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 && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 378                         $text = MWTidy::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
 654          * @deprecated Use MWTidy::tidy()
 655          */
 656         public static function tidy( $text ) {
 657                 wfDeprecated( __METHOD__ );
 658                 return MWTidy::tidy( $text );
 659         }
 660
 661         /**
 662          * parse the wiki syntax used to render tables
 663          *
 664          * @private
 665          */
 666         function doTableStuff ( $text ) {
 667                 wfProfileIn( __METHOD__ );
 668
 669                 $lines = StringUtils::explode( "\n", $text );
 670                 $out = '';
 671                 $td_history = array (); // Is currently a td tag open?
 672                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 673                 $tr_history = array (); // Is currently a tr tag open?
 674                 $tr_attributes = array (); // history of tr attributes
 675                 $has_opened_tr = array(); // Did this table open a <tr> element?
 676                 $indent_level = 0; // indent level of the table
 677
 678                 foreach ( $lines as $outLine ) {
 679                         $line = trim( $outLine );
 680
 681                         if( $line == '' ) { // empty line, go to next line
 682                                 $out .= $outLine."\n";
 683                                 continue;
 684                         }
 685                         $first_character = $line[0];
 686                         $matches = array();
 687
 688                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 689                                 // First check if we are starting a new table
 690                                 $indent_level = strlen( $matches[1] );
 691
 692                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 693                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 694
 695                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 696                                 array_push ( $td_history , false );
 697                                 array_push ( $last_tag_history , '' );
 698                                 array_push ( $tr_history , false );
 699                                 array_push ( $tr_attributes , '' );
 700                                 array_push ( $has_opened_tr , false );
 701                         } else if ( count ( $td_history ) == 0 ) {
 702                                 // Don't do any of the following
 703                                 $out .= $outLine."\n";
 704                                 continue;
 705                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 706                                 // We are ending a table
 707                                 $line = '</table>' . substr ( $line , 2 );
 708                                 $last_tag = array_pop ( $last_tag_history );
 709
 710                                 if ( !array_pop ( $has_opened_tr ) ) {
 711                                         $line = "<tr><td></td></tr>{$line}";
 712                                 }
 713
 714                                 if ( array_pop ( $tr_history ) ) {
 715                                         $line = "</tr>{$line}";
 716                                 }
 717
 718                                 if ( array_pop ( $td_history ) ) {
 719                                         $line = "</{$last_tag}>{$line}";
 720                                 }
 721                                 array_pop ( $tr_attributes );
 722                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 723                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 724                                 // Now we have a table row
 725                                 $line = preg_replace( '#^\|-+#', '', $line );
 726
 727                                 // Whats after the tag is now only attributes
 728                                 $attributes = $this->mStripState->unstripBoth( $line );
 729                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 730                                 array_pop ( $tr_attributes );
 731                                 array_push ( $tr_attributes , $attributes );
 732
 733                                 $line = '';
 734                                 $last_tag = array_pop ( $last_tag_history );
 735                                 array_pop ( $has_opened_tr );
 736                                 array_push ( $has_opened_tr , true );
 737
 738                                 if ( array_pop ( $tr_history ) ) {
 739                                         $line = '</tr>';
 740                                 }
 741
 742                                 if ( array_pop ( $td_history ) ) {
 743                                         $line = "</{$last_tag}>{$line}";
 744                                 }
 745
 746                                 $outLine = $line;
 747                                 array_push ( $tr_history , false );
 748                                 array_push ( $td_history , false );
 749                                 array_push ( $last_tag_history , '' );
 750                         }
 751                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 752                                 // This might be cell elements, td, th or captions
 753                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 754                                         $first_character = '+';
 755                                         $line = substr ( $line , 1 );
 756                                 }
 757
 758                                 $line = substr ( $line , 1 );
 759
 760                                 if ( $first_character === '!' ) {
 761                                         $line = str_replace ( '!!' , '||' , $line );
 762                                 }
 763
 764                                 // Split up multiple cells on the same line.
 765                                 // FIXME : This can result in improper nesting of tags processed
 766                                 // by earlier parser steps, but should avoid splitting up eg
 767                                 // attribute values containing literal "||".
 768                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 769
 770                                 $outLine = '';
 771
 772                                 // Loop through each table cell
 773                                 foreach ( $cells as $cell )
 774                                 {
 775                                         $previous = '';
 776                                         if ( $first_character !== '+' )
 777                                         {
 778                                                 $tr_after = array_pop ( $tr_attributes );
 779                                                 if ( !array_pop ( $tr_history ) ) {
 780                                                         $previous = "<tr{$tr_after}>\n";
 781                                                 }
 782                                                 array_push ( $tr_history , true );
 783                                                 array_push ( $tr_attributes , '' );
 784                                                 array_pop ( $has_opened_tr );
 785                                                 array_push ( $has_opened_tr , true );
 786                                         }
 787
 788                                         $last_tag = array_pop ( $last_tag_history );
 789
 790                                         if ( array_pop ( $td_history ) ) {
 791                                                 $previous = "</{$last_tag}>{$previous}";
 792                                         }
 793
 794                                         if ( $first_character === '|' ) {
 795                                                 $last_tag = 'td';
 796                                         } else if ( $first_character === '!' ) {
 797                                                 $last_tag = 'th';
 798                                         } else if ( $first_character === '+' ) {
 799                                                 $last_tag = 'caption';
 800                                         } else {
 801                                                 $last_tag = '';
 802                                         }
 803
 804                                         array_push ( $last_tag_history , $last_tag );
 805
 806                                         // A cell could contain both parameters and data
 807                                         $cell_data = explode ( '|' , $cell , 2 );
 808
 809                                         // Bug 553: Note that a '|' inside an invalid link should not
 810                                         // be mistaken as delimiting cell parameters
 811                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 812                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 813                                         } else if ( count ( $cell_data ) == 1 )
 814                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 815                                         else {
 816                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 817                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 818                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 819                                         }
 820
 821                                         $outLine .= $cell;
 822                                         array_push ( $td_history , true );
 823                                 }
 824                         }
 825                         $out .= $outLine . "\n";
 826                 }
 827
 828                 // Closing open td, tr && table
 829                 while ( count ( $td_history ) > 0 )
 830                 {
 831                         if ( array_pop ( $td_history ) ) {
 832                                 $out .= "</td>\n";
 833                         }
 834                         if ( array_pop ( $tr_history ) ) {
 835                                 $out .= "</tr>\n";
 836                         }
 837                         if ( !array_pop ( $has_opened_tr ) ) {
 838                                 $out .= "<tr><td></td></tr>\n" ;
 839                         }
 840
 841                         $out .= "</table>\n";
 842                 }
 843
 844                 // Remove trailing line-ending (b/c)
 845                 if ( substr( $out, -1 ) === "\n" ) {
 846                         $out = substr( $out, 0, -1 );
 847                 }
 848
 849                 // special case: don't return empty table
 850                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 851                         $out = '';
 852                 }
 853
 854                 wfProfileOut( __METHOD__ );
 855
 856                 return $out;
 857         }
 858
 859         /**
 860          * Helper function for parse() that transforms wiki markup into
 861          * HTML. Only called for $mOutputType == self::OT_HTML.
 862          *
 863          * @private
 864          */
 865         function internalParse( $text ) {
 866                 $isMain = true;
 867                 wfProfileIn( __METHOD__ );
 868
 869                 # Hook to suspend the parser in this state
 870                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 871                         wfProfileOut( __METHOD__ );
 872                         return $text ;
 873                 }
 874
 875                 $text = $this->replaceVariables( $text );
 876                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 877                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 878
 879                 // Tables need to come after variable replacement for things to work
 880                 // properly; putting them before other transformations should keep
 881                 // exciting things like link expansions from showing up in surprising
 882                 // places.
 883                 $text = $this->doTableStuff( $text );
 884
 885                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 886
 887                 $text = $this->doDoubleUnderscore( $text );
 888                 $text = $this->doHeadings( $text );
 889                 if( $this->mOptions->getUseDynamicDates() ) {
 890                         $df = DateFormatter::getInstance();
 891                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 892                 }
 893                 $text = $this->doAllQuotes( $text );
 894                 $text = $this->replaceInternalLinks( $text );
 895                 $text = $this->replaceExternalLinks( $text );
 896
 897                 # replaceInternalLinks may sometimes leave behind
 898                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 899                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 900
 901                 $text = $this->doMagicLinks( $text );
 902                 $text = $this->formatHeadings( $text, $isMain );
 903
 904                 wfProfileOut( __METHOD__ );
 905                 return $text;
 906         }
 907
 908         /**
 909          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 910          * magic external links.
 911          *
 912          * DML
 913          * @private
 914          */
 915         function doMagicLinks( $text ) {
 916                 wfProfileIn( __METHOD__ );
 917                 $prots = $this->mUrlProtocols;
 918                 $urlChar = self::EXT_LINK_URL_CLASS;
 919                 $text = preg_replace_callback(
 920                         '!(?:                           # Start cases
 921                                 (<a.*?</a>) |               # m[1]: Skip link text
 922                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 923                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 924                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 925                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 926                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 927                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 928                                     [0-9Xx]                 # check digit
 929                                     \b)
 930                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 931                 wfProfileOut( __METHOD__ );
 932                 return $text;
 933         }
 934
 935         function magicLinkCallback( $m ) {
 936                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 937                         # Skip anchor
 938                         return $m[0];
 939                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 940                         # Skip HTML element
 941                         return $m[0];
 942                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 943                         # Free external link
 944                         return $this->makeFreeExternalLink( $m[0] );
 945                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 946                         # RFC or PMID
 947                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 948                                 $keyword = 'RFC';
 949                                 $urlmsg = 'rfcurl';
 950                                 $id = $m[4];
 951                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
 952                                 $keyword = 'PMID';
 953                                 $urlmsg = 'pubmedurl';
 954                                 $id = $m[4];
 955                         } else {
 956                                 throw new MWException( __METHOD__.': unrecognised match type "' .
 957                                         substr($m[0], 0, 20 ) . '"' );
 958                         }
 959                         $url = wfMsg( $urlmsg, $id);
 960                         $sk = $this->mOptions->getSkin();
 961                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
 962                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
 963                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
 964                         # ISBN
 965                         $isbn = $m[5];
 966                         $num = strtr( $isbn, array(
 967                                 '-' => '',
 968                                 ' ' => '',
 969                                 'x' => 'X',
 970                         ));
 971                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
 972                         return'<a href="' .
 973                                 $titleObj->escapeLocalUrl() .
 974                                 "\" class=\"internal\">ISBN $isbn</a>";
 975                 } else {
 976                         return $m[0];
 977                 }
 978         }
 979
 980         /**
 981          * Make a free external link, given a user-supplied URL
 982          * @return HTML
 983          * @private
 984          */
 985         function makeFreeExternalLink( $url ) {
 986                 global $wgContLang;
 987                 wfProfileIn( __METHOD__ );
 988
 989                 $sk = $this->mOptions->getSkin();
 990                 $trail = '';
 991
 992                 # The characters '<' and '>' (which were escaped by
 993                 # removeHTMLtags()) should not be included in
 994                 # URLs, per RFC 2396.
 995                 $m2 = array();
 996                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
 997                         $trail = substr($url, $m2[0][1]) . $trail;
 998                         $url = substr($url, 0, $m2[0][1]);
 999                 }
1000
1001                 # Move trailing punctuation to $trail
1002                 $sep = ',;\.:!?';
1003                 # If there is no left bracket, then consider right brackets fair game too
1004                 if ( strpos( $url, '(' ) === false ) {
1005                         $sep .= ')';
1006                 }
1007
1008                 $numSepChars = strspn( strrev( $url ), $sep );
1009                 if ( $numSepChars ) {
1010                         $trail = substr( $url, -$numSepChars ) . $trail;
1011                         $url = substr( $url, 0, -$numSepChars );
1012                 }
1013
1014                 $url = Sanitizer::cleanUrl( $url );
1015
1016                 # Is this an external image?
1017                 $text = $this->maybeMakeExternalImage( $url );
1018                 if ( $text === false ) {
1019                         # Not an image, make a link
1020                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1021                                 $this->getExternalLinkAttribs( $url ) );
1022                         # Register it in the output object...
1023                         # Replace unnecessary URL escape codes with their equivalent characters
1024                         $pasteurized = self::replaceUnusualEscapes( $url );
1025                         $this->mOutput->addExternalLink( $pasteurized );
1026                 }
1027                 wfProfileOut( __METHOD__ );
1028                 return $text . $trail;
1029         }
1030
1031
1032         /**
1033          * Parse headers and return html
1034          *
1035          * @private
1036          */
1037         function doHeadings( $text ) {
1038                 wfProfileIn( __METHOD__ );
1039                 for ( $i = 6; $i >= 1; --$i ) {
1040                         $h = str_repeat( '=', $i );
1041                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1042                           "<h$i>\\1</h$i>", $text );
1043                 }
1044                 wfProfileOut( __METHOD__ );
1045                 return $text;
1046         }
1047
1048         /**
1049          * Replace single quotes with HTML markup
1050          * @private
1051          * @return string the altered text
1052          */
1053         function doAllQuotes( $text ) {
1054                 wfProfileIn( __METHOD__ );
1055                 $outtext = '';
1056                 $lines = StringUtils::explode( "\n", $text );
1057                 foreach ( $lines as $line ) {
1058                         $outtext .= $this->doQuotes( $line ) . "\n";
1059                 }
1060                 $outtext = substr($outtext, 0,-1);
1061                 wfProfileOut( __METHOD__ );
1062                 return $outtext;
1063         }
1064
1065         /**
1066          * Helper function for doAllQuotes()
1067          */
1068         public function doQuotes( $text ) {
1069                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1070                 if ( count( $arr ) == 1 )
1071                         return $text;
1072                 else
1073                 {
1074                         # First, do some preliminary work. This may shift some apostrophes from
1075                         # being mark-up to being text. It also counts the number of occurrences
1076                         # of bold and italics mark-ups.
