includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * preprocess()
  24  *   removes HTML comments and expands templates
  25  * cleanSig()
  26  *   Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *   Extracts sections from an article for section editing
  29  *
  30  * Globals used:
  31  *    objects:   $wgLang, $wgContLang
  32  *
  33  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  34  *
  35  * settings:
  36  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  37  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  38  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  39  *  $wgMaxArticleSize*
  40  *
  41  *  * only within ParserOptions
  42  * </pre>
  43  *
  44  * @ingroup Parser
  45  */
  46 class Parser
  47 {
  48         /**
  49          * Update this version number when the ParserOutput format
  50          * changes in an incompatible way, so the parser cache
  51          * can automatically discard old data.
  52          */
  53         const VERSION = '1.6.4';
  54
  55         # Flags for Parser::setFunctionHook
  56         # Also available as global constants from Defines.php
  57         const SFH_NO_HASH = 1;
  58         const SFH_OBJECT_ARGS = 2;
  59
  60         # Constants needed for external link processing
  61         # Everything except bracket, space, or control characters
  62         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  63         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  64                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  65
  66         // State constants for the definition list colon extraction
  67         const COLON_STATE_TEXT = 0;
  68         const COLON_STATE_TAG = 1;
  69         const COLON_STATE_TAGSTART = 2;
  70         const COLON_STATE_CLOSETAG = 3;
  71         const COLON_STATE_TAGSLASH = 4;
  72         const COLON_STATE_COMMENT = 5;
  73         const COLON_STATE_COMMENTDASH = 6;
  74         const COLON_STATE_COMMENTDASHDASH = 7;
  75
  76         // Flags for preprocessToDom
  77         const PTD_FOR_INCLUSION = 1;
  78
  79         // Allowed values for $this->mOutputType
  80         // Parameter to startExternalParse().
  81         const OT_HTML = 1;
  82         const OT_WIKI = 2;
  83         const OT_PREPROCESS = 3;
  84         const OT_MSG = 3;
  85
  86         // Marker Suffix needs to be accessible staticly.
  87         const MARKER_SUFFIX = "-QINU\x7f";
  88
  89         /**#@+
  90          * @private
  91          */
  92         # Persistent:
  93         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  94                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex, $mPreprocessor,
  95                 $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList, $mVarCache, $mConf;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mLinkHolders, $mLinkID;
 102         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 103         var $mTplExpandCache; // empty-frame expansion cache
 104         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 105         var $mExpensiveFunctionCount; // number of expensive parser function calls
 106         var $mFileCache;
 107
 108         # Temporary
 109         # These are variables reset at least once per parse regardless of $clearState
 110         var $mOptions,      // ParserOptions object
 111                 $mTitle,        // Title context, used for self-link rendering and similar things
 112                 $mOutputType,   // Output type, one of the OT_xxx constants
 113                 $ot,            // Shortcut alias, see setOutputType()
 114                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 115                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 116                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 117
 118         /**#@-*/
 119
 120         /**
 121          * Constructor
 122          *
 123          * @public
 124          */
 125         function __construct( $conf = array() ) {
 126                 $this->mConf = $conf;
 127                 $this->mTagHooks = array();
 128                 $this->mTransparentTagHooks = array();
 129                 $this->mFunctionHooks = array();
 130                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 131                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 132                 $this->mUrlProtocols = wfUrlProtocols();
 133                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 134                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 135                 $this->mVarCache = array();
 136                 if ( isset( $conf['preprocessorClass'] ) ) {
 137                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 138                 } elseif ( extension_loaded( 'domxml' ) ) {
 139                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 140                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 141                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 142                 } elseif ( extension_loaded( 'dom' ) ) {
 143                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 144                 } else {
 145                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 146                 }
 147                 $this->mMarkerIndex = 0;
 148                 $this->mFirstCall = true;
 149         }
 150
 151         /**
 152          * Reduce memory usage to reduce the impact of circular references
 153          */
 154         function __destruct() {
 155                 if ( isset( $this->mLinkHolders ) ) {
 156                         $this->mLinkHolders->__destruct();
 157                 }
 158                 foreach ( $this as $name => $value ) {
 159                         unset( $this->$name );
 160                 }
 161         }
 162
 163         /**
 164          * Do various kinds of initialisation on the first call of the parser
 165          */
 166         function firstCallInit() {
 167                 if ( !$this->mFirstCall ) {
 168                         return;
 169                 }
 170                 $this->mFirstCall = false;
 171
 172                 wfProfileIn( __METHOD__ );
 173
 174                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 175                 CoreParserFunctions::register( $this );
 176                 $this->initialiseVariables();
 177
 178                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 179                 wfProfileOut( __METHOD__ );
 180         }
 181
 182         /**
 183          * Clear Parser state
 184          *
 185          * @private
 186          */
 187         function clearState() {
 188                 wfProfileIn( __METHOD__ );
 189                 if ( $this->mFirstCall ) {
 190                         $this->firstCallInit();
 191                 }
 192                 $this->mOutput = new ParserOutput;
 193                 $this->mAutonumber = 0;
 194                 $this->mLastSection = '';
 195                 $this->mDTopen = false;
 196                 $this->mIncludeCount = array();
 197                 $this->mStripState = new StripState;
 198                 $this->mArgStack = false;
 199                 $this->mInPre = false;
 200                 $this->mLinkHolders = new LinkHolderArray( $this );
 201                 $this->mLinkID = 0;
 202                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 203
 204                 /**
 205                  * Prefix for temporary replacement strings for the multipass parser.
 206                  * \x07 should never appear in input as it's disallowed in XML.
 207                  * Using it at the front also gives us a little extra robustness
 208                  * since it shouldn't match when butted up against identifier-like
 209                  * string constructs.
 210                  *
 211                  * Must not consist of all title characters, or else it will change
 212                  * the behaviour of <nowiki> in a link.
 213                  */
 214                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 215                 # Changed to \x7f to allow XML double-parsing -- TS
 216                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 217
 218
 219                 # Clear these on every parse, bug 4549
 220                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 221
 222                 $this->mShowToc = true;
 223                 $this->mForceTocPosition = false;
 224                 $this->mIncludeSizes = array(
 225                         'post-expand' => 0,
 226                         'arg' => 0,
 227                 );
 228                 $this->mPPNodeCount = 0;
 229                 $this->mDefaultSort = false;
 230                 $this->mHeadings = array();
 231                 $this->mDoubleUnderscores = array();
 232                 $this->mExpensiveFunctionCount = 0;
 233                 $this->mFileCache = array();
 234
 235                 # Fix cloning
 236                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 237                         $this->mPreprocessor = null;
 238                 }
 239
 240                 wfRunHooks( 'ParserClearState', array( &$this ) );
 241                 wfProfileOut( __METHOD__ );
 242         }
 243
 244         function setOutputType( $ot ) {
 245                 $this->mOutputType = $ot;
 246                 // Shortcut alias
 247                 $this->ot = array(
 248                         'html' => $ot == self::OT_HTML,
 249                         'wiki' => $ot == self::OT_WIKI,
 250                         'pre' => $ot == self::OT_PREPROCESS,
 251                 );
 252         }
 253
 254         /**
 255          * Set the context title
 256          */
 257         function setTitle( $t ) {
 258                 if ( !$t || $t instanceof FakeTitle ) {
 259                         $t = Title::newFromText( 'NO TITLE' );
 260                 }
 261                 if ( strval( $t->getFragment() ) !== '' ) {
 262                         # Strip the fragment to avoid various odd effects
 263                         $this->mTitle = clone $t;
 264                         $this->mTitle->setFragment( '' );
 265                 } else {
 266                         $this->mTitle = $t;
 267                 }
 268         }
 269
 270         /**
 271          * Accessor for mUniqPrefix.
 272          *
 273          * @public
 274          */
 275         function uniqPrefix() {
 276                 if( !isset( $this->mUniqPrefix ) ) {
 277                         // @fixme this is probably *horribly wrong*
 278                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 279                         // if this is called for a parser cache hit, the parser may not
 280                         // have ever been initialized in the first place.
 281                         // Not really sure what the heck is supposed to be going on here.
 282                         return '';
 283                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 284                 }
 285                 return $this->mUniqPrefix;
 286         }
 287
 288         /**
 289          * Convert wikitext to HTML
 290          * Do not call this function recursively.
 291          *
 292          * @param $text String: text we want to parse
 293          * @param $title A title object
 294          * @param $options ParserOptions
 295          * @param $linestart boolean
 296          * @param $clearState boolean
 297          * @param $revid Int: number to pass in {{REVISIONID}}
 298          * @return ParserOutput a ParserOutput
 299          */
 300         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 301                 /**
 302                  * First pass--just handle <nowiki> sections, pass the rest off
 303                  * to internalParse() which does all the real work.
 304                  */
 305
 306                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 307                 $fname = __METHOD__.'-' . wfGetCaller();
 308                 wfProfileIn( __METHOD__ );
 309                 wfProfileIn( $fname );
 310
 311                 if ( $clearState ) {
 312                         $this->clearState();
 313                 }
 314
 315                 $this->mOptions = $options;
 316                 $this->setTitle( $title );
 317                 $oldRevisionId = $this->mRevisionId;
 318                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 319                 if( $revid !== null ) {
 320                         $this->mRevisionId = $revid;
 321                         $this->mRevisionTimestamp = null;
 322                 }
 323                 $this->setOutputType( self::OT_HTML );
 324                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 325                 # No more strip!
 326                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 327                 $text = $this->internalParse( $text );
 328                 $text = $this->mStripState->unstripGeneral( $text );
 329
 330                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 331                 $fixtags = array(
 332                         # french spaces, last one Guillemet-left
 333                         # only if there is something before the space
 334                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 335                         # french spaces, Guillemet-right
 336                         '/(\\302\\253) /' => '\\1&nbsp;',
 337                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 338                 );
 339                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 340
 341                 $text = $this->doBlockLevels( $text, $linestart );
 342
 343                 $this->replaceLinkHolders( $text );
 344
 345                 # the position of the parserConvert() call should not be changed. it
 346                 # assumes that the links are all replaced and the only thing left
 347                 # is the <nowiki> mark.
 348                 # Side-effects: this calls $this->mOutput->setTitleText()
 349                 $text = $wgContLang->parserConvert( $text, $this );
 350
 351                 $text = $this->mStripState->unstripNoWiki( $text );
 352
 353                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 354
 355 //!JF Move to its own function
 356
 357                 $uniq_prefix = $this->mUniqPrefix;
 358                 $matches = array();
 359                 $elements = array_keys( $this->mTransparentTagHooks );
 360                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 361
 362                 foreach( $matches as $marker => $data ) {
 363                         list( $element, $content, $params, $tag ) = $data;
 364                         $tagName = strtolower( $element );
 365                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 366                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 367                                         array( $content, $params, $this ) );
 368                         } else {
 369                                 $output = $tag;
 370                         }
 371                         $this->mStripState->general->setPair( $marker, $output );
 372                 }
 373                 $text = $this->mStripState->unstripGeneral( $text );
 374
 375                 $text = Sanitizer::normalizeCharReferences( $text );
 376
 377                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 378                         $text = self::tidy($text);
 379                 } else {
 380                         # attempt to sanitize at least some nesting problems
 381                         # (bug #2702 and quite a few others)
 382                         $tidyregs = array(
 383                                 # ''Something [http://www.cool.com cool''] -->
 384                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 385                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 386                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 387                                 # fix up an anchor inside another anchor, only
 388                                 # at least for a single single nested link (bug 3695)
 389                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 390                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 391                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 392                                 # contains a div, so fix it up here; replace
 393                                 # div with escaped text
 394                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 395                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 396                                 # remove empty italic or bold tag pairs, some
 397                                 # introduced by rules above
 398                                 '/<([bi])><\/\\1>/' => '',
 399                         );
 400
 401                         $text = preg_replace(
 402                                 array_keys( $tidyregs ),
 403                                 array_values( $tidyregs ),
 404                                 $text );
 405                 }
 406                 global $wgExpensiveParserFunctionLimit;
 407                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 408                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 409                 }
 410
 411                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 412
 413                 # Information on include size limits, for the benefit of users who try to skirt them
 414                 if ( $this->mOptions->getEnableLimitReport() ) {
 415                         global $wgExpensiveParserFunctionLimit;
 416                         $max = $this->mOptions->getMaxIncludeSize();
 417                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 418                         $limitReport =
 419                                 "NewPP limit report\n" .
 420                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 421                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 422                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 423                                 $PFreport;
 424                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 425                         $text .= "\n<!-- \n$limitReport-->\n";
 426                 }
 427                 $this->mOutput->setText( $text );
 428                 $this->mRevisionId = $oldRevisionId;
 429                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 430                 wfProfileOut( $fname );
 431                 wfProfileOut( __METHOD__ );
 432
 433                 return $this->mOutput;
 434         }
 435
 436         /**
 437          * Recursive parser entry point that can be called from an extension tag
 438          * hook.
 439          */
 440         function recursiveTagParse( $text ) {
 441                 wfProfileIn( __METHOD__ );
 442                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 443                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 444                 $text = $this->internalParse( $text );
 445                 wfProfileOut( __METHOD__ );
 446                 return $text;
 447         }
 448
 449         /**
 450          * Expand templates and variables in the text, producing valid, static wikitext.
 451          * Also removes comments.
 452          */
 453         function preprocess( $text, $title, $options, $revid = null ) {
 454                 wfProfileIn( __METHOD__ );
 455                 $this->clearState();
 456                 $this->setOutputType( self::OT_PREPROCESS );
 457                 $this->mOptions = $options;
 458                 $this->setTitle( $title );
 459                 if( $revid !== null ) {
 460                         $this->mRevisionId = $revid;
 461                 }
 462                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 463                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 464                 $text = $this->replaceVariables( $text );
 465                 $text = $this->mStripState->unstripBoth( $text );
 466                 wfProfileOut( __METHOD__ );
 467                 return $text;
 468         }
 469
 470         /**
 471          * Get a random string
 472          *
 473          * @private
 474          * @static
 475          */
 476         function getRandomString() {
 477                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 478         }
 479
 480         function &getTitle() { return $this->mTitle; }
 481         function getOptions() { return $this->mOptions; }
 482         function getRevisionId() { return $this->mRevisionId; }
 483         function getOutput() { return $this->mOutput; }
 484         function nextLinkID() { return $this->mLinkID++; }
 485
 486         function getFunctionLang() {
 487                 global $wgLang, $wgContLang;
 488
 489                 $target = $this->mOptions->getTargetLanguage();
 490                 if ( $target !== null ) {
 491                         return $target;
 492                 } else {
 493                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 494                 }
 495         }
 496
 497         /**
 498          * Get a preprocessor object
 499          */
 500         function getPreprocessor() {
 501                 if ( !isset( $this->mPreprocessor ) ) {
 502                         $class = $this->mPreprocessorClass;
 503                         $this->mPreprocessor = new $class( $this );
 504                 }
 505                 return $this->mPreprocessor;
 506         }
 507
 508         /**
 509          * Replaces all occurrences of HTML-style comments and the given tags
 510          * in the text with a random marker and returns the next text. The output
 511          * parameter $matches will be an associative array filled with data in
 512          * the form:
 513          *   'UNIQ-xxxxx' => array(
 514          *     'element',
 515          *     'tag content',
 516          *     array( 'param' => 'x' ),
 517          *     '<element param="x">tag content</element>' ) )
 518          *
 519          * @param $elements list of element names. Comments are always extracted.
 520          * @param $text Source text string.
 521          * @param $uniq_prefix
 522          *
 523          * @public
 524          * @static
 525          */
 526         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 527                 static $n = 1;
 528                 $stripped = '';
 529                 $matches = array();
 530
 531                 $taglist = implode( '|', $elements );
 532                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 533
 534                 while ( '' != $text ) {
 535                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 536                         $stripped .= $p[0];
 537                         if( count( $p ) < 5 ) {
 538                                 break;
 539                         }
 540                         if( count( $p ) > 5 ) {
 541                                 // comment
 542                                 $element    = $p[4];
 543                                 $attributes = '';
 544                                 $close      = '';
 545                                 $inside     = $p[5];
 546                         } else {
 547                                 // tag
 548                                 $element    = $p[1];
 549                                 $attributes = $p[2];
 550                                 $close      = $p[3];
 551                                 $inside     = $p[4];
 552                         }
 553
 554                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 555                         $stripped .= $marker;
 556
 557                         if ( $close === '/>' ) {
 558                                 // Empty element tag, <tag />
 559                                 $content = null;
 560                                 $text = $inside;
 561                                 $tail = null;
 562                         } else {
 563                                 if( $element === '!--' ) {
 564                                         $end = '/(-->)/';
 565                                 } else {
 566                                         $end = "/(<\\/$element\\s*>)/i";
 567                                 }
 568                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 569                                 $content = $q[0];
 570                                 if( count( $q ) < 3 ) {
 571                                         # No end tag -- let it run out to the end of the text.
 572                                         $tail = '';
 573                                         $text = '';
 574                                 } else {
 575                                         $tail = $q[1];
 576                                         $text = $q[2];
 577                                 }
 578                         }
 579
 580                         $matches[$marker] = array( $element,
 581                                 $content,
 582                                 Sanitizer::decodeTagAttributes( $attributes ),
 583                                 "<$element$attributes$close$content$tail" );
 584                 }
 585                 return $stripped;
 586         }
 587
 588         /**
 589          * Get a list of strippable XML-like elements
 590          */
 591         function getStripList() {
 592                 global $wgRawHtml;
 593                 $elements = $this->mStripList;
 594                 if( $wgRawHtml ) {
 595                         $elements[] = 'html';
 596                 }
 597                 if( $this->mOptions->getUseTeX() ) {
 598                         $elements[] = 'math';
 599                 }
 600                 return $elements;
 601         }
 602
 603         /**
 604          * @deprecated use replaceVariables
 605          */
 606         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 607                 return $text;
 608         }
 609
 610         /**
 611          * Restores pre, math, and other extensions removed by strip()
 612          *
 613          * always call unstripNoWiki() after this one
 614          * @private
 615          * @deprecated use $this->mStripState->unstrip()
 616          */
 617         function unstrip( $text, $state ) {
 618                 return $state->unstripGeneral( $text );
 619         }
 620
 621         /**
 622          * Always call this after unstrip() to preserve the order
 623          *
 624          * @private
 625          * @deprecated use $this->mStripState->unstrip()
 626          */
 627         function unstripNoWiki( $text, $state ) {
 628                 return $state->unstripNoWiki( $text );
 629         }
 630
 631         /**
 632          * @deprecated use $this->mStripState->unstripBoth()
 633          */
 634         function unstripForHTML( $text ) {
 635                 return $this->mStripState->unstripBoth( $text );
 636         }
 637
 638         /**
 639          * Add an item to the strip state
 640          * Returns the unique tag which must be inserted into the stripped text
 641          * The tag will be replaced with the original text in unstrip()
 642          *
 643          * @private
 644          */
 645         function insertStripItem( $text ) {
 646                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 647                 $this->mMarkerIndex++;
 648                 $this->mStripState->general->setPair( $rnd, $text );
 649                 return $rnd;
 650         }
 651
 652         /**
 653          * Interface with html tidy, used if $wgUseTidy = true.
 654          * If tidy isn't able to correct the markup, the original will be
 655          * returned in all its glory with a warning comment appended.
 656          *
 657          * Either the external tidy program or the in-process tidy extension
 658          * will be used depending on availability. Override the default
 659          * $wgTidyInternal setting to disable the internal if it's not working.
