                 if( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ){
3483                         $this->mOutput->setIndexPolicy( 'index' );
3484                         $this->addTrackingCategory( 'index-category' );
3485                 }
3486
3487                 wfProfileOut( __METHOD__ );
3488                 return $text;
3489         }
3490
3491         /**
3492          * Add a tracking category, getting the title from a system message,
3493          * or print a debug message if the title is invalid.
3494          * @param $msg String message key
3495          * @return Bool whether the addition was successful
3496          */
3497         protected function addTrackingCategory( $msg ){
3498                 $cat = wfMsgForContent( $msg );
3499
3500                 # Allow tracking categories to be disabled by setting them to "-"
3501                 if( $cat === '-' ) return false;
3502
3503                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3504                 if ( $containerCategory ) {
3505                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3506                         return true;
3507                 } else {
3508                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3509                         return false;
3510                 }
3511         }
3512
3513         /**
3514          * This function accomplishes several tasks:
3515          * 1) Auto-number headings if that option is enabled
3516          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3517          * 3) Add a Table of contents on the top for users who have enabled the option
3518          * 4) Auto-anchor headings
3519          *
3520          * It loops through all headlines, collects the necessary data, then splits up the
3521          * string and re-inserts the newly formatted headlines.
3522          *
3523          * @param string $text
3524          * @param string $origText Original, untouched wikitext
3525          * @param boolean $isMain
3526          * @private
3527          */
3528         function formatHeadings( $text, $origText, $isMain=true ) {
3529                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3530
3531                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3532                 $showEditLink = $this->mOptions->getEditSection();
3533
3534                 // Do not call quickUserCan unless necessary
3535                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3536                         $showEditLink = 0;
3537                 }
3538
3539                 # Inhibit editsection links if requested in the page
3540                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3541                         $showEditLink = 0;
3542                 }
3543
3544                 # Get all headlines for numbering them and adding funky stuff like [edit]
3545                 # links - this is for later, but we need the number of headlines right now
3546                 $matches = array();
3547                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3548
3549                 # if there are fewer than 4 headlines in the article, do not show TOC
3550                 # unless it's been explicitly enabled.
3551                 $enoughToc = $this->mShowToc &&
3552                         (($numMatches >= 4) || $this->mForceTocPosition);
3553
3554                 # Allow user to stipulate that a page should have a "new section"
3555                 # link added via __NEWSECTIONLINK__
3556                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3557                         $this->mOutput->setNewSection( true );
3558                 }
3559
3560                 # Allow user to remove the "new section"
3561                 # link via __NONEWSECTIONLINK__
3562                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3563                         $this->mOutput->hideNewSection( true );
3564                 }
3565
3566                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3567                 # override above conditions and always show TOC above first header
3568                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3569                         $this->mShowToc = true;
3570                         $enoughToc = true;
3571                 }
3572
3573                 # We need this to perform operations on the HTML
3574                 $sk = $this->mOptions->getSkin();
3575
3576                 # headline counter
3577                 $headlineCount = 0;
3578                 $numVisible = 0;
3579
3580                 # Ugh .. the TOC should have neat indentation levels which can be
3581                 # passed to the skin functions. These are determined here
3582                 $toc = '';
3583                 $full = '';
3584                 $head = array();
3585                 $sublevelCount = array();
3586                 $levelCount = array();
3587                 $toclevel = 0;
3588                 $level = 0;
3589                 $prevlevel = 0;
3590                 $toclevel = 0;
3591                 $prevtoclevel = 0;
3592                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3593                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3594                 $oldType = $this->mOutputType;
3595                 $this->setOutputType( self::OT_WIKI );
3596                 $frame = $this->getPreprocessor()->newFrame();
3597                 $root = $this->preprocessToDom( $origText );
3598                 $node = $root->getFirstChild();
3599                 $byteOffset = 0;
3600                 $tocraw = array();
3601
3602                 foreach( $matches[3] as $headline ) {
3603                         $isTemplate = false;
3604                         $titleText = false;
3605                         $sectionIndex = false;
3606                         $numbering = '';
3607                         $markerMatches = array();
3608                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3609                                 $serial = $markerMatches[1];
3610                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3611                                 $isTemplate = ($titleText != $baseTitleText);
3612                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3613                         }
3614
3615                         if( $toclevel ) {
3616                                 $prevlevel = $level;
3617                                 $prevtoclevel = $toclevel;
3618                         }
3619                         $level = $matches[1][$headlineCount];
3620
3621                         if ( $level > $prevlevel ) {
3622                                 # Increase TOC level
3623                                 $toclevel++;
3624                                 $sublevelCount[$toclevel] = 0;
3625                                 if( $toclevel<$wgMaxTocLevel ) {
3626                                         $prevtoclevel = $toclevel;
3627                                         $toc .= $sk->tocIndent();
3628                                         $numVisible++;
3629                                 }
3630                         }
3631                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3632                                 # Decrease TOC level, find level to jump to
3633
3634                                 for ($i = $toclevel; $i > 0; $i--) {
3635                                         if ( $levelCount[$i] == $level ) {
3636                                                 # Found last matching level
3637                                                 $toclevel = $i;
3638                                                 break;
3639                                         }
3640                                         elseif ( $levelCount[$i] < $level ) {
3641                                                 # Found first matching level below current level
3642                                                 $toclevel = $i + 1;
3643                                                 break;
3644                                         }
3645                                 }
3646                                 if( $i == 0 ) $toclevel = 1;
3647                                 if( $toclevel<$wgMaxTocLevel ) {
3648                                         if($prevtoclevel < $wgMaxTocLevel) {
3649                                                 # Unindent only if the previous toc level was shown :p
3650                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3651                                                 $prevtoclevel = $toclevel;
3652                                         } else {
3653                                                 $toc .= $sk->tocLineEnd();
3654                                         }
3655                                 }
3656                         }
3657                         else {
3658                                 # No change in level, end TOC line
3659                                 if( $toclevel<$wgMaxTocLevel ) {
3660                                         $toc .= $sk->tocLineEnd();
3661                                 }
3662                         }
3663
3664                         $levelCount[$toclevel] = $level;
3665
3666                         # count number of headlines for each level
3667                         @$sublevelCount[$toclevel]++;
3668                         $dot = 0;
3669                         for( $i = 1; $i <= $toclevel; $i++ ) {
3670                                 if( !empty( $sublevelCount[$i] ) ) {
3671                                         if( $dot ) {
3672                                                 $numbering .= '.';
3673                                         }
3674                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3675                                         $dot = 1;
3676                                 }
3677                         }
3678
3679                         # The safe header is a version of the header text safe to use for links
3680                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3681                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3682
3683                         # Remove link placeholders by the link text.
3684                         #     <!--LINK number-->
3685                         # turns into
3686                         #     link text with suffix
3687                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3688
3689                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3690                         $tocline = preg_replace(
3691                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3692                                 array( '',                          '<$1>'),
3693                                 $safeHeadline
3694                         );
3695                         $tocline = trim( $tocline );
3696
3697                         # For the anchor, strip out HTML-y stuff period
3698                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3699                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3700                         $safeHeadline = trim( $safeHeadline );
3701
3702                         # Save headline for section edit hint before it's escaped
3703                         $headlineHint = $safeHeadline;
3704
3705                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3706                                 # For reverse compatibility, provide an id that's
3707                                 # HTML4-compatible, like we used to.