1077                         $i = 0;
1078                         $numbold = 0;
1079                         $numitalics = 0;
1080                         foreach ( $arr as $r )
1081                         {
1082                                 if ( ( $i % 2 ) == 1 )
1083                                 {
1084                                         # If there are ever four apostrophes, assume the first is supposed to
1085                                         # be text, and the remaining three constitute mark-up for bold text.
1086                                         if ( strlen( $arr[$i] ) == 4 )
1087                                         {
1088                                                 $arr[$i-1] .= "'";
1089                                                 $arr[$i] = "'''";
1090                                         }
1091                                         # If there are more than 5 apostrophes in a row, assume they're all
1092                                         # text except for the last 5.
1093                                         else if ( strlen( $arr[$i] ) > 5 )
1094                                         {
1095                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1096                                                 $arr[$i] = "'''''";
1097                                         }
1098                                         # Count the number of occurrences of bold and italics mark-ups.
1099                                         # We are not counting sequences of five apostrophes.
1100                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1101                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1102                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1103                                 }
1104                                 $i++;
1105                         }
1106
1107                         # If there is an odd number of both bold and italics, it is likely
1108                         # that one of the bold ones was meant to be an apostrophe followed
1109                         # by italics. Which one we cannot know for certain, but it is more
1110                         # likely to be one that has a single-letter word before it.
1111                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1112                         {
1113                                 $i = 0;
1114                                 $firstsingleletterword = -1;
1115                                 $firstmultiletterword = -1;
1116                                 $firstspace = -1;
1117                                 foreach ( $arr as $r )
1118                                 {
1119                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1120                                         {
1121                                                 $x1 = substr ($arr[$i-1], -1);
1122                                                 $x2 = substr ($arr[$i-1], -2, 1);
1123                                                 if ($x1 === ' ') {
1124                                                         if ($firstspace == -1) $firstspace = $i;
1125                                                 } else if ($x2 === ' ') {
1126                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1127                                                 } else {
1128                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1129                                                 }
1130                                         }
1131                                         $i++;
1132                                 }
1133
1134                                 # If there is a single-letter word, use it!
1135                                 if ($firstsingleletterword > -1)
1136                                 {
1137                                         $arr [ $firstsingleletterword ] = "''";
1138                                         $arr [ $firstsingleletterword-1 ] .= "'";
1139                                 }
1140                                 # If not, but there's a multi-letter word, use that one.
1141                                 else if ($firstmultiletterword > -1)
1142                                 {
1143                                         $arr [ $firstmultiletterword ] = "''";
1144                                         $arr [ $firstmultiletterword-1 ] .= "'";
1145                                 }
1146                                 # ... otherwise use the first one that has neither.
1147                                 # (notice that it is possible for all three to be -1 if, for example,
1148                                 # there is only one pentuple-apostrophe in the line)
1149                                 else if ($firstspace > -1)
1150                                 {
1151                                         $arr [ $firstspace ] = "''";
1152                                         $arr [ $firstspace-1 ] .= "'";
1153                                 }
1154                         }
1155
1156                         # Now let's actually convert our apostrophic mush to HTML!
1157                         $output = '';
1158                         $buffer = '';
1159                         $state = '';
1160                         $i = 0;
1161                         foreach ($arr as $r)
1162                         {
1163                                 if (($i % 2) == 0)
1164                                 {
1165                                         if ($state === 'both')
1166                                                 $buffer .= $r;
1167                                         else
1168                                                 $output .= $r;
1169                                 }
1170                                 else
1171                                 {
1172                                         if (strlen ($r) == 2)
1173                                         {
1174                                                 if ($state === 'i')
1175                                                 { $output .= '</i>'; $state = ''; }
1176                                                 else if ($state === 'bi')
1177                                                 { $output .= '</i>'; $state = 'b'; }
1178                                                 else if ($state === 'ib')
1179                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1180                                                 else if ($state === 'both')
1181                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1182                                                 else # $state can be 'b' or ''
1183                                                 { $output .= '<i>'; $state .= 'i'; }
1184                                         }
1185                                         else if (strlen ($r) == 3)
1186                                         {
1187                                                 if ($state === 'b')
1188                                                 { $output .= '</b>'; $state = ''; }
1189                                                 else if ($state === 'bi')
1190                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1191                                                 else if ($state === 'ib')
1192                                                 { $output .= '</b>'; $state = 'i'; }
1193                                                 else if ($state === 'both')
1194                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1195                                                 else # $state can be 'i' or ''
1196                                                 { $output .= '<b>'; $state .= 'b'; }
1197                                         }
1198                                         else if (strlen ($r) == 5)
1199                                         {
1200                                                 if ($state === 'b')
1201                                                 { $output .= '</b><i>'; $state = 'i'; }
1202                                                 else if ($state === 'i')
1203                                                 { $output .= '</i><b>'; $state = 'b'; }
1204                                                 else if ($state === 'bi')
1205                                                 { $output .= '</i></b>'; $state = ''; }
1206                                                 else if ($state === 'ib')
1207                                                 { $output .= '</b></i>'; $state = ''; }
1208                                                 else if ($state === 'both')
1209                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1210                                                 else # ($state == '')
1211                                                 { $buffer = ''; $state = 'both'; }
1212                                         }
1213                                 }
1214                                 $i++;
1215                         }
1216                         # Now close all remaining tags.  Notice that the order is important.
1217                         if ($state === 'b' || $state === 'ib')
1218                                 $output .= '</b>';
1219                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1220                                 $output .= '</i>';
1221                         if ($state === 'bi')
1222                                 $output .= '</b>';
1223                         # There might be lonely ''''', so make sure we have a buffer
1224                         if ($state === 'both' && $buffer)
1225                                 $output .= '<b><i>'.$buffer.'</i></b>';
1226                         return $output;
1227                 }
1228         }
1229
1230         /**
1231          * Replace external links (REL)
1232          *
1233          * Note: this is all very hackish and the order of execution matters a lot.
1234          * Make sure to run maintenance/parserTests.php if you change this code.
1235          *
1236          * @private
1237          */
1238         function replaceExternalLinks( $text ) {
1239                 global $wgContLang;
1240                 wfProfileIn( __METHOD__ );
1241
1242                 $sk = $this->mOptions->getSkin();
1243
1244                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1245                 $s = array_shift( $bits );
1246
1247                 $i = 0;
1248                 while ( $i<count( $bits ) ) {
1249                         $url = $bits[$i++];
1250                         $protocol = $bits[$i++];
1251                         $text = $bits[$i++];
1252                         $trail = $bits[$i++];
1253
1254                         # The characters '<' and '>' (which were escaped by
1255                         # removeHTMLtags()) should not be included in
1256                         # URLs, per RFC 2396.
1257                         $m2 = array();
1258                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1259                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1260                                 $url = substr($url, 0, $m2[0][1]);
1261                         }
1262
1263                         # If the link text is an image URL, replace it with an <img> tag
1264                         # This happened by accident in the original parser, but some people used it extensively
1265                         $img = $this->maybeMakeExternalImage( $text );
1266                         if ( $img !== false ) {
1267                                 $text = $img;
1268                         }
1269
1270                         $dtrail = '';
1271
1272                         # Set linktype for CSS - if URL==text, link is essentially free
1273                         $linktype = ($text === $url) ? 'free' : 'text';
1274
1275                         # No link text, e.g. [http://domain.tld/some.link]
1276                         if ( $text == '' ) {
1277                                 # Autonumber if allowed. See bug #5918
1278                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1279                                         $langObj = $this->getFunctionLang();
1280                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1281                                         $linktype = 'autonumber';
1282                                 } else {
1283                                         # Otherwise just use the URL
1284                                         $text = htmlspecialchars( $url );
1285                                         $linktype = 'free';
1286                                 }
1287                         } else {
1288                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1289                                 # Check for trail
1290                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1291                         }
1292
1293                         $text = $wgContLang->markNoConversion($text);
1294
1295                         $url = Sanitizer::cleanUrl( $url );
1296
1297                         # Use the encoded URL
1298                         # This means that users can paste URLs directly into the text
1299                         # Funny characters like &ouml; aren't valid in URLs anyway
1300                         # This was changed in August 2004
1301                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1302                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1303
1304                         # Register link in the output object.
1305                         # Replace unnecessary URL escape codes with the referenced character
1306                         # This prevents spammers from hiding links from the filters
1307                         $pasteurized = self::replaceUnusualEscapes( $url );
1308                         $this->mOutput->addExternalLink( $pasteurized );
1309                 }
1310
1311                 wfProfileOut( __METHOD__ );
1312                 return $s;
1313         }
1314
1315         /**
1316          * Get an associative array of additional HTML attributes appropriate for a
1317          * particular external link.  This currently may include rel => nofollow
1318          * (depending on configuration, namespace, and the URL's domain) and/or a
1319          * target attribute (depending on configuration).
1320          *
1321          * @param string $url Optional URL, to extract the domain from for rel =>
1322          *   nofollow if appropriate
1323          * @return array Associative array of HTML attributes
1324          */
1325         function getExternalLinkAttribs( $url = false ) {
1326                 $attribs = array();
1327                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1328                 $ns = $this->mTitle->getNamespace();
1329                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1330                         $attribs['rel'] = 'nofollow';
1331
1332                         global $wgNoFollowDomainExceptions;
1333                         if ( $wgNoFollowDomainExceptions ) {
1334                                 $bits = wfParseUrl( $url );
1335                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1336                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1337                                                 if( substr( $bits['host'], -strlen( $domain ) )
1338                                                 == $domain ) {
1339                                                         unset( $attribs['rel'] );
1340                                                         break;
1341                                                 }
1342                                         }
1343                                 }
1344                         }
1345                 }
1346                 if ( $this->mOptions->getExternalLinkTarget() ) {
1347                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1348                 }
1349                 return $attribs;
1350         }
1351
1352
1353         /**
1354          * Replace unusual URL escape codes with their equivalent characters
1355          * @param string
1356          * @return string
1357          * @static
1358          * @todo  This can merge genuinely required bits in the path or query string,
1359          *        breaking legit URLs. A proper fix would treat the various parts of
1360          *        the URL differently; as a workaround, just use the output for
1361          *        statistical records, not for actual linking/output.
1362          */
1363         static function replaceUnusualEscapes( $url ) {
1364                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1365                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1366         }
1367
1368         /**
1369          * Callback function used in replaceUnusualEscapes().
1370          * Replaces unusual URL escape codes with their equivalent character
1371          * @static
1372          * @private
1373          */
1374         private static function replaceUnusualEscapesCallback( $matches ) {
1375                 $char = urldecode( $matches[0] );
1376                 $ord = ord( $char );
1377                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1378                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1379                         // No, shouldn't be escaped
1380                         return $char;
1381                 } else {
1382                         // Yes, leave it escaped
1383                         return $matches[0];
1384                 }
1385         }
1386
1387         /**
1388          * make an image if it's allowed, either through the global
1389          * option, through the exception, or through the on-wiki whitelist
1390          * @private
1391          */
1392         function maybeMakeExternalImage( $url ) {
1393                 $sk = $this->mOptions->getSkin();
1394                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1395                 $imagesexception = !empty($imagesfrom);
1396                 $text = false;
1397                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1398                 if( $imagesexception && is_array( $imagesfrom ) ) {
1399                         $imagematch = false;
1400                         foreach( $imagesfrom as $match ) {
1401                                 if( strpos( $url, $match ) === 0 ) {
1402                                         $imagematch = true;
1403                                         break;
1404                                 }
1405                         }
1406                 } elseif( $imagesexception ) {
1407                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1408                 } else {
1409                         $imagematch = false;
1410                 }
1411                 if ( $this->mOptions->getAllowExternalImages()
1412                      || ( $imagesexception && $imagematch ) ) {
1413                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1414                                 # Image found
1415                                 $text = $sk->makeExternalImage( $url );
1416                         }
1417                 }
1418                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1419                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1420                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1421                         foreach( $whitelist as $entry ) {
1422                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1423                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1424                                         continue;
1425                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1426                                         # Image matches a whitelist entry
1427                                         $text = $sk->makeExternalImage( $url );
1428                                         break;
1429                                 }
1430                         }
1431                 }
1432                 return $text;
1433         }
1434
1435         /**
1436          * Process [[ ]] wikilinks
1437          * @return processed text
1438          *
1439          * @private
1440          */
1441         function replaceInternalLinks( $s ) {
1442                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1443                 return $s;
1444         }
1445
1446         /**
1447          * Process [[ ]] wikilinks (RIL)
1448          * @return LinkHolderArray
1449          *
1450          * @private
1451          */
1452         function replaceInternalLinks2( &$s ) {
1453                 global $wgContLang;
1454
1455                 wfProfileIn( __METHOD__ );
1456
1457                 wfProfileIn( __METHOD__.'-setup' );
1458                 static $tc = FALSE, $e1, $e1_img;
1459                 # the % is needed to support urlencoded titles as well
1460                 if ( !$tc ) {
1461                         $tc = Title::legalChars() . '#%';
1462                         # Match a link having the form [[namespace:link|alternate]]trail
1463                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1464                         # Match cases where there is no "]]", which might still be images
1465                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1466                 }
1467
1468                 $sk = $this->mOptions->getSkin();
1469                 $holders = new LinkHolderArray( $this );
1470
1471                 #split the entire text string on occurences of [[
1472                 $a = StringUtils::explode( '[[', ' ' . $s );
1473                 #get the first element (all text up to first [[), and remove the space we added
1474                 $s = $a->current();
1475                 $a->next();
1476                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1477                 $s = substr( $s, 1 );
1478
1479                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1480                 $e2 = null;
1481                 if ( $useLinkPrefixExtension ) {
1482                         # Match the end of a line for a word that's not followed by whitespace,
1483                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1484                         $e2 = wfMsgForContent( 'linkprefix' );
1485                 }
1486
1487                 if( is_null( $this->mTitle ) ) {
1488                         wfProfileOut( __METHOD__.'-setup' );
1489                         wfProfileOut( __METHOD__ );
1490                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1491                 }
1492                 $nottalk = !$this->mTitle->isTalkPage();
1493
1494                 if ( $useLinkPrefixExtension ) {
1495                         $m = array();
1496                         if ( preg_match( $e2, $s, $m ) ) {
1497                                 $first_prefix = $m[2];
1498                         } else {
1499                                 $first_prefix = false;
1500                         }
1501                 } else {
1502                         $prefix = '';
1503                 }
1504
1505                 if($wgContLang->hasVariants()) {
1506                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1507                 } else {
1508                         $selflink = array($this->mTitle->getPrefixedText());
1509                 }
1510                 $useSubpages = $this->areSubpagesAllowed();
1511                 wfProfileOut( __METHOD__.'-setup' );
1512
1513                 # Loop for each link
1514                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1515                         # Check for excessive memory usage
1516                         if ( $holders->isBig() ) {
1517                                 # Too big
1518                                 # Do the existence check, replace the link holders and clear the array
1519                                 $holders->replace( $s );
1520                                 $holders->clear();
1521                         }
1522
1523                         if ( $useLinkPrefixExtension ) {
1524                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1525                                 if ( preg_match( $e2, $s, $m ) ) {
1526                                         $prefix = $m[2];
1527                                         $s = $m[1];
1528                                 } else {
1529                                         $prefix='';
1530                                 }
1531                                 # first link
1532                                 if($first_prefix) {
1533                                         $prefix = $first_prefix;
1534                                         $first_prefix = false;
1535                                 }
1536                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1537                         }
1538
1539                         $might_be_img = false;
1540
1541                         wfProfileIn( __METHOD__."-e1" );
1542                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1543                                 $text = $m[2];
1544                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1545                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1546                                 # the real problem is with the $e1 regex
1547                                 # See bug 1300.