 660          *
 661          * @param string $text Hideous HTML input
 662          * @return string Corrected HTML output
 663          * @public
 664          * @static
 665          */
 666         function tidy( $text ) {
 667                 global $wgTidyInternal;
 668
 669                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 670 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 671 '<head><title>test</title></head><body>'.$text.'</body></html>';
 672
 673                 # Tidy is known to clobber tabs; convert 'em to entities
 674                 $wrappedtext = str_replace("\t", '&#9;', $wrappedtext);
 675
 676                 if( $wgTidyInternal ) {
 677                         $correctedtext = self::internalTidy( $wrappedtext );
 678                 } else {
 679                         $correctedtext = self::externalTidy( $wrappedtext );
 680                 }
 681                 if( is_null( $correctedtext ) ) {
 682                         wfDebug( "Tidy error detected!\n" );
 683                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 684                 }
 685
 686                 # Convert the tabs back from entities
 687                 $correctedtext = str_replace('&#9;', "\t", $correctedtext);
 688
 689                 return $correctedtext;
 690         }
 691
 692         /**
 693          * Spawn an external HTML tidy process and get corrected markup back from it.
 694          *
 695          * @private
 696          * @static
 697          */
 698         function externalTidy( $text ) {
 699                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 700                 wfProfileIn( __METHOD__ );
 701
 702                 $cleansource = '';
 703                 $opts = ' -utf8';
 704
 705                 $descriptorspec = array(
 706                         0 => array('pipe', 'r'),
 707                         1 => array('pipe', 'w'),
 708                         2 => array('file', wfGetNull(), 'a')
 709                 );
 710                 $pipes = array();
 711                 if( function_exists('proc_open') ) {
 712                         $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 713                         if (is_resource($process)) {
 714                                 // Theoretically, this style of communication could cause a deadlock
 715                                 // here. If the stdout buffer fills up, then writes to stdin could
 716                                 // block. This doesn't appear to happen with tidy, because tidy only
 717                                 // writes to stdout after it's finished reading from stdin. Search
 718                                 // for tidyParseStdin and tidySaveStdout in console/tidy.c
 719                                 fwrite($pipes[0], $text);
 720                                 fclose($pipes[0]);
 721                                 while (!feof($pipes[1])) {
 722                                         $cleansource .= fgets($pipes[1], 1024);
 723                                 }
 724                                 fclose($pipes[1]);
 725                                 proc_close($process);
 726                         }
 727                 }
 728
 729                 wfProfileOut( __METHOD__ );
 730
 731                 if( $cleansource == '' && $text != '') {
 732                         // Some kind of error happened, so we couldn't get the corrected text.
 733                         // Just give up; we'll use the source text and append a warning.
 734                         return null;
 735                 } else {
 736                         return $cleansource;
 737                 }
 738         }
 739
 740         /**
 741          * Use the HTML tidy PECL extension to use the tidy library in-process,
 742          * saving the overhead of spawning a new process.
 743          *
 744          * 'pear install tidy' should be able to compile the extension module.
 745          *
 746          * @private
 747          * @static
 748          */
 749         function internalTidy( $text ) {
 750                 global $wgTidyConf, $IP, $wgDebugTidy;
 751                 wfProfileIn( __METHOD__ );
 752
 753                 $tidy = new tidy;
 754                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 755                 $tidy->cleanRepair();
 756                 if( $tidy->getStatus() == 2 ) {
 757                         // 2 is magic number for fatal error
 758                         // http://www.php.net/manual/en/function.tidy-get-status.php
 759                         $cleansource = null;
 760                 } else {
 761                         $cleansource = tidy_get_output( $tidy );
 762                 }
 763                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 764                         $cleansource .= "<!--\nTidy reports:\n" .
 765                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 766                                 "\n-->";
 767                 }
 768
 769                 wfProfileOut( __METHOD__ );
 770                 return $cleansource;
 771         }
 772
 773         /**
 774          * parse the wiki syntax used to render tables
 775          *
 776          * @private
 777          */
 778         function doTableStuff ( $text ) {
 779                 wfProfileIn( __METHOD__ );
 780
 781                 $lines = StringUtils::explode( "\n", $text );
 782                 $out = '';
 783                 $td_history = array (); // Is currently a td tag open?
 784                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 785                 $tr_history = array (); // Is currently a tr tag open?
 786                 $tr_attributes = array (); // history of tr attributes
 787                 $has_opened_tr = array(); // Did this table open a <tr> element?
 788                 $indent_level = 0; // indent level of the table
 789
 790                 foreach ( $lines as $outLine ) {
 791                         $line = trim( $outLine );
 792
 793                         if( $line == '' ) { // empty line, go to next line
 794                                 $out .= $outLine."\n";
 795                                 continue;
 796                         }
 797                         $first_character = $line[0];
 798                         $matches = array();
 799
 800                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 801                                 // First check if we are starting a new table
 802                                 $indent_level = strlen( $matches[1] );
 803
 804                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 805                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 806
 807                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 808                                 array_push ( $td_history , false );
 809                                 array_push ( $last_tag_history , '' );
 810                                 array_push ( $tr_history , false );
 811                                 array_push ( $tr_attributes , '' );
 812                                 array_push ( $has_opened_tr , false );
 813                         } else if ( count ( $td_history ) == 0 ) {
 814                                 // Don't do any of the following
 815                                 $out .= $outLine."\n";
 816                                 continue;
 817                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 818                                 // We are ending a table
 819                                 $line = '</table>' . substr ( $line , 2 );
 820                                 $last_tag = array_pop ( $last_tag_history );
 821
 822                                 if ( !array_pop ( $has_opened_tr ) ) {
 823                                         $line = "<tr><td></td></tr>{$line}";
 824                                 }
 825
 826                                 if ( array_pop ( $tr_history ) ) {
 827                                         $line = "</tr>{$line}";
 828                                 }
 829
 830                                 if ( array_pop ( $td_history ) ) {
 831                                         $line = "</{$last_tag}>{$line}";
 832                                 }
 833                                 array_pop ( $tr_attributes );
 834                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 835                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 836                                 // Now we have a table row
 837                                 $line = preg_replace( '#^\|-+#', '', $line );
 838
 839                                 // Whats after the tag is now only attributes
 840                                 $attributes = $this->mStripState->unstripBoth( $line );
 841                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 842                                 array_pop ( $tr_attributes );
 843                                 array_push ( $tr_attributes , $attributes );
 844
 845                                 $line = '';
 846                                 $last_tag = array_pop ( $last_tag_history );
 847                                 array_pop ( $has_opened_tr );
 848                                 array_push ( $has_opened_tr , true );
 849
 850                                 if ( array_pop ( $tr_history ) ) {
 851                                         $line = '</tr>';
 852                                 }
 853
 854                                 if ( array_pop ( $td_history ) ) {
 855                                         $line = "</{$last_tag}>{$line}";
 856                                 }
 857
 858                                 $outLine = $line;
 859                                 array_push ( $tr_history , false );
 860                                 array_push ( $td_history , false );
 861                                 array_push ( $last_tag_history , '' );
 862                         }
 863                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 864                                 // This might be cell elements, td, th or captions
 865                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 866                                         $first_character = '+';
 867                                         $line = substr ( $line , 1 );
 868                                 }
 869
 870                                 $line = substr ( $line , 1 );
 871
 872                                 if ( $first_character === '!' ) {
 873                                         $line = str_replace ( '!!' , '||' , $line );
 874                                 }
 875
 876                                 // Split up multiple cells on the same line.
 877                                 // FIXME : This can result in improper nesting of tags processed
 878                                 // by earlier parser steps, but should avoid splitting up eg
 879                                 // attribute values containing literal "||".
 880                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 881
 882                                 $outLine = '';
 883
 884                                 // Loop through each table cell
 885                                 foreach ( $cells as $cell )
 886                                 {
 887                                         $previous = '';
 888                                         if ( $first_character !== '+' )
 889                                         {
 890                                                 $tr_after = array_pop ( $tr_attributes );
 891                                                 if ( !array_pop ( $tr_history ) ) {
 892                                                         $previous = "<tr{$tr_after}>\n";
 893                                                 }
 894                                                 array_push ( $tr_history , true );
 895                                                 array_push ( $tr_attributes , '' );
 896                                                 array_pop ( $has_opened_tr );
 897                                                 array_push ( $has_opened_tr , true );
 898                                         }
 899
 900                                         $last_tag = array_pop ( $last_tag_history );
 901
 902                                         if ( array_pop ( $td_history ) ) {
 903                                                 $previous = "</{$last_tag}>{$previous}";
 904                                         }
 905
 906                                         if ( $first_character === '|' ) {
 907                                                 $last_tag = 'td';
 908                                         } else if ( $first_character === '!' ) {
 909                                                 $last_tag = 'th';
 910                                         } else if ( $first_character === '+' ) {
 911                                                 $last_tag = 'caption';
 912                                         } else {
 913                                                 $last_tag = '';
 914                                         }
 915
 916                                         array_push ( $last_tag_history , $last_tag );
 917
 918                                         // A cell could contain both parameters and data
 919                                         $cell_data = explode ( '|' , $cell , 2 );
 920
 921                                         // Bug 553: Note that a '|' inside an invalid link should not
 922                                         // be mistaken as delimiting cell parameters
 923                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 924                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 925                                         } else if ( count ( $cell_data ) == 1 )
 926                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 927                                         else {
 928                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 929                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 930                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 931                                         }
 932
 933                                         $outLine .= $cell;
 934                                         array_push ( $td_history , true );
 935                                 }
 936                         }
 937                         $out .= $outLine . "\n";
 938                 }
 939
 940                 // Closing open td, tr && table
 941                 while ( count ( $td_history ) > 0 )
 942                 {
 943                         if ( array_pop ( $td_history ) ) {
 944                                 $out .= "</td>\n";
 945                         }
 946                         if ( array_pop ( $tr_history ) ) {
 947                                 $out .= "</tr>\n";
 948                         }
 949                         if ( !array_pop ( $has_opened_tr ) ) {
 950                                 $out .= "<tr><td></td></tr>\n" ;
 951                         }
 952
 953                         $out .= "</table>\n";
 954                 }
 955
 956                 // Remove trailing line-ending (b/c)
 957                 if ( substr( $out, -1 ) === "\n" ) {
 958                         $out = substr( $out, 0, -1 );
 959                 }
 960
 961                 // special case: don't return empty table
 962                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 963                         $out = '';
 964                 }
 965
 966                 wfProfileOut( __METHOD__ );
 967
 968                 return $out;
 969         }
 970
 971         /**
 972          * Helper function for parse() that transforms wiki markup into
 973          * HTML. Only called for $mOutputType == self::OT_HTML.
 974          *
 975          * @private
 976          */
 977         function internalParse( $text ) {
 978                 $isMain = true;
 979                 wfProfileIn( __METHOD__ );
 980
 981                 # Hook to suspend the parser in this state
 982                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 983                         wfProfileOut( __METHOD__ );
 984                         return $text ;
 985                 }
 986
 987                 $text = $this->replaceVariables( $text );
 988                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 989                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 990
 991                 // Tables need to come after variable replacement for things to work
 992                 // properly; putting them before other transformations should keep
 993                 // exciting things like link expansions from showing up in surprising
 994                 // places.
 995                 $text = $this->doTableStuff( $text );
 996
 997                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 998
 999                 $text = $this->doDoubleUnderscore( $text );
1000                 $text = $this->doHeadings( $text );
1001                 if($this->mOptions->getUseDynamicDates()) {
1002                         $df = DateFormatter::getInstance();
1003                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1004                 }
1005                 $text = $this->doAllQuotes( $text );
1006                 $text = $this->replaceInternalLinks( $text );
1007                 $text = $this->replaceExternalLinks( $text );
1008
1009                 # replaceInternalLinks may sometimes leave behind
1010                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1011                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1012
1013                 $text = $this->doMagicLinks( $text );
1014                 $text = $this->formatHeadings( $text, $isMain );
1015
1016                 wfProfileOut( __METHOD__ );
1017                 return $text;
1018         }
1019
1020         /**
1021          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1022          * magic external links.
1023          *
1024          * DML
1025          * @private
1026          */
1027         function doMagicLinks( $text ) {
1028                 wfProfileIn( __METHOD__ );
1029                 $prots = $this->mUrlProtocols;
1030                 $urlChar = self::EXT_LINK_URL_CLASS;
1031                 $text = preg_replace_callback(
1032                         '!(?:                           # Start cases
1033                                 (<a.*?</a>) |               # m[1]: Skip link text
1034                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1035                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1036                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1037                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1038                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1039                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1040                                     [0-9Xx]                 # check digit
1041                                     \b)
1042                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1043                 wfProfileOut( __METHOD__ );
1044                 return $text;
1045         }
1046
1047         function magicLinkCallback( $m ) {
1048                 if ( isset( $m[1] ) && strval( $m[1] ) !== '' ) {
1049                         # Skip anchor
1050                         return $m[0];
1051                 } elseif ( isset( $m[2] ) && strval( $m[2] ) !== '' ) {
1052                         # Skip HTML element
1053                         return $m[0];
1054                 } elseif ( isset( $m[3] ) && strval( $m[3] ) !== '' ) {
1055                         # Free external link
1056                         return $this->makeFreeExternalLink( $m[0] );
1057                 } elseif ( isset( $m[4] ) && strval( $m[4] ) !== '' ) {
1058                         # RFC or PMID
1059                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1060                                 $keyword = 'RFC';
1061                                 $urlmsg = 'rfcurl';
1062                                 $id = $m[4];
1063                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1064                                 $keyword = 'PMID';
1065                                 $urlmsg = 'pubmedurl';
1066                                 $id = $m[4];
1067                         } else {
1068                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1069                                         substr($m[0], 0, 20 ) . '"' );
1070                         }
1071                         $url = wfMsg( $urlmsg, $id);
1072                         $sk = $this->mOptions->getSkin();
1073                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1074                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1075                 } elseif ( isset( $m[5] ) && strval( $m[5] ) !== '' ) {
1076                         # ISBN
1077                         $isbn = $m[5];
1078                         $num = strtr( $isbn, array(
1079                                 '-' => '',
1080                                 ' ' => '',
1081                                 'x' => 'X',
1082                         ));
1083                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1084                         return'<a href="' .
1085                                 $titleObj->escapeLocalUrl() .
1086                                 "\" class=\"internal\">ISBN $isbn</a>";
1087                 } else {
1088                         return $m[0];
1089                 }
1090         }
1091
1092         /**
1093          * Make a free external link, given a user-supplied URL
1094          * @return HTML
1095          * @private
1096          */
1097         function makeFreeExternalLink( $url ) {
1098                 global $wgContLang;
1099                 wfProfileIn( __METHOD__ );
1100
1101                 $sk = $this->mOptions->getSkin();
1102                 $trail = '';
1103
1104                 # The characters '<' and '>' (which were escaped by
1105                 # removeHTMLtags()) should not be included in
1106                 # URLs, per RFC 2396.
1107                 $m2 = array();
1108                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1109                         $trail = substr($url, $m2[0][1]) . $trail;
1110                         $url = substr($url, 0, $m2[0][1]);
1111                 }
1112
1113                 # Move trailing punctuation to $trail
1114                 $sep = ',;\.:!?';
1115                 # If there is no left bracket, then consider right brackets fair game too
1116                 if ( strpos( $url, '(' ) === false ) {
1117                         $sep .= ')';
1118                 }
1119
1120                 $numSepChars = strspn( strrev( $url ), $sep );
1121                 if ( $numSepChars ) {
1122                         $trail = substr( $url, -$numSepChars ) . $trail;
1123                         $url = substr( $url, 0, -$numSepChars );
1124                 }
1125
1126                 $url = Sanitizer::cleanUrl( $url );
1127
1128                 # Is this an external image?
1129                 $text = $this->maybeMakeExternalImage( $url );
1130                 if ( $text === false ) {
1131                         # Not an image, make a link
1132                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1133                                 $this->getExternalLinkAttribs() );
1134                         # Register it in the output object...
1135                         # Replace unnecessary URL escape codes with their equivalent characters
1136                         $pasteurized = self::replaceUnusualEscapes( $url );
1137                         $this->mOutput->addExternalLink( $pasteurized );
1138                 }
1139                 wfProfileOut( __METHOD__ );
1140                 return $text . $trail;
1141         }
1142
1143
1144         /**
1145          * Parse headers and return html
1146          *
1147          * @private
1148          */
1149         function doHeadings( $text ) {
1150                 wfProfileIn( __METHOD__ );
1151                 for ( $i = 6; $i >= 1; --$i ) {
1152                         $h = str_repeat( '=', $i );
1153                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1154                           "<h$i>\\1</h$i>", $text );
1155                 }
1156                 wfProfileOut( __METHOD__ );
1157                 return $text;
1158         }
1159
1160         /**
1161          * Replace single quotes with HTML markup
1162          * @private
1163          * @return string the altered text
1164          */
1165         function doAllQuotes( $text ) {
1166                 wfProfileIn( __METHOD__ );
1167                 $outtext = '';
1168                 $lines = StringUtils::explode( "\n", $text );
1169                 foreach ( $lines as $line ) {
1170                         $outtext .= $this->doQuotes( $line ) . "\n";
1171                 }
1172                 $outtext = substr($outtext, 0,-1);
1173                 wfProfileOut( __METHOD__ );
1174                 return $outtext;
1175         }
1176
1177         /**
1178          * Helper function for doAllQuotes()
1179          */
1180         public function doQuotes( $text ) {
1181                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1182                 if ( count( $arr ) == 1 )
1183                         return $text;
1184                 else
1185                 {
1186                         # First, do some preliminary work. This may shift some apostrophes from
1187                         # being mark-up to being text. It also counts the number of occurrences
1188                         # of bold and italics mark-ups.
1189                         $i = 0;
1190                         $numbold = 0;
1191                         $numitalics = 0;
1192                         foreach ( $arr as $r )
1193                         {
1194                                 if ( ( $i % 2 ) == 1 )
1195                                 {
1196                                         # If there are ever four apostrophes, assume the first is supposed to
1197                                         # be text, and the remaining three constitute mark-up for bold text.
1198                                         if ( strlen( $arr[$i] ) == 4 )
1199                                         {
1200                                                 $arr[$i-1] .= "'";
1201                                                 $arr[$i] = "'''";
1202                                         }
1203                                         # If there are more than 5 apostrophes in a row, assume they're all
1204                                         # text except for the last 5.
1205                                         else if ( strlen( $arr[$i] ) > 5 )
1206                                         {
1207                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1208                                                 $arr[$i] = "'''''";
1209                                         }
1210                                         # Count the number of occurrences of bold and italics mark-ups.
1211                                         # We are not counting sequences of five apostrophes.
1212                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1213                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1214                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1215                                 }
1216                                 $i++;
1217                         }
1218
1219                         # If there is an odd number of both bold and italics, it is likely
1220                         # that one of the bold ones was meant to be an apostrophe followed
1221                         # by italics. Which one we cannot know for certain, but it is more
1222                         # likely to be one that has a single-letter word before it.
1223                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1224                         {
1225                                 $i = 0;
1226                                 $firstsingleletterword = -1;
1227                                 $firstmultiletterword = -1;
1228                                 $firstspace = -1;
1229                                 foreach ( $arr as $r )
1230                                 {
1231                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1232                                         {
1233                                                 $x1 = substr ($arr[$i-1], -1);
1234                                                 $x2 = substr ($arr[$i-1], -2, 1);
1235                                                 if ($x1 === ' ') {
1236                                                         if ($firstspace == -1) $firstspace = $i;
1237                                                 } else if ($x2 === ' ') {
1238                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1239                                                 } else {
1240                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1241                                                 }
1242                                         }
1243                                         $i++;
1244                                 }
1245
1246                                 # If there is a single-letter word, use it!