3708                                 #
3709                                 # It may be worth noting, academically, that it's possible for
3710                                 # the legacy anchor to conflict with a non-legacy headline
3711                                 # anchor on the page.  In this case likely the "correct" thing
3712                                 # would be to either drop the legacy anchors or make sure
3713                                 # they're numbered first.  However, this would require people
3714                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3715                                 # manually, so let's not bother worrying about it.
3716                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3717                                         array( 'noninitial', 'legacy' ) );
3718                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3719
3720                                 if ( $legacyHeadline == $safeHeadline ) {
3721                                         # No reason to have both (in fact, we can't)
3722                                         $legacyHeadline = false;
3723                                 }
3724                         } else {
3725                                 $legacyHeadline = false;
3726                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3727                                         'noninitial' );
3728                         }
3729
3730                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3731                         # Does this apply to Unicode characters?  Because we aren't
3732                         # handling those here.
3733                         $arrayKey = strtolower( $safeHeadline );
3734                         if ( $legacyHeadline === false ) {
3735                                 $legacyArrayKey = false;
3736                         } else {
3737                                 $legacyArrayKey = strtolower( $legacyHeadline );
3738                         }
3739
3740                         # count how many in assoc. array so we can track dupes in anchors
3741                         if ( isset( $refers[$arrayKey] ) ) {
3742                                 $refers[$arrayKey]++;
3743                         } else {
3744                                 $refers[$arrayKey] = 1;
3745                         }
3746                         if ( isset( $refers[$legacyArrayKey] ) ) {
3747                                 $refers[$legacyArrayKey]++;
3748                         } else {
3749                                 $refers[$legacyArrayKey] = 1;
3750                         }
3751
3752                         # Don't number the heading if it is the only one (looks silly)
3753                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3754                                 # the two are different if the line contains a link
3755                                 $headline = $numbering . ' ' . $headline;
3756                         }
3757
3758                         # Create the anchor for linking from the TOC to the section
3759                         $anchor = $safeHeadline;
3760                         $legacyAnchor = $legacyHeadline;
3761                         if ( $refers[$arrayKey] > 1 ) {
3762                                 $anchor .= '_' . $refers[$arrayKey];
3763                         }
3764                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3765                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3766                         }
3767                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3768                                 $toc .= $sk->tocLine($anchor, $tocline,
3769                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3770                         }
3771
3772                         # Add the section to the section tree
3773                         # Find the DOM node for this header
3774                         while ( $node && !$isTemplate ) {
3775                                 if ( $node->getName() === 'h' ) {
3776                                         $bits = $node->splitHeading();
3777                                         if ( $bits['i'] == $sectionIndex )
3778                                                 break;
3779                                 }
3780                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3781                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3782                                 $node = $node->getNextSibling();
3783                         }
3784                         $tocraw[] = array(
3785                                 'toclevel' => $toclevel,
3786                                 'level' => $level,
3787                                 'line' => $tocline,
3788                                 'number' => $numbering,
3789                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3790                                 'fromtitle' => $titleText,
3791                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3792                                 'anchor' => $anchor,
3793                         );
3794
3795                         # give headline the correct <h#> tag
3796                         if( $showEditLink && $sectionIndex !== false ) {
3797                                 if( $isTemplate ) {
3798                                         # Put a T flag in the section identifier, to indicate to extractSections()
3799                                         # that sections inside <includeonly> should be counted.
3800                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3801                                 } else {
3802                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3803                                 }
3804                         } else {
3805                                 $editlink = '';
3806                         }
3807                         $head[$headlineCount] = $sk->makeHeadline( $level,
3808                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3809                                 $editlink, $legacyAnchor );
3810
3811                         $headlineCount++;
3812                 }
3813
3814                 $this->setOutputType( $oldType );
3815
3816                 # Never ever show TOC if no headers
3817                 if( $numVisible < 1 ) {
3818                         $enoughToc = false;
3819                 }
3820
3821                 if( $enoughToc ) {
3822                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3823                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3824                         }
3825                         $toc = $sk->tocList( $toc );
3826                         $this->mOutput->setTOCHTML( $toc );
3827                 }
3828
3829                 if ( $isMain ) {
3830                         $this->mOutput->setSections( $tocraw );
3831                 }
3832
3833                 # split up and insert constructed headlines
3834
3835                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3836                 $i = 0;
3837
3838                 foreach( $blocks as $block ) {
3839                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3840                                 # This is the [edit] link that appears for the top block of text when
3841                                 # section editing is enabled
3842
3843                                 # Disabled because it broke block formatting
3844                                 # For example, a bullet point in the top line
3845                                 # $full .= $sk->editSectionLink(0);
3846                         }
3847                         $full .= $block;
3848                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3849                                 # Top anchor now in skin
3850                                 $full = $full.$toc;
3851                         }
3852
3853                         if( !empty( $head[$i] ) ) {
3854                                 $full .= $head[$i];
3855                         }
3856                         $i++;
3857                 }
3858                 if( $this->mForceTocPosition ) {
3859                         return str_replace( '<!--MWTOC-->', $toc, $full );
3860                 } else {
3861                         return $full;
3862                 }
3863         }
3864
3865         /**
3866          * Merge $tree2 into $tree1 by replacing the section with index
3867          * $section in $tree1 and its descendants with the sections in $tree2.
3868          * Note that in the returned section tree, only the 'index' and
3869          * 'byteoffset' fields are guaranteed to be correct.