1548                                 #
1549                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1550                                 # and no image is in sight. See bug 2095.
1551                                 #
1552                                 if( $text !== '' &&
1553                                         substr( $m[3], 0, 1 ) === ']' &&
1554                                         strpos($text, '[') !== false
1555                                 )
1556                                 {
1557                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1558                                         $m[3] = substr( $m[3], 1 );
1559                                 }
1560                                 # fix up urlencoded title texts
1561                                 if( strpos( $m[1], '%' ) !== false ) {
1562                                         # Should anchors '#' also be rejected?
1563                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1564                                 }
1565                                 $trail = $m[3];
1566                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1567                                 $might_be_img = true;
1568                                 $text = $m[2];
1569                                 if ( strpos( $m[1], '%' ) !== false ) {
1570                                         $m[1] = urldecode($m[1]);
1571                                 }
1572                                 $trail = "";
1573                         } else { # Invalid form; output directly
1574                                 $s .= $prefix . '[[' . $line ;
1575                                 wfProfileOut( __METHOD__."-e1" );
1576                                 continue;
1577                         }
1578                         wfProfileOut( __METHOD__."-e1" );
1579                         wfProfileIn( __METHOD__."-misc" );
1580
1581                         # Don't allow internal links to pages containing
1582                         # PROTO: where PROTO is a valid URL protocol; these
1583                         # should be external links.
1584                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1585                                 $s .= $prefix . '[[' . $line ;
1586                                 wfProfileOut( __METHOD__."-misc" );
1587                                 continue;
1588                         }
1589
1590                         # Make subpage if necessary
1591                         if( $useSubpages ) {
1592                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1593                         } else {
1594                                 $link = $m[1];
1595                         }
1596
1597                         $noforce = (substr($m[1], 0, 1) !== ':');
1598                         if (!$noforce) {
1599                                 # Strip off leading ':'
1600                                 $link = substr($link, 1);
1601                         }
1602
1603                         wfProfileOut( __METHOD__."-misc" );
1604                         wfProfileIn( __METHOD__."-title" );
1605                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1606                         if( $nt === NULL ) {
1607                                 $s .= $prefix . '[[' . $line;
1608                                 wfProfileOut( __METHOD__."-title" );
1609                                 continue;
1610                         }
1611
1612                         $ns = $nt->getNamespace();
1613                         $iw = $nt->getInterWiki();
1614                         wfProfileOut( __METHOD__."-title" );
1615
1616                         if ($might_be_img) { # if this is actually an invalid link
1617                                 wfProfileIn( __METHOD__."-might_be_img" );
1618                                 if ($ns == NS_FILE && $noforce) { #but might be an image
1619                                         $found = false;
1620                                         while ( true ) {
1621                                                 #look at the next 'line' to see if we can close it there
1622                                                 $a->next();
1623                                                 $next_line = $a->current();
1624                                                 if ( $next_line === false || $next_line === null ) {
1625                                                         break;
1626                                                 }
1627                                                 $m = explode( ']]', $next_line, 3 );
1628                                                 if ( count( $m ) == 3 ) {
1629                                                         # the first ]] closes the inner link, the second the image
1630                                                         $found = true;
1631                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1632                                                         $trail = $m[2];
1633                                                         break;
1634                                                 } elseif ( count( $m ) == 2 ) {
1635                                                         #if there's exactly one ]] that's fine, we'll keep looking
1636                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1637                                                 } else {
1638                                                         #if $next_line is invalid too, we need look no further
1639                                                         $text .= '[[' . $next_line;
1640                                                         break;
1641                                                 }
1642                                         }
1643                                         if ( !$found ) {
1644                                                 # we couldn't find the end of this imageLink, so output it raw
1645                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1646                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1647                                                 $s .= "{$prefix}[[$link|$text";
1648                                                 # note: no $trail, because without an end, there *is* no trail
1649                                                 wfProfileOut( __METHOD__."-might_be_img" );
1650                                                 continue;
1651                                         }
1652                                 } else { #it's not an image, so output it raw
1653                                         $s .= "{$prefix}[[$link|$text";
1654                                         # note: no $trail, because without an end, there *is* no trail
1655                                         wfProfileOut( __METHOD__."-might_be_img" );
1656                                         continue;
1657                                 }
1658                                 wfProfileOut( __METHOD__."-might_be_img" );
1659                         }
1660
1661                         $wasblank = ( '' == $text );
1662                         if( $wasblank ) $text = $link;
1663
1664                         # Link not escaped by : , create the various objects
1665                         if( $noforce ) {
1666
1667                                 # Interwikis
1668                                 wfProfileIn( __METHOD__."-interwiki" );
1669                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1670                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1671                                         $s = rtrim($s . $prefix);
1672                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1673                                         wfProfileOut( __METHOD__."-interwiki" );
1674                                         continue;
1675                                 }
1676                                 wfProfileOut( __METHOD__."-interwiki" );
1677
1678                                 if ( $ns == NS_FILE ) {
1679                                         wfProfileIn( __METHOD__."-image" );
1680                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1681                                                 # recursively parse links inside the image caption
1682                                                 # actually, this will parse them in any other parameters, too,
1683                                                 # but it might be hard to fix that, and it doesn't matter ATM
1684                                                 $text = $this->replaceExternalLinks($text);
1685                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1686
1687                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1688                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1689                                         }
1690                                         $this->mOutput->addImage( $nt->getDBkey() );
1691                                         wfProfileOut( __METHOD__."-image" );
1692                                         continue;
1693
1694                                 }
1695
1696                                 if ( $ns == NS_CATEGORY ) {
1697                                         wfProfileIn( __METHOD__."-category" );
1698                                         $s = rtrim($s . "\n"); # bug 87
1699
1700                                         if ( $wasblank ) {
1701                                                 $sortkey = $this->getDefaultSort();
1702                                         } else {
1703                                                 $sortkey = $text;
1704                                         }
1705                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1706                                         $sortkey = str_replace( "\n", '', $sortkey );
1707                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1708                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1709
1710                                         /**
1711                                          * Strip the whitespace Category links produce, see bug 87
1712                                          * @todo We might want to use trim($tmp, "\n") here.
1713                                          */
1714                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1715
1716                                         wfProfileOut( __METHOD__."-category" );
1717                                         continue;
1718                                 }
1719                         }
1720
1721                         # Self-link checking
1722                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1723                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1724                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1725                                         continue;
1726                                 }
1727                         }
1728
1729                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1730                         # FIXME: Should do batch file existence checks, see comment below
1731                         if( $ns == NS_MEDIA ) {
1732                                 wfProfileIn( __METHOD__."-media" );
1733                                 # Give extensions a chance to select the file revision for us
1734                                 $skip = $time = false;
1735                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1736                                 if ( $skip ) {
1737                                         $link = $sk->link( $nt );
1738                                 } else {
1739                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1740                                 }
1741                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1742                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1743                                 $this->mOutput->addImage( $nt->getDBkey() );
1744                                 wfProfileOut( __METHOD__."-media" );
1745                                 continue;
1746                         }
1747
1748                         wfProfileIn( __METHOD__."-always_known" );
1749                         # Some titles, such as valid special pages or files in foreign repos, should
1750                         # be shown as bluelinks even though they're not included in the page table
1751                         #
1752                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1753                         # batch file existence checks for NS_FILE and NS_MEDIA
1754                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1755                                 $this->mOutput->addLink( $nt );
1756                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1757                         } else {
1758                                 # Links will be added to the output link list after checking
1759                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1760                         }
1761                         wfProfileOut( __METHOD__."-always_known" );
1762                 }
1763                 wfProfileOut( __METHOD__ );
1764                 return $holders;
1765         }
1766
1767         /**
1768          * Make a link placeholder. The text returned can be later resolved to a real link with
1769          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1770          * parsing of interwiki links, and secondly to allow all existence checks and
1771          * article length checks (for stub links) to be bundled into a single query.
1772          *
1773          * @deprecated
1774          */
1775         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1776                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1777         }
1778
1779         /**
1780          * Render a forced-blue link inline; protect against double expansion of
1781          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1782          * Since this little disaster has to split off the trail text to avoid
1783          * breaking URLs in the following text without breaking trails on the
1784          * wiki links, it's been made into a horrible function.
1785          *
1786          * @param Title $nt
1787          * @param string $text
1788          * @param string $query
1789          * @param string $trail
1790          * @param string $prefix
1791          * @return string HTML-wikitext mix oh yuck
1792          */
1793         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1794                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1795                 $sk = $this->mOptions->getSkin();
1796                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1797                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1798                 return $this->armorLinks( $link ) . $trail;
1799         }
1800
1801         /**
1802          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1803          * going to go through further parsing steps before inline URL expansion.
1804          *
1805          * Not needed quite as much as it used to be since free links are a bit
1806          * more sensible these days. But bracketed links are still an issue.
1807          *
1808          * @param string more-or-less HTML
1809          * @return string less-or-more HTML with NOPARSE bits
1810          */
1811         function armorLinks( $text ) {
1812                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1813                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1814         }
1815
1816         /**
1817          * Return true if subpage links should be expanded on this page.
1818          * @return bool
1819          */
1820         function areSubpagesAllowed() {
1821                 # Some namespaces don't allow subpages
1822                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1823         }
1824
1825         /**
1826          * Handle link to subpage if necessary
1827          * @param string $target the source of the link
1828          * @param string &$text the link text, modified as necessary
1829          * @return string the full name of the link
1830          * @private
1831          */
1832         function maybeDoSubpageLink($target, &$text) {
1833                 # Valid link forms:
1834                 # Foobar -- normal
1835                 # :Foobar -- override special treatment of prefix (images, language links)
1836                 # /Foobar -- convert to CurrentPage/Foobar
1837                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1838                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1839                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1840
1841                 wfProfileIn( __METHOD__ );
1842                 $ret = $target; # default return value is no change
1843
1844                 # Some namespaces don't allow subpages,
1845                 # so only perform processing if subpages are allowed
1846                 if( $this->areSubpagesAllowed() ) {
1847                         $hash = strpos( $target, '#' );
1848                         if( $hash !== false ) {
1849                                 $suffix = substr( $target, $hash );
1850                                 $target = substr( $target, 0, $hash );
1851                         } else {
1852                                 $suffix = '';
1853                         }
1854                         # bug 7425
1855                         $target = trim( $target );
1856                         # Look at the first character
1857                         if( $target != '' && $target{0} === '/' ) {
1858                                 # / at end means we don't want the slash to be shown
1859                                 $m = array();
1860                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1861                                 if( $trailingSlashes ) {
1862                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1863                                 } else {
1864                                         $noslash = substr( $target, 1 );
1865                                 }
1866
1867                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1868                                 if( '' === $text ) {
1869                                         $text = $target . $suffix;
1870                                 } # this might be changed for ugliness reasons
1871                         } else {
1872                                 # check for .. subpage backlinks
1873                                 $dotdotcount = 0;
1874                                 $nodotdot = $target;
1875                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1876                                         ++$dotdotcount;
1877                                         $nodotdot = substr( $nodotdot, 3 );
1878                                 }
1879                                 if($dotdotcount > 0) {
1880                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1881                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1882                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1883                                                 # / at the end means don't show full path
1884                                                 if( substr( $nodotdot, -1, 1 ) === '/' ) {
1885                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1886                                                         if( '' === $text ) {
1887                                                                 $text = $nodotdot . $suffix;
1888                                                         }
1889                                                 }
1890                                                 $nodotdot = trim( $nodotdot );
1891                                                 if( $nodotdot != '' ) {
1892                                                         $ret .= '/' . $nodotdot;
1893                                                 }
1894                                                 $ret .= $suffix;
1895                                         }
1896                                 }
1897                         }
1898                 }
1899
1900                 wfProfileOut( __METHOD__ );
1901                 return $ret;
1902         }
1903
1904         /**#@+
1905          * Used by doBlockLevels()
1906          * @private
1907          */
1908         /* private */ function closeParagraph() {
1909                 $result = '';
1910                 if ( '' != $this->mLastSection ) {
1911                         $result = '</' . $this->mLastSection  . ">\n";
1912                 }
1913                 $this->mInPre = false;
1914                 $this->mLastSection = '';
1915                 return $result;
1916         }
1917         # getCommon() returns the length of the longest common substring
1918         # of both arguments, starting at the beginning of both.