1247                                 if ($firstsingleletterword > -1)
1248                                 {
1249                                         $arr [ $firstsingleletterword ] = "''";
1250                                         $arr [ $firstsingleletterword-1 ] .= "'";
1251                                 }
1252                                 # If not, but there's a multi-letter word, use that one.
1253                                 else if ($firstmultiletterword > -1)
1254                                 {
1255                                         $arr [ $firstmultiletterword ] = "''";
1256                                         $arr [ $firstmultiletterword-1 ] .= "'";
1257                                 }
1258                                 # ... otherwise use the first one that has neither.
1259                                 # (notice that it is possible for all three to be -1 if, for example,
1260                                 # there is only one pentuple-apostrophe in the line)
1261                                 else if ($firstspace > -1)
1262                                 {
1263                                         $arr [ $firstspace ] = "''";
1264                                         $arr [ $firstspace-1 ] .= "'";
1265                                 }
1266                         }
1267
1268                         # Now let's actually convert our apostrophic mush to HTML!
1269                         $output = '';
1270                         $buffer = '';
1271                         $state = '';
1272                         $i = 0;
1273                         foreach ($arr as $r)
1274                         {
1275                                 if (($i % 2) == 0)
1276                                 {
1277                                         if ($state === 'both')
1278                                                 $buffer .= $r;
1279                                         else
1280                                                 $output .= $r;
1281                                 }
1282                                 else
1283                                 {
1284                                         if (strlen ($r) == 2)
1285                                         {
1286                                                 if ($state === 'i')
1287                                                 { $output .= '</i>'; $state = ''; }
1288                                                 else if ($state === 'bi')
1289                                                 { $output .= '</i>'; $state = 'b'; }
1290                                                 else if ($state === 'ib')
1291                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1292                                                 else if ($state === 'both')
1293                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1294                                                 else # $state can be 'b' or ''
1295                                                 { $output .= '<i>'; $state .= 'i'; }
1296                                         }
1297                                         else if (strlen ($r) == 3)
1298                                         {
1299                                                 if ($state === 'b')
1300                                                 { $output .= '</b>'; $state = ''; }
1301                                                 else if ($state === 'bi')
1302                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1303                                                 else if ($state === 'ib')
1304                                                 { $output .= '</b>'; $state = 'i'; }
1305                                                 else if ($state === 'both')
1306                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1307                                                 else # $state can be 'i' or ''
1308                                                 { $output .= '<b>'; $state .= 'b'; }
1309                                         }
1310                                         else if (strlen ($r) == 5)
1311                                         {
1312                                                 if ($state === 'b')
1313                                                 { $output .= '</b><i>'; $state = 'i'; }
1314                                                 else if ($state === 'i')
1315                                                 { $output .= '</i><b>'; $state = 'b'; }
1316                                                 else if ($state === 'bi')
1317                                                 { $output .= '</i></b>'; $state = ''; }
1318                                                 else if ($state === 'ib')
1319                                                 { $output .= '</b></i>'; $state = ''; }
1320                                                 else if ($state === 'both')
1321                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1322                                                 else # ($state == '')
1323                                                 { $buffer = ''; $state = 'both'; }
1324                                         }
1325                                 }
1326                                 $i++;
1327                         }
1328                         # Now close all remaining tags.  Notice that the order is important.
1329                         if ($state === 'b' || $state === 'ib')
1330                                 $output .= '</b>';
1331                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1332                                 $output .= '</i>';
1333                         if ($state === 'bi')
1334                                 $output .= '</b>';
1335                         # There might be lonely ''''', so make sure we have a buffer
1336                         if ($state === 'both' && $buffer)
1337                                 $output .= '<b><i>'.$buffer.'</i></b>';
1338                         return $output;
1339                 }
1340         }
1341
1342         /**
1343          * Replace external links (REL)
1344          *
1345          * Note: this is all very hackish and the order of execution matters a lot.
1346          * Make sure to run maintenance/parserTests.php if you change this code.
1347          *
1348          * @private
1349          */
1350         function replaceExternalLinks( $text ) {
1351                 global $wgContLang;
1352                 wfProfileIn( __METHOD__ );
1353
1354                 $sk = $this->mOptions->getSkin();
1355
1356                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1357                 $s = array_shift( $bits );
1358
1359                 $i = 0;
1360                 while ( $i<count( $bits ) ) {
1361                         $url = $bits[$i++];
1362                         $protocol = $bits[$i++];
1363                         $text = $bits[$i++];
1364                         $trail = $bits[$i++];
1365
1366                         # The characters '<' and '>' (which were escaped by
1367                         # removeHTMLtags()) should not be included in
1368                         # URLs, per RFC 2396.
1369                         $m2 = array();
1370                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1371                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1372                                 $url = substr($url, 0, $m2[0][1]);
1373                         }
1374
1375                         # If the link text is an image URL, replace it with an <img> tag
1376                         # This happened by accident in the original parser, but some people used it extensively
1377                         $img = $this->maybeMakeExternalImage( $text );
1378                         if ( $img !== false ) {
1379                                 $text = $img;
1380                         }
1381
1382                         $dtrail = '';
1383
1384                         # Set linktype for CSS - if URL==text, link is essentially free
1385                         $linktype = ($text === $url) ? 'free' : 'text';
1386
1387                         # No link text, e.g. [http://domain.tld/some.link]
1388                         if ( $text == '' ) {
1389                                 # Autonumber if allowed. See bug #5918
1390                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1391                                         $langObj = $this->getFunctionLang();
1392                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1393                                         $linktype = 'autonumber';
1394                                 } else {
1395                                         # Otherwise just use the URL
1396                                         $text = htmlspecialchars( $url );
1397                                         $linktype = 'free';
1398                                 }
1399                         } else {
1400                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1401                                 # Check for trail
1402                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1403                         }
1404
1405                         $text = $wgContLang->markNoConversion($text);
1406
1407                         $url = Sanitizer::cleanUrl( $url );
1408
1409                         if ( $this->mOptions->mExternalLinkTarget ) {
1410                                 $attribs = array( 'target' => $this->mOptions->mExternalLinkTarget );
1411                         } else {
1412                                 $attribs = array();
1413                         }
1414
1415                         # Use the encoded URL
1416                         # This means that users can paste URLs directly into the text
1417                         # Funny characters like &ouml; aren't valid in URLs anyway
1418                         # This was changed in August 2004
1419                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->getExternalLinkAttribs() )
1420                                 . $dtrail . $trail;
1421
1422                         # Register link in the output object.
1423                         # Replace unnecessary URL escape codes with the referenced character
1424                         # This prevents spammers from hiding links from the filters
1425                         $pasteurized = self::replaceUnusualEscapes( $url );
1426                         $this->mOutput->addExternalLink( $pasteurized );
1427                 }
1428
1429                 wfProfileOut( __METHOD__ );
1430                 return $s;
1431         }
1432
1433         function getExternalLinkAttribs() {
1434                 $attribs = array();
1435                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1436                 $ns = $this->mTitle->getNamespace();
1437                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1438                         $attribs['rel'] = 'nofollow';
1439                 }
1440                 if ( $this->mOptions->getExternalLinkTarget() ) {
1441                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1442                 }
1443                 return $attribs;
1444         }
1445
1446
1447         /**
1448          * Replace unusual URL escape codes with their equivalent characters
1449          * @param string
1450          * @return string
1451          * @static
1452          * @todo  This can merge genuinely required bits in the path or query string,
1453          *        breaking legit URLs. A proper fix would treat the various parts of
1454          *        the URL differently; as a workaround, just use the output for
1455          *        statistical records, not for actual linking/output.
1456          */
1457         static function replaceUnusualEscapes( $url ) {
1458                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1459                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1460         }
1461
1462         /**
1463          * Callback function used in replaceUnusualEscapes().
1464          * Replaces unusual URL escape codes with their equivalent character
1465          * @static
1466          * @private
1467          */
1468         private static function replaceUnusualEscapesCallback( $matches ) {
1469                 $char = urldecode( $matches[0] );
1470                 $ord = ord( $char );
1471                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1472                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1473                         // No, shouldn't be escaped
1474                         return $char;
1475                 } else {
1476                         // Yes, leave it escaped
1477                         return $matches[0];
1478                 }
1479         }
1480
1481         /**
1482          * make an image if it's allowed, either through the global
1483          * option, through the exception, or through the on-wiki whitelist
1484          * @private
1485          */
1486         function maybeMakeExternalImage( $url ) {
1487                 $sk = $this->mOptions->getSkin();
1488                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1489                 $imagesexception = !empty($imagesfrom);
1490                 $text = false;
1491                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1492                 if( $imagesexception && is_array( $imagesfrom ) ) {
1493                         $imagematch = false;
1494                         foreach( $imagesfrom as $match ) {
1495                                 if( strpos( $url, $match ) === 0 ) {
1496                                         $imagematch = true;
1497                                         break;
1498                                 }
1499                         }
1500                 } elseif( $imagesexception ) {
1501                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1502                 } else {
1503                         $imagematch = false;
1504                 }
1505                 if ( $this->mOptions->getAllowExternalImages()
1506                      || ( $imagesexception && $imagematch ) ) {
1507                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1508                                 # Image found
1509                                 $text = $sk->makeExternalImage( $url );
1510                         }
1511                 }
1512                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1513                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1514                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1515                         foreach( $whitelist as $entry ) {
1516                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1517                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1518                                         continue;
1519                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1520                                         # Image matches a whitelist entry
1521                                         $text = $sk->makeExternalImage( $url );
1522                                         break;
1523                                 }
1524                         }
1525                 }
1526                 return $text;
1527         }
1528
1529         /**
1530          * Process [[ ]] wikilinks
1531          * @return processed text
1532          *
1533          * @private
1534          */
1535         function replaceInternalLinks( $s ) {
1536                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1537                 return $s;
1538         }
1539
1540         /**
1541          * Process [[ ]] wikilinks (RIL)
1542          * @return LinkHolderArray
1543          *
1544          * @private
1545          */
1546         function replaceInternalLinks2( &$s ) {
1547                 global $wgContLang;
1548
1549                 wfProfileIn( __METHOD__ );
1550
1551                 wfProfileIn( __METHOD__.'-setup' );
1552                 static $tc = FALSE, $e1, $e1_img;
1553                 # the % is needed to support urlencoded titles as well
1554                 if ( !$tc ) {
1555                         $tc = Title::legalChars() . '#%';
1556                         # Match a link having the form [[namespace:link|alternate]]trail
1557                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1558                         # Match cases where there is no "]]", which might still be images
1559                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1560                 }
1561
1562                 $sk = $this->mOptions->getSkin();
1563                 $holders = new LinkHolderArray( $this );
1564
1565                 #split the entire text string on occurences of [[
1566                 $a = StringUtils::explode( '[[', ' ' . $s );
1567                 #get the first element (all text up to first [[), and remove the space we added
1568                 $s = $a->current();
1569                 $a->next();
1570                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1571                 $s = substr( $s, 1 );
1572
1573                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1574                 $e2 = null;
1575                 if ( $useLinkPrefixExtension ) {
1576                         # Match the end of a line for a word that's not followed by whitespace,
1577                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1578                         $e2 = wfMsgForContent( 'linkprefix' );
1579                 }
1580
1581                 if( is_null( $this->mTitle ) ) {
1582                         wfProfileOut( __METHOD__.'-setup' );
1583                         wfProfileOut( __METHOD__ );
1584                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1585                 }
1586                 $nottalk = !$this->mTitle->isTalkPage();
1587
1588                 if ( $useLinkPrefixExtension ) {
1589                         $m = array();
1590                         if ( preg_match( $e2, $s, $m ) ) {
1591                                 $first_prefix = $m[2];
1592                         } else {
1593                                 $first_prefix = false;
1594                         }
1595                 } else {
1596                         $prefix = '';
1597                 }
1598
1599                 if($wgContLang->hasVariants()) {
1600                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1601                 } else {
1602                         $selflink = array($this->mTitle->getPrefixedText());
1603                 }
1604                 $useSubpages = $this->areSubpagesAllowed();
1605                 wfProfileOut( __METHOD__.'-setup' );
1606
1607                 # Loop for each link
1608                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1609                         # Check for excessive memory usage
1610                         if ( $holders->isBig() ) {
1611                                 # Too big
1612                                 # Do the existence check, replace the link holders and clear the array
1613                                 $holders->replace( $s );
1614                                 $holders->clear();
1615                         }
1616
1617                         if ( $useLinkPrefixExtension ) {
1618                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1619                                 if ( preg_match( $e2, $s, $m ) ) {
1620                                         $prefix = $m[2];
1621                                         $s = $m[1];
1622                                 } else {
1623                                         $prefix='';
1624                                 }
1625                                 # first link
1626                                 if($first_prefix) {
1627                                         $prefix = $first_prefix;
1628                                         $first_prefix = false;
1629                                 }
1630                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1631                         }
1632
1633                         $might_be_img = false;
1634
1635                         wfProfileIn( __METHOD__."-e1" );
1636                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1637                                 $text = $m[2];
1638                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1639                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1640                                 # the real problem is with the $e1 regex
1641                                 # See bug 1300.
1642                                 #
1643                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1644                                 # and no image is in sight. See bug 2095.
1645                                 #
1646                                 if( $text !== '' &&
1647                                         substr( $m[3], 0, 1 ) === ']' &&
1648                                         strpos($text, '[') !== false
1649                                 )
1650                                 {
1651                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1652                                         $m[3] = substr( $m[3], 1 );
1653                                 }
1654                                 # fix up urlencoded title texts
1655                                 if( strpos( $m[1], '%' ) !== false ) {
1656                                         # Should anchors '#' also be rejected?
1657                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1658                                 }
1659                                 $trail = $m[3];
1660                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1661                                 $might_be_img = true;
1662                                 $text = $m[2];
1663                                 if ( strpos( $m[1], '%' ) !== false ) {
1664                                         $m[1] = urldecode($m[1]);
1665                                 }
1666                                 $trail = "";
1667                         } else { # Invalid form; output directly
1668                                 $s .= $prefix . '[[' . $line ;
1669                                 wfProfileOut( __METHOD__."-e1" );
1670                                 continue;
1671                         }
1672                         wfProfileOut( __METHOD__."-e1" );
1673                         wfProfileIn( __METHOD__."-misc" );
1674
1675                         # Don't allow internal links to pages containing
1676                         # PROTO: where PROTO is a valid URL protocol; these
1677                         # should be external links.
1678                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1679                                 $s .= $prefix . '[[' . $line ;
1680                                 wfProfileOut( __METHOD__."-misc" );
1681                                 continue;
1682                         }
1683
1684                         # Make subpage if necessary
1685                         if( $useSubpages ) {
1686                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1687                         } else {
1688                                 $link = $m[1];
1689                         }
1690
1691                         $noforce = (substr($m[1], 0, 1) !== ':');
1692                         if (!$noforce) {
1693                                 # Strip off leading ':'
1694                                 $link = substr($link, 1);
1695                         }
1696
1697                         wfProfileOut( __METHOD__."-misc" );
1698                         wfProfileIn( __METHOD__."-title" );
1699                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1700                         if( !$nt ) {
1701                                 $s .= $prefix . '[[' . $line;
1702                                 wfProfileOut( __METHOD__."-title" );
1703                                 continue;
1704                         }
1705
1706                         $ns = $nt->getNamespace();
1707                         $iw = $nt->getInterWiki();
1708                         wfProfileOut( __METHOD__."-title" );
1709
1710                         if ($might_be_img) { # if this is actually an invalid link
1711                                 wfProfileIn( __METHOD__."-might_be_img" );
1712                                 if ($ns == NS_FILE && $noforce) { #but might be an image
1713                                         $found = false;
1714                                         while ( true ) {
1715                                                 #look at the next 'line' to see if we can close it there
1716                                                 $a->next();
1717                                                 $next_line = $a->current();
1718                                                 if ( $next_line === false || $next_line === null ) {
1719                                                         break;
1720                                                 }
1721                                                 $m = explode( ']]', $next_line, 3 );
1722                                                 if ( count( $m ) == 3 ) {
1723                                                         # the first ]] closes the inner link, the second the image
1724                                                         $found = true;
1725                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1726                                                         $trail = $m[2];
1727                                                         break;
1728                                                 } elseif ( count( $m ) == 2 ) {
1729                                                         #if there's exactly one ]] that's fine, we'll keep looking
1730                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1731                                                 } else {
1732                                                         #if $next_line is invalid too, we need look no further
1733                                                         $text .= '[[' . $next_line;
1734                                                         break;
1735                                                 }
1736                                         }
1737                                         if ( !$found ) {
1738                                                 # we couldn't find the end of this imageLink, so output it raw
1739                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1740                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1741                                                 $s .= "{$prefix}[[$link|$text";
1742                                                 # note: no $trail, because without an end, there *is* no trail
1743                                                 wfProfileOut( __METHOD__."-might_be_img" );
1744                                                 continue;
1745                                         }
1746                                 } else { #it's not an image, so output it raw
1747                                         $s .= "{$prefix}[[$link|$text";
1748                                         # note: no $trail, because without an end, there *is* no trail
1749                                         wfProfileOut( __METHOD__."-might_be_img" );
1750                                         continue;
1751                                 }
1752                                 wfProfileOut( __METHOD__."-might_be_img" );
1753                         }
1754
1755                         $wasblank = ( '' == $text );
1756                         if( $wasblank ) $text = $link;
1757
1758                         # Link not escaped by : , create the various objects
1759                         if( $noforce ) {
1760
1761                                 # Interwikis
1762                                 wfProfileIn( __METHOD__."-interwiki" );
1763                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1764                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1765                                         $s = rtrim($s . $prefix);
1766                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1767                                         wfProfileOut( __METHOD__."-interwiki" );
1768                                         continue;
1769                                 }
1770                                 wfProfileOut( __METHOD__."-interwiki" );
1771
1772                                 if ( $ns == NS_FILE ) {
1773                                         wfProfileIn( __METHOD__."-image" );
1774                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1775                                                 # recursively parse links inside the image caption
1776                                                 # actually, this will parse them in any other parameters, too,
1777                                                 # but it might be hard to fix that, and it doesn't matter ATM
1778                                                 $text = $this->replaceExternalLinks($text);
1779                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1780
1781                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1782                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1783                                         }
1784                                         $this->mOutput->addImage( $nt->getDBkey() );
1785                                         wfProfileOut( __METHOD__."-image" );
1786                                         continue;
1787
1788                                 }
1789
1790                                 if ( $ns == NS_CATEGORY ) {
1791                                         wfProfileIn( __METHOD__."-category" );
1792                                         $s = rtrim($s . "\n"); # bug 87
1793
1794                                         if ( $wasblank ) {
1795                                                 $sortkey = $this->getDefaultSort();
1796                                         } else {
1797                                                 $sortkey = $text;
1798                                         }
1799                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1800                                         $sortkey = str_replace( "\n", '', $sortkey );
1801                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1802                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1803
1804                                         /**
1805                                          * Strip the whitespace Category links produce, see bug 87
1806                                          * @todo We might want to use trim($tmp, "\n") here.