3870          * @param $tree1 array Section tree from ParserOutput::getSectons()
3871         *  @param $tree2 array Section tree
3872          * @param $section int Section index
3873          * @param $title Title Title both section trees come from
3874          * @param $len2 int Length of the original wikitext for $tree2
3875          * @return array Merged section tree
3876          */
3877         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3878                 global $wgContLang;
3879                 $newTree = array();
3880                 $targetLevel = false;
3881                 $merged = false;
3882                 $lastLevel = 1;
3883                 $nextIndex = 1;
3884                 $numbering = array( 0 );
3885                 $titletext = $title->getPrefixedDBkey();
3886                 foreach ( $tree1 as $s ) {
3887                         if ( $targetLevel !== false ) {
3888                                 if ( $s['level'] <= $targetLevel )
3889                                         // We've skipped enough
3890                                         $targetLevel = false;
3891                                 else
3892                                         continue;
3893                         }
3894                         if ( $s['index'] != $section ||
3895                                         $s['fromtitle'] != $titletext ) {
3896                                 self::incrementNumbering( $numbering,
3897                                         $s['toclevel'], $lastLevel );
3898
3899                                 // Rewrite index, byteoffset and number
3900                                 if ( $s['fromtitle'] == $titletext ) {
3901                                         $s['index'] = $nextIndex++;
3902                                         if ( $merged )
3903                                                 $s['byteoffset'] += $len2;
3904                                 }
3905                                 $s['number']  = implode( '.', array_map(
3906                                         array( $wgContLang, 'formatnum' ),
3907                                         $numbering ) );
3908                                 $lastLevel = $s['toclevel'];
3909                                 $newTree[] = $s;
3910                         } else {
3911                                 // We're at $section
3912                                 // Insert sections from $tree2 here
3913                                 foreach ( $tree2 as $s2 ) {
3914                                         // Rewrite the fields in $s2
3915                                         // before inserting it
3916                                         $s2['toclevel'] += $s['toclevel'] - 1;
3917                                         $s2['level'] += $s['level'] - 1;
3918                                         $s2['index'] = $nextIndex++;
3919                                         $s2['byteoffset'] += $s['byteoffset'];
3920
3921                                         self::incrementNumbering( $numbering,
3922                                                 $s2['toclevel'], $lastLevel );
3923                                         $s2['number']  = implode( '.', array_map(
3924                                                 array( $wgContLang, 'formatnum' ),
3925                                                 $numbering ) );
3926                                         $lastLevel = $s2['toclevel'];
3927                                         $newTree[] = $s2;
3928                                 }
3929                                 // Skip all descendants of $section in $tree1
3930                                 $targetLevel = $s['level'];
3931                                 $merged = true;
3932                         }
3933                 }
3934                 return $newTree;
3935         }
3936
3937         /**
3938          * Increment a section number. Helper function for mergeSectionTrees()
3939          * @param $number array Array representing a section number
3940          * @param $level int Current TOC level (depth)
3941          * @param $lastLevel int Level of previous TOC entry
3942          */
3943         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3944                 if ( $level > $lastLevel )
3945                         $number[$level - 1] = 1;
3946                 else if ( $level < $lastLevel ) {
3947                         foreach ( $number as $key => $unused )
3948                                 if ( $key >= $level )
3949                                         unset( $number[$key] );
3950                         $number[$level - 1]++;
3951                 } else
3952                         $number[$level - 1]++;
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                  *
3997                  * (see also bug 12815)
3998                  */
3999                 $ts = $this->mOptions->getTimestamp();
4000                 if ( isset( $wgLocaltimezone ) ) {
4001                         $tz = $wgLocaltimezone;
4002                 } else {
4003                         $tz = date_default_timezone_get();
4004                 }
4005
4006                 $unixts = wfTimestamp( TS_UNIX, $ts );
4007                 $oldtz = date_default_timezone_get();
4008                 date_default_timezone_set( $tz );
4009                 $ts = date( 'YmdHis', $unixts );
4010                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4011
4012                 /* Allow translation of timezones trough wiki. date() can return
4013                  * whatever crap the system uses, localised or not, so we cannot
4014                  * ship premade translations.
4015                  */
4016                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4017                 $value = wfMsgForContent( $key );
4018                 if ( !wfEmptyMsg( $key, $value ) ) $tzMsg = $value;
4019
4020                 date_default_timezone_set( $oldtz );
4021
4022                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4023
4024                 # Variable replacement
4025                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4026                 $text = $this->replaceVariables( $text );
4027
4028                 # Signatures
4029                 $sigText = $this->getUserSig( $user );
4030                 $text = strtr( $text, array(
4031                         '~~~~~' => $d,
4032                         '~~~~' => "$sigText $d",
4033                         '~~~' => $sigText
4034                 ) );
4035
4036                 # Links of the form [[|<blah>]] or [[<blah>|]] perform pipe tricks
4037                 # Note this only allows the # in the position it works.
4038                 global $wgLegalTitleChars;
4039                 $pipeTrickRe = "/\[\[(?:(\\|)([$wgLegalTitleChars]+)|([#$wgLegalTitleChars]+)\\|)\]\]/";
4040                 $text = preg_replace_callback( $pipeTrickRe, array( $this, 'pstPipeTrickCallback' ), $text );
4041
4042                 # Trim trailing whitespace
4043                 $text = rtrim( $text );
4044
4045                 return $text;
4046         }
4047
4048         /**
4049          * Called from pstPass2 to perform the pipe trick on links.
4050          * Original was either [[|text]] or [[link|]]
4051          *
4052          * @param Array ("|" or "", text, link) $m
4053          */
4054         function pstPipeTrickCallback( $m )
4055         {
4056                 if( $m[1] ) { # [[|<blah>]]
4057                         $text = $m[2];
4058                         $link = $this->getPipeTrickLink( $text );
4059                 } else { # [[<blah>|]]
4060                         $link = $m[3];
4061                         $text = $this->getPipeTrickText( $link );
4062                 }
4063
4064                 return $link === $text ? "[[$link]]" : "[[$link|$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                 $nickname = $nickname == null ? $username : $nickname;
4089
4090                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4091                         $nickname = $username;
4092                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4093                 } elseif( $fancySig !== false ) {
4094                         # Sig. might contain markup; validate this
4095                         if( $this->validateSig( $nickname ) !== false ) {
4096                                 # Validated; clean up (if needed) and return it
4097                                 return $this->cleanSig( $nickname, true );
4098                         } else {
4099                                 # Failed to validate; fall back to the default
4100                                 $nickname = $username;
4101                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4102                         }
4103                 }
4104
4105                 // Make sure nickname doesnt get a sig in a sig
4106                 $nickname = $this->cleanSigInSig( $nickname );
4107
4108                 # If we're still here, make it a link to the user page
4109                 $userText = wfEscapeWikiText( $username );
4110                 $nickText = wfEscapeWikiText( $nickname );
4111                 if ( $user->isAnon() )  {
4112                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4113                 } else {
4114                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4115                 }
4116         }
4117
4118         /**
4119          * Check that the user's signature contains no bad XML
4120          *
4121          * @param string $text
4122          * @return mixed An expanded string, or false if invalid.