1919         #
1920         /* private */ function getCommon( $st1, $st2 ) {
1921                 $fl = strlen( $st1 );
1922                 $shorter = strlen( $st2 );
1923                 if ( $fl < $shorter ) { $shorter = $fl; }
1924
1925                 for ( $i = 0; $i < $shorter; ++$i ) {
1926                         if ( $st1{$i} != $st2{$i} ) { break; }
1927                 }
1928                 return $i;
1929         }
1930         # These next three functions open, continue, and close the list
1931         # element appropriate to the prefix character passed into them.
1932         #
1933         /* private */ function openList( $char ) {
1934                 $result = $this->closeParagraph();
1935
1936                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1937                 else if ( '#' === $char ) { $result .= '<ol><li>'; }
1938                 else if ( ':' === $char ) { $result .= '<dl><dd>'; }
1939                 else if ( ';' === $char ) {
1940                         $result .= '<dl><dt>';
1941                         $this->mDTopen = true;
1942                 }
1943                 else { $result = '<!-- ERR 1 -->'; }
1944
1945                 return $result;
1946         }
1947
1948         /* private */ function nextItem( $char ) {
1949                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1950                 else if ( ':' === $char || ';' === $char ) {
1951                         $close = '</dd>';
1952                         if ( $this->mDTopen ) { $close = '</dt>'; }
1953                         if ( ';' === $char ) {
1954                                 $this->mDTopen = true;
1955                                 return $close . '<dt>';
1956                         } else {
1957                                 $this->mDTopen = false;
1958                                 return $close . '<dd>';
1959                         }
1960                 }
1961                 return '<!-- ERR 2 -->';
1962         }
1963
1964         /* private */ function closeList( $char ) {
1965                 if ( '*' === $char ) { $text = '</li></ul>'; }
1966                 else if ( '#' === $char ) { $text = '</li></ol>'; }
1967                 else if ( ':' === $char ) {
1968                         if ( $this->mDTopen ) {
1969                                 $this->mDTopen = false;
1970                                 $text = '</dt></dl>';
1971                         } else {
1972                                 $text = '</dd></dl>';
1973                         }
1974                 }
1975                 else {  return '<!-- ERR 3 -->'; }
1976                 return $text."\n";
1977         }
1978         /**#@-*/
1979
1980         /**
1981          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1982          *
1983          * @private
1984          * @return string the lists rendered as HTML
1985          */
1986         function doBlockLevels( $text, $linestart ) {
1987                 wfProfileIn( __METHOD__ );
1988
1989                 # Parsing through the text line by line.  The main thing
1990                 # happening here is handling of block-level elements p, pre,
1991                 # and making lists from lines starting with * # : etc.
1992                 #
1993                 $textLines = StringUtils::explode( "\n", $text );
1994
1995                 $lastPrefix = $output = '';
1996                 $this->mDTopen = $inBlockElem = false;
1997                 $prefixLength = 0;
1998                 $paragraphStack = false;
1999
2000                 foreach ( $textLines as $oLine ) {
2001                         # Fix up $linestart
2002                         if ( !$linestart ) {
2003                                 $output .= $oLine;
2004                                 $linestart = true;
2005                                 continue;
2006                         }
2007
2008                         $lastPrefixLength = strlen( $lastPrefix );
2009                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2010                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2011                         if ( !$this->mInPre ) {
2012                                 # Multiple prefixes may abut each other for nested lists.
2013                                 $prefixLength = strspn( $oLine, '*#:;' );
2014                                 $prefix = substr( $oLine, 0, $prefixLength );
2015
2016                                 # eh?
2017                                 $prefix2 = str_replace( ';', ':', $prefix );
2018                                 $t = substr( $oLine, $prefixLength );
2019                                 $this->mInPre = (bool)$preOpenMatch;
2020                         } else {
2021                                 # Don't interpret any other prefixes in preformatted text
2022                                 $prefixLength = 0;
2023                                 $prefix = $prefix2 = '';
2024                                 $t = $oLine;
2025                         }
2026
2027                         # List generation
2028                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2029                                 # Same as the last item, so no need to deal with nesting or opening stuff
2030                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2031                                 $paragraphStack = false;
2032
2033                                 if ( substr( $prefix, -1 ) === ';') {
2034                                         # The one nasty exception: definition lists work like this:
2035                                         # ; title : definition text
2036                                         # So we check for : in the remainder text to split up the
2037                                         # title and definition, without b0rking links.
2038                                         $term = $t2 = '';
2039                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2040                                                 $t = $t2;
2041                                                 $output .= $term . $this->nextItem( ':' );
2042                                         }
2043                                 }
2044                         } elseif( $prefixLength || $lastPrefixLength ) {
2045                                 # Either open or close a level...
2046                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2047                                 $paragraphStack = false;
2048
2049                                 while( $commonPrefixLength < $lastPrefixLength ) {
2050                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2051                                         --$lastPrefixLength;
2052                                 }
2053                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2054                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2055                                 }
2056                                 while ( $prefixLength > $commonPrefixLength ) {
2057                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2058                                         $output .= $this->openList( $char );
2059
2060                                         if ( ';' === $char ) {
2061                                                 # FIXME: This is dupe of code above
2062                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2063                                                         $t = $t2;
2064                                                         $output .= $term . $this->nextItem( ':' );
2065                                                 }
2066                                         }
2067                                         ++$commonPrefixLength;
2068                                 }
2069                                 $lastPrefix = $prefix2;
2070                         }
2071                         if( 0 == $prefixLength ) {
2072                                 wfProfileIn( __METHOD__."-paragraph" );
2073                                 # No prefix (not in list)--go to paragraph mode
2074                                 // XXX: use a stack for nestable elements like span, table and div
2075                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2076                                 $closematch = preg_match(
2077                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2078                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2079                                 if ( $openmatch or $closematch ) {
2080                                         $paragraphStack = false;
2081                                         # TODO bug 5718: paragraph closed
2082                                         $output .= $this->closeParagraph();
2083                                         if ( $preOpenMatch and !$preCloseMatch ) {
2084                                                 $this->mInPre = true;
2085                                         }
2086                                         if ( $closematch ) {
2087                                                 $inBlockElem = false;
2088                                         } else {
2089                                                 $inBlockElem = true;
2090                                         }
2091                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2092                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2093                                                 // pre
2094                                                 if ($this->mLastSection !== 'pre') {
2095                                                         $paragraphStack = false;
2096                                                         $output .= $this->closeParagraph().'<pre>';
2097                                                         $this->mLastSection = 'pre';
2098                                                 }
2099                                                 $t = substr( $t, 1 );
2100                                         } else {
2101                                                 // paragraph
2102                                                 if ( '' == trim($t) ) {
2103                                                         if ( $paragraphStack ) {
2104                                                                 $output .= $paragraphStack.'<br />';
2105                                                                 $paragraphStack = false;
2106                                                                 $this->mLastSection = 'p';
2107                                                         } else {
2108                                                                 if ($this->mLastSection !== 'p' ) {
2109                                                                         $output .= $this->closeParagraph();
2110                                                                         $this->mLastSection = '';
2111                                                                         $paragraphStack = '<p>';
2112                                                                 } else {
2113                                                                         $paragraphStack = '</p><p>';
2114                                                                 }
2115                                                         }
2116                                                 } else {
2117                                                         if ( $paragraphStack ) {
2118                                                                 $output .= $paragraphStack;
2119                                                                 $paragraphStack = false;
2120                                                                 $this->mLastSection = 'p';
2121                                                         } else if ($this->mLastSection !== 'p') {
2122                                                                 $output .= $this->closeParagraph().'<p>';
2123                                                                 $this->mLastSection = 'p';
2124                                                         }
2125                                                 }
2126                                         }
2127                                 }
2128                                 wfProfileOut( __METHOD__."-paragraph" );
2129                         }
2130                         // somewhere above we forget to get out of pre block (bug 785)
2131                         if($preCloseMatch && $this->mInPre) {
2132                                 $this->mInPre = false;
2133                         }
2134                         if ($paragraphStack === false) {
2135                                 $output .= $t."\n";
2136                         }
2137                 }
2138                 while ( $prefixLength ) {
2139                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2140                         --$prefixLength;
2141                 }
2142                 if ( '' != $this->mLastSection ) {
2143                         $output .= '</' . $this->mLastSection . '>';
2144                         $this->mLastSection = '';
2145                 }
2146
2147                 wfProfileOut( __METHOD__ );
2148                 return $output;
2149         }
2150
2151         /**
2152          * Split up a string on ':', ignoring any occurences inside tags
2153          * to prevent illegal overlapping.
2154          * @param string $str the string to split
2155          * @param string &$before set to everything before the ':'
2156          * @param string &$after set to everything after the ':'
2157          * return string the position of the ':', or false if none found
2158          */
2159         function findColonNoLinks($str, &$before, &$after) {
2160                 wfProfileIn( __METHOD__ );
2161
2162                 $pos = strpos( $str, ':' );
2163                 if( $pos === false ) {
2164                         // Nothing to find!
2165                         wfProfileOut( __METHOD__ );
2166                         return false;
2167                 }
2168
2169                 $lt = strpos( $str, '<' );
2170                 if( $lt === false || $lt > $pos ) {
2171                         // Easy; no tag nesting to worry about
2172                         $before = substr( $str, 0, $pos );
2173                         $after = substr( $str, $pos+1 );
2174                         wfProfileOut( __METHOD__ );
2175                         return $pos;
2176                 }
2177
2178                 // Ugly state machine to walk through avoiding tags.
2179                 $state = self::COLON_STATE_TEXT;
2180                 $stack = 0;
2181                 $len = strlen( $str );
2182                 for( $i = 0; $i < $len; $i++ ) {
2183                         $c = $str{$i};
2184
2185                         switch( $state ) {
2186                         // (Using the number is a performance hack for common cases)
2187                         case 0: // self::COLON_STATE_TEXT:
2188                                 switch( $c ) {
2189                                 case "<":
2190                                         // Could be either a <start> tag or an </end> tag
2191                                         $state = self::COLON_STATE_TAGSTART;
2192                                         break;
2193                                 case ":":
2194                                         if( $stack == 0 ) {
2195                                                 // We found it!
2196                                                 $before = substr( $str, 0, $i );
2197                                                 $after = substr( $str, $i + 1 );
2198                                                 wfProfileOut( __METHOD__ );
2199                                                 return $i;
2200                                         }
2201                                         // Embedded in a tag; don't break it.
2202                                         break;
2203                                 default:
2204                                         // Skip ahead looking for something interesting
2205                                         $colon = strpos( $str, ':', $i );
2206                                         if( $colon === false ) {
2207                                                 // Nothing else interesting
2208                                                 wfProfileOut( __METHOD__ );
2209                                                 return false;
2210                                         }
2211                                         $lt = strpos( $str, '<', $i );
2212                                         if( $stack === 0 ) {
2213                                                 if( $lt === false || $colon < $lt ) {
2214                                                         // We found it!
2215                                                         $before = substr( $str, 0, $colon );
2216                                                         $after = substr( $str, $colon + 1 );
2217                                                         wfProfileOut( __METHOD__ );
2218                                                         return $i;
2219                                                 }
2220                                         }
2221                                         if( $lt === false ) {
2222                                                 // Nothing else interesting to find; abort!
2223                                                 // We're nested, but there's no close tags left. Abort!
2224                                                 break 2;
2225                                         }
2226                                         // Skip ahead to next tag start
2227                                         $i = $lt;
2228                                         $state = self::COLON_STATE_TAGSTART;
2229                                 }
2230                                 break;
2231                         case 1: // self::COLON_STATE_TAG:
2232                                 // In a <tag>
2233                                 switch( $c ) {
2234                                 case ">":
2235                                         $stack++;
2236                                         $state = self::COLON_STATE_TEXT;
2237                                         break;
2238                                 case "/":
2239                                         // Slash may be followed by >?