1807                                          */
1808                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1809
1810                                         wfProfileOut( __METHOD__."-category" );
1811                                         continue;
1812                                 }
1813                         }
1814
1815                         # Self-link checking
1816                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1817                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1818                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1819                                         continue;
1820                                 }
1821                         }
1822
1823                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1824                         # FIXME: Should do batch file existence checks, see comment below
1825                         if( $ns == NS_MEDIA ) {
1826                                 # Give extensions a chance to select the file revision for us
1827                                 $skip = $time = false;
1828                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1829                                 if ( $skip ) {
1830                                         $link = $sk->link( $nt );
1831                                 } else {
1832                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1833                                 }
1834                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1835                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1836                                 $this->mOutput->addImage( $nt->getDBkey() );
1837                                 continue;
1838                         }
1839
1840                         # Some titles, such as valid special pages or files in foreign repos, should
1841                         # be shown as bluelinks even though they're not included in the page table
1842                         #
1843                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1844                         # batch file existence checks for NS_FILE and NS_MEDIA
1845                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1846                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1847                         } else {
1848                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1849                         }
1850                 }
1851                 wfProfileOut( __METHOD__ );
1852                 return $holders;
1853         }
1854
1855         /**
1856          * Make a link placeholder. The text returned can be later resolved to a real link with
1857          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1858          * parsing of interwiki links, and secondly to allow all existence checks and
1859          * article length checks (for stub links) to be bundled into a single query.
1860          *
1861          * @deprecated
1862          */
1863         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1864                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1865         }
1866
1867         /**
1868          * Render a forced-blue link inline; protect against double expansion of
1869          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1870          * Since this little disaster has to split off the trail text to avoid
1871          * breaking URLs in the following text without breaking trails on the
1872          * wiki links, it's been made into a horrible function.
1873          *
1874          * @param Title $nt
1875          * @param string $text
1876          * @param string $query
1877          * @param string $trail
1878          * @param string $prefix
1879          * @return string HTML-wikitext mix oh yuck
1880          */
1881         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1882                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1883                 $sk = $this->mOptions->getSkin();
1884                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1885                 return $this->armorLinks( $link ) . $trail;
1886         }
1887
1888         /**
1889          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1890          * going to go through further parsing steps before inline URL expansion.
1891          *
1892          * Not needed quite as much as it used to be since free links are a bit
1893          * more sensible these days. But bracketed links are still an issue.
1894          *
1895          * @param string more-or-less HTML
1896          * @return string less-or-more HTML with NOPARSE bits
1897          */
1898         function armorLinks( $text ) {
1899                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1900                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1901         }
1902
1903         /**
1904          * Return true if subpage links should be expanded on this page.
1905          * @return bool
1906          */
1907         function areSubpagesAllowed() {
1908                 # Some namespaces don't allow subpages
1909                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1910         }
1911
1912         /**
1913          * Handle link to subpage if necessary
1914          * @param string $target the source of the link
1915          * @param string &$text the link text, modified as necessary
1916          * @return string the full name of the link
1917          * @private
1918          */
1919         function maybeDoSubpageLink($target, &$text) {
1920                 # Valid link forms:
1921                 # Foobar -- normal
1922                 # :Foobar -- override special treatment of prefix (images, language links)
1923                 # /Foobar -- convert to CurrentPage/Foobar
1924                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1925                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1926                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1927
1928                 wfProfileIn( __METHOD__ );
1929                 $ret = $target; # default return value is no change
1930
1931                 # Some namespaces don't allow subpages,
1932                 # so only perform processing if subpages are allowed
1933                 if( $this->areSubpagesAllowed() ) {
1934                         $hash = strpos( $target, '#' );
1935                         if( $hash !== false ) {
1936                                 $suffix = substr( $target, $hash );
1937                                 $target = substr( $target, 0, $hash );
1938                         } else {
1939                                 $suffix = '';
1940                         }
1941                         # bug 7425
1942                         $target = trim( $target );
1943                         # Look at the first character
1944                         if( $target != '' && $target{0} === '/' ) {
1945                                 # / at end means we don't want the slash to be shown
1946                                 $m = array();
1947                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1948                                 if( $trailingSlashes ) {
1949                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1950                                 } else {
1951                                         $noslash = substr( $target, 1 );
1952                                 }
1953
1954                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1955                                 if( '' === $text ) {
1956                                         $text = $target . $suffix;
1957                                 } # this might be changed for ugliness reasons
1958                         } else {
1959                                 # check for .. subpage backlinks
1960                                 $dotdotcount = 0;
1961                                 $nodotdot = $target;
1962                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1963                                         ++$dotdotcount;
1964                                         $nodotdot = substr( $nodotdot, 3 );
1965                                 }
1966                                 if($dotdotcount > 0) {
1967                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1968                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1969                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1970                                                 # / at the end means don't show full path
1971                                                 if( substr( $nodotdot, -1, 1 ) === '/' ) {
1972                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1973                                                         if( '' === $text ) {
1974                                                                 $text = $nodotdot . $suffix;
1975                                                         }
1976                                                 }
1977                                                 $nodotdot = trim( $nodotdot );
1978                                                 if( $nodotdot != '' ) {
1979                                                         $ret .= '/' . $nodotdot;
1980                                                 }
1981                                                 $ret .= $suffix;
1982                                         }
1983                                 }
1984                         }
1985                 }
1986
1987                 wfProfileOut( __METHOD__ );
1988                 return $ret;
1989         }
1990
1991         /**#@+
1992          * Used by doBlockLevels()
1993          * @private
1994          */
1995         /* private */ function closeParagraph() {
1996                 $result = '';
1997                 if ( '' != $this->mLastSection ) {
1998                         $result = '</' . $this->mLastSection  . ">\n";
1999                 }
2000                 $this->mInPre = false;
2001                 $this->mLastSection = '';
2002                 return $result;
2003         }
2004         # getCommon() returns the length of the longest common substring
2005         # of both arguments, starting at the beginning of both.
2006         #
2007         /* private */ function getCommon( $st1, $st2 ) {
2008                 $fl = strlen( $st1 );
2009                 $shorter = strlen( $st2 );
2010                 if ( $fl < $shorter ) { $shorter = $fl; }
2011
2012                 for ( $i = 0; $i < $shorter; ++$i ) {
2013                         if ( $st1{$i} != $st2{$i} ) { break; }
2014                 }
2015                 return $i;
2016         }
2017         # These next three functions open, continue, and close the list
2018         # element appropriate to the prefix character passed into them.
2019         #
2020         /* private */ function openList( $char ) {
2021                 $result = $this->closeParagraph();
2022
2023                 if ( '*' === $char ) { $result .= '<ul><li>'; }
2024                 else if ( '#' === $char ) { $result .= '<ol><li>'; }
2025                 else if ( ':' === $char ) { $result .= '<dl><dd>'; }
2026                 else if ( ';' === $char ) {
2027                         $result .= '<dl><dt>';
2028                         $this->mDTopen = true;
2029                 }
2030                 else { $result = '<!-- ERR 1 -->'; }
2031
2032                 return $result;
2033         }
2034
2035         /* private */ function nextItem( $char ) {
2036                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
2037                 else if ( ':' === $char || ';' === $char ) {
2038                         $close = '</dd>';
2039                         if ( $this->mDTopen ) { $close = '</dt>'; }
2040                         if ( ';' === $char ) {
2041                                 $this->mDTopen = true;
2042                                 return $close . '<dt>';
2043                         } else {
2044                                 $this->mDTopen = false;
2045                                 return $close . '<dd>';
2046                         }
2047                 }
2048                 return '<!-- ERR 2 -->';
2049         }
2050
2051         /* private */ function closeList( $char ) {
2052                 if ( '*' === $char ) { $text = '</li></ul>'; }
2053                 else if ( '#' === $char ) { $text = '</li></ol>'; }
2054                 else if ( ':' === $char ) {
2055                         if ( $this->mDTopen ) {
2056                                 $this->mDTopen = false;
2057                                 $text = '</dt></dl>';
2058                         } else {
2059                                 $text = '</dd></dl>';
2060                         }
2061                 }
2062                 else {  return '<!-- ERR 3 -->'; }
2063                 return $text."\n";
2064         }
2065         /**#@-*/
2066
2067         /**
2068          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2069          *
2070          * @private
2071          * @return string the lists rendered as HTML
2072          */
2073         function doBlockLevels( $text, $linestart ) {
2074                 wfProfileIn( __METHOD__ );
2075
2076                 # Parsing through the text line by line.  The main thing
2077                 # happening here is handling of block-level elements p, pre,
2078                 # and making lists from lines starting with * # : etc.
2079                 #
2080                 $textLines = StringUtils::explode( "\n", $text );
2081
2082                 $lastPrefix = $output = '';
2083                 $this->mDTopen = $inBlockElem = false;
2084                 $prefixLength = 0;
2085                 $paragraphStack = false;
2086
2087                 foreach ( $textLines as $oLine ) {
2088                         # Fix up $linestart
2089                         if ( !$linestart ) {
2090                                 $output .= $oLine;
2091                                 $linestart = true;
2092                                 continue;
2093                         }
2094
2095                         $lastPrefixLength = strlen( $lastPrefix );
2096                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2097                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2098                         if ( !$this->mInPre ) {
2099                                 # Multiple prefixes may abut each other for nested lists.
2100                                 $prefixLength = strspn( $oLine, '*#:;' );
2101                                 $prefix = substr( $oLine, 0, $prefixLength );
2102
2103                                 # eh?
2104                                 $prefix2 = str_replace( ';', ':', $prefix );
2105                                 $t = substr( $oLine, $prefixLength );
2106                                 $this->mInPre = (bool)$preOpenMatch;
2107                         } else {
2108                                 # Don't interpret any other prefixes in preformatted text
2109                                 $prefixLength = 0;
2110                                 $prefix = $prefix2 = '';
2111                                 $t = $oLine;
2112                         }
2113
2114                         # List generation
2115                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2116                                 # Same as the last item, so no need to deal with nesting or opening stuff
2117                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2118                                 $paragraphStack = false;
2119
2120                                 if ( substr( $prefix, -1 ) === ';') {
2121                                         # The one nasty exception: definition lists work like this:
2122                                         # ; title : definition text
2123                                         # So we check for : in the remainder text to split up the
2124                                         # title and definition, without b0rking links.
2125                                         $term = $t2 = '';
2126                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2127                                                 $t = $t2;
2128                                                 $output .= $term . $this->nextItem( ':' );
2129                                         }
2130                                 }
2131                         } elseif( $prefixLength || $lastPrefixLength ) {
2132                                 # Either open or close a level...
2133                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2134                                 $paragraphStack = false;
2135
2136                                 while( $commonPrefixLength < $lastPrefixLength ) {
2137                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2138                                         --$lastPrefixLength;
2139                                 }
2140                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2141                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2142                                 }
2143                                 while ( $prefixLength > $commonPrefixLength ) {
2144                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2145                                         $output .= $this->openList( $char );
2146
2147                                         if ( ';' === $char ) {
2148                                                 # FIXME: This is dupe of code above
2149                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2150                                                         $t = $t2;
2151                                                         $output .= $term . $this->nextItem( ':' );
2152                                                 }
2153                                         }
2154                                         ++$commonPrefixLength;
2155                                 }
2156                                 $lastPrefix = $prefix2;
2157                         }
2158                         if( 0 == $prefixLength ) {
2159                                 wfProfileIn( __METHOD__."-paragraph" );
2160                                 # No prefix (not in list)--go to paragraph mode
2161                                 // XXX: use a stack for nestable elements like span, table and div
2162                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2163                                 $closematch = preg_match(
2164                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2165                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2166                                 if ( $openmatch or $closematch ) {
2167                                         $paragraphStack = false;
2168                                         # TODO bug 5718: paragraph closed
2169                                         $output .= $this->closeParagraph();
2170                                         if ( $preOpenMatch and !$preCloseMatch ) {
2171                                                 $this->mInPre = true;
2172                                         }
2173                                         if ( $closematch ) {
2174                                                 $inBlockElem = false;
2175                                         } else {
2176                                                 $inBlockElem = true;
2177                                         }
2178                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2179                                         if ( ' ' == $t{0} and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2180                                                 // pre
2181                                                 if ($this->mLastSection !== 'pre') {
2182                                                         $paragraphStack = false;
2183                                                         $output .= $this->closeParagraph().'<pre>';
2184                                                         $this->mLastSection = 'pre';
2185                                                 }
2186                                                 $t = substr( $t, 1 );
2187                                         } else {
2188                                                 // paragraph
2189                                                 if ( '' == trim($t) ) {
2190                                                         if ( $paragraphStack ) {
2191                                                                 $output .= $paragraphStack.'<br />';
2192                                                                 $paragraphStack = false;
2193                                                                 $this->mLastSection = 'p';
2194                                                         } else {
2195                                                                 if ($this->mLastSection !== 'p' ) {
2196                                                                         $output .= $this->closeParagraph();
2197                                                                         $this->mLastSection = '';
2198                                                                         $paragraphStack = '<p>';
2199                                                                 } else {
2200                                                                         $paragraphStack = '</p><p>';
2201                                                                 }
2202                                                         }
2203                                                 } else {
2204                                                         if ( $paragraphStack ) {
2205                                                                 $output .= $paragraphStack;
2206                                                                 $paragraphStack = false;
2207                                                                 $this->mLastSection = 'p';
2208                                                         } else if ($this->mLastSection !== 'p') {
2209                                                                 $output .= $this->closeParagraph().'<p>';
2210                                                                 $this->mLastSection = 'p';
2211                                                         }
2212                                                 }
2213                                         }
2214                                 }
2215                                 wfProfileOut( __METHOD__."-paragraph" );
2216                         }
2217                         // somewhere above we forget to get out of pre block (bug 785)
2218                         if($preCloseMatch && $this->mInPre) {
2219                                 $this->mInPre = false;
2220                         }
2221                         if ($paragraphStack === false) {
2222                                 $output .= $t."\n";
2223                         }
2224                 }
2225                 while ( $prefixLength ) {
2226                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2227                         --$prefixLength;
2228                 }
2229                 if ( '' != $this->mLastSection ) {
2230                         $output .= '</' . $this->mLastSection . '>';
2231                         $this->mLastSection = '';
2232                 }
2233
2234                 wfProfileOut( __METHOD__ );
2235                 return $output;
2236         }
2237
2238         /**
2239          * Split up a string on ':', ignoring any occurences inside tags
2240          * to prevent illegal overlapping.
2241          * @param string $str the string to split
2242          * @param string &$before set to everything before the ':'
2243          * @param string &$after set to everything after the ':'
2244          * return string the position of the ':', or false if none found
2245          */
2246         function findColonNoLinks($str, &$before, &$after) {
2247                 wfProfileIn( __METHOD__ );
2248
2249                 $pos = strpos( $str, ':' );
2250                 if( $pos === false ) {
2251                         // Nothing to find!
2252                         wfProfileOut( __METHOD__ );
2253                         return false;
2254                 }
2255
2256                 $lt = strpos( $str, '<' );
2257                 if( $lt === false || $lt > $pos ) {
2258                         // Easy; no tag nesting to worry about
2259                         $before = substr( $str, 0, $pos );
2260                         $after = substr( $str, $pos+1 );
2261                         wfProfileOut( __METHOD__ );
2262                         return $pos;
2263                 }
2264
2265                 // Ugly state machine to walk through avoiding tags.
2266                 $state = self::COLON_STATE_TEXT;
2267                 $stack = 0;
2268                 $len = strlen( $str );
2269                 for( $i = 0; $i < $len; $i++ ) {
2270                         $c = $str{$i};
2271
2272                         switch( $state ) {
2273                         // (Using the number is a performance hack for common cases)
2274                         case 0: // self::COLON_STATE_TEXT:
2275                                 switch( $c ) {
2276                                 case "<":
2277                                         // Could be either a <start> tag or an </end> tag
2278                                         $state = self::COLON_STATE_TAGSTART;
2279                                         break;
2280                                 case ":":
2281                                         if( $stack == 0 ) {
2282                                                 // We found it!
2283                                                 $before = substr( $str, 0, $i );
2284                                                 $after = substr( $str, $i + 1 );
2285                                                 wfProfileOut( __METHOD__ );
2286                                                 return $i;
2287                                         }
2288                                         // Embedded in a tag; don't break it.
2289                                         break;
2290                                 default:
2291                                         // Skip ahead looking for something interesting
2292                                         $colon = strpos( $str, ':', $i );
2293                                         if( $colon === false ) {
2294                                                 // Nothing else interesting
2295                                                 wfProfileOut( __METHOD__ );
2296                                                 return false;
2297                                         }
2298                                         $lt = strpos( $str, '<', $i );
2299                                         if( $stack === 0 ) {
2300                                                 if( $lt === false || $colon < $lt ) {
2301                                                         // We found it!
2302                                                         $before = substr( $str, 0, $colon );
2303                                                         $after = substr( $str, $colon + 1 );
2304                                                         wfProfileOut( __METHOD__ );
2305                                                         return $i;
2306                                                 }
2307                                         }
2308                                         if( $lt === false ) {
2309                                                 // Nothing else interesting to find; abort!
2310                                                 // We're nested, but there's no close tags left. Abort!
2311                                                 break 2;
2312                                         }
2313                                         // Skip ahead to next tag start
2314                                         $i = $lt;
2315                                         $state = self::COLON_STATE_TAGSTART;
2316                                 }
2317                                 break;
2318                         case 1: // self::COLON_STATE_TAG:
2319                                 // In a <tag>
2320                                 switch( $c ) {
2321                                 case ">":
2322                                         $stack++;
2323                                         $state = self::COLON_STATE_TEXT;
2324                                         break;
2325                                 case "/":
2326                                         // Slash may be followed by >?
2327                                         $state = self::COLON_STATE_TAGSLASH;
2328                                         break;
2329                                 default:
2330                                         // ignore
2331                                 }
2332                                 break;
2333                         case 2: // self::COLON_STATE_TAGSTART:
2334                                 switch( $c ) {
2335                                 case "/":
2336                                         $state = self::COLON_STATE_CLOSETAG;
2337                                         break;
2338                                 case "!":
2339                                         $state = self::COLON_STATE_COMMENT;
2340                                         break;
2341                                 case ">":
2342                                         // Illegal early close? This shouldn't happen D:
2343                                         $state = self::COLON_STATE_TEXT;
2344                                         break;
2345                                 default:
2346                                         $state = self::COLON_STATE_TAG;
2347                                 }
2348                                 break;
2349                         case 3: // self::COLON_STATE_CLOSETAG:
2350                                 // In a </tag>
2351                                 if( $c === ">" ) {
2352                                         $stack--;
2353                                         if( $stack < 0 ) {
2354                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2355                                                 wfProfileOut( __METHOD__ );
2356                                                 return false;
2357                                         }
2358                                         $state = self::COLON_STATE_TEXT;
2359                                 }
2360                                 break;
2361                         case self::COLON_STATE_TAGSLASH:
2362                                 if( $c === ">" ) {
2363                                         // Yes, a self-closed tag <blah/>
2364                                         $state = self::COLON_STATE_TEXT;
2365                                 } else {
2366                                         // Probably we're jumping the gun, and this is an attribute
2367                                         $state = self::COLON_STATE_TAG;
2368                                 }
2369                                 break;
2370                         case 5: // self::COLON_STATE_COMMENT:
2371                                 if( $c === "-" ) {
2372                                         $state = self::COLON_STATE_COMMENTDASH;
2373                                 }
2374                                 break;
2375                         case self::COLON_STATE_COMMENTDASH:
2376                                 if( $c === "-" ) {
2377                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2378                                 } else {
2379                                         $state = self::COLON_STATE_COMMENT;
2380                                 }
2381                                 break;
2382                         case self::COLON_STATE_COMMENTDASHDASH:
2383                                 if( $c === ">" ) {
2384                                         $state = self::COLON_STATE_TEXT;
2385                                 } else {
2386                                         $state = self::COLON_STATE_COMMENT;
2387                                 }
2388                                 break;
2389                         default:
2390                                 throw new MWException( "State machine error in " . __METHOD__ );
2391                         }
2392                 }
2393                 if( $stack > 0 ) {
2394                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2395                         return false;
2396                 }
2397                 wfProfileOut( __METHOD__ );
2398                 return false;
2399         }
2400
2401         /**
2402          * Return value of a magic variable (like PAGENAME)
2403          *
2404          * @private
2405          */
2406         function getVariableValue( $index ) {
2407                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2408
2409                 /**
2410                  * Some of these require message or data lookups and can be
2411                  * expensive to check many times.