4123          */
4124         function validateSig( $text ) {
4125                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4126         }
4127
4128         /**
4129          * Clean up signature text
4130          *
4131          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4132          * 2) Substitute all transclusions
4133          *
4134          * @param string $text
4135          * @param $parsing Whether we're cleaning (preferences save) or parsing
4136          * @return string Signature text
4137          */
4138         function cleanSig( $text, $parsing = false ) {
4139                 if ( !$parsing ) {
4140                         global $wgTitle;
4141                         $this->clearState();
4142                         $this->setTitle( $wgTitle );
4143                         $this->mOptions = new ParserOptions;
4144                         $this->setOutputType = self::OT_PREPROCESS;
4145                 }
4146
4147                 # Option to disable this feature
4148                 if ( !$this->mOptions->getCleanSignatures() ) {
4149                         return $text;
4150                 }
4151
4152                 # FIXME: regex doesn't respect extension tags or nowiki
4153                 #  => Move this logic to braceSubstitution()
4154                 $substWord = MagicWord::get( 'subst' );
4155                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4156                 $substText = '{{' . $substWord->getSynonym( 0 );
4157
4158                 $text = preg_replace( $substRegex, $substText, $text );
4159                 $text = $this->cleanSigInSig( $text );
4160                 $dom = $this->preprocessToDom( $text );
4161                 $frame = $this->getPreprocessor()->newFrame();
4162                 $text = $frame->expand( $dom );
4163
4164                 if ( !$parsing ) {
4165                         $text = $this->mStripState->unstripBoth( $text );
4166                 }
4167
4168                 return $text;
4169         }
4170
4171         /**
4172          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4173          * @param string $text
4174          * @return string Signature text with /~{3,5}/ removed
4175          */
4176         function cleanSigInSig( $text ) {
4177                 $text = preg_replace( '/~{3,5}/', '', $text );
4178                 return $text;
4179         }
4180
4181         /**
4182          * Set up some variables which are usually set up in parse()
4183          * so that an external function can call some class members with confidence
4184          * @public
4185          */
4186         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4187                 $this->setTitle( $title );
4188                 $this->mOptions = $options;
4189                 $this->setOutputType( $outputType );
4190                 if ( $clearState ) {
4191                         $this->clearState();
4192                 }
4193         }
4194
4195         /**
4196          * Wrapper for preprocess()
4197          *
4198          * @param string $text the text to preprocess
4199          * @param ParserOptions $options  options
4200          * @return string
4201          * @public
4202          */
4203         function transformMsg( $text, $options ) {
4204                 global $wgTitle;
4205                 static $executing = false;
4206
4207                 # Guard against infinite recursion
4208                 if ( $executing ) {
4209                         return $text;
4210                 }
4211                 $executing = true;
4212
4213                 wfProfileIn(__METHOD__);
4214                 $text = $this->preprocess( $text, $wgTitle, $options );
4215
4216                 $executing = false;
4217                 wfProfileOut(__METHOD__);
4218                 return $text;
4219         }
4220
4221         /**
4222          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4223          * The callback should have the following form:
4224          *    function myParserHook( $text, $params, &$parser ) { ... }
4225          *
4226          * Transform and return $text. Use $parser for any required context, e.g. use
4227          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4228          *
4229          * @public
4230          *
4231          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4232          * @param mixed $callback The callback function (and object) to use for the tag
4233          *
4234          * @return The old value of the mTagHooks array associated with the hook
4235          */
4236         function setHook( $tag, $callback ) {
4237                 $tag = strtolower( $tag );
4238                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4239                 $this->mTagHooks[$tag] = $callback;
4240                 if( !in_array( $tag, $this->mStripList ) ) {
4241                         $this->mStripList[] = $tag;
4242                 }
4243
4244                 return $oldVal;
4245         }
4246
4247         /* An old work-around for bug 2257 - deprecated 2010-02-13 */
4248         function setTransparentTagHook( $tag, $callback ) {
4249                 wfDeprecated( __METHOD__ );
4250                 $tag = strtolower( $tag );
4251                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4252                 $this->mTransparentTagHooks[$tag] = $callback;
4253
4254                 return $oldVal;
4255         }
4256
4257         /**
4258          * Remove all tag hooks
4259          */
4260         function clearTagHooks() {
4261                 $this->mTagHooks = array();
4262                 $this->mStripList = $this->mDefaultStripList;
4263         }
4264
4265         /**
4266          * Create a function, e.g. {{sum:1|2|3}}
4267          * The callback function should have the form:
4268          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4269          *
4270          * Or with SFH_OBJECT_ARGS:
4271          *    function myParserFunction( $parser, $frame, $args ) { ... }
4272          *
4273          * The callback may either return the text result of the function, or an array with the text
4274          * in element 0, and a number of flags in the other elements. The names of the flags are
4275          * specified in the keys. Valid flags are:
4276          *   found                     The text returned is valid, stop processing the template. This
4277          *                             is on by default.
4278          *   nowiki                    Wiki markup in the return value should be escaped
4279          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4280          *
4281          * @public
4282          *
4283          * @param string $id The magic word ID
4284          * @param mixed $callback The callback function (and object) to use
4285          * @param integer $flags a combination of the following flags:
4286          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4287          *
4288          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4289          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4290          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4291          *     the arguments, and to control the way they are expanded.
4292          *
4293          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4294          *     arguments, for instance:
4295          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4296          *
4297          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4298          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4299          *     working if/when this is changed.
4300          *
4301          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4302          *     expansion.
4303          *
4304          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4305          *     about the methods available in PPFrame and PPNode.
4306          *
4307          * @return The old callback function for this name, if any
4308          */
4309         function setFunctionHook( $id, $callback, $flags = 0 ) {
4310                 global $wgContLang;
4311
4312                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4313                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4314
4315                 # Add to function cache
4316                 $mw = MagicWord::get( $id );
4317                 if( !$mw )
4318                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4319
4320                 $synonyms = $mw->getSynonyms();
4321                 $sensitive = intval( $mw->isCaseSensitive() );
4322
4323                 foreach ( $synonyms as $syn ) {
4324                         # Case
4325                         if ( !$sensitive ) {
4326                                 $syn = $wgContLang->lc( $syn );
4327                         }
4328                         # Add leading hash
4329                         if ( !( $flags & SFH_NO_HASH ) ) {
4330                                 $syn = '#' . $syn;
4331                         }
4332                         # Remove trailing colon
4333                         if ( substr( $syn, -1, 1 ) === ':' ) {
4334                                 $syn = substr( $syn, 0, -1 );
4335                         }
4336                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4337                 }
4338                 return $oldVal;
4339         }
4340
4341         /**
4342          * Get all registered function hook identifiers
4343          *
4344          * @return array
4345          */
4346         function getFunctionHooks() {
4347                 return array_keys( $this->mFunctionHooks );
4348         }
4349
4350         /**
4351          * Create a tag function, e.g. <test>some stuff</test>.
4352          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4353          * Unlike parser functions, their content is not preprocessed.
4354          */
4355         function setFunctionTagHook( $tag, $callback, $flags ) {
4356                 $tag = strtolower( $tag );
4357                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4358                         $this->mFunctionTagHooks[$tag] : null;
4359                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4360
4361                 if( !in_array( $tag, $this->mStripList ) ) {
4362                         $this->mStripList[] = $tag;
4363                 }
4364
4365                 return $old;
4366         }
4367
4368         /**
4369          * FIXME: update documentation. makeLinkObj() is deprecated.
4370          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4371          * Placeholders created in Skin::makeLinkObj()
4372          * Returns an array of link CSS classes, indexed by PDBK.
4373          */
4374         function replaceLinkHolders( &$text, $options = 0 ) {
4375                 return $this->mLinkHolders->replace( $text );
4376         }
4377
4378         /**
4379          * Replace <!--LINK--> link placeholders with plain text of links
4380          * (not HTML-formatted).
4381          * @param string $text
4382          * @return string
4383          */
4384         function replaceLinkHoldersText( $text ) {
4385                 return $this->mLinkHolders->replaceText( $text );
4386         }
4387
4388         /**
4389          * Renders an image gallery from a text with one line per image.
4390          * text labels may be given by using |-style alternative text. E.g.
4391          *   Image:one.jpg|The number "1"
4392          *   Image:tree.jpg|A tree
4393          * given as text will return the HTML of a gallery with two images,
4394          * labeled 'The number "1"' and
4395          * 'A tree'.