2240                                         $state = self::COLON_STATE_TAGSLASH;
2241                                         break;
2242                                 default:
2243                                         // ignore
2244                                 }
2245                                 break;
2246                         case 2: // self::COLON_STATE_TAGSTART:
2247                                 switch( $c ) {
2248                                 case "/":
2249                                         $state = self::COLON_STATE_CLOSETAG;
2250                                         break;
2251                                 case "!":
2252                                         $state = self::COLON_STATE_COMMENT;
2253                                         break;
2254                                 case ">":
2255                                         // Illegal early close? This shouldn't happen D:
2256                                         $state = self::COLON_STATE_TEXT;
2257                                         break;
2258                                 default:
2259                                         $state = self::COLON_STATE_TAG;
2260                                 }
2261                                 break;
2262                         case 3: // self::COLON_STATE_CLOSETAG:
2263                                 // In a </tag>
2264                                 if( $c === ">" ) {
2265                                         $stack--;
2266                                         if( $stack < 0 ) {
2267                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2268                                                 wfProfileOut( __METHOD__ );
2269                                                 return false;
2270                                         }
2271                                         $state = self::COLON_STATE_TEXT;
2272                                 }
2273                                 break;
2274                         case self::COLON_STATE_TAGSLASH:
2275                                 if( $c === ">" ) {
2276                                         // Yes, a self-closed tag <blah/>
2277                                         $state = self::COLON_STATE_TEXT;
2278                                 } else {
2279                                         // Probably we're jumping the gun, and this is an attribute
2280                                         $state = self::COLON_STATE_TAG;
2281                                 }
2282                                 break;
2283                         case 5: // self::COLON_STATE_COMMENT:
2284                                 if( $c === "-" ) {
2285                                         $state = self::COLON_STATE_COMMENTDASH;
2286                                 }
2287                                 break;
2288                         case self::COLON_STATE_COMMENTDASH:
2289                                 if( $c === "-" ) {
2290                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2291                                 } else {
2292                                         $state = self::COLON_STATE_COMMENT;
2293                                 }
2294                                 break;
2295                         case self::COLON_STATE_COMMENTDASHDASH:
2296                                 if( $c === ">" ) {
2297                                         $state = self::COLON_STATE_TEXT;
2298                                 } else {
2299                                         $state = self::COLON_STATE_COMMENT;
2300                                 }
2301                                 break;
2302                         default:
2303                                 throw new MWException( "State machine error in " . __METHOD__ );
2304                         }
2305                 }
2306                 if( $stack > 0 ) {
2307                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2308                         return false;
2309                 }
2310                 wfProfileOut( __METHOD__ );
2311                 return false;
2312         }
2313
2314         /**
2315          * Return value of a magic variable (like PAGENAME)
2316          *
2317          * @private
2318          */
2319         function getVariableValue( $index ) {
2320                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2321
2322                 /**
2323                  * Some of these require message or data lookups and can be
2324                  * expensive to check many times.
2325                  */
2326                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2327                         if ( isset( $this->mVarCache[$index] ) ) {
2328                                 return $this->mVarCache[$index];
2329                         }
2330                 }
2331
2332                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2333                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2334
2335                 # Use the time zone
2336                 global $wgLocaltimezone;
2337                 if ( isset( $wgLocaltimezone ) ) {
2338                         $oldtz = getenv( 'TZ' );
2339                         putenv( 'TZ='.$wgLocaltimezone );
2340                 }
2341
2342                 wfSuppressWarnings(); // E_STRICT system time bitching
2343                 $localTimestamp = date( 'YmdHis', $ts );
2344                 $localMonth = date( 'm', $ts );
2345                 $localMonth1 = date( 'n', $ts );
2346                 $localMonthName = date( 'n', $ts );
2347                 $localDay = date( 'j', $ts );
2348                 $localDay2 = date( 'd', $ts );
2349                 $localDayOfWeek = date( 'w', $ts );
2350                 $localWeek = date( 'W', $ts );
2351                 $localYear = date( 'Y', $ts );
2352                 $localHour = date( 'H', $ts );
2353                 if ( isset( $wgLocaltimezone ) ) {
2354                         putenv( 'TZ='.$oldtz );
2355                 }
2356                 wfRestoreWarnings();
2357
2358                 switch ( $index ) {
2359                         case 'currentmonth':
2360                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2361                         case 'currentmonth1':
2362                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2363                         case 'currentmonthname':
2364                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2365                         case 'currentmonthnamegen':
2366                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2367                         case 'currentmonthabbrev':
2368                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2369                         case 'currentday':
2370                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2371                         case 'currentday2':
2372                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2373                         case 'localmonth':
2374                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2375                         case 'localmonth1':
2376                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth1 );
2377                         case 'localmonthname':
2378                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2379                         case 'localmonthnamegen':
2380                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2381                         case 'localmonthabbrev':
2382                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2383                         case 'localday':
2384                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2385                         case 'localday2':
2386                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2387                         case 'pagename':
2388                                 return wfEscapeWikiText( $this->mTitle->getText() );
2389                         case 'pagenamee':
2390                                 return $this->mTitle->getPartialURL();
2391                         case 'fullpagename':
2392                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2393                         case 'fullpagenamee':
2394                                 return $this->mTitle->getPrefixedURL();
2395                         case 'subpagename':
2396                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2397                         case 'subpagenamee':
2398                                 return $this->mTitle->getSubpageUrlForm();
2399                         case 'basepagename':
2400                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2401                         case 'basepagenamee':
2402                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2403                         case 'talkpagename':
2404                                 if( $this->mTitle->canTalk() ) {
2405                                         $talkPage = $this->mTitle->getTalkPage();
2406                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2407                                 } else {
2408                                         return '';
2409                                 }
2410                         case 'talkpagenamee':
2411                                 if( $this->mTitle->canTalk() ) {
2412                                         $talkPage = $this->mTitle->getTalkPage();
2413                                         return $talkPage->getPrefixedUrl();
2414                                 } else {
2415                                         return '';
2416                                 }
2417                         case 'subjectpagename':
2418                                 $subjPage = $this->mTitle->getSubjectPage();
2419                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2420                         case 'subjectpagenamee':
2421                                 $subjPage = $this->mTitle->getSubjectPage();
2422                                 return $subjPage->getPrefixedUrl();
2423                         case 'revisionid':
2424                                 // Let the edit saving system know we should parse the page
2425                                 // *after* a revision ID has been assigned.
2426                                 $this->mOutput->setFlag( 'vary-revision' );
2427                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2428                                 return $this->mRevisionId;
2429                         case 'revisionday':
2430                                 // Let the edit saving system know we should parse the page
2431                                 // *after* a revision ID has been assigned. This is for null edits.
2432                                 $this->mOutput->setFlag( 'vary-revision' );
2433                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2434                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2435                         case 'revisionday2':
2436                                 // Let the edit saving system know we should parse the page
2437                                 // *after* a revision ID has been assigned. This is for null edits.
2438                                 $this->mOutput->setFlag( 'vary-revision' );
2439                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2440                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2441                         case 'revisionmonth':
2442                                 // Let the edit saving system know we should parse the page
2443                                 // *after* a revision ID has been assigned. This is for null edits.
2444                                 $this->mOutput->setFlag( 'vary-revision' );
2445                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2446                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2447                         case 'revisionyear':
2448                                 // Let the edit saving system know we should parse the page
2449                                 // *after* a revision ID has been assigned. This is for null edits.
2450                                 $this->mOutput->setFlag( 'vary-revision' );
2451                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2452                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2453                         case 'revisiontimestamp':
2454                                 // Let the edit saving system know we should parse the page
2455                                 // *after* a revision ID has been assigned. This is for null edits.
2456                                 $this->mOutput->setFlag( 'vary-revision' );
2457                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2458                                 return $this->getRevisionTimestamp();
2459                         case 'revisionuser':
2460                                 // Let the edit saving system know we should parse the page
2461                                 // *after* a revision ID has been assigned. This is for null edits.
2462                                 $this->mOutput->setFlag( 'vary-revision' );
2463                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2464                                 return $this->getRevisionUser();
2465                         case 'namespace':
2466                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2467                         case 'namespacee':
2468                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2469                         case 'talkspace':
2470                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2471                         case 'talkspacee':
2472                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2473                         case 'subjectspace':
2474                                 return $this->mTitle->getSubjectNsText();
2475                         case 'subjectspacee':
2476                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2477                         case 'currentdayname':
2478                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2479                         case 'currentyear':
2480                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2481                         case 'currenttime':
2482                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2483                         case 'currenthour':
2484                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2485                         case 'currentweek':
2486                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2487                                 // int to remove the padding
2488                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2489                         case 'currentdow':
2490                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2491                         case 'localdayname':
2492                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2493                         case 'localyear':
2494                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2495                         case 'localtime':
2496                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2497                         case 'localhour':
2498                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2499                         case 'localweek':
2500                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2501                                 // int to remove the padding
2502                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2503                         case 'localdow':
2504                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2505                         case 'numberofarticles':
2506                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2507                         case 'numberoffiles':
2508                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2509                         case 'numberofusers':
2510                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2511                         case 'numberofactiveusers':
2512                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::activeUsers() );
2513                         case 'numberofpages':
2514                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2515                         case 'numberofadmins':
2516                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2517                         case 'numberofedits':
2518                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2519                         case 'numberofviews':
2520                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::views() );
2521                         case 'currenttimestamp':
2522                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2523                         case 'localtimestamp':
2524                                 return $this->mVarCache[$index] = $localTimestamp;
2525                         case 'currentversion':
2526                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2527                         case 'sitename':
2528                                 return $wgSitename;
2529                         case 'server':
2530                                 return $wgServer;
2531                         case 'servername':
2532                                 return $wgServerName;
2533                         case 'scriptpath':
2534                                 return $wgScriptPath;
2535                         case 'directionmark':
2536                                 return $wgContLang->getDirMark();
2537                         case 'contentlanguage':
2538                                 global $wgContLanguageCode;
2539                                 return $wgContLanguageCode;
2540                         default:
2541                                 $ret = null;
2542                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2543                                         return $ret;
2544                                 else
2545                                         return null;
2546                 }
2547         }
2548
2549         /**
2550          * initialise the magic variables (like CURRENTMONTHNAME)
2551          *
2552          * @private
2553          */
2554         function initialiseVariables() {
2555                 wfProfileIn( __METHOD__ );
2556                 $variableIDs = MagicWord::getVariableIDs();
2557
2558                 $this->mVariables = new MagicWordArray( $variableIDs );
2559                 wfProfileOut( __METHOD__ );
2560         }
2561
2562         /**
2563          * Preprocess some wikitext and return the document tree.
2564          * This is the ghost of replace_variables().
2565          *
2566          * @param string $text The text to parse
2567          * @param integer flags Bitwise combination of:
2568          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2569          *                                     included. Default is to assume a direct page view.
2570          *
2571          * The generated DOM tree must depend only on the input text and the flags.
2572          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2573          *
2574          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2575          * change in the DOM tree for a given text, must be passed through the section identifier
2576          * in the section edit link and thus back to extractSections().
2577          *
2578          * The output of this function is currently only cached in process memory, but a persistent
2579          * cache may be implemented at a later date which takes further advantage of these strict
2580          * dependency requirements.
2581          *
2582          * @private
2583          */
2584         function preprocessToDom ( $text, $flags = 0 ) {
2585                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2586                 return $dom;
2587         }
2588
2589         /*
2590          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2591          */
2592         public static function splitWhitespace( $s ) {
2593                 $ltrimmed = ltrim( $s );
2594                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2595                 $trimmed = rtrim( $ltrimmed );
2596                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2597                 if ( $diff > 0 ) {
2598                         $w2 = substr( $ltrimmed, -$diff );
2599                 } else {
2600                         $w2 = '';
2601                 }
2602                 return array( $w1, $trimmed, $w2 );
2603         }
2604
2605         /**
2606          * Replace magic variables, templates, and template arguments
2607          * with the appropriate text. Templates are substituted recursively,
2608          * taking care to avoid infinite loops.
2609          *
2610          * Note that the substitution depends on value of $mOutputType:
2611          *  self::OT_WIKI: only {{subst:}} templates
2612          *  self::OT_PREPROCESS: templates but not extension tags
2613          *  self::OT_HTML: all templates and extension tags
2614          *
2615          * @param string $tex The text to transform
2616          * @param PPFrame $frame Object describing the arguments passed to the template.
2617          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2618          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2619          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2620          * @private
2621          */
2622         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2623                 # Is there any text? Also, Prevent too big inclusions!
2624                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2625                         return $text;
2626                 }
2627                 wfProfileIn( __METHOD__ );
2628
2629                 if ( $frame === false ) {
2630                         $frame = $this->getPreprocessor()->newFrame();
2631                 } elseif ( !( $frame instanceof PPFrame ) ) {
2632                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2633                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2634                 }
2635
2636                 $dom = $this->preprocessToDom( $text );
2637                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2638                 $text = $frame->expand( $dom, $flags );
2639
2640                 wfProfileOut( __METHOD__ );
2641                 return $text;
2642         }
2643
2644         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2645         static function createAssocArgs( $args ) {
2646                 $assocArgs = array();
2647                 $index = 1;
2648                 foreach( $args as $arg ) {
2649                         $eqpos = strpos( $arg, '=' );
2650                         if ( $eqpos === false ) {
2651                                 $assocArgs[$index++] = $arg;
2652                         } else {
2653                                 $name = trim( substr( $arg, 0, $eqpos ) );
2654                                 $value = trim( substr( $arg, $eqpos+1 ) );
2655                                 if ( $value === false ) {
2656                                         $value = '';
2657                                 }
2658                                 if ( $name !== false ) {
2659                                         $assocArgs[$name] = $value;
2660                                 }
2661                         }
2662                 }
2663
2664                 return $assocArgs;
2665         }
2666
2667         /**
2668          * Warn the user when a parser limitation is reached
2669          * Will warn at most once the user per limitation type
2670          *
2671          * @param string $limitationType, should be one of:
2672          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2673          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2674          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2675          * @params int $current, $max When an explicit limit has been
2676          *       exceeded, provide the values (optional)
2677          */
2678         function limitationWarn( $limitationType, $current=null, $max=null) {
2679                 $msgName = $limitationType . '-warning';
2680                 //does no harm if $current and $max are present but are unnecessary for the message
2681                 $warning = wfMsgExt( $msgName, array( 'parsemag', 'escape' ), $current, $max );
2682                 $this->mOutput->addWarning( $warning );
2683                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2684                 if ( $cat ) {
2685                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2686                 }
2687         }
2688
2689         /**
2690          * Return the text of a template, after recursively
2691          * replacing any variables or templates within the template.