2412                  */
2413                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2414                         if ( isset( $this->mVarCache[$index] ) ) {
2415                                 return $this->mVarCache[$index];
2416                         }
2417                 }
2418
2419                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2420                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2421
2422                 # Use the time zone
2423                 global $wgLocaltimezone;
2424                 if ( isset( $wgLocaltimezone ) ) {
2425                         $oldtz = getenv( 'TZ' );
2426                         putenv( 'TZ='.$wgLocaltimezone );
2427                 }
2428
2429                 wfSuppressWarnings(); // E_STRICT system time bitching
2430                 $localTimestamp = date( 'YmdHis', $ts );
2431                 $localMonth = date( 'm', $ts );
2432                 $localMonthName = date( 'n', $ts );
2433                 $localDay = date( 'j', $ts );
2434                 $localDay2 = date( 'd', $ts );
2435                 $localDayOfWeek = date( 'w', $ts );
2436                 $localWeek = date( 'W', $ts );
2437                 $localYear = date( 'Y', $ts );
2438                 $localHour = date( 'H', $ts );
2439                 if ( isset( $wgLocaltimezone ) ) {
2440                         putenv( 'TZ='.$oldtz );
2441                 }
2442                 wfRestoreWarnings();
2443
2444                 switch ( $index ) {
2445                         case 'currentmonth':
2446                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2447                         case 'currentmonthname':
2448                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2449                         case 'currentmonthnamegen':
2450                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2451                         case 'currentmonthabbrev':
2452                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2453                         case 'currentday':
2454                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2455                         case 'currentday2':
2456                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2457                         case 'localmonth':
2458                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2459                         case 'localmonthname':
2460                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2461                         case 'localmonthnamegen':
2462                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2463                         case 'localmonthabbrev':
2464                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2465                         case 'localday':
2466                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2467                         case 'localday2':
2468                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2469                         case 'pagename':
2470                                 return wfEscapeWikiText( $this->mTitle->getText() );
2471                         case 'pagenamee':
2472                                 return $this->mTitle->getPartialURL();
2473                         case 'fullpagename':
2474                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2475                         case 'fullpagenamee':
2476                                 return $this->mTitle->getPrefixedURL();
2477                         case 'subpagename':
2478                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2479                         case 'subpagenamee':
2480                                 return $this->mTitle->getSubpageUrlForm();
2481                         case 'basepagename':
2482                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2483                         case 'basepagenamee':
2484                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2485                         case 'talkpagename':
2486                                 if( $this->mTitle->canTalk() ) {
2487                                         $talkPage = $this->mTitle->getTalkPage();
2488                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2489                                 } else {
2490                                         return '';
2491                                 }
2492                         case 'talkpagenamee':
2493                                 if( $this->mTitle->canTalk() ) {
2494                                         $talkPage = $this->mTitle->getTalkPage();
2495                                         return $talkPage->getPrefixedUrl();
2496                                 } else {
2497                                         return '';
2498                                 }
2499                         case 'subjectpagename':
2500                                 $subjPage = $this->mTitle->getSubjectPage();
2501                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2502                         case 'subjectpagenamee':
2503                                 $subjPage = $this->mTitle->getSubjectPage();
2504                                 return $subjPage->getPrefixedUrl();
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                                 return $this->mRevisionId;
2511                         case 'revisionday':
2512                                 // Let the edit saving system know we should parse the page
2513                                 // *after* a revision ID has been assigned. This is for null edits.
2514                                 $this->mOutput->setFlag( 'vary-revision' );
2515                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2516                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2517                         case 'revisionday2':
2518                                 // Let the edit saving system know we should parse the page
2519                                 // *after* a revision ID has been assigned. This is for null edits.
2520                                 $this->mOutput->setFlag( 'vary-revision' );
2521                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2522                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2523                         case 'revisionmonth':
2524                                 // Let the edit saving system know we should parse the page
2525                                 // *after* a revision ID has been assigned. This is for null edits.
2526                                 $this->mOutput->setFlag( 'vary-revision' );
2527                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2528                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2529                         case 'revisionyear':
2530                                 // Let the edit saving system know we should parse the page
2531                                 // *after* a revision ID has been assigned. This is for null edits.
2532                                 $this->mOutput->setFlag( 'vary-revision' );
2533                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2534                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2535                         case 'revisiontimestamp':
2536                                 // Let the edit saving system know we should parse the page
2537                                 // *after* a revision ID has been assigned. This is for null edits.
2538                                 $this->mOutput->setFlag( 'vary-revision' );
2539                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2540                                 return $this->getRevisionTimestamp();
2541                         case 'namespace':
2542                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2543                         case 'namespacee':
2544                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2545                         case 'talkspace':
2546                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2547                         case 'talkspacee':
2548                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2549                         case 'subjectspace':
2550                                 return $this->mTitle->getSubjectNsText();
2551                         case 'subjectspacee':
2552                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2553                         case 'currentdayname':
2554                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2555                         case 'currentyear':
2556                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2557                         case 'currenttime':
2558                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2559                         case 'currenthour':
2560                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2561                         case 'currentweek':
2562                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2563                                 // int to remove the padding
2564                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2565                         case 'currentdow':
2566                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2567                         case 'localdayname':
2568                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2569                         case 'localyear':
2570                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2571                         case 'localtime':
2572                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2573                         case 'localhour':
2574                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2575                         case 'localweek':
2576                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2577                                 // int to remove the padding
2578                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2579                         case 'localdow':
2580                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2581                         case 'numberofarticles':
2582                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2583                         case 'numberoffiles':
2584                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2585                         case 'numberofusers':
2586                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2587                         case 'numberofpages':
2588                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2589                         case 'numberofadmins':
2590                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2591                         case 'numberofedits':
2592                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2593                         case 'numberofviews':
2594                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::views() );
2595                         case 'currenttimestamp':
2596                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2597                         case 'localtimestamp':
2598                                 return $this->mVarCache[$index] = $localTimestamp;
2599                         case 'currentversion':
2600                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2601                         case 'sitename':
2602                                 return $wgSitename;
2603                         case 'server':
2604                                 return $wgServer;
2605                         case 'servername':
2606                                 return $wgServerName;
2607                         case 'scriptpath':
2608                                 return $wgScriptPath;
2609                         case 'directionmark':
2610                                 return $wgContLang->getDirMark();
2611                         case 'contentlanguage':
2612                                 global $wgContLanguageCode;
2613                                 return $wgContLanguageCode;
2614                         default:
2615                                 $ret = null;
2616                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2617                                         return $ret;
2618                                 else
2619                                         return null;
2620                 }
2621         }
2622
2623         /**
2624          * initialise the magic variables (like CURRENTMONTHNAME)
2625          *
2626          * @private
2627          */
2628         function initialiseVariables() {
2629                 wfProfileIn( __METHOD__ );
2630                 $variableIDs = MagicWord::getVariableIDs();
2631
2632                 $this->mVariables = new MagicWordArray( $variableIDs );
2633                 wfProfileOut( __METHOD__ );
2634         }
2635
2636         /**
2637          * Preprocess some wikitext and return the document tree.
2638          * This is the ghost of replace_variables().
2639          *
2640          * @param string $text The text to parse
2641          * @param integer flags Bitwise combination of:
2642          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2643          *                                     included. Default is to assume a direct page view.
2644          *
2645          * The generated DOM tree must depend only on the input text and the flags.
2646          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2647          *
2648          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2649          * change in the DOM tree for a given text, must be passed through the section identifier
2650          * in the section edit link and thus back to extractSections().
2651          *
2652          * The output of this function is currently only cached in process memory, but a persistent
2653          * cache may be implemented at a later date which takes further advantage of these strict
2654          * dependency requirements.
2655          *
2656          * @private
2657          */
2658         function preprocessToDom ( $text, $flags = 0 ) {
2659                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2660                 return $dom;
2661         }
2662
2663         /*
2664          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2665          */
2666         public static function splitWhitespace( $s ) {
2667                 $ltrimmed = ltrim( $s );
2668                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2669                 $trimmed = rtrim( $ltrimmed );
2670                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2671                 if ( $diff > 0 ) {
2672                         $w2 = substr( $ltrimmed, -$diff );
2673                 } else {
2674                         $w2 = '';
2675                 }
2676                 return array( $w1, $trimmed, $w2 );
2677         }
2678
2679         /**
2680          * Replace magic variables, templates, and template arguments
2681          * with the appropriate text. Templates are substituted recursively,
2682          * taking care to avoid infinite loops.
2683          *
2684          * Note that the substitution depends on value of $mOutputType:
2685          *  self::OT_WIKI: only {{subst:}} templates
2686          *  self::OT_PREPROCESS: templates but not extension tags
2687          *  self::OT_HTML: all templates and extension tags
2688          *
2689          * @param string $tex The text to transform
2690          * @param PPFrame $frame Object describing the arguments passed to the template.
2691          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2692          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2693          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2694          * @private
2695          */
2696         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2697                 # Prevent too big inclusions
2698                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2699                         return $text;
2700                 }
2701
2702                 wfProfileIn( __METHOD__ );
2703
2704                 if ( $frame === false ) {
2705                         $frame = $this->getPreprocessor()->newFrame();
2706                 } elseif ( !( $frame instanceof PPFrame ) ) {
2707                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2708                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2709                 }
2710
2711                 $dom = $this->preprocessToDom( $text );
2712                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2713                 $text = $frame->expand( $dom, $flags );
2714
2715                 wfProfileOut( __METHOD__ );
2716                 return $text;
2717         }
2718
2719         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2720         static function createAssocArgs( $args ) {
2721                 $assocArgs = array();
2722                 $index = 1;
2723                 foreach( $args as $arg ) {
2724                         $eqpos = strpos( $arg, '=' );
2725                         if ( $eqpos === false ) {
2726                                 $assocArgs[$index++] = $arg;
2727                         } else {
2728                                 $name = trim( substr( $arg, 0, $eqpos ) );
2729                                 $value = trim( substr( $arg, $eqpos+1 ) );
2730                                 if ( $value === false ) {
2731                                         $value = '';
2732                                 }
2733                                 if ( $name !== false ) {
2734                                         $assocArgs[$name] = $value;
2735                                 }
2736                         }
2737                 }
2738
2739                 return $assocArgs;
2740         }
2741
2742         /**
2743          * Warn the user when a parser limitation is reached
2744          * Will warn at most once the user per limitation type
2745          *
2746          * @param string $limitationType, should be one of:
2747          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2748          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2749          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2750          * @params int $current, $max When an explicit limit has been
2751          *       exceeded, provide the values (optional)
2752          */
2753         function limitationWarn( $limitationType, $current=null, $max=null) {
2754                 $msgName = $limitationType . '-warning';
2755                 //does no harm if $current and $max are present but are unnecessary for the message
2756                 $warning = wfMsgExt( $msgName, array( 'parsemag', 'escape' ), $current, $max );
2757                 $this->mOutput->addWarning( $warning );
2758                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2759                 if ( $cat ) {
2760                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2761                 }
2762         }
2763
2764         /**
2765          * Return the text of a template, after recursively
2766          * replacing any variables or templates within the template.
2767          *
2768          * @param array $piece The parts of the template
2769          *  $piece['title']: the title, i.e. the part before the |
2770          *  $piece['parts']: the parameter array
2771          *  $piece['lineStart']: whether the brace was at the start of a line
2772          * @param PPFrame The current frame, contains template arguments
2773          * @return string the text of the template
2774          * @private
2775          */
2776         function braceSubstitution( $piece, $frame ) {
2777                 global $wgContLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2778                 wfProfileIn( __METHOD__ );
2779                 wfProfileIn( __METHOD__.'-setup' );
2780
2781                 # Flags
2782                 $found = false;             # $text has been filled
2783                 $nowiki = false;            # wiki markup in $text should be escaped
2784                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2785                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2786                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2787                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2788
2789                 # Title object, where $text came from
2790                 $title = NULL;
2791
2792                 # $part1 is the bit before the first |, and must contain only title characters.
2793                 # Various prefixes will be stripped from it later.
2794                 $titleWithSpaces = $frame->expand( $piece['title'] );
2795                 $part1 = trim( $titleWithSpaces );
2796                 $titleText = false;
2797
2798                 # Original title text preserved for various purposes
2799                 $originalTitle = $part1;
2800
2801                 # $args is a list of argument nodes, starting from index 0, not including $part1
2802                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2803                 wfProfileOut( __METHOD__.'-setup' );
2804
2805                 # SUBST
2806                 wfProfileIn( __METHOD__.'-modifiers' );
2807                 if ( !$found ) {
2808                         $mwSubst = MagicWord::get( 'subst' );
2809                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2810                                 # One of two possibilities is true:
2811                                 # 1) Found SUBST but not in the PST phase
2812                                 # 2) Didn't find SUBST and in the PST phase
2813                                 # In either case, return without further processing
2814                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2815                                 $isLocalObj = true;
2816                                 $found = true;
2817                         }
2818                 }
2819
2820                 # Variables
2821                 if ( !$found && $args->getLength() == 0 ) {
2822                         $id = $this->mVariables->matchStartToEnd( $part1 );
2823                         if ( $id !== false ) {
2824                                 $text = $this->getVariableValue( $id );
2825                                 if (MagicWord::getCacheTTL($id)>-1)
2826                                         $this->mOutput->mContainsOldMagic = true;
2827                                 $found = true;
2828                         }
2829                 }
2830
2831                 # MSG, MSGNW and RAW
2832                 if ( !$found ) {
2833                         # Check for MSGNW:
2834                         $mwMsgnw = MagicWord::get( 'msgnw' );
2835                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2836                                 $nowiki = true;
2837                         } else {
2838                                 # Remove obsolete MSG:
2839                                 $mwMsg = MagicWord::get( 'msg' );
2840                                 $mwMsg->matchStartAndRemove( $part1 );
2841                         }
2842
2843                         # Check for RAW:
2844                         $mwRaw = MagicWord::get( 'raw' );
2845                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2846                                 $forceRawInterwiki = true;
2847                         }
2848                 }
2849                 wfProfileOut( __METHOD__.'-modifiers' );
2850
2851                 # Parser functions
2852                 if ( !$found ) {
2853                         wfProfileIn( __METHOD__ . '-pfunc' );
2854
2855                         $colonPos = strpos( $part1, ':' );
2856                         if ( $colonPos !== false ) {
2857                                 # Case sensitive functions
2858                                 $function = substr( $part1, 0, $colonPos );
2859                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2860                                         $function = $this->mFunctionSynonyms[1][$function];
2861                                 } else {
2862                                         # Case insensitive functions
2863                                         $function = strtolower( $function );
2864                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2865                                                 $function = $this->mFunctionSynonyms[0][$function];
2866                                         } else {
2867                                                 $function = false;
2868                                         }
2869                                 }
2870                                 if ( $function ) {
2871                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2872                                         $initialArgs = array( &$this );
2873                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2874                                         if ( $flags & SFH_OBJECT_ARGS ) {
2875                                                 # Add a frame parameter, and pass the arguments as an array
2876                                                 $allArgs = $initialArgs;
2877                                                 $allArgs[] = $frame;
2878                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2879                                                         $funcArgs[] = $args->item( $i );
2880                                                 }
2881                                                 $allArgs[] = $funcArgs;
2882                                         } else {
2883                                                 # Convert arguments to plain text
2884                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2885                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2886                                                 }
2887                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2888                                         }
2889
2890                                         # Workaround for PHP bug 35229 and similar
2891                                         if ( !is_callable( $callback ) ) {
2892                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2893                                         }
2894                                         $result = call_user_func_array( $callback, $allArgs );
2895                                         $found = true;
2896                                         $noparse = true;
2897                                         $preprocessFlags = 0;
2898
2899                                         if ( is_array( $result ) ) {
2900                                                 if ( isset( $result[0] ) ) {
2901                                                         $text = $result[0];
2902                                                         unset( $result[0] );
2903                                                 }
2904
2905                                                 // Extract flags into the local scope
2906                                                 // This allows callers to set flags such as nowiki, found, etc.
2907                                                 extract( $result );
2908                                         } else {
2909                                                 $text = $result;
2910                                         }
2911                                         if ( !$noparse ) {
2912                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2913                                                 $isChildObj = true;
2914                                         }
2915                                 }
2916                         }
2917                         wfProfileOut( __METHOD__ . '-pfunc' );
2918                 }
2919
2920                 # Finish mangling title and then check for loops.
2921                 # Set $title to a Title object and $titleText to the PDBK
2922                 if ( !$found ) {
2923                         $ns = NS_TEMPLATE;
2924                         # Split the title into page and subpage
2925                         $subpage = '';
2926                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2927                         if ($subpage !== '') {
2928                                 $ns = $this->mTitle->getNamespace();
2929                         }
2930                         $title = Title::newFromText( $part1, $ns );
2931                         if ( $title ) {
2932                                 $titleText = $title->getPrefixedText();
2933                                 # Check for language variants if the template is not found
2934                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2935                                         $wgContLang->findVariantLink( $part1, $title, true );
2936                                 }
2937                                 # Do infinite loop check
2938                                 if ( !$frame->loopCheck( $title ) ) {
2939                                         $found = true;
2940                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2941                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2942                                 }
2943                                 # Do recursion depth check
2944                                 $limit = $this->mOptions->getMaxTemplateDepth();
2945                                 if ( $frame->depth >= $limit ) {
2946                                         $found = true;
2947                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2948                                 }
2949                         }
2950                 }
2951
2952                 # Load from database
2953                 if ( !$found && $title ) {
2954                         wfProfileIn( __METHOD__ . '-loadtpl' );
2955                         if ( !$title->isExternal() ) {
2956                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2957                                         $text = SpecialPage::capturePath( $title );
2958                                         if ( is_string( $text ) ) {
2959                                                 $found = true;
2960                                                 $isHTML = true;
2961                                                 $this->disableCache();
2962                                         }
2963                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2964                                         $found = false; //access denied
2965                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2966                                 } else {
2967                                         list( $text, $title ) = $this->getTemplateDom( $title );
2968                                         if ( $text !== false ) {
2969                                                 $found = true;
2970                                                 $isChildObj = true;
2971                                         }
2972                                 }
2973
2974                                 # If the title is valid but undisplayable, make a link to it
2975                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2976                                         $text = "[[:$titleText]]";
2977                                         $found = true;
2978                                 }
2979                         } elseif ( $title->isTrans() ) {
2980                                 // Interwiki transclusion
2981                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2982                                         $text = $this->interwikiTransclude( $title, 'render' );
2983                                         $isHTML = true;
2984                                 } else {
2985                                         $text = $this->interwikiTransclude( $title, 'raw' );
2986                                         // Preprocess it like a template
2987                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2988                                         $isChildObj = true;
2989                                 }
2990                                 $found = true;
2991                         }
2992                         wfProfileOut( __METHOD__ . '-loadtpl' );
2993                 }
2994
2995                 # If we haven't found text to substitute by now, we're done
2996                 # Recover the source wikitext and return it
2997                 if ( !$found ) {
2998                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2999                         wfProfileOut( __METHOD__ );
3000                         return array( 'object' => $text );
3001                 }
3002
3003                 # Expand DOM-style return values in a child frame
3004                 if ( $isChildObj ) {
3005                         # Clean up argument array
3006                         $newFrame = $frame->newChild( $args, $title );
3007
3008                         if ( $nowiki ) {
3009                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3010                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3011                                 # Expansion is eligible for the empty-frame cache
3012                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3013                                         $text = $this->mTplExpandCache[$titleText];
3014                                 } else {
3015                                         $text = $newFrame->expand( $text );
3016                                         $this->mTplExpandCache[$titleText] = $text;
3017                                 }
3018                         } else {
3019                                 # Uncached expansion
3020                                 $text = $newFrame->expand( $text );
3021                         }
3022                 }
3023                 if ( $isLocalObj && $nowiki ) {
3024                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3025                         $isLocalObj = false;
3026                 }
3027
3028                 # Replace raw HTML by a placeholder
3029                 # Add a blank line preceding, to prevent it from mucking up
3030                 # immediately preceding headings
3031                 if ( $isHTML ) {
3032                         $text = "\n\n" . $this->insertStripItem( $text );
3033                 }
3034                 # Escape nowiki-style return values
3035                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3036                         $text = wfEscapeWikiText( $text );
3037                 }
3038                 # Bug 529: if the template begins with a table or block-level
3039                 # element, it should be treated as beginning a new line.