4396          */
4397         function renderImageGallery( $text, $params ) {
4398                 $ig = new ImageGallery();
4399                 $ig->setContextTitle( $this->mTitle );
4400                 $ig->setShowBytes( false );
4401                 $ig->setShowFilename( false );
4402                 $ig->setParser( $this );
4403                 $ig->setHideBadImages();
4404                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4405                 $ig->useSkin( $this->mOptions->getSkin() );
4406                 $ig->mRevisionId = $this->mRevisionId;
4407
4408                 if( isset( $params['caption'] ) ) {
4409                         $caption = $params['caption'];
4410                         $caption = htmlspecialchars( $caption );
4411                         $caption = $this->replaceInternalLinks( $caption );
4412                         $ig->setCaptionHtml( $caption );
4413                 }
4414                 if( isset( $params['perrow'] ) ) {
4415                         $ig->setPerRow( $params['perrow'] );
4416                 }
4417                 if( isset( $params['widths'] ) ) {
4418                         $ig->setWidths( $params['widths'] );
4419                 }
4420                 if( isset( $params['heights'] ) ) {
4421                         $ig->setHeights( $params['heights'] );
4422                 }
4423
4424                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4425
4426                 $lines = StringUtils::explode( "\n", $text );
4427                 foreach ( $lines as $line ) {
4428                         # match lines like these:
4429                         # Image:someimage.jpg|This is some image
4430                         $matches = array();
4431                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4432                         # Skip empty lines
4433                         if ( count( $matches ) == 0 ) {
4434                                 continue;
4435                         }
4436
4437                         if ( strpos( $matches[0], '%' ) !== false )
4438                                 $matches[1] = urldecode( $matches[1] );
4439                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4440                         $nt =& $tp;
4441                         if( is_null( $nt ) ) {
4442                                 # Bogus title. Ignore these so we don't bomb out later.
4443                                 continue;
4444                         }
4445                         if ( isset( $matches[3] ) ) {
4446                                 $label = $matches[3];
4447                         } else {
4448                                 $label = '';
4449                         }
4450
4451                         $html = $this->recursiveTagParse( trim( $label ) );
4452
4453                         $ig->add( $nt, $html );
4454
4455                         # Only add real images (bug #5586)
4456                         if ( $nt->getNamespace() == NS_FILE ) {
4457                                 $this->mOutput->addImage( $nt->getDBkey() );
4458                         }
4459                 }
4460                 return $ig->toHTML();
4461         }
4462
4463         function getImageParams( $handler ) {
4464                 if ( $handler ) {
4465                         $handlerClass = get_class( $handler );
4466                 } else {
4467                         $handlerClass = '';
4468                 }
4469                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4470                         // Initialise static lists
4471                         static $internalParamNames = array(
4472                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4473                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4474                                         'bottom', 'text-bottom' ),
4475                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4476                                         'upright', 'border', 'link', 'alt' ),
4477                         );
4478                         static $internalParamMap;
4479                         if ( !$internalParamMap ) {
4480                                 $internalParamMap = array();
4481                                 foreach ( $internalParamNames as $type => $names ) {
4482                                         foreach ( $names as $name ) {
4483                                                 $magicName = str_replace( '-', '_', "img_$name" );
4484                                                 $internalParamMap[$magicName] = array( $type, $name );
4485                                         }
4486                                 }
4487                         }
4488
4489                         // Add handler params
4490                         $paramMap = $internalParamMap;
4491                         if ( $handler ) {
4492                                 $handlerParamMap = $handler->getParamMap();
4493                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4494                                         $paramMap[$magic] = array( 'handler', $paramName );
4495                                 }
4496                         }
4497                         $this->mImageParams[$handlerClass] = $paramMap;
4498                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4499                 }
4500                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4501         }
4502
4503         /**
4504          * Parse image options text and use it to make an image
4505          * @param Title $title
4506          * @param string $options
4507          * @param LinkHolderArray $holders
4508          */
4509         function makeImage( $title, $options, $holders = false ) {
4510                 # Check if the options text is of the form "options|alt text"
4511                 # Options are:
4512                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4513                 #  * left       no resizing, just left align. label is used for alt= only
4514                 #  * right      same, but right aligned
4515                 #  * none       same, but not aligned
4516                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4517                 #  * center     center the image
4518                 #  * frame      Keep original image size, no magnify-button.
4519                 #  * framed     Same as "frame"
4520                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4521                 #  * upright    reduce width for upright images, rounded to full __0 px
4522                 #  * border     draw a 1px border around the image
4523                 #  * alt        Text for HTML alt attribute (defaults to empty)
4524                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4525                 # vertical-align values (no % or length right now):
4526                 #  * baseline
4527                 #  * sub
4528                 #  * super
4529                 #  * top
4530                 #  * text-top
4531                 #  * middle
4532                 #  * bottom
4533                 #  * text-bottom
4534
4535                 $parts = StringUtils::explode( "|", $options );
4536                 $sk = $this->mOptions->getSkin();
4537
4538                 # Give extensions a chance to select the file revision for us
4539                 $skip = $time = $descQuery = false;
4540                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4541
4542                 if ( $skip ) {
4543                         return $sk->link( $title );
4544                 }
4545
4546                 # Get the file
4547                 $imagename = $title->getDBkey();
4548                 $file = wfFindFile( $title, array( 'time' => $time ) );
4549                 # Get parameter map
4550                 $handler = $file ? $file->getHandler() : false;
4551
4552                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4553
4554                 # Process the input parameters
4555                 $caption = '';
4556                 $params = array( 'frame' => array(), 'handler' => array(),
4557                         'horizAlign' => array(), 'vertAlign' => array() );
4558                 foreach( $parts as $part ) {
4559                         $part = trim( $part );
4560                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4561                         $validated = false;
4562                         if( isset( $paramMap[$magicName] ) ) {
4563                                 list( $type, $paramName ) = $paramMap[$magicName];
4564
4565                                 // Special case; width and height come in one variable together
4566                                 if( $type === 'handler' && $paramName === 'width' ) {
4567                                         $m = array();
4568                                         # (bug 13500) In both cases (width/height and width only),
4569                                         # permit trailing "px" for backward compatibility.
4570                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4571                                                 $width = intval( $m[1] );
4572                                                 $height = intval( $m[2] );
4573                                                 if ( $handler->validateParam( 'width', $width ) ) {
4574                                                         $params[$type]['width'] = $width;
4575                                                         $validated = true;
4576                                                 }
4577                                                 if ( $handler->validateParam( 'height', $height ) ) {
4578                                                         $params[$type]['height'] = $height;
4579                                                         $validated = true;
4580                                                 }
4581                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4582                                                 $width = intval( $value );
4583                                                 if ( $handler->validateParam( 'width', $width ) ) {
4584                                                         $params[$type]['width'] = $width;
4585                                                         $validated = true;
4586                                                 }
4587                                         } // else no validation -- bug 13436
4588                                 } else {
4589                                         if ( $type === 'handler' ) {
4590                                                 # Validate handler parameter
4591                                                 $validated = $handler->validateParam( $paramName, $value );
4592                                         } else {
4593                                                 # Validate internal parameters
4594                                                 switch( $paramName ) {
4595                                                 case 'manualthumb':
4596                                                 case 'alt':
4597                                                         // @todo Fixme: possibly check validity here for
4598                                                         // manualthumb? downstream behavior seems odd with
4599                                                         // missing manual thumbs.