2692          *
2693          * @param array $piece The parts of the template
2694          *  $piece['title']: the title, i.e. the part before the |
2695          *  $piece['parts']: the parameter array
2696          *  $piece['lineStart']: whether the brace was at the start of a line
2697          * @param PPFrame The current frame, contains template arguments
2698          * @return string the text of the template
2699          * @private
2700          */
2701         function braceSubstitution( $piece, $frame ) {
2702                 global $wgContLang, $wgNonincludableNamespaces;
2703                 wfProfileIn( __METHOD__ );
2704                 wfProfileIn( __METHOD__.'-setup' );
2705
2706                 # Flags
2707                 $found = false;             # $text has been filled
2708                 $nowiki = false;            # wiki markup in $text should be escaped
2709                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2710                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2711                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2712                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2713
2714                 # Title object, where $text came from
2715                 $title = NULL;
2716
2717                 # $part1 is the bit before the first |, and must contain only title characters.
2718                 # Various prefixes will be stripped from it later.
2719                 $titleWithSpaces = $frame->expand( $piece['title'] );
2720                 $part1 = trim( $titleWithSpaces );
2721                 $titleText = false;
2722
2723                 # Original title text preserved for various purposes
2724                 $originalTitle = $part1;
2725
2726                 # $args is a list of argument nodes, starting from index 0, not including $part1
2727                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2728                 wfProfileOut( __METHOD__.'-setup' );
2729
2730                 # SUBST
2731                 wfProfileIn( __METHOD__.'-modifiers' );
2732                 if ( !$found ) {
2733                         $mwSubst = MagicWord::get( 'subst' );
2734                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2735                                 # One of two possibilities is true:
2736                                 # 1) Found SUBST but not in the PST phase
2737                                 # 2) Didn't find SUBST and in the PST phase
2738                                 # In either case, return without further processing
2739                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2740                                 $isLocalObj = true;
2741                                 $found = true;
2742                         }
2743                 }
2744
2745                 # Variables
2746                 if ( !$found && $args->getLength() == 0 ) {
2747                         $id = $this->mVariables->matchStartToEnd( $part1 );
2748                         if ( $id !== false ) {
2749                                 $text = $this->getVariableValue( $id );
2750                                 if (MagicWord::getCacheTTL($id)>-1)
2751                                         $this->mOutput->mContainsOldMagic = true;
2752                                 $found = true;
2753                         }
2754                 }
2755
2756                 # MSG, MSGNW and RAW
2757                 if ( !$found ) {
2758                         # Check for MSGNW:
2759                         $mwMsgnw = MagicWord::get( 'msgnw' );
2760                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2761                                 $nowiki = true;
2762                         } else {
2763                                 # Remove obsolete MSG:
2764                                 $mwMsg = MagicWord::get( 'msg' );
2765                                 $mwMsg->matchStartAndRemove( $part1 );
2766                         }
2767
2768                         # Check for RAW:
2769                         $mwRaw = MagicWord::get( 'raw' );
2770                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2771                                 $forceRawInterwiki = true;
2772                         }
2773                 }
2774                 wfProfileOut( __METHOD__.'-modifiers' );
2775
2776                 # Parser functions
2777                 if ( !$found ) {
2778                         wfProfileIn( __METHOD__ . '-pfunc' );
2779
2780                         $colonPos = strpos( $part1, ':' );
2781                         if ( $colonPos !== false ) {
2782                                 # Case sensitive functions
2783                                 $function = substr( $part1, 0, $colonPos );
2784                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2785                                         $function = $this->mFunctionSynonyms[1][$function];
2786                                 } else {
2787                                         # Case insensitive functions
2788                                         $function = strtolower( $function );
2789                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2790                                                 $function = $this->mFunctionSynonyms[0][$function];
2791                                         } else {
2792                                                 $function = false;
2793                                         }
2794                                 }
2795                                 if ( $function ) {
2796                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2797                                         $initialArgs = array( &$this );
2798                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2799                                         if ( $flags & SFH_OBJECT_ARGS ) {
2800                                                 # Add a frame parameter, and pass the arguments as an array
2801                                                 $allArgs = $initialArgs;
2802                                                 $allArgs[] = $frame;
2803                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2804                                                         $funcArgs[] = $args->item( $i );
2805                                                 }
2806                                                 $allArgs[] = $funcArgs;
2807                                         } else {
2808                                                 # Convert arguments to plain text
2809                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2810                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2811                                                 }
2812                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2813                                         }
2814
2815                                         # Workaround for PHP bug 35229 and similar
2816                                         if ( !is_callable( $callback ) ) {
2817                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2818                                                 wfProfileOut( __METHOD__ );
2819                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2820                                         }
2821                                         $result = call_user_func_array( $callback, $allArgs );
2822                                         $found = true;
2823                                         $noparse = true;
2824                                         $preprocessFlags = 0;
2825
2826                                         if ( is_array( $result ) ) {
2827                                                 if ( isset( $result[0] ) ) {
2828                                                         $text = $result[0];
2829                                                         unset( $result[0] );
2830                                                 }
2831
2832                                                 // Extract flags into the local scope
2833                                                 // This allows callers to set flags such as nowiki, found, etc.
2834                                                 extract( $result );
2835                                         } else {
2836                                                 $text = $result;
2837                                         }
2838                                         if ( !$noparse ) {
2839                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2840                                                 $isChildObj = true;
2841                                         }
2842                                 }
2843                         }
2844                         wfProfileOut( __METHOD__ . '-pfunc' );
2845                 }
2846
2847                 # Finish mangling title and then check for loops.
2848                 # Set $title to a Title object and $titleText to the PDBK
2849                 if ( !$found ) {
2850                         $ns = NS_TEMPLATE;
2851                         # Split the title into page and subpage
2852                         $subpage = '';
2853                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2854                         if ($subpage !== '') {
2855                                 $ns = $this->mTitle->getNamespace();
2856                         }
2857                         $title = Title::newFromText( $part1, $ns );
2858                         if ( $title ) {
2859                                 $titleText = $title->getPrefixedText();
2860                                 # Check for language variants if the template is not found
2861                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2862                                         $wgContLang->findVariantLink( $part1, $title, true );
2863                                 }
2864                                 # Do recursion depth check
2865                                 $limit = $this->mOptions->getMaxTemplateDepth();
2866                                 if ( $frame->depth >= $limit ) {
2867                                         $found = true;
2868                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2869                                 }
2870                         }
2871                 }
2872
2873                 # Load from database
2874                 if ( !$found && $title ) {
2875                         wfProfileIn( __METHOD__ . '-loadtpl' );
2876                         if ( !$title->isExternal() ) {
2877                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2878                                         $text = SpecialPage::capturePath( $title );
2879                                         if ( is_string( $text ) ) {
2880                                                 $found = true;
2881                                                 $isHTML = true;
2882                                                 $this->disableCache();
2883                                         }
2884                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2885                                         $found = false; //access denied
2886                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2887                                 } else {
2888                                         list( $text, $title ) = $this->getTemplateDom( $title );
2889                                         if ( $text !== false ) {
2890                                                 $found = true;
2891                                                 $isChildObj = true;
2892                                         }
2893                                 }
2894
2895                                 # If the title is valid but undisplayable, make a link to it
2896                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2897                                         $text = "[[:$titleText]]";
2898                                         $found = true;
2899                                 }
2900                         } elseif ( $title->isTrans() ) {
2901                                 // Interwiki transclusion
2902                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2903                                         $text = $this->interwikiTransclude( $title, 'render' );
2904                                         $isHTML = true;
2905                                 } else {
2906                                         $text = $this->interwikiTransclude( $title, 'raw' );
2907                                         // Preprocess it like a template
2908                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2909                                         $isChildObj = true;
2910                                 }
2911                                 $found = true;
2912                         }
2913
2914                         # Do infinite loop check
2915                         # This has to be done after redirect resolution to avoid infinite loops via redirects
2916                         if ( !$frame->loopCheck( $title ) ) {
2917                                 $found = true;
2918                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2919                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2920                         }
2921                         wfProfileOut( __METHOD__ . '-loadtpl' );
2922                 }
2923
2924                 # If we haven't found text to substitute by now, we're done
2925                 # Recover the source wikitext and return it
2926                 if ( !$found ) {
2927                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2928                         wfProfileOut( __METHOD__ );
2929                         return array( 'object' => $text );
2930                 }
2931
2932                 # Expand DOM-style return values in a child frame
2933                 if ( $isChildObj ) {
2934                         # Clean up argument array
2935                         $newFrame = $frame->newChild( $args, $title );
2936
2937                         if ( $nowiki ) {
2938                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2939                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2940                                 # Expansion is eligible for the empty-frame cache
2941                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2942                                         $text = $this->mTplExpandCache[$titleText];
2943                                 } else {
2944                                         $text = $newFrame->expand( $text );
2945                                         $this->mTplExpandCache[$titleText] = $text;
2946                                 }
2947                         } else {
2948                                 # Uncached expansion
2949                                 $text = $newFrame->expand( $text );
2950                         }
2951                 }
2952                 if ( $isLocalObj && $nowiki ) {
2953                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2954                         $isLocalObj = false;
2955                 }
2956
2957                 # Replace raw HTML by a placeholder
2958                 # Add a blank line preceding, to prevent it from mucking up
2959                 # immediately preceding headings
2960                 if ( $isHTML ) {
2961                         $text = "\n\n" . $this->insertStripItem( $text );
2962                 }
2963                 # Escape nowiki-style return values
2964                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2965                         $text = wfEscapeWikiText( $text );
2966                 }
2967                 # Bug 529: if the template begins with a table or block-level
2968                 # element, it should be treated as beginning a new line.
2969                 # This behaviour is somewhat controversial.
2970                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
2971                         $text = "\n" . $text;
2972                 }
2973
2974                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
2975                         # Error, oversize inclusion
2976                         $text = "[[$originalTitle]]" .
2977                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
2978                         $this->limitationWarn( 'post-expand-template-inclusion' );
2979                 }
2980
2981                 if ( $isLocalObj ) {
2982                         $ret = array( 'object' => $text );
2983                 } else {
2984                         $ret = array( 'text' => $text );
2985                 }
2986
2987                 wfProfileOut( __METHOD__ );
2988                 return $ret;
2989         }
2990
2991         /**
2992          * Get the semi-parsed DOM representation of a template with a given title,
2993          * and its redirect destination title. Cached.
2994          */
2995         function getTemplateDom( $title ) {
2996                 $cacheTitle = $title;
2997                 $titleText = $title->getPrefixedDBkey();
2998
2999                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3000                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3001                         $title = Title::makeTitle( $ns, $dbk );
3002                         $titleText = $title->getPrefixedDBkey();
3003                 }
3004                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3005                         return array( $this->mTplDomCache[$titleText], $title );
3006                 }
3007
3008                 // Cache miss, go to the database
3009                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3010
3011                 if ( $text === false ) {
3012                         $this->mTplDomCache[$titleText] = false;
3013                         return array( false, $title );
3014                 }
3015
3016                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3017                 $this->mTplDomCache[ $titleText ] = $dom;
3018
3019                 if (! $title->equals($cacheTitle)) {
3020                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3021                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3022                 }
3023
3024                 return array( $dom, $title );
3025         }
3026
3027         /**
3028          * Fetch the unparsed text of a template and register a reference to it.
3029          */
3030         function fetchTemplateAndTitle( $title ) {
3031                 $templateCb = $this->mOptions->getTemplateCallback();
3032                 $stuff = call_user_func( $templateCb, $title, $this );
3033                 $text = $stuff['text'];
3034                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3035                 if ( isset( $stuff['deps'] ) ) {
3036                         foreach ( $stuff['deps'] as $dep ) {
3037                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3038                         }
3039                 }
3040                 return array($text,$finalTitle);
3041         }
3042
3043         function fetchTemplate( $title ) {
3044                 $rv = $this->fetchTemplateAndTitle($title);
3045                 return $rv[0];
3046         }
3047
3048         /**
3049          * Static function to get a template
3050          * Can be overridden via ParserOptions::setTemplateCallback().
3051          */
3052         static function statelessFetchTemplate( $title, $parser=false ) {
3053                 $text = $skip = false;
3054                 $finalTitle = $title;
3055                 $deps = array();
3056
3057                 // Loop to fetch the article, with up to 1 redirect
3058                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3059                         # Give extensions a chance to select the revision instead
3060                         $id = false; // Assume current
3061                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3062
3063                         if( $skip ) {
3064                                 $text = false;
3065                                 $deps[] = array(
3066                                         'title' => $title,
3067                                         'page_id' => $title->getArticleID(),
3068                                         'rev_id' => null );
3069                                 break;
3070                         }
3071                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3072                         $rev_id = $rev ? $rev->getId() : 0;
3073                         // If there is no current revision, there is no page
3074                         if( $id === false && !$rev ) {
3075                                 $linkCache = LinkCache::singleton();
3076                                 $linkCache->addBadLinkObj( $title );
3077                         }
3078
3079                         $deps[] = array(
3080                                 'title' => $title,
3081                                 'page_id' => $title->getArticleID(),
3082                                 'rev_id' => $rev_id );
3083
3084                         if( $rev ) {
3085                                 $text = $rev->getText();
3086                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3087                                 global $wgContLang;
3088                                 $message = $wgContLang->lcfirst( $title->getText() );
3089                                 $text = wfMsgForContentNoTrans( $message );
3090                                 if( wfEmptyMsg( $message, $text ) ) {
3091                                         $text = false;
3092                                         break;
3093                                 }
3094                         } else {
3095                                 break;
3096                         }
3097                         if ( $text === false ) {
3098                                 break;
3099                         }
3100                         // Redirect?