3040                 # This behaviour is somewhat controversial.
3041                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3042                         $text = "\n" . $text;
3043                 }
3044
3045                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3046                         # Error, oversize inclusion
3047                         $text = "[[$originalTitle]]" .
3048                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3049                         $this->limitationWarn( 'post-expand-template-inclusion' );
3050                 }
3051
3052                 if ( $isLocalObj ) {
3053                         $ret = array( 'object' => $text );
3054                 } else {
3055                         $ret = array( 'text' => $text );
3056                 }
3057
3058                 wfProfileOut( __METHOD__ );
3059                 return $ret;
3060         }
3061
3062         /**
3063          * Get the semi-parsed DOM representation of a template with a given title,
3064          * and its redirect destination title. Cached.
3065          */
3066         function getTemplateDom( $title ) {
3067                 $cacheTitle = $title;
3068                 $titleText = $title->getPrefixedDBkey();
3069
3070                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3071                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3072                         $title = Title::makeTitle( $ns, $dbk );
3073                         $titleText = $title->getPrefixedDBkey();
3074                 }
3075                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3076                         return array( $this->mTplDomCache[$titleText], $title );
3077                 }
3078
3079                 // Cache miss, go to the database
3080                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3081
3082                 if ( $text === false ) {
3083                         $this->mTplDomCache[$titleText] = false;
3084                         return array( false, $title );
3085                 }
3086
3087                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3088                 $this->mTplDomCache[ $titleText ] = $dom;
3089
3090                 if (! $title->equals($cacheTitle)) {
3091                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3092                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3093                 }
3094
3095                 return array( $dom, $title );
3096         }
3097
3098         /**
3099          * Fetch the unparsed text of a template and register a reference to it.
3100          */
3101         function fetchTemplateAndTitle( $title ) {
3102                 $templateCb = $this->mOptions->getTemplateCallback();
3103                 $stuff = call_user_func( $templateCb, $title, $this );
3104                 $text = $stuff['text'];
3105                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3106                 if ( isset( $stuff['deps'] ) ) {
3107                         foreach ( $stuff['deps'] as $dep ) {
3108                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3109                         }
3110                 }
3111                 return array($text,$finalTitle);
3112         }
3113
3114         function fetchTemplate( $title ) {
3115                 $rv = $this->fetchTemplateAndTitle($title);
3116                 return $rv[0];
3117         }
3118
3119         /**
3120          * Static function to get a template
3121          * Can be overridden via ParserOptions::setTemplateCallback().
3122          */
3123         static function statelessFetchTemplate( $title, $parser=false ) {
3124                 $text = $skip = false;
3125                 $finalTitle = $title;
3126                 $deps = array();
3127
3128                 // Loop to fetch the article, with up to 1 redirect
3129                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3130                         # Give extensions a chance to select the revision instead
3131                         $id = false; // Assume current
3132                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3133
3134                         if( $skip ) {
3135                                 $text = false;
3136                                 $deps[] = array(
3137                                         'title' => $title,
3138                                         'page_id' => $title->getArticleID(),
3139                                         'rev_id' => null );
3140                                 break;
3141                         }
3142                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3143                         $rev_id = $rev ? $rev->getId() : 0;
3144                         // If there is no current revision, there is no page
3145                         if( $id === false && !$rev ) {
3146                                 $linkCache = LinkCache::singleton();
3147                                 $linkCache->addBadLinkObj( $title );
3148                         }
3149
3150                         $deps[] = array(
3151                                 'title' => $title,
3152                                 'page_id' => $title->getArticleID(),
3153                                 'rev_id' => $rev_id );
3154
3155                         if( $rev ) {
3156                                 $text = $rev->getText();
3157                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3158                                 global $wgContLang;
3159                                 $message = $wgContLang->lcfirst( $title->getText() );
3160                                 $text = wfMsgForContentNoTrans( $message );
3161                                 if( wfEmptyMsg( $message, $text ) ) {
3162                                         $text = false;
3163                                         break;
3164                                 }
3165                         } else {
3166                                 break;
3167                         }
3168                         if ( $text === false ) {
3169                                 break;
3170                         }
3171                         // Redirect?
3172                         $finalTitle = $title;
3173                         $title = Title::newFromRedirect( $text );
3174                 }
3175                 return array(
3176                         'text' => $text,
3177                         'finalTitle' => $finalTitle,
3178                         'deps' => $deps );
3179         }
3180
3181         /**
3182          * Transclude an interwiki link.
3183          */
3184         function interwikiTransclude( $title, $action ) {
3185                 global $wgEnableScaryTranscluding;
3186
3187                 if (!$wgEnableScaryTranscluding)
3188                         return wfMsg('scarytranscludedisabled');
3189
3190                 $url = $title->getFullUrl( "action=$action" );
3191
3192                 if (strlen($url) > 255)
3193                         return wfMsg('scarytranscludetoolong');
3194                 return $this->fetchScaryTemplateMaybeFromCache($url);
3195         }
3196
3197         function fetchScaryTemplateMaybeFromCache($url) {
3198                 global $wgTranscludeCacheExpiry;
3199                 $dbr = wfGetDB(DB_SLAVE);
3200                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3201                                 array('tc_url' => $url));
3202                 if ($obj) {
3203                         $time = $obj->tc_time;
3204                         $text = $obj->tc_contents;
3205                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3206                                 return $text;
3207                         }
3208                 }
3209
3210                 $text = Http::get($url);
3211                 if (!$text)
3212                         return wfMsg('scarytranscludefailed', $url);
3213
3214                 $dbw = wfGetDB(DB_MASTER);
3215                 $dbw->replace('transcache', array('tc_url'), array(
3216                         'tc_url' => $url,
3217                         'tc_time' => time(),
3218                         'tc_contents' => $text));
3219                 return $text;
3220         }
3221
3222
3223         /**
3224          * Triple brace replacement -- used for template arguments
3225          * @private
3226          */
3227         function argSubstitution( $piece, $frame ) {
3228                 wfProfileIn( __METHOD__ );
3229
3230                 $error = false;
3231                 $parts = $piece['parts'];
3232                 $nameWithSpaces = $frame->expand( $piece['title'] );
3233                 $argName = trim( $nameWithSpaces );
3234                 $object = false;
3235                 $text = $frame->getArgument( $argName );
3236                 if (  $text === false && $parts->getLength() > 0
3237                   && (
3238                     $this->ot['html']
3239                     || $this->ot['pre']
3240                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3241                   )
3242                 ) {
3243                         # No match in frame, use the supplied default
3244                         $object = $parts->item( 0 )->getChildren();
3245                 }
3246                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3247                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3248                         $this->limitationWarn( 'post-expand-template-argument' );
3249                 }
3250
3251                 if ( $text === false && $object === false ) {
3252                         # No match anywhere
3253                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3254                 }
3255                 if ( $error !== false ) {
3256                         $text .= $error;
3257                 }
3258                 if ( $object !== false ) {
3259                         $ret = array( 'object' => $object );
3260                 } else {
3261                         $ret = array( 'text' => $text );
3262                 }
3263
3264                 wfProfileOut( __METHOD__ );
3265                 return $ret;
3266         }
3267
3268         /**
3269          * Return the text to be used for a given extension tag.
3270          * This is the ghost of strip().
3271          *
3272          * @param array $params Associative array of parameters:
3273          *     name       PPNode for the tag name
3274          *     attr       PPNode for unparsed text where tag attributes are thought to be
3275          *     attributes Optional associative array of parsed attributes
3276          *     inner      Contents of extension element
3277          *     noClose    Original text did not have a close tag
3278          * @param PPFrame $frame
3279          */
3280         function extensionSubstitution( $params, $frame ) {
3281                 global $wgRawHtml, $wgContLang;
3282
3283                 $name = $frame->expand( $params['name'] );
3284                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3285                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3286
3287                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3288
3289                 if ( $this->ot['html'] ) {
3290                         $name = strtolower( $name );
3291
3292                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3293                         if ( isset( $params['attributes'] ) ) {
3294                                 $attributes = $attributes + $params['attributes'];
3295                         }
3296                         switch ( $name ) {
3297                                 case 'html':
3298                                         if( $wgRawHtml ) {
3299                                                 $output = $content;
3300                                                 break;
3301                                         } else {
3302                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3303                                         }
3304                                 case 'nowiki':
3305                                         $output = Xml::escapeTagsOnly( $content );
3306                                         break;
3307                                 case 'math':
3308                                         $output = $wgContLang->armourMath(
3309                                                 MathRenderer::renderMath( $content, $attributes ) );
3310                                         break;
3311                                 case 'gallery':
3312                                         $output = $this->renderImageGallery( $content, $attributes );
3313                                         break;
3314                                 default:
3315                                         if( isset( $this->mTagHooks[$name] ) ) {
3316                                                 # Workaround for PHP bug 35229 and similar
3317                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3318                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3319                                                 }
3320                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3321                                                         array( $content, $attributes, $this ) );
3322                                         } else {
3323                                                 $output = '<span class="error">Invalid tag extension name: ' .
3324                                                         htmlspecialchars( $name ) . '</span>';
3325                                         }
3326                         }
3327                 } else {
3328                         if ( is_null( $attrText ) ) {
3329                                 $attrText = '';
3330                         }
3331                         if ( isset( $params['attributes'] ) ) {
3332                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3333                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3334                                                 htmlspecialchars( $attrValue ) . '"';
3335                                 }
3336                         }
3337                         if ( $content === null ) {
3338                                 $output = "<$name$attrText/>";
3339                         } else {
3340                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3341                                 $output = "<$name$attrText>$content$close";
3342                         }
3343                 }
3344
3345                 if ( $name === 'html' || $name === 'nowiki' ) {
3346                         $this->mStripState->nowiki->setPair( $marker, $output );
3347                 } else {
3348                         $this->mStripState->general->setPair( $marker, $output );
3349                 }
3350                 return $marker;
3351         }
3352
3353         /**
3354          * Increment an include size counter
3355          *
3356          * @param string $type The type of expansion
3357          * @param integer $size The size of the text
3358          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3359          */
3360         function incrementIncludeSize( $type, $size ) {
3361                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3362                         return false;
3363                 } else {
3364                         $this->mIncludeSizes[$type] += $size;
3365                         return true;
3366                 }
3367         }
3368
3369         /**
3370          * Increment the expensive function count
3371          *
3372          * @return boolean False if the limit has been exceeded
3373          */
3374         function incrementExpensiveFunctionCount() {
3375                 global $wgExpensiveParserFunctionLimit;
3376                 $this->mExpensiveFunctionCount++;
3377                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3378                         return true;
3379                 }
3380                 return false;
3381         }
3382
3383         /**
3384          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3385          * Fills $this->mDoubleUnderscores, returns the modified text
3386          */
3387         function doDoubleUnderscore( $text ) {
3388                 // The position of __TOC__ needs to be recorded
3389                 $mw = MagicWord::get( 'toc' );
3390                 if( $mw->match( $text ) ) {
3391                         $this->mShowToc = true;
3392                         $this->mForceTocPosition = true;
3393
3394                         // Set a placeholder. At the end we'll fill it in with the TOC.
3395                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3396
3397                         // Only keep the first one.
3398                         $text = $mw->replace( '', $text );
3399                 }
3400
3401                 // Now match and remove the rest of them
3402                 $mwa = MagicWord::getDoubleUnderscoreArray();
3403                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3404
3405                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3406                         $this->mOutput->mNoGallery = true;
3407                 }
3408                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3409                         $this->mShowToc = false;
3410                 }
3411                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3412                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3413
3414                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3415                         if ( $containerCategory ) {
3416                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3417                         } else {
3418                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3419                         }
3420                 }
3421                 # (bug 8068) Allow control over whether robots index a page.
3422                 #
3423                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3424                 # is not desirable, the last one on the page should win.
3425                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3426                         $this->mOutput->setIndexPolicy( 'noindex' );
3427                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3428                         $this->mOutput->setIndexPolicy( 'index' );
3429                 }
3430
3431                 return $text;
3432         }
3433
3434         /**
3435          * This function accomplishes several tasks:
3436          * 1) Auto-number headings if that option is enabled
3437          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3438          * 3) Add a Table of contents on the top for users who have enabled the option
3439          * 4) Auto-anchor headings
3440          *
3441          * It loops through all headlines, collects the necessary data, then splits up the
3442          * string and re-inserts the newly formatted headlines.
3443          *
3444          * @param string $text
3445          * @param boolean $isMain
3446          * @private
3447          */
3448         function formatHeadings( $text, $isMain=true ) {
3449                 global $wgMaxTocLevel, $wgContLang;
3450
3451                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3452                 $showEditLink = $this->mOptions->getEditSection();
3453
3454                 // Do not call quickUserCan unless necessary
3455                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3456                         $showEditLink = 0;
3457                 }
3458
3459                 # Inhibit editsection links if requested in the page
3460                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3461                         $showEditLink = 0;
3462                 }
3463
3464                 # Get all headlines for numbering them and adding funky stuff like [edit]
3465                 # links - this is for later, but we need the number of headlines right now
3466                 $matches = array();
3467                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3468
3469                 # if there are fewer than 4 headlines in the article, do not show TOC
3470                 # unless it's been explicitly enabled.
3471                 $enoughToc = $this->mShowToc &&
3472                         (($numMatches >= 4) || $this->mForceTocPosition);
3473
3474                 # Allow user to stipulate that a page should have a "new section"
3475                 # link added via __NEWSECTIONLINK__
3476                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3477                         $this->mOutput->setNewSection( true );
3478                 }
3479
3480                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3481                 # override above conditions and always show TOC above first header
3482                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3483                         $this->mShowToc = true;
3484                         $enoughToc = true;
3485                 }
3486
3487                 # We need this to perform operations on the HTML
3488                 $sk = $this->mOptions->getSkin();
3489
3490                 # headline counter
3491                 $headlineCount = 0;
3492                 $numVisible = 0;
3493
3494                 # Ugh .. the TOC should have neat indentation levels which can be
3495                 # passed to the skin functions. These are determined here
3496                 $toc = '';
3497                 $full = '';
3498                 $head = array();
3499                 $sublevelCount = array();
3500                 $levelCount = array();
3501                 $toclevel = 0;
3502                 $level = 0;
3503                 $prevlevel = 0;
3504                 $toclevel = 0;
3505                 $prevtoclevel = 0;
3506                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3507                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3508                 $tocraw = array();
3509
3510                 foreach( $matches[3] as $headline ) {
3511                         $isTemplate = false;
3512                         $titleText = false;
3513                         $sectionIndex = false;
3514                         $numbering = '';
3515                         $markerMatches = array();
3516                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3517                                 $serial = $markerMatches[1];
3518                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3519                                 $isTemplate = ($titleText != $baseTitleText);
3520                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3521                         }
3522
3523                         if( $toclevel ) {
3524                                 $prevlevel = $level;
3525                                 $prevtoclevel = $toclevel;
3526                         }
3527                         $level = $matches[1][$headlineCount];
3528
3529                         if( $doNumberHeadings || $enoughToc ) {
3530
3531                                 if ( $level > $prevlevel ) {
3532                                         # Increase TOC level
3533                                         $toclevel++;
3534                                         $sublevelCount[$toclevel] = 0;
3535                                         if( $toclevel<$wgMaxTocLevel ) {
3536                                                 $prevtoclevel = $toclevel;
3537                                                 $toc .= $sk->tocIndent();
3538                                                 $numVisible++;
3539                                         }
3540                                 }
3541                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3542                                         # Decrease TOC level, find level to jump to
3543
3544                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3545                                                 # Can only go down to level 1
3546                                                 $toclevel = 1;
3547                                         } else {
3548                                                 for ($i = $toclevel; $i > 0; $i--) {
3549                                                         if ( $levelCount[$i] == $level ) {
3550                                                                 # Found last matching level
3551                                                                 $toclevel = $i;
3552                                                                 break;
3553                                                         }
3554                                                         elseif ( $levelCount[$i] < $level ) {
3555                                                                 # Found first matching level below current level
3556                                                                 $toclevel = $i + 1;
3557                                                                 break;
3558                                                         }
3559                                                 }
3560                                         }
3561                                         if( $toclevel<$wgMaxTocLevel ) {
3562                                                 if($prevtoclevel < $wgMaxTocLevel) {
3563                                                         # Unindent only if the previous toc level was shown :p
3564                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3565                                                         $prevtoclevel = $toclevel;
3566                                                 } else {
3567                                                         $toc .= $sk->tocLineEnd();
3568                                                 }
3569                                         }
3570                                 }
3571                                 else {
3572                                         # No change in level, end TOC line
3573                                         if( $toclevel<$wgMaxTocLevel ) {
3574                                                 $toc .= $sk->tocLineEnd();
3575                                         }
3576                                 }
3577
3578                                 $levelCount[$toclevel] = $level;
3579
3580                                 # count number of headlines for each level
3581                                 @$sublevelCount[$toclevel]++;
3582                                 $dot = 0;
3583                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3584                                         if( !empty( $sublevelCount[$i] ) ) {
3585                                                 if( $dot ) {
3586                                                         $numbering .= '.';
3587                                                 }
3588                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3589                                                 $dot = 1;
3590                                         }
3591                                 }
3592                         }
3593
3594                         # The safe header is a version of the header text safe to use for links
3595                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3596                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3597
3598                         # Remove link placeholders by the link text.