4600                                                         $validated = true;
4601                                                         $value = $this->stripAltText( $value, $holders );
4602                                                         break;
4603                                                 case 'link':
4604                                                         $chars = self::EXT_LINK_URL_CLASS;
4605                                                         $prots = $this->mUrlProtocols;
4606                                                         if ( $value === '' ) {
4607                                                                 $paramName = 'no-link';
4608                                                                 $value = true;
4609                                                                 $validated = true;
4610                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4611                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4612                                                                         $paramName = 'link-url';
4613                                                                         $this->mOutput->addExternalLink( $value );
4614                                                                         $validated = true;
4615                                                                 }
4616                                                         } else {
4617                                                                 $linkTitle = Title::newFromText( $value );
4618                                                                 if ( $linkTitle ) {
4619                                                                         $paramName = 'link-title';
4620                                                                         $value = $linkTitle;
4621                                                                         $this->mOutput->addLink( $linkTitle );
4622                                                                         $validated = true;
4623                                                                 }
4624                                                         }
4625                                                         break;
4626                                                 default:
4627                                                         // Most other things appear to be empty or numeric...
4628                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4629                                                 }
4630                                         }
4631
4632                                         if ( $validated ) {
4633                                                 $params[$type][$paramName] = $value;
4634                                         }
4635                                 }
4636                         }
4637                         if ( !$validated ) {
4638                                 $caption = $part;
4639                         }
4640                 }
4641
4642                 # Process alignment parameters
4643                 if ( $params['horizAlign'] ) {
4644                         $params['frame']['align'] = key( $params['horizAlign'] );
4645                 }
4646                 if ( $params['vertAlign'] ) {
4647                         $params['frame']['valign'] = key( $params['vertAlign'] );
4648                 }
4649
4650                 $params['frame']['caption'] = $caption;
4651
4652                 # Will the image be presented in a frame, with the caption below?
4653                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4654                                  isset( $params['frame']['framed'] ) ||
4655                                  isset( $params['frame']['thumbnail'] ) ||
4656                                  isset( $params['frame']['manualthumb'] );
4657
4658                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4659                 # came to also set the caption, ordinary text after the image -- which
4660                 # makes no sense, because that just repeats the text multiple times in
4661                 # screen readers.  It *also* came to set the title attribute.
4662                 #
4663                 # Now that we have an alt attribute, we should not set the alt text to
4664                 # equal the caption: that's worse than useless, it just repeats the
4665                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4666                 # use the unnamed parameter for alt text as well, just for the time be-
4667                 # ing, if the unnamed param is set and the alt param is not.
4668                 #
4669                 # For the future, we need to figure out if we want to tweak this more,
4670                 # e.g., introducing a title= parameter for the title; ignoring the un-
4671                 # named parameter entirely for images without a caption; adding an ex-
4672                 # plicit caption= parameter and preserving the old magic unnamed para-
4673                 # meter for BC; ...
4674                 if ( $imageIsFramed ) { # Framed image
4675                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4676                                 # No caption or alt text, add the filename as the alt text so
4677                                 # that screen readers at least get some description of the image
4678                                 $params['frame']['alt'] = $title->getText();
4679                         }
4680                         # Do not set $params['frame']['title'] because tooltips don't make sense
4681                         # for framed images
4682                 } else { # Inline image
4683                         if ( !isset( $params['frame']['alt'] ) ) {
4684                                 # No alt text, use the "caption" for the alt text
4685                                 if ( $caption !== '') {
4686                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4687                                 } else {
4688                                         # No caption, fall back to using the filename for the
4689                                         # alt text
4690                                         $params['frame']['alt'] = $title->getText();
4691                                 }
4692                         }
4693                         # Use the "caption" for the tooltip text
4694                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4695                 }
4696
4697                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4698
4699                 # Linker does the rest
4700                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4701
4702                 # Give the handler a chance to modify the parser object
4703                 if ( $handler ) {
4704                         $handler->parserTransformHook( $this, $file );
4705                 }
4706
4707                 return $ret;
4708         }
4709
4710         protected function stripAltText( $caption, $holders ) {
4711                 # Strip bad stuff out of the title (tooltip).  We can't just use
4712                 # replaceLinkHoldersText() here, because if this function is called
4713                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4714                 if ( $holders ) {
4715                         $tooltip = $holders->replaceText( $caption );
4716                 } else {
4717                         $tooltip = $this->replaceLinkHoldersText( $caption );
4718                 }
4719
4720                 # make sure there are no placeholders in thumbnail attributes
4721                 # that are later expanded to html- so expand them now and
4722                 # remove the tags
4723                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4724                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4725
4726                 return $tooltip;
4727         }
4728
4729         /**
4730          * Set a flag in the output object indicating that the content is dynamic and
4731          * shouldn't be cached.
4732          */
4733         function disableCache() {
4734                 wfDebug( "Parser output marked as uncacheable.\n" );
4735                 $this->mOutput->mCacheTime = -1;
4736         }
4737
4738         /**#@+
4739          * Callback from the Sanitizer for expanding items found in HTML attribute
4740          * values, so they can be safely tested and escaped.
4741          * @param string $text
4742          * @param PPFrame $frame
4743          * @return string
4744          * @private
4745          */
4746         function attributeStripCallback( &$text, $frame = false ) {
4747                 $text = $this->replaceVariables( $text, $frame );
4748                 $text = $this->mStripState->unstripBoth( $text );
4749                 return $text;
4750         }
4751
4752         /**#@-*/
4753
4754         /**#@+
4755          * Accessor/mutator
4756          */
4757         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4758         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4759         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4760         /**#@-*/
4761
4762         /**#@+
4763          * Accessor
4764          */
4765         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4766         /**#@-*/
4767
4768
4769         /**
4770          * Break wikitext input into sections, and either pull or replace
4771          * some particular section's text.
4772          *
4773          * External callers should use the getSection and replaceSection methods.
4774          *
4775          * @param string $text Page wikitext
4776          * @param string $section A section identifier string of the form:
4777          *   <flag1> - <flag2> - ... - <section number>
4778          *
4779          * Currently the only recognised flag is "T", which means the target section number
4780          * was derived during a template inclusion parse, in other words this is a template
4781          * section edit link. If no flags are given, it was an ordinary section edit link.
4782          * This flag is required to avoid a section numbering mismatch when a section is
4783          * enclosed by <includeonly> (bug 6563).
4784          *
4785          * The section number 0 pulls the text before the first heading; other numbers will
4786          * pull the given section along with its lower-level subsections. If the section is
4787          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4788          *
4789          * @param string $mode One of "get" or "replace"
4790          * @param string $newText Replacement text for section data.
4791          * @return string for "get", the extracted section text.
4792          *                for "replace", the whole page with the section replaced.