3101                         $finalTitle = $title;
3102                         $title = Title::newFromRedirect( $text );
3103                 }
3104                 return array(
3105                         'text' => $text,
3106                         'finalTitle' => $finalTitle,
3107                         'deps' => $deps );
3108         }
3109
3110         /**
3111          * Transclude an interwiki link.
3112          */
3113         function interwikiTransclude( $title, $action ) {
3114                 global $wgEnableScaryTranscluding;
3115
3116                 if (!$wgEnableScaryTranscluding)
3117                         return wfMsg('scarytranscludedisabled');
3118
3119                 $url = $title->getFullUrl( "action=$action" );
3120
3121                 if (strlen($url) > 255)
3122                         return wfMsg('scarytranscludetoolong');
3123                 return $this->fetchScaryTemplateMaybeFromCache($url);
3124         }
3125
3126         function fetchScaryTemplateMaybeFromCache($url) {
3127                 global $wgTranscludeCacheExpiry;
3128                 $dbr = wfGetDB(DB_SLAVE);
3129                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3130                                 array('tc_url' => $url));
3131                 if ($obj) {
3132                         $time = $obj->tc_time;
3133                         $text = $obj->tc_contents;
3134                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3135                                 return $text;
3136                         }
3137                 }
3138
3139                 $text = Http::get($url);
3140                 if (!$text)
3141                         return wfMsg('scarytranscludefailed', $url);
3142
3143                 $dbw = wfGetDB(DB_MASTER);
3144                 $dbw->replace('transcache', array('tc_url'), array(
3145                         'tc_url' => $url,
3146                         'tc_time' => time(),
3147                         'tc_contents' => $text));
3148                 return $text;
3149         }
3150
3151
3152         /**
3153          * Triple brace replacement -- used for template arguments
3154          * @private
3155          */
3156         function argSubstitution( $piece, $frame ) {
3157                 wfProfileIn( __METHOD__ );
3158
3159                 $error = false;
3160                 $parts = $piece['parts'];
3161                 $nameWithSpaces = $frame->expand( $piece['title'] );
3162                 $argName = trim( $nameWithSpaces );
3163                 $object = false;
3164                 $text = $frame->getArgument( $argName );
3165                 if (  $text === false && $parts->getLength() > 0
3166                   && (
3167                     $this->ot['html']
3168                     || $this->ot['pre']
3169                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3170                   )
3171                 ) {
3172                         # No match in frame, use the supplied default
3173                         $object = $parts->item( 0 )->getChildren();
3174                 }
3175                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3176                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3177                         $this->limitationWarn( 'post-expand-template-argument' );
3178                 }
3179
3180                 if ( $text === false && $object === false ) {
3181                         # No match anywhere
3182                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3183                 }
3184                 if ( $error !== false ) {
3185                         $text .= $error;
3186                 }
3187                 if ( $object !== false ) {
3188                         $ret = array( 'object' => $object );
3189                 } else {
3190                         $ret = array( 'text' => $text );
3191                 }
3192
3193                 wfProfileOut( __METHOD__ );
3194                 return $ret;
3195         }
3196
3197         /**
3198          * Return the text to be used for a given extension tag.
3199          * This is the ghost of strip().
3200          *
3201          * @param array $params Associative array of parameters:
3202          *     name       PPNode for the tag name
3203          *     attr       PPNode for unparsed text where tag attributes are thought to be
3204          *     attributes Optional associative array of parsed attributes
3205          *     inner      Contents of extension element
3206          *     noClose    Original text did not have a close tag
3207          * @param PPFrame $frame
3208          */
3209         function extensionSubstitution( $params, $frame ) {
3210                 global $wgRawHtml, $wgContLang;
3211
3212                 $name = $frame->expand( $params['name'] );
3213                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3214                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3215
3216                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3217
3218                 if ( $this->ot['html'] ) {
3219                         $name = strtolower( $name );
3220
3221                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3222                         if ( isset( $params['attributes'] ) ) {
3223                                 $attributes = $attributes + $params['attributes'];
3224                         }
3225                         switch ( $name ) {
3226                                 case 'html':
3227                                         if( $wgRawHtml ) {
3228                                                 $output = $content;
3229                                                 break;
3230                                         } else {
3231                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3232                                         }
3233                                 case 'nowiki':
3234                                         $content = strtr($content, array('-{' => '-&#123;', '}-' => '&#125;-'));
3235                                         $output = Xml::escapeTagsOnly( $content );
3236                                         break;
3237                                 case 'math':
3238                                         $output = $wgContLang->armourMath(
3239                                                 MathRenderer::renderMath( $content, $attributes ) );
3240                                         break;
3241                                 case 'gallery':
3242                                         $output = $this->renderImageGallery( $content, $attributes );
3243                                         break;
3244                                 default:
3245                                         if( isset( $this->mTagHooks[$name] ) ) {
3246                                                 # Workaround for PHP bug 35229 and similar
3247                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3248                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3249                                                 }
3250                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3251                                                         array( $content, $attributes, $this ) );
3252                                         } else {
3253                                                 $output = '<span class="error">Invalid tag extension name: ' .
3254                                                         htmlspecialchars( $name ) . '</span>';
3255                                         }
3256                         }
3257                 } else {
3258                         if ( is_null( $attrText ) ) {
3259                                 $attrText = '';
3260                         }
3261                         if ( isset( $params['attributes'] ) ) {
3262                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3263                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3264                                                 htmlspecialchars( $attrValue ) . '"';
3265                                 }
3266                         }
3267                         if ( $content === null ) {
3268                                 $output = "<$name$attrText/>";
3269                         } else {
3270                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3271                                 $output = "<$name$attrText>$content$close";
3272                         }
3273                 }
3274
3275                 if ( $name === 'html' || $name === 'nowiki' ) {
3276                         $this->mStripState->nowiki->setPair( $marker, $output );
3277                 } else {
3278                         $this->mStripState->general->setPair( $marker, $output );
3279                 }
3280                 return $marker;
3281         }
3282
3283         /**
3284          * Increment an include size counter
3285          *
3286          * @param string $type The type of expansion
3287          * @param integer $size The size of the text
3288          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3289          */
3290         function incrementIncludeSize( $type, $size ) {
3291                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3292                         return false;
3293                 } else {
3294                         $this->mIncludeSizes[$type] += $size;
3295                         return true;
3296                 }
3297         }
3298
3299         /**
3300          * Increment the expensive function count
3301          *
3302          * @return boolean False if the limit has been exceeded
3303          */
3304         function incrementExpensiveFunctionCount() {
3305                 global $wgExpensiveParserFunctionLimit;
3306                 $this->mExpensiveFunctionCount++;
3307                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3308                         return true;
3309                 }
3310                 return false;
3311         }
3312
3313         /**
3314          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3315          * Fills $this->mDoubleUnderscores, returns the modified text
3316          */
3317         function doDoubleUnderscore( $text ) {
3318                 wfProfileIn( __METHOD__ );
3319                 // The position of __TOC__ needs to be recorded
3320                 $mw = MagicWord::get( 'toc' );
3321                 if( $mw->match( $text ) ) {
3322                         $this->mShowToc = true;
3323                         $this->mForceTocPosition = true;
3324
3325                         // Set a placeholder. At the end we'll fill it in with the TOC.
3326                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3327
3328                         // Only keep the first one.
3329                         $text = $mw->replace( '', $text );
3330                 }
3331
3332                 // Now match and remove the rest of them
3333                 $mwa = MagicWord::getDoubleUnderscoreArray();
3334                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3335
3336                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3337                         $this->mOutput->mNoGallery = true;
3338                 }
3339                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3340                         $this->mShowToc = false;
3341                 }
3342                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3343                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3344
3345                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3346                         if ( $containerCategory ) {
3347                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3348                         } else {
3349                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3350                         }
3351                 }
3352                 # (bug 8068) Allow control over whether robots index a page.
3353                 #
3354                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3355                 # is not desirable, the last one on the page should win.
3356                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3357                         $this->mOutput->setIndexPolicy( 'noindex' );
3358                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3359                         $this->mOutput->setIndexPolicy( 'index' );
3360                 }
3361                 wfProfileOut( __METHOD__ );
3362                 return $text;
3363         }
3364
3365         /**
3366          * This function accomplishes several tasks:
3367          * 1) Auto-number headings if that option is enabled
3368          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3369          * 3) Add a Table of contents on the top for users who have enabled the option
3370          * 4) Auto-anchor headings
3371          *
3372          * It loops through all headlines, collects the necessary data, then splits up the
3373          * string and re-inserts the newly formatted headlines.
3374          *
3375          * @param string $text
3376          * @param boolean $isMain
3377          * @private
3378          */
3379         function formatHeadings( $text, $isMain=true ) {
3380                 global $wgMaxTocLevel, $wgContLang, $wgEnforceHtmlIds;
3381
3382                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3383                 $showEditLink = $this->mOptions->getEditSection();
3384
3385                 // Do not call quickUserCan unless necessary
3386                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3387                         $showEditLink = 0;
3388                 }
3389
3390                 # Inhibit editsection links if requested in the page
3391                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3392                         $showEditLink = 0;
3393                 }
3394
3395                 # Get all headlines for numbering them and adding funky stuff like [edit]
3396                 # links - this is for later, but we need the number of headlines right now
3397                 $matches = array();
3398                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3399
3400                 # if there are fewer than 4 headlines in the article, do not show TOC
3401                 # unless it's been explicitly enabled.
3402                 $enoughToc = $this->mShowToc &&
3403                         (($numMatches >= 4) || $this->mForceTocPosition);
3404
3405                 # Allow user to stipulate that a page should have a "new section"
3406                 # link added via __NEWSECTIONLINK__
3407                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3408                         $this->mOutput->setNewSection( true );
3409                 }
3410
3411                 # Allow user to remove the "new section"
3412                 # link via __NONEWSECTIONLINK__
3413                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3414                         $this->mOutput->hideNewSection( true );
3415                 }
3416
3417                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3418                 # override above conditions and always show TOC above first header
3419                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3420                         $this->mShowToc = true;
3421                         $enoughToc = true;
3422                 }
3423
3424                 # We need this to perform operations on the HTML
3425                 $sk = $this->mOptions->getSkin();
3426
3427                 # headline counter
3428                 $headlineCount = 0;
3429                 $numVisible = 0;
3430
3431                 # Ugh .. the TOC should have neat indentation levels which can be
3432                 # passed to the skin functions. These are determined here
3433                 $toc = '';
3434                 $full = '';
3435                 $head = array();
3436                 $sublevelCount = array();
3437                 $levelCount = array();
3438                 $toclevel = 0;
3439                 $level = 0;
3440                 $prevlevel = 0;
3441                 $toclevel = 0;
3442                 $prevtoclevel = 0;
3443                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3444                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3445                 $tocraw = array();
3446
3447                 foreach( $matches[3] as $headline ) {
3448                         $isTemplate = false;
3449                         $titleText = false;
3450                         $sectionIndex = false;
3451                         $numbering = '';
3452                         $markerMatches = array();
3453                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3454                                 $serial = $markerMatches[1];
3455                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3456                                 $isTemplate = ($titleText != $baseTitleText);
3457                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3458                         }
3459
3460                         if( $toclevel ) {
3461                                 $prevlevel = $level;
3462                                 $prevtoclevel = $toclevel;
3463                         }
3464                         $level = $matches[1][$headlineCount];
3465
3466                         if( $doNumberHeadings || $enoughToc ) {
3467
3468                                 if ( $level > $prevlevel ) {
3469                                         # Increase TOC level
3470                                         $toclevel++;
3471                                         $sublevelCount[$toclevel] = 0;
3472                                         if( $toclevel<$wgMaxTocLevel ) {
3473                                                 $prevtoclevel = $toclevel;
3474                                                 $toc .= $sk->tocIndent();
3475                                                 $numVisible++;
3476                                         }
3477                                 }
3478                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3479                                         # Decrease TOC level, find level to jump to
3480
3481                                         for ($i = $toclevel; $i > 0; $i--) {
3482                                                 if ( $levelCount[$i] == $level ) {
3483                                                         # Found last matching level
3484                                                         $toclevel = $i;
3485                                                         break;
3486                                                 }
3487                                                 elseif ( $levelCount[$i] < $level ) {
3488                                                         # Found first matching level below current level
3489                                                         $toclevel = $i + 1;
3490                                                         break;
3491                                                 }
3492                                         }
3493                                         if( $i == 0 ) $toclevel = 1;
3494                                         if( $toclevel<$wgMaxTocLevel ) {
3495                                                 if($prevtoclevel < $wgMaxTocLevel) {
3496                                                         # Unindent only if the previous toc level was shown :p
3497                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3498                                                         $prevtoclevel = $toclevel;
3499                                                 } else {
3500                                                         $toc .= $sk->tocLineEnd();
3501                                                 }
3502                                         }
3503                                 }
3504                                 else {
3505                                         # No change in level, end TOC line
3506                                         if( $toclevel<$wgMaxTocLevel ) {
3507                                                 $toc .= $sk->tocLineEnd();
3508                                         }
3509                                 }
3510
3511                                 $levelCount[$toclevel] = $level;
3512
3513                                 # count number of headlines for each level
3514                                 @$sublevelCount[$toclevel]++;
3515                                 $dot = 0;
3516                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3517                                         if( !empty( $sublevelCount[$i] ) ) {
3518                                                 if( $dot ) {
3519                                                         $numbering .= '.';
3520                                                 }
3521                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3522                                                 $dot = 1;
3523                                         }
3524                                 }
3525                         }
3526
3527                         # The safe header is a version of the header text safe to use for links
3528                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3529                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3530
3531                         # Remove link placeholders by the link text.