3599                         #     <!--LINK number-->
3600                         # turns into
3601                         #     link text with suffix
3602                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3603
3604                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3605                         $tocline = preg_replace(
3606                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3607                                 array( '',                          '<$1>'),
3608                                 $safeHeadline
3609                         );
3610                         $tocline = trim( $tocline );
3611
3612                         # For the anchor, strip out HTML-y stuff period
3613                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3614                         $safeHeadline = trim( $safeHeadline );
3615
3616                         # Save headline for section edit hint before it's escaped
3617                         $headlineHint = $safeHeadline;
3618                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3619                         # HTML names must be case-insensitively unique (bug 10721)
3620                         $arrayKey = strtolower( $safeHeadline );
3621
3622                         # count how many in assoc. array so we can track dupes in anchors
3623                         isset( $refers[$arrayKey] ) ? $refers[$arrayKey]++ : $refers[$arrayKey] = 1;
3624                         $refcount[$headlineCount] = $refers[$arrayKey];
3625
3626                         # Don't number the heading if it is the only one (looks silly)
3627                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3628                                 # the two are different if the line contains a link
3629                                 $headline=$numbering . ' ' . $headline;
3630                         }
3631
3632                         # Create the anchor for linking from the TOC to the section
3633                         $anchor = $safeHeadline;
3634                         if($refcount[$headlineCount] > 1 ) {
3635                                 $anchor .= '_' . $refcount[$headlineCount];
3636                         }
3637                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3638                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3639                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3640                         }
3641                         # give headline the correct <h#> tag
3642                         if( $showEditLink && $sectionIndex !== false ) {
3643                                 if( $isTemplate ) {
3644                                         # Put a T flag in the section identifier, to indicate to extractSections()
3645                                         # that sections inside <includeonly> should be counted.
3646                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3647                                 } else {
3648                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3649                                 }
3650                         } else {
3651                                 $editlink = '';
3652                         }
3653                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3654
3655                         $headlineCount++;
3656                 }
3657
3658                 $this->mOutput->setSections( $tocraw );
3659
3660                 # Never ever show TOC if no headers
3661                 if( $numVisible < 1 ) {
3662                         $enoughToc = false;
3663                 }
3664
3665                 if( $enoughToc ) {
3666                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3667                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3668                         }
3669                         $toc = $sk->tocList( $toc );
3670                 }
3671
3672                 # split up and insert constructed headlines
3673
3674                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3675                 $i = 0;
3676
3677                 foreach( $blocks as $block ) {
3678                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3679                                 # This is the [edit] link that appears for the top block of text when
3680                                 # section editing is enabled
3681
3682                                 # Disabled because it broke block formatting
3683                                 # For example, a bullet point in the top line
3684                                 # $full .= $sk->editSectionLink(0);
3685                         }
3686                         $full .= $block;
3687                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3688                                 # Top anchor now in skin
3689                                 $full = $full.$toc;
3690                         }
3691
3692                         if( !empty( $head[$i] ) ) {
3693                                 $full .= $head[$i];
3694                         }
3695                         $i++;
3696                 }
3697                 if( $this->mForceTocPosition ) {
3698                         return str_replace( '<!--MWTOC-->', $toc, $full );
3699                 } else {
3700                         return $full;
3701                 }
3702         }
3703
3704         /**
3705          * Transform wiki markup when saving a page by doing \r\n -> \n
3706          * conversion, substitting signatures, {{subst:}} templates, etc.
3707          *
3708          * @param string $text the text to transform
3709          * @param Title &$title the Title object for the current article
3710          * @param User &$user the User object describing the current user
3711          * @param ParserOptions $options parsing options
3712          * @param bool $clearState whether to clear the parser state first
3713          * @return string the altered wiki markup
3714          * @public
3715          */
3716         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3717                 $this->mOptions = $options;
3718                 $this->setTitle( $title );
3719                 $this->setOutputType( self::OT_WIKI );
3720
3721                 if ( $clearState ) {
3722                         $this->clearState();
3723                 }
3724
3725                 $pairs = array(
3726                         "\r\n" => "\n",
3727                 );
3728                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3729                 $text = $this->pstPass2( $text, $user );
3730                 $text = $this->mStripState->unstripBoth( $text );
3731                 return $text;
3732         }
3733
3734         /**
3735          * Pre-save transform helper function
3736          * @private
3737          */
3738         function pstPass2( $text, $user ) {
3739                 global $wgContLang, $wgLocaltimezone;
3740
3741                 /* Note: This is the timestamp saved as hardcoded wikitext to
3742                  * the database, we use $wgContLang here in order to give
3743                  * everyone the same signature and use the default one rather
3744                  * than the one selected in each user's preferences.
3745                  *
3746                  * (see also bug 12815)
3747                  */
3748                 $ts = $this->mOptions->getTimestamp();
3749                 $tz = wfMsgForContent( 'timezone-utc' );
3750                 if ( isset( $wgLocaltimezone ) ) {
3751                         $unixts = wfTimestamp( TS_UNIX, $ts );
3752                         $oldtz = getenv( 'TZ' );
3753                         putenv( 'TZ='.$wgLocaltimezone );
3754                         $ts = date( 'YmdHis', $unixts );
3755                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3756                         putenv( 'TZ='.$oldtz );
3757                 }
3758
3759                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3760
3761                 # Variable replacement
3762                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3763                 $text = $this->replaceVariables( $text );
3764
3765                 # Signatures
3766                 $sigText = $this->getUserSig( $user );
3767                 $text = strtr( $text, array(
3768                         '~~~~~' => $d,
3769                         '~~~~' => "$sigText $d",
3770                         '~~~' => $sigText
3771                 ) );
3772
3773                 # Context links: [[|name]] and [[name (context)|]]
3774                 #
3775                 global $wgLegalTitleChars;
3776                 $tc = "[$wgLegalTitleChars]";
3777                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3778
3779                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3780                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3781                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3782                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3783
3784                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3785                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3786                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3787                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3788
3789                 $t = $this->mTitle->getText();
3790                 $m = array();
3791                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3792                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3793                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3794                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3795                 } else {
3796                         # if there's no context, don't bother duplicating the title
3797                         $text = preg_replace( $p2, '[[\\1]]', $text );
3798                 }
3799
3800                 # Trim trailing whitespace
3801                 $text = rtrim( $text );
3802
3803                 return $text;
3804         }
3805
3806         /**
3807          * Fetch the user's signature text, if any, and normalize to
3808          * validated, ready-to-insert wikitext.
3809          *
3810          * @param User $user
3811          * @return string
3812          * @private
3813          */
3814         function getUserSig( &$user ) {
3815                 global $wgMaxSigChars;
3816
3817                 $username = $user->getName();
3818                 $nickname = $user->getOption( 'nickname' );
3819                 $nickname = $nickname === '' ? $username : $nickname;
3820
3821                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3822                         $nickname = $username;
3823                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3824                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3825                         # Sig. might contain markup; validate this
3826                         if( $this->validateSig( $nickname ) !== false ) {
3827                                 # Validated; clean up (if needed) and return it
3828                                 return $this->cleanSig( $nickname, true );
3829                         } else {
3830                                 # Failed to validate; fall back to the default
3831                                 $nickname = $username;
3832                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3833                         }
3834                 }
3835
3836                 // Make sure nickname doesnt get a sig in a sig
3837                 $nickname = $this->cleanSigInSig( $nickname );
3838
3839                 # If we're still here, make it a link to the user page
3840                 $userText = wfEscapeWikiText( $username );
3841                 $nickText = wfEscapeWikiText( $nickname );
3842                 if ( $user->isAnon() )  {
3843                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3844                 } else {
3845                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3846                 }
3847         }
3848
3849         /**
3850          * Check that the user's signature contains no bad XML
3851          *
3852          * @param string $text
3853          * @return mixed An expanded string, or false if invalid.
3854          */
3855         function validateSig( $text ) {
3856                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3857         }
3858
3859         /**
3860          * Clean up signature text
3861          *
3862          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3863          * 2) Substitute all transclusions
3864          *
3865          * @param string $text
3866          * @param $parsing Whether we're cleaning (preferences save) or parsing
3867          * @return string Signature text
3868          */
3869         function cleanSig( $text, $parsing = false ) {
3870                 if ( !$parsing ) {
3871                         global $wgTitle;
3872                         $this->clearState();
3873                         $this->setTitle( $wgTitle );
3874                         $this->mOptions = new ParserOptions;
3875                         $this->setOutputType = self::OT_PREPROCESS;
3876                 }
3877
3878                 # Option to disable this feature
3879                 if ( !$this->mOptions->getCleanSignatures() ) {
3880                         return $text;
3881                 }
3882
3883                 # FIXME: regex doesn't respect extension tags or nowiki
3884                 #  => Move this logic to braceSubstitution()
3885                 $substWord = MagicWord::get( 'subst' );
3886                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3887                 $substText = '{{' . $substWord->getSynonym( 0 );
3888
3889                 $text = preg_replace( $substRegex, $substText, $text );
3890                 $text = $this->cleanSigInSig( $text );
3891                 $dom = $this->preprocessToDom( $text );
3892                 $frame = $this->getPreprocessor()->newFrame();
3893                 $text = $frame->expand( $dom );
3894
3895                 if ( !$parsing ) {
3896                         $text = $this->mStripState->unstripBoth( $text );
3897                 }
3898
3899                 return $text;
3900         }
3901
3902         /**
3903          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3904          * @param string $text
3905          * @return string Signature text with /~{3,5}/ removed
3906          */
3907         function cleanSigInSig( $text ) {
3908                 $text = preg_replace( '/~{3,5}/', '', $text );
3909                 return $text;
3910         }
3911
3912         /**
3913          * Set up some variables which are usually set up in parse()
3914          * so that an external function can call some class members with confidence
3915          * @public
3916          */
3917         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3918                 $this->setTitle( $title );
3919                 $this->mOptions = $options;
3920                 $this->setOutputType( $outputType );
3921                 if ( $clearState ) {
3922                         $this->clearState();
3923                 }
3924         }
3925
3926         /**
3927          * Wrapper for preprocess()
3928          *
3929          * @param string $text the text to preprocess
3930          * @param ParserOptions $options  options
3931          * @return string
3932          * @public
3933          */
3934         function transformMsg( $text, $options ) {
3935                 global $wgTitle;
3936                 static $executing = false;
3937
3938                 # Guard against infinite recursion
3939                 if ( $executing ) {
3940                         return $text;
3941                 }
3942                 $executing = true;
3943
3944                 wfProfileIn(__METHOD__);
3945                 $text = $this->preprocess( $text, $wgTitle, $options );
3946
3947                 $executing = false;
3948                 wfProfileOut(__METHOD__);
3949                 return $text;
3950         }
3951
3952         /**
3953          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3954          * The callback should have the following form:
3955          *    function myParserHook( $text, $params, &$parser ) { ... }
3956          *
3957          * Transform and return $text. Use $parser for any required context, e.g. use
3958          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3959          *
3960          * @public
3961          *
3962          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3963          * @param mixed $callback The callback function (and object) to use for the tag
3964          *
3965          * @return The old value of the mTagHooks array associated with the hook
3966          */
3967         function setHook( $tag, $callback ) {
3968                 $tag = strtolower( $tag );
3969                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3970                 $this->mTagHooks[$tag] = $callback;
3971                 if( !in_array( $tag, $this->mStripList ) ) {
3972                         $this->mStripList[] = $tag;
3973                 }
3974
3975                 return $oldVal;
3976         }
3977
3978         function setTransparentTagHook( $tag, $callback ) {
3979                 $tag = strtolower( $tag );
3980                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3981                 $this->mTransparentTagHooks[$tag] = $callback;
3982
3983                 return $oldVal;
3984         }
3985
3986         /**
3987          * Remove all tag hooks
3988          */
3989         function clearTagHooks() {
3990                 $this->mTagHooks = array();
3991                 $this->mStripList = $this->mDefaultStripList;
3992         }
3993
3994         /**
3995          * Create a function, e.g. {{sum:1|2|3}}
3996          * The callback function should have the form:
3997          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3998          *
3999          * Or with SFH_OBJECT_ARGS:
4000          *    function myParserFunction( $parser, $frame, $args ) { ... }
4001          *
4002          * The callback may either return the text result of the function, or an array with the text
4003          * in element 0, and a number of flags in the other elements. The names of the flags are
4004          * specified in the keys. Valid flags are:
4005          *   found                     The text returned is valid, stop processing the template. This
4006          *                             is on by default.
4007          *   nowiki                    Wiki markup in the return value should be escaped
4008          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4009          *
4010          * @public
4011          *
4012          * @param string $id The magic word ID
4013          * @param mixed $callback The callback function (and object) to use
4014          * @param integer $flags a combination of the following flags:
4015          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4016          *
4017          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4018          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4019          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4020          *     the arguments, and to control the way they are expanded.
4021          *
4022          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4023          *     arguments, for instance:
4024          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4025          *
4026          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4027          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4028          *     working if/when this is changed.
4029          *
4030          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4031          *     expansion.
4032          *
4033          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4034          *     about the methods available in PPFrame and PPNode.
4035          *
4036          * @return The old callback function for this name, if any
4037          */
4038         function setFunctionHook( $id, $callback, $flags = 0 ) {
4039                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4040                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4041
4042                 # Add to function cache
4043                 $mw = MagicWord::get( $id );
4044                 if( !$mw )
4045                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4046
4047                 $synonyms = $mw->getSynonyms();
4048                 $sensitive = intval( $mw->isCaseSensitive() );
4049
4050                 foreach ( $synonyms as $syn ) {
4051                         # Case
4052                         if ( !$sensitive ) {
4053                                 $syn = strtolower( $syn );
4054                         }
4055                         # Add leading hash
4056                         if ( !( $flags & SFH_NO_HASH ) ) {
4057                                 $syn = '#' . $syn;
4058                         }
4059                         # Remove trailing colon
4060                         if ( substr( $syn, -1, 1 ) === ':' ) {
4061                                 $syn = substr( $syn, 0, -1 );
4062                         }
4063                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4064                 }
4065                 return $oldVal;
4066         }
4067
4068         /**
4069          * Get all registered function hook identifiers
4070          *
4071          * @return array
4072          */
4073         function getFunctionHooks() {
4074                 return array_keys( $this->mFunctionHooks );
4075         }
4076
4077         /**
4078          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4079          * Placeholders created in Skin::makeLinkObj()
4080          * Returns an array of link CSS classes, indexed by PDBK.
4081          */
4082         function replaceLinkHolders( &$text, $options = 0 ) {
4083                 return $this->mLinkHolders->replace( $text );
4084         }
4085
4086         /**
4087          * Replace <!--LINK--> link placeholders with plain text of links
4088          * (not HTML-formatted).
4089          * @param string $text
4090          * @return string
4091          */
4092         function replaceLinkHoldersText( $text ) {
4093                 return $this->mLinkHolders->replaceText( $text );
4094         }
4095
4096         /**
4097          * Tag hook handler for 'pre'.
4098          */
4099         function renderPreTag( $text, $attribs ) {
4100                 // Backwards-compatibility hack
4101                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4102
4103                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4104                 return Xml::openElement( 'pre', $attribs ) .
4105                         Xml::escapeTagsOnly( $content ) .
4106                         '</pre>';
4107         }
4108
4109         /**
4110          * Renders an image gallery from a text with one line per image.
4111          * text labels may be given by using |-style alternative text. E.g.
4112          *   Image:one.jpg|The number "1"
4113          *   Image:tree.jpg|A tree
4114          * given as text will return the HTML of a gallery with two images,
4115          * labeled 'The number "1"' and
4116          * 'A tree'.
4117          */
4118         function renderImageGallery( $text, $params ) {
4119                 $ig = new ImageGallery();
4120                 $ig->setContextTitle( $this->mTitle );
4121                 $ig->setShowBytes( false );
4122                 $ig->setShowFilename( false );
4123                 $ig->setParser( $this );
4124                 $ig->setHideBadImages();
4125                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4126                 $ig->useSkin( $this->mOptions->getSkin() );
4127                 $ig->mRevisionId = $this->mRevisionId;
4128
4129                 if( isset( $params['caption'] ) ) {
4130                         $caption = $params['caption'];
4131                         $caption = htmlspecialchars( $caption );
4132                         $caption = $this->replaceInternalLinks( $caption );
4133                         $ig->setCaptionHtml( $caption );
4134                 }
4135                 if( isset( $params['perrow'] ) ) {
4136                         $ig->setPerRow( $params['perrow'] );
4137                 }
4138                 if( isset( $params['widths'] ) ) {
4139                         $ig->setWidths( $params['widths'] );
4140                 }
4141                 if( isset( $params['heights'] ) ) {
4142                         $ig->setHeights( $params['heights'] );
4143                 }
4144
4145                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4146
4147                 $lines = StringUtils::explode( "\n", $text );
4148                 foreach ( $lines as $line ) {
4149                         # match lines like these:
4150                         # Image:someimage.jpg|This is some image
4151                         $matches = array();
4152                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4153                         # Skip empty lines
4154                         if ( count( $matches ) == 0 ) {
4155                                 continue;
4156                         }
4157
4158                         if ( strpos( $matches[0], '%' ) !== false )
4159                                 $matches[1] = urldecode( $matches[1] );
4160                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4161                         $nt =& $tp;
4162                         if( is_null( $nt ) ) {
4163                                 # Bogus title. Ignore these so we don't bomb out later.
4164                                 continue;
4165                         }
4166                         if ( isset( $matches[3] ) ) {
4167                                 $label = $matches[3];
4168                         } else {
4169                                 $label = '';
4170                         }
4171
4172                         $html = $this->recursiveTagParse( trim( $label ) );
4173
4174                         $ig->add( $nt, $html );
4175
4176                         # Only add real images (bug #5586)
4177                         if ( $nt->getNamespace() == NS_FILE ) {
4178                                 $this->mOutput->addImage( $nt->getDBkey() );
4179                         }
4180                 }
4181                 return $ig->toHTML();
4182         }
4183
4184         function getImageParams( $handler ) {
4185                 if ( $handler ) {
4186                         $handlerClass = get_class( $handler );
4187                 } else {
4188                         $handlerClass = '';
4189                 }
4190                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4191                         // Initialise static lists
4192                         static $internalParamNames = array(
4193                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4194                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4195                                         'bottom', 'text-bottom' ),
4196                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4197                                         'upright', 'border', 'link', 'alt' ),
4198                         );
4199                         static $internalParamMap;
4200                         if ( !$internalParamMap ) {
4201                                 $internalParamMap = array();
4202                                 foreach ( $internalParamNames as $type => $names ) {
4203                                         foreach ( $names as $name ) {
4204                                                 $magicName = str_replace( '-', '_', "img_$name" );
4205                                                 $internalParamMap[$magicName] = array( $type, $name );
4206                                         }
4207                                 }
4208                         }
4209
4210                         // Add handler params
4211                         $paramMap = $internalParamMap;
4212                         if ( $handler ) {
4213                                 $handlerParamMap = $handler->getParamMap();
4214                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4215                                         $paramMap[$magic] = array( 'handler', $paramName );
4216                                 }
4217                         }
4218                         $this->mImageParams[$handlerClass] = $paramMap;
4219                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4220                 }
4221                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4222         }
4223
4224         /**
4225          * Parse image options text and use it to make an image
4226          * @param Title $title
4227          * @param string $options
4228          * @param LinkHolderArray $holders
4229          */
4230         function makeImage( $title, $options, $holders = false ) {
4231                 # Check if the options text is of the form "options|alt text"
4232                 # Options are:
4233                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4234                 #  * left       no resizing, just left align. label is used for alt= only
4235                 #  * right      same, but right aligned
4236                 #  * none       same, but not aligned
4237                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4238                 #  * center     center the image
4239                 #  * framed     Keep original image size, no magnify-button.