4793          */
4794         private function extractSections( $text, $section, $mode, $newText='' ) {
4795                 global $wgTitle;
4796                 $this->clearState();
4797                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4798                 $this->mOptions = new ParserOptions;
4799                 $this->setOutputType( self::OT_WIKI );
4800                 $outText = '';
4801                 $frame = $this->getPreprocessor()->newFrame();
4802
4803                 // Process section extraction flags
4804                 $flags = 0;
4805                 $sectionParts = explode( '-', $section );
4806                 $sectionIndex = array_pop( $sectionParts );
4807                 foreach ( $sectionParts as $part ) {
4808                         if ( $part === 'T' ) {
4809                                 $flags |= self::PTD_FOR_INCLUSION;
4810                         }
4811                 }
4812                 // Preprocess the text
4813                 $root = $this->preprocessToDom( $text, $flags );
4814
4815                 // <h> nodes indicate section breaks
4816                 // They can only occur at the top level, so we can find them by iterating the root's children
4817                 $node = $root->getFirstChild();
4818
4819                 // Find the target section
4820                 if ( $sectionIndex == 0 ) {
4821                         // Section zero doesn't nest, level=big
4822                         $targetLevel = 1000;
4823                 } else {
4824             while ( $node ) {
4825                 if ( $node->getName() === 'h' ) {
4826                     $bits = $node->splitHeading();
4827                                         if ( $bits['i'] == $sectionIndex ) {
4828                                         $targetLevel = $bits['level'];
4829                                                 break;
4830                                         }
4831                                 }
4832                                 if ( $mode === 'replace' ) {
4833                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4834                                 }
4835                                 $node = $node->getNextSibling();
4836                         }
4837                 }
4838
4839                 if ( !$node ) {
4840                         // Not found
4841                         if ( $mode === 'get' ) {
4842                                 return $newText;
4843                         } else {
4844                                 return $text;
4845                         }
4846                 }
4847
4848                 // Find the end of the section, including nested sections
4849                 do {
4850                         if ( $node->getName() === 'h' ) {
4851                                 $bits = $node->splitHeading();
4852                                 $curLevel = $bits['level'];
4853                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4854                                         break;
4855                                 }
4856                         }
4857                         if ( $mode === 'get' ) {
4858                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4859                         }
4860                         $node = $node->getNextSibling();
4861                 } while ( $node );
4862
4863                 // Write out the remainder (in replace mode only)
4864                 if ( $mode === 'replace' ) {
4865                         // Output the replacement text
4866                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4867                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4868                         // Only add trailing whitespace if there is newText
4869                         if($newText != "") {
4870                                 $outText .= $newText . "\n\n";
4871                         }
4872
4873                         while ( $node ) {
4874                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4875                                 $node = $node->getNextSibling();
4876                         }
4877                 }
4878
4879                 if ( is_string( $outText ) ) {
4880                         // Re-insert stripped tags
4881                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4882                 }
4883
4884                 return $outText;
4885         }
4886
4887         /**
4888          * This function returns the text of a section, specified by a number ($section).
4889          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4890          * the first section before any such heading (section 0).
4891          *
4892          * If a section contains subsections, these are also returned.
4893          *
4894          * @param string $text text to look in
4895          * @param string $section section identifier
4896          * @param string $deftext default to return if section is not found
4897          * @return string text of the requested section
4898          */
4899         public function getSection( $text, $section, $deftext='' ) {
4900                 return $this->extractSections( $text, $section, "get", $deftext );
4901         }
4902
4903         public function replaceSection( $oldtext, $section, $text ) {
4904                 return $this->extractSections( $oldtext, $section, "replace", $text );
4905         }
4906
4907         /**
4908          * Get the timestamp associated with the current revision, adjusted for
4909          * the default server-local timestamp
4910          */
4911         function getRevisionTimestamp() {
4912                 if ( is_null( $this->mRevisionTimestamp ) ) {
4913                         wfProfileIn( __METHOD__ );
4914                         global $wgContLang;
4915                         $dbr = wfGetDB( DB_SLAVE );
4916                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4917                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4918
4919                         // Normalize timestamp to internal MW format for timezone processing.
4920                         // This has the added side-effect of replacing a null value with
4921                         // the current time, which gives us more sensible behavior for
4922                         // previews.
4923                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4924
4925                         // The cryptic '' timezone parameter tells to use the site-default
4926                         // timezone offset instead of the user settings.
4927                         //
4928                         // Since this value will be saved into the parser cache, served
4929                         // to other users, and potentially even used inside links and such,
4930                         // it needs to be consistent for all visitors.
4931                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4932
4933                         wfProfileOut( __METHOD__ );
4934                 }
4935                 return $this->mRevisionTimestamp;
4936         }
4937
4938         /**
4939          * Get the name of the user that edited the last revision
4940          */
4941         function getRevisionUser() {
4942                 // if this template is subst: the revision id will be blank,
4943                 // so just use the current user's name
4944                 if( $this->mRevisionId ) {
4945                         $revision = Revision::newFromId( $this->mRevisionId );
4946                         $revuser = $revision->getUserText();
4947                 } else {
4948                         global $wgUser;
4949                         $revuser = $wgUser->getName();
4950                 }
4951                 return $revuser;
4952         }
4953
4954         /**
4955          * Mutator for $mDefaultSort
4956          *
4957          * @param $sort New value
4958          */
4959         public function setDefaultSort( $sort ) {
4960                 $this->mDefaultSort = $sort;
4961         }
4962
4963         /**
4964          * Accessor for $mDefaultSort
4965          * Will use the title/prefixed title if none is set
4966          *
4967          * @return string
4968          */
4969         public function getDefaultSort() {
4970                 global $wgCategoryPrefixedDefaultSortkey;
4971                 if( $this->mDefaultSort !== false ) {
4972                         return $this->mDefaultSort;
4973                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4974                         !$wgCategoryPrefixedDefaultSortkey) {
4975                         return $this->mTitle->getText();
4976                 } else {
4977                         return $this->mTitle->getPrefixedText();
4978                 }
4979         }
4980
4981         /**
4982          * Accessor for $mDefaultSort
4983          * Unlike getDefaultSort(), will return false if none is set
4984          *
4985          * @return string or false
4986          */
4987         public function getCustomDefaultSort() {
4988                 return $this->mDefaultSort;
4989         }
4990
4991         /**
4992          * Try to guess the section anchor name based on a wikitext fragment
4993          * presumably extracted from a heading, for example "Header" from
4994          * "== Header ==".
4995          */
4996         public function guessSectionNameFromWikiText( $text ) {
4997                 # Strip out wikitext links(they break the anchor)
4998                 $text = $this->stripSectionName( $text );
4999                 $headline = Sanitizer::decodeCharReferences( $text );
5000                 # strip out HTML
5001                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5002                 $headline = trim( $headline );
5003                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5004                 $replacearray = array(
5005                         '%3A' => ':',
5006                         '%' => '.'