3532                         #     <!--LINK number-->
3533                         # turns into
3534                         #     link text with suffix
3535                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3536
3537                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3538                         $tocline = preg_replace(
3539                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3540                                 array( '',                          '<$1>'),
3541                                 $safeHeadline
3542                         );
3543                         $tocline = trim( $tocline );
3544
3545                         # For the anchor, strip out HTML-y stuff period
3546                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3547                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3548                         $safeHeadline = trim( $safeHeadline );
3549
3550                         # Save headline for section edit hint before it's escaped
3551                         $headlineHint = $safeHeadline;
3552
3553                         if ( $wgEnforceHtmlIds ) {
3554                                 $legacyHeadline = false;
3555                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3556                                         'noninitial' );
3557                         } else {
3558                                 # For reverse compatibility, provide an id that's
3559                                 # HTML4-compatible, like we used to.
3560                                 #
3561                                 # It may be worth noting, academically, that it's possible for
3562                                 # the legacy anchor to conflict with a non-legacy headline
3563                                 # anchor on the page.  In this case likely the "correct" thing
3564                                 # would be to either drop the legacy anchors or make sure
3565                                 # they're numbered first.  However, this would require people
3566                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3567                                 # manually, so let's not bother worrying about it.
3568                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3569                                         'noninitial' );
3570                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline, 'xml' );
3571
3572                                 if ( $legacyHeadline == $safeHeadline ) {
3573                                         # No reason to have both (in fact, we can't)
3574                                         $legacyHeadline = false;
3575                                 } elseif ( $legacyHeadline != Sanitizer::escapeId(
3576                                 $legacyHeadline, 'xml' ) ) {
3577                                         # The legacy id is invalid XML.  We used to allow this, but
3578                                         # there's no reason to do so anymore.  Backward
3579                                         # compatibility will fail slightly in this case, but it's
3580                                         # no big deal.
3581                                         $legacyHeadline = false;
3582                                 }
3583                         }
3584
3585                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3586                         # Does this apply to Unicode characters?  Because we aren't
3587                         # handling those here.
3588                         $arrayKey = strtolower( $safeHeadline );
3589                         if ( $legacyHeadline === false ) {
3590                                 $legacyArrayKey = false;
3591                         } else {
3592                                 $legacyArrayKey = strtolower( $legacyHeadline );
3593                         }
3594
3595                         # count how many in assoc. array so we can track dupes in anchors
3596                         if ( isset( $refers[$arrayKey] ) ) {
3597                                 $refers[$arrayKey]++;
3598                         } else {
3599                                 $refers[$arrayKey] = 1;
3600                         }
3601                         if ( isset( $refers[$legacyArrayKey] ) ) {
3602                                 $refers[$legacyArrayKey]++;
3603                         } else {
3604                                 $refers[$legacyArrayKey] = 1;
3605                         }
3606
3607                         # Don't number the heading if it is the only one (looks silly)
3608                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3609                                 # the two are different if the line contains a link
3610                                 $headline=$numbering . ' ' . $headline;
3611                         }
3612
3613                         # Create the anchor for linking from the TOC to the section
3614                         $anchor = $safeHeadline;
3615                         $legacyAnchor = $legacyHeadline;
3616                         if ( $refers[$arrayKey] > 1 ) {
3617                                 $anchor .= '_' . $refers[$arrayKey];
3618                         }
3619                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3620                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3621                         }
3622                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3623                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3624                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3625                         }
3626                         # give headline the correct <h#> tag
3627                         if( $showEditLink && $sectionIndex !== false ) {
3628                                 if( $isTemplate ) {
3629                                         # Put a T flag in the section identifier, to indicate to extractSections()
3630                                         # that sections inside <includeonly> should be counted.
3631                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3632                                 } else {
3633                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3634                                 }
3635                         } else {
3636                                 $editlink = '';
3637                         }
3638                         $head[$headlineCount] = $sk->makeHeadline( $level,
3639                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3640                                 $editlink, $legacyAnchor );
3641
3642                         $headlineCount++;
3643                 }
3644
3645                 $this->mOutput->setSections( $tocraw );
3646
3647                 # Never ever show TOC if no headers
3648                 if( $numVisible < 1 ) {
3649                         $enoughToc = false;
3650                 }
3651
3652                 if( $enoughToc ) {
3653                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3654                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3655                         }
3656                         $toc = $sk->tocList( $toc );
3657                 }
3658
3659                 # split up and insert constructed headlines
3660
3661                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3662                 $i = 0;
3663
3664                 foreach( $blocks as $block ) {
3665                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3666                                 # This is the [edit] link that appears for the top block of text when
3667                                 # section editing is enabled
3668
3669                                 # Disabled because it broke block formatting
3670                                 # For example, a bullet point in the top line
3671                                 # $full .= $sk->editSectionLink(0);
3672                         }
3673                         $full .= $block;
3674                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3675                                 # Top anchor now in skin
3676                                 $full = $full.$toc;
3677                         }
3678
3679                         if( !empty( $head[$i] ) ) {
3680                                 $full .= $head[$i];
3681                         }
3682                         $i++;
3683                 }
3684                 if( $this->mForceTocPosition ) {
3685                         return str_replace( '<!--MWTOC-->', $toc, $full );
3686                 } else {
3687                         return $full;
3688                 }
3689         }
3690
3691         /**
3692          * Transform wiki markup when saving a page by doing \r\n -> \n
3693          * conversion, substitting signatures, {{subst:}} templates, etc.
3694          *
3695          * @param string $text the text to transform
3696          * @param Title &$title the Title object for the current article
3697          * @param User $user the User object describing the current user
3698          * @param ParserOptions $options parsing options
3699          * @param bool $clearState whether to clear the parser state first
3700          * @return string the altered wiki markup
3701          * @public
3702          */
3703         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3704                 $this->mOptions = $options;
3705                 $this->setTitle( $title );
3706                 $this->setOutputType( self::OT_WIKI );
3707
3708                 if ( $clearState ) {
3709                         $this->clearState();
3710                 }
3711
3712                 $pairs = array(
3713                         "\r\n" => "\n",
3714                 );
3715                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3716                 $text = $this->pstPass2( $text, $user );
3717                 $text = $this->mStripState->unstripBoth( $text );
3718                 return $text;
3719         }
3720
3721         /**
3722          * Pre-save transform helper function
3723          * @private
3724          */
3725         function pstPass2( $text, $user ) {
3726                 global $wgContLang, $wgLocaltimezone;
3727
3728                 /* Note: This is the timestamp saved as hardcoded wikitext to
3729                  * the database, we use $wgContLang here in order to give
3730                  * everyone the same signature and use the default one rather
3731                  * than the one selected in each user's preferences.
3732                  *
3733                  * (see also bug 12815)
3734                  */
3735                 $ts = $this->mOptions->getTimestamp();
3736                 $tz = wfMsgForContent( 'timezone-utc' );
3737                 if ( isset( $wgLocaltimezone ) ) {
3738                         $unixts = wfTimestamp( TS_UNIX, $ts );
3739                         $oldtz = getenv( 'TZ' );
3740                         putenv( 'TZ='.$wgLocaltimezone );
3741                         $ts = date( 'YmdHis', $unixts );
3742                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3743
3744                         /* Allow translation of timezones trough wiki. date() can return
3745                          * whatever crap the system uses, localised or not, so we cannot
3746                          * ship premade translations.
3747                          */
3748                         $key = 'timezone-' . strtolower( trim( $tz ) );
3749                         $value = wfMsgForContent( $key );
3750                         if ( !wfEmptyMsg( $key, $value ) ) $tz = $value;
3751
3752                         putenv( 'TZ='.$oldtz );
3753                 }
3754
3755                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3756
3757                 # Variable replacement
3758                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3759                 $text = $this->replaceVariables( $text );
3760
3761                 # Signatures
3762                 $sigText = $this->getUserSig( $user );
3763                 $text = strtr( $text, array(
3764                         '~~~~~' => $d,
3765                         '~~~~' => "$sigText $d",
3766                         '~~~' => $sigText
3767                 ) );
3768
3769                 # Context links: [[|name]] and [[name (context)|]]
3770                 #
3771                 global $wgLegalTitleChars;
3772                 $tc = "[$wgLegalTitleChars]";
3773                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3774
3775                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3776                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3777                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3778                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3779
3780                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3781                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3782                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3783                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3784
3785                 $t = $this->mTitle->getText();
3786                 $m = array();
3787                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3788                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3789                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3790                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3791                 } else {
3792                         # if there's no context, don't bother duplicating the title
3793                         $text = preg_replace( $p2, '[[\\1]]', $text );
3794                 }
3795
3796                 # Trim trailing whitespace
3797                 $text = rtrim( $text );
3798
3799                 return $text;
3800         }
3801
3802         /**
3803          * Fetch the user's signature text, if any, and normalize to
3804          * validated, ready-to-insert wikitext.
3805          *
3806          * @param User $user
3807          * @return string
3808          * @private
3809          */
3810         function getUserSig( &$user ) {
3811                 global $wgMaxSigChars;
3812
3813                 $username = $user->getName();
3814                 $nickname = $user->getOption( 'nickname' );
3815                 $nickname = $nickname === null ? $username : $nickname;
3816
3817                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3818                         $nickname = $username;
3819                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3820                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3821                         # Sig. might contain markup; validate this
3822                         if( $this->validateSig( $nickname ) !== false ) {
3823                                 # Validated; clean up (if needed) and return it
3824                                 return $this->cleanSig( $nickname, true );
3825                         } else {
3826                                 # Failed to validate; fall back to the default
3827                                 $nickname = $username;
3828                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3829                         }
3830                 }
3831
3832                 // Make sure nickname doesnt get a sig in a sig
3833                 $nickname = $this->cleanSigInSig( $nickname );
3834
3835                 # If we're still here, make it a link to the user page
3836                 $userText = wfEscapeWikiText( $username );
3837                 $nickText = wfEscapeWikiText( $nickname );
3838                 if ( $user->isAnon() )  {
3839                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3840                 } else {
3841                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3842                 }
3843         }
3844
3845         /**
3846          * Check that the user's signature contains no bad XML
3847          *
3848          * @param string $text
3849          * @return mixed An expanded string, or false if invalid.
3850          */
3851         function validateSig( $text ) {
3852                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3853         }
3854
3855         /**
3856          * Clean up signature text
3857          *
3858          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3859          * 2) Substitute all transclusions
3860          *
3861          * @param string $text
3862          * @param $parsing Whether we're cleaning (preferences save) or parsing
3863          * @return string Signature text
3864          */
3865         function cleanSig( $text, $parsing = false ) {
3866                 if ( !$parsing ) {
3867                         global $wgTitle;
3868                         $this->clearState();
3869                         $this->setTitle( $wgTitle );
3870                         $this->mOptions = new ParserOptions;
3871                         $this->setOutputType = self::OT_PREPROCESS;
3872                 }
3873
3874                 # Option to disable this feature
3875                 if ( !$this->mOptions->getCleanSignatures() ) {
3876                         return $text;
3877                 }
3878
3879                 # FIXME: regex doesn't respect extension tags or nowiki
3880                 #  => Move this logic to braceSubstitution()
3881                 $substWord = MagicWord::get( 'subst' );
3882                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3883                 $substText = '{{' . $substWord->getSynonym( 0 );
3884
3885                 $text = preg_replace( $substRegex, $substText, $text );
3886                 $text = $this->cleanSigInSig( $text );
3887                 $dom = $this->preprocessToDom( $text );
3888                 $frame = $this->getPreprocessor()->newFrame();
3889                 $text = $frame->expand( $dom );
3890
3891                 if ( !$parsing ) {
3892                         $text = $this->mStripState->unstripBoth( $text );
3893                 }
3894
3895                 return $text;
3896         }
3897
3898         /**
3899          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3900          * @param string $text
3901          * @return string Signature text with /~{3,5}/ removed
3902          */
3903         function cleanSigInSig( $text ) {
3904                 $text = preg_replace( '/~{3,5}/', '', $text );
3905                 return $text;
3906         }
3907
3908         /**
3909          * Set up some variables which are usually set up in parse()
3910          * so that an external function can call some class members with confidence
3911          * @public
3912          */
3913         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3914                 $this->setTitle( $title );
3915                 $this->mOptions = $options;
3916                 $this->setOutputType( $outputType );
3917                 if ( $clearState ) {
3918                         $this->clearState();
3919                 }
3920         }
3921
3922         /**
3923          * Wrapper for preprocess()
3924          *
3925          * @param string $text the text to preprocess
3926          * @param ParserOptions $options  options
3927          * @return string
3928          * @public
3929          */
3930         function transformMsg( $text, $options ) {
3931                 global $wgTitle;
3932                 static $executing = false;
3933
3934                 # Guard against infinite recursion
3935                 if ( $executing ) {