4240                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4241                 #  * upright    reduce width for upright images, rounded to full __0 px
4242                 #  * border     draw a 1px border around the image
4243                 #  * alt        Text for HTML alt attribute (defaults to empty)
4244                 # vertical-align values (no % or length right now):
4245                 #  * baseline
4246                 #  * sub
4247                 #  * super
4248                 #  * top
4249                 #  * text-top
4250                 #  * middle
4251                 #  * bottom
4252                 #  * text-bottom
4253
4254                 $parts = StringUtils::explode( "|", $options );
4255                 $sk = $this->mOptions->getSkin();
4256
4257                 # Give extensions a chance to select the file revision for us
4258                 $skip = $time = $descQuery = false;
4259                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4260
4261                 if ( $skip ) {
4262                         return $sk->link( $title );
4263                 }
4264
4265                 # Get the file
4266                 $imagename = $title->getDBkey();
4267                 if ( isset( $this->mFileCache[$imagename][$time] ) ) {
4268                         $file = $this->mFileCache[$imagename][$time];
4269                 } else {
4270                         $file = wfFindFile( $title, $time );
4271                         if ( count( $this->mFileCache ) > 1000 ) {
4272                                 $this->mFileCache = array();
4273                         }
4274                         $this->mFileCache[$imagename][$time] = $file;
4275                 }
4276                 # Get parameter map
4277                 $handler = $file ? $file->getHandler() : false;
4278
4279                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4280
4281                 # Process the input parameters
4282                 $caption = '';
4283                 $params = array( 'frame' => array(), 'handler' => array(),
4284                         'horizAlign' => array(), 'vertAlign' => array() );
4285                 foreach( $parts as $part ) {
4286                         $part = trim( $part );
4287                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4288                         $validated = false;
4289                         if( isset( $paramMap[$magicName] ) ) {
4290                                 list( $type, $paramName ) = $paramMap[$magicName];
4291
4292                                 // Special case; width and height come in one variable together
4293                                 if( $type === 'handler' && $paramName === 'width' ) {
4294                                         $m = array();
4295                                         # (bug 13500) In both cases (width/height and width only),
4296                                         # permit trailing "px" for backward compatibility.
4297                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4298                                                 $width = intval( $m[1] );
4299                                                 $height = intval( $m[2] );
4300                                                 if ( $handler->validateParam( 'width', $width ) ) {
4301                                                         $params[$type]['width'] = $width;
4302                                                         $validated = true;
4303                                                 }
4304                                                 if ( $handler->validateParam( 'height', $height ) ) {
4305                                                         $params[$type]['height'] = $height;
4306                                                         $validated = true;
4307                                                 }
4308                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4309                                                 $width = intval( $value );
4310                                                 if ( $handler->validateParam( 'width', $width ) ) {
4311                                                         $params[$type]['width'] = $width;
4312                                                         $validated = true;
4313                                                 }
4314                                         } // else no validation -- bug 13436
4315                                 } else {
4316                                         if ( $type === 'handler' ) {
4317                                                 # Validate handler parameter
4318                                                 $validated = $handler->validateParam( $paramName, $value );
4319                                         } else {
4320                                                 # Validate internal parameters
4321                                                 switch( $paramName ) {
4322                                                 case 'manualthumb':
4323                                                 case 'alt':
4324                                                         // @fixme - possibly check validity here for
4325                                                         // manualthumb? downstream behavior seems odd with
4326                                                         // missing manual thumbs.
4327                                                         $validated = true;
4328                                                         $value = $this->stripAltText( $value, $holders );
4329                                                         break;
4330                                                 case 'link':
4331                                                         $chars = self::EXT_LINK_URL_CLASS;
4332                                                         $prots = $this->mUrlProtocols;
4333                                                         if ( $value === '' ) {
4334                                                                 $paramName = 'no-link';
4335                                                                 $value = true;
4336                                                                 $validated = true;
4337                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4338                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4339                                                                         $paramName = 'link-url';
4340                                                                         $this->mOutput->addExternalLink( $value );
4341                                                                         $validated = true;
4342                                                                 }
4343                                                         } else {
4344                                                                 $linkTitle = Title::newFromText( $value );
4345                                                                 if ( $linkTitle ) {
4346                                                                         $paramName = 'link-title';
4347                                                                         $value = $linkTitle;
4348                                                                         $this->mOutput->addLink( $linkTitle );
4349                                                                         $validated = true;
4350                                                                 }
4351                                                         }
4352                                                         break;
4353                                                 default:
4354                                                         // Most other things appear to be empty or numeric...
4355                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4356                                                 }
4357                                         }
4358
4359                                         if ( $validated ) {
4360                                                 $params[$type][$paramName] = $value;
4361                                         }
4362                                 }
4363                         }
4364                         if ( !$validated ) {
4365                                 $caption = $part;
4366                         }
4367                 }
4368
4369                 # Process alignment parameters
4370                 if ( $params['horizAlign'] ) {
4371                         $params['frame']['align'] = key( $params['horizAlign'] );
4372                 }
4373                 if ( $params['vertAlign'] ) {
4374                         $params['frame']['valign'] = key( $params['vertAlign'] );
4375                 }
4376
4377                 $params['frame']['caption'] = $caption;
4378
4379                 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4380
4381                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4382                 # came to also set the caption, ordinary text after the image -- which
4383                 # makes no sense, because that just repeats the text multiple times in
4384                 # screen readers.  It *also* came to set the title attribute.
4385                 #
4386                 # Now that we have an alt attribute, we should not set the alt text to
4387                 # equal the caption: that's worse than useless, it just repeats the
4388                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4389                 # use the unnamed parameter for alt text as well, just for the time be-
4390                 # ing, if the unnamed param is set and the alt param is not.
4391                 #
4392                 # For the future, we need to figure out if we want to tweak this more,
4393                 # e.g., introducing a title= parameter for the title; ignoring the un-
4394                 # named parameter entirely for images without a caption; adding an ex-
4395                 # plicit caption= parameter and preserving the old magic unnamed para-
4396                 # meter for BC; ...
4397                 if( $caption !== '' && !isset( $params['frame']['alt'] )
4398                 && !isset( $params['frame']['framed'] )
4399                 && !isset( $params['frame']['thumbnail'] )
4400                 && !isset( $params['frame']['manualthumb'] ) ) {
4401                         $params['frame']['alt'] = $params['frame']['title'];
4402                 }
4403
4404                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4405
4406                 # Linker does the rest
4407                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4408
4409                 # Give the handler a chance to modify the parser object
4410                 if ( $handler ) {
4411                         $handler->parserTransformHook( $this, $file );
4412                 }
4413
4414                 return $ret;
4415         }
4416
4417         protected function stripAltText( $caption, $holders ) {
4418                 # Strip bad stuff out of the title (tooltip).  We can't just use
4419                 # replaceLinkHoldersText() here, because if this function is called
4420                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4421                 if ( $holders ) {
4422                         $tooltip = $holders->replaceText( $caption );
4423                 } else {
4424                         $tooltip = $this->replaceLinkHoldersText( $caption );
4425                 }
4426
4427                 # make sure there are no placeholders in thumbnail attributes
4428                 # that are later expanded to html- so expand them now and
4429                 # remove the tags
4430                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4431                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4432
4433                 return $tooltip;
4434         }
4435
4436         /**
4437          * Set a flag in the output object indicating that the content is dynamic and
4438          * shouldn't be cached.
4439          */
4440         function disableCache() {
4441                 wfDebug( "Parser output marked as uncacheable.\n" );
4442                 $this->mOutput->mCacheTime = -1;
4443         }
4444
4445         /**#@+
4446          * Callback from the Sanitizer for expanding items found in HTML attribute
4447          * values, so they can be safely tested and escaped.
4448          * @param string $text
4449          * @param PPFrame $frame
4450          * @return string
4451          * @private
4452          */
4453         function attributeStripCallback( &$text, $frame = false ) {
4454                 $text = $this->replaceVariables( $text, $frame );
4455                 $text = $this->mStripState->unstripBoth( $text );
4456                 return $text;
4457         }
4458
4459         /**#@-*/
4460
4461         /**#@+
4462          * Accessor/mutator
4463          */
4464         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4465         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4466         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4467         /**#@-*/
4468
4469         /**#@+
4470          * Accessor
4471          */
4472         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4473         /**#@-*/
4474
4475
4476         /**
4477          * Break wikitext input into sections, and either pull or replace
4478          * some particular section's text.
4479          *
4480          * External callers should use the getSection and replaceSection methods.
4481          *
4482          * @param string $text Page wikitext
4483          * @param string $section A section identifier string of the form:
4484          *   <flag1> - <flag2> - ... - <section number>
4485          *
4486          * Currently the only recognised flag is "T", which means the target section number
4487          * was derived during a template inclusion parse, in other words this is a template
4488          * section edit link. If no flags are given, it was an ordinary section edit link.
4489          * This flag is required to avoid a section numbering mismatch when a section is
4490          * enclosed by <includeonly> (bug 6563).
4491          *
4492          * The section number 0 pulls the text before the first heading; other numbers will
4493          * pull the given section along with its lower-level subsections. If the section is
4494          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4495          *
4496          * @param string $mode One of "get" or "replace"
4497          * @param string $newText Replacement text for section data.
4498          * @return string for "get", the extracted section text.
4499          *                for "replace", the whole page with the section replaced.
4500          */
4501         private function extractSections( $text, $section, $mode, $newText='' ) {
4502                 global $wgTitle;
4503                 $this->clearState();
4504                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4505                 $this->mOptions = new ParserOptions;
4506                 $this->setOutputType( self::OT_WIKI );
4507                 $outText = '';
4508                 $frame = $this->getPreprocessor()->newFrame();
4509
4510                 // Process section extraction flags
4511                 $flags = 0;
4512                 $sectionParts = explode( '-', $section );
4513                 $sectionIndex = array_pop( $sectionParts );
4514                 foreach ( $sectionParts as $part ) {
4515                         if ( $part === 'T' ) {
4516                                 $flags |= self::PTD_FOR_INCLUSION;
4517                         }
4518                 }
4519                 // Preprocess the text
4520                 $root = $this->preprocessToDom( $text, $flags );
4521
4522                 // <h> nodes indicate section breaks
4523                 // They can only occur at the top level, so we can find them by iterating the root's children
4524                 $node = $root->getFirstChild();
4525
4526                 // Find the target section
4527                 if ( $sectionIndex == 0 ) {
4528                         // Section zero doesn't nest, level=big
4529                         $targetLevel = 1000;
4530                 } else {
4531             while ( $node ) {
4532                 if ( $node->getName() === 'h' ) {
4533                     $bits = $node->splitHeading();
4534                                         if ( $bits['i'] == $sectionIndex ) {
4535                                         $targetLevel = $bits['level'];
4536                                                 break;
4537                                         }
4538                                 }
4539                                 if ( $mode === 'replace' ) {
4540                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4541                                 }
4542                                 $node = $node->getNextSibling();
4543                         }
4544                 }
4545
4546                 if ( !$node ) {
4547                         // Not found
4548                         if ( $mode === 'get' ) {
4549                                 return $newText;
4550                         } else {
4551                                 return $text;
4552                         }
4553                 }
4554
4555                 // Find the end of the section, including nested sections
4556                 do {
4557                         if ( $node->getName() === 'h' ) {
4558                                 $bits = $node->splitHeading();
4559                                 $curLevel = $bits['level'];
4560                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4561                                         break;
4562                                 }
4563                         }
4564                         if ( $mode === 'get' ) {
4565                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4566                         }
4567                         $node = $node->getNextSibling();
4568                 } while ( $node );
4569
4570                 // Write out the remainder (in replace mode only)
4571                 if ( $mode === 'replace' ) {
4572                         // Output the replacement text
4573                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4574                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4575                         $outText .= $newText . "\n\n";
4576                         while ( $node ) {
4577                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4578                                 $node = $node->getNextSibling();
4579                         }
4580                 }
4581
4582                 if ( is_string( $outText ) ) {
4583                         // Re-insert stripped tags
4584                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4585                 }
4586
4587                 return $outText;
4588         }
4589
4590         /**
4591          * This function returns the text of a section, specified by a number ($section).
4592          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4593          * the first section before any such heading (section 0).
4594          *
4595          * If a section contains subsections, these are also returned.
4596          *
4597          * @param string $text text to look in
4598          * @param string $section section identifier
4599          * @param string $deftext default to return if section is not found
4600          * @return string text of the requested section
4601          */
4602         public function getSection( $text, $section, $deftext='' ) {
4603                 return $this->extractSections( $text, $section, "get", $deftext );
4604         }
4605
4606         public function replaceSection( $oldtext, $section, $text ) {
4607                 return $this->extractSections( $oldtext, $section, "replace", $text );
4608         }
4609
4610         /**
4611          * Get the timestamp associated with the current revision, adjusted for
4612          * the default server-local timestamp
4613          */
4614         function getRevisionTimestamp() {
4615                 if ( is_null( $this->mRevisionTimestamp ) ) {
4616                         wfProfileIn( __METHOD__ );
4617                         global $wgContLang;
4618                         $dbr = wfGetDB( DB_SLAVE );
4619                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4620                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4621
4622                         // Normalize timestamp to internal MW format for timezone processing.
4623                         // This has the added side-effect of replacing a null value with
4624                         // the current time, which gives us more sensible behavior for
4625                         // previews.
4626                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4627
4628                         // The cryptic '' timezone parameter tells to use the site-default
4629                         // timezone offset instead of the user settings.
4630                         //
4631                         // Since this value will be saved into the parser cache, served
4632                         // to other users, and potentially even used inside links and such,
4633                         // it needs to be consistent for all visitors.
4634                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4635
4636                         wfProfileOut( __METHOD__ );
4637                 }
4638                 return $this->mRevisionTimestamp;
4639         }
4640
4641         /**
4642          * Mutator for $mDefaultSort
4643          *
4644          * @param $sort New value
4645          */
4646         public function setDefaultSort( $sort ) {
4647                 $this->mDefaultSort = $sort;
4648         }
4649
4650         /**
4651          * Accessor for $mDefaultSort
4652          * Will use the title/prefixed title if none is set
4653          *
4654          * @return string
4655          */
4656         public function getDefaultSort() {
4657                 global $wgCategoryPrefixedDefaultSortkey;
4658                 if( $this->mDefaultSort !== false ) {
4659                         return $this->mDefaultSort;
4660                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4661                         !$wgCategoryPrefixedDefaultSortkey) {
4662                         return $this->mTitle->getText();
4663                 } else {
4664                         return $this->mTitle->getPrefixedText();
4665                 }
4666         }
4667
4668         /**
4669          * Accessor for $mDefaultSort
4670          * Unlike getDefaultSort(), will return false if none is set
4671          *
4672          * @return string or false
4673          */
4674         public function getCustomDefaultSort() {
4675                 return $this->mDefaultSort;
4676         }
4677
4678         /**
4679          * Try to guess the section anchor name based on a wikitext fragment
4680          * presumably extracted from a heading, for example "Header" from
4681          * "== Header ==".
4682          */
4683         public function guessSectionNameFromWikiText( $text ) {
4684                 # Strip out wikitext links(they break the anchor)
4685                 $text = $this->stripSectionName( $text );
4686                 $headline = Sanitizer::decodeCharReferences( $text );
4687                 # strip out HTML
4688                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4689                 $headline = trim( $headline );
4690                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4691                 $replacearray = array(
4692                         '%3A' => ':',
4693                         '%' => '.'
4694                 );
4695                 return str_replace(
4696                         array_keys( $replacearray ),
4697                         array_values( $replacearray ),
4698                         $sectionanchor );
4699         }
4700
4701         /**
4702          * Strips a text string of wikitext for use in a section anchor
4703          *
4704          * Accepts a text string and then removes all wikitext from the
4705          * string and leaves only the resultant text (i.e. the result of
4706          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4707          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4708          * to create valid section anchors by mimicing the output of the
4709          * parser when headings are parsed.
4710          *
4711          * @param $text string Text string to be stripped of wikitext
4712          * for use in a Section anchor
4713          * @return Filtered text string
4714          */
4715         public function stripSectionName( $text ) {
4716                 # Strip internal link markup
4717                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4718                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4719
4720                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4721                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4722                 # on how many empty links there are on the page - need to figure that out.
4723                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4724
4725                 # Parse wikitext quotes (italics & bold)
4726                 $text = $this->doQuotes($text);
4727
4728                 # Strip HTML tags
4729                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4730                 return $text;
4731         }
4732
4733         function srvus( $text ) {
4734                 return $this->testSrvus( $text, $this->mOutputType );
4735         }
4736
4737         /**
4738          * strip/replaceVariables/unstrip for preprocessor regression testing
4739          */
4740         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4741                 $this->clearState();
4742                 if ( ! ( $title instanceof Title ) ) {
4743                         $title = Title::newFromText( $title );
4744                 }
4745                 $this->mTitle = $title;
4746                 $this->mOptions = $options;
4747                 $this->setOutputType( $outputType );
4748                 $text = $this->replaceVariables( $text );
4749                 $text = $this->mStripState->unstripBoth( $text );
4750                 $text = Sanitizer::removeHTMLtags( $text );
4751                 return $text;
4752         }
4753
4754         function testPst( $text, $title, $options ) {
4755                 global $wgUser;
4756                 if ( ! ( $title instanceof Title ) ) {
4757                         $title = Title::newFromText( $title );
4758                 }
4759                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4760         }
4761
4762         function testPreprocess( $text, $title, $options ) {
4763                 if ( ! ( $title instanceof Title ) ) {
4764                         $title = Title::newFromText( $title );
4765                 }
4766                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4767         }
4768
4769         function markerSkipCallback( $s, $callback ) {
4770                 $i = 0;
4771                 $out = '';
4772                 while ( $i < strlen( $s ) ) {
4773                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4774                         if ( $markerStart === false ) {
4775                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4776                                 break;
4777                         } else {
4778                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4779                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4780                                 if ( $markerEnd === false ) {
4781                                         $out .= substr( $s, $markerStart );
4782                                         break;
4783                                 } else {
4784                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4785                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4786                                         $i = $markerEnd;
4787                                 }
4788                         }
4789                 }
4790                 return $out;
4791         }
4792 }
4793
4794 /**
4795  * @todo document, briefly.
4796  * @ingroup Parser
4797  */
4798 class StripState {
4799         var $general, $nowiki;
4800
4801         function __construct() {
4802                 $this->general = new ReplacementArray;
4803                 $this->nowiki = new ReplacementArray;
4804         }
4805
4806         function unstripGeneral( $text ) {
4807                 wfProfileIn( __METHOD__ );
4808                 do {
4809                         $oldText = $text;
4810                         $text = $this->general->replace( $text );
4811                 } while ( $text !== $oldText );
4812                 wfProfileOut( __METHOD__ );
4813                 return $text;
4814         }
4815
4816         function unstripNoWiki( $text ) {
4817                 wfProfileIn( __METHOD__ );
4818                 do {
4819                         $oldText = $text;
4820                         $text = $this->nowiki->replace( $text );
4821                 } while ( $text !== $oldText );
4822                 wfProfileOut( __METHOD__ );
4823                 return $text;
4824         }
4825
4826         function unstripBoth( $text ) {
4827                 wfProfileIn( __METHOD__ );
4828                 do {
4829                         $oldText = $text;
4830                         $text = $this->general->replace( $text );
4831                         $text = $this->nowiki->replace( $text );
4832                 } while ( $text !== $oldText );
4833                 wfProfileOut( __METHOD__ );
4834                 return $text;
4835         }
4836 }
4837
4838 /**
4839  * @todo document, briefly.
4840  * @ingroup Parser
4841  */
4842 class OnlyIncludeReplacer {
4843         var $output = '';
4844
4845         function replace( $matches ) {
4846                 if ( substr( $matches[1], -1 ) === "\n" ) {
4847                         $this->output .= substr( $matches[1], 0, -1 );
4848                 } else {
4849                         $this->output .= $matches[1];
4850                 }
4851         }
4852 }