5007                 );
5008                 return str_replace(
5009                         array_keys( $replacearray ),
5010                         array_values( $replacearray ),
5011                         $sectionanchor );
5012         }
5013
5014         /**
5015          * Strips a text string of wikitext for use in a section anchor
5016          *
5017          * Accepts a text string and then removes all wikitext from the
5018          * string and leaves only the resultant text (i.e. the result of
5019          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5020          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5021          * to create valid section anchors by mimicing the output of the
5022          * parser when headings are parsed.
5023          *
5024          * @param $text string Text string to be stripped of wikitext
5025          * for use in a Section anchor
5026          * @return Filtered text string
5027          */
5028         public function stripSectionName( $text ) {
5029                 # Strip internal link markup
5030                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5031                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5032
5033                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5034                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5035                 # on how many empty links there are on the page - need to figure that out.
5036                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5037
5038                 # Parse wikitext quotes (italics & bold)
5039                 $text = $this->doQuotes($text);
5040
5041                 # Strip HTML tags
5042                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5043                 return $text;
5044         }
5045
5046         function srvus( $text ) {
5047                 return $this->testSrvus( $text, $this->mOutputType );
5048         }
5049
5050         /**
5051          * strip/replaceVariables/unstrip for preprocessor regression testing
5052          */
5053         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5054                 $this->clearState();
5055                 if ( ! ( $title instanceof Title ) ) {
5056                         $title = Title::newFromText( $title );
5057                 }
5058                 $this->mTitle = $title;
5059                 $this->mOptions = $options;
5060                 $this->setOutputType( $outputType );
5061                 $text = $this->replaceVariables( $text );
5062                 $text = $this->mStripState->unstripBoth( $text );
5063                 $text = Sanitizer::removeHTMLtags( $text );
5064                 return $text;
5065         }
5066
5067         function testPst( $text, $title, $options ) {
5068                 global $wgUser;
5069                 if ( ! ( $title instanceof Title ) ) {
5070                         $title = Title::newFromText( $title );
5071                 }
5072                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5073         }
5074
5075         function testPreprocess( $text, $title, $options ) {
5076                 if ( ! ( $title instanceof Title ) ) {
5077                         $title = Title::newFromText( $title );
5078                 }
5079                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5080         }
5081
5082         function markerSkipCallback( $s, $callback ) {
5083                 $i = 0;
5084                 $out = '';
5085                 while ( $i < strlen( $s ) ) {
5086                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5087                         if ( $markerStart === false ) {
5088                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5089                                 break;
5090                         } else {
5091                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5092                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5093                                 if ( $markerEnd === false ) {
5094                                         $out .= substr( $s, $markerStart );
5095                                         break;
5096                                 } else {
5097                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5098                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5099                                         $i = $markerEnd;
5100                                 }
5101                         }
5102                 }
5103                 return $out;
5104         }
5105
5106         function serialiseHalfParsedText( $text ) {
5107                 $data = array();
5108                 $data['text'] = $text;
5109
5110                 // First, find all strip markers, and store their
5111                 //  data in an array.
5112                 $stripState = new StripState;
5113                 $pos = 0;
5114                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
5115                         $end_pos += strlen( self::MARKER_SUFFIX );
5116                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5117
5118                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5119                                 $replaceArray = $stripState->general;
5120                                 $stripText = $this->mStripState->general->data[$marker];
5121                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5122                                 $replaceArray = $stripState->nowiki;
5123                                 $stripText = $this->mStripState->nowiki->data[$marker];
5124                         } else {
5125                                 throw new MWException( "Hanging strip marker: '$marker'." );
5126                         }
5127
5128                         $replaceArray->setPair( $marker, $stripText );
5129                         $pos = $end_pos;
5130                 }
5131                 $data['stripstate'] = $stripState;
5132
5133                 // Now, find all of our links, and store THEIR
5134                 //  data in an array! :)
5135                 $links = array( 'internal' => array(), 'interwiki' => array() );
5136                 $pos = 0;
5137
5138                 // Internal links
5139                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5140                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5141
5142                         $ns = trim($ns);
5143                         if (empty( $links['internal'][$ns] )) {
5144                                 $links['internal'][$ns] = array();
5145                         }
5146
5147                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5148                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5149                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5150                 }
5151
5152                 $pos = 0;
5153
5154                 // Interwiki links
5155                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5156                         $data = substr( $text, $start_pos );
5157                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5158                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5159                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5160                 }
5161
5162                 $data['linkholder'] = $links;
5163
5164                 return $data;
5165         }
5166
5167         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5168                 if (!$intPrefix)
5169                         $intPrefix = $this->getRandomString();
5170
5171                 // First, extract the strip state.
5172                 $stripState = $data['stripstate'];
5173                 $this->mStripState->general->merge( $stripState->general );
5174                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5175
5176                 // Now, extract the text, and renumber links
5177                 $text = $data['text'];
5178                 $links = $data['linkholder'];
5179
5180                 // Internal...
5181                 foreach( $links['internal'] as $ns => $nsLinks ) {
5182                         foreach( $nsLinks as $key => $entry ) {
5183                                 $newKey = $intPrefix . '-' . $key;
5184                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5185
5186                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5187                         }
5188                 }
5189
5190                 // Interwiki...
5191                 foreach( $links['interwiki'] as $key => $entry ) {
5192                         $newKey = "$intPrefix-$key";
5193                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5194
5195                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5196                 }
5197
5198                 // Should be good to go.
5199                 return $text;
5200         }
5201 }
5202
5203 /**
5204  * @todo document, briefly.
5205  * @ingroup Parser
5206  */
5207 class StripState {
5208         var $general, $nowiki;
5209
5210         function __construct() {
5211                 $this->general = new ReplacementArray;
5212                 $this->nowiki = new ReplacementArray;
5213         }
5214
5215         function unstripGeneral( $text ) {
5216                 wfProfileIn( __METHOD__ );
5217                 do {
5218                         $oldText = $text;
5219                         $text = $this->general->replace( $text );
5220                 } while ( $text !== $oldText );
5221                 wfProfileOut( __METHOD__ );
5222                 return $text;
5223         }
5224
5225         function unstripNoWiki( $text ) {
5226                 wfProfileIn( __METHOD__ );
5227                 do {
5228                         $oldText = $text;
5229                         $text = $this->nowiki->replace( $text );
5230                 } while ( $text !== $oldText );
5231                 wfProfileOut( __METHOD__ );
5232                 return $text;
5233         }
5234
5235         function unstripBoth( $text ) {
5236                 wfProfileIn( __METHOD__ );
5237                 do {
5238                         $oldText = $text;
5239                         $text = $this->general->replace( $text );
5240                         $text = $this->nowiki->replace( $text );
5241                 } while ( $text !== $oldText );
5242                 wfProfileOut( __METHOD__ );
5243                 return $text;
5244         }
5245 }
5246
5247 /**
5248  * @todo document, briefly.
5249  * @ingroup Parser
5250  */
5251 class OnlyIncludeReplacer {
5252         var $output = '';
5253
5254         function replace( $matches ) {
5255                 if ( substr( $matches[1], -1 ) === "\n" ) {
5256                         $this->output .= substr( $matches[1], 0, -1 );
5257                 } else {
5258                         $this->output .= $matches[1];
5259                 }
5260         }
5261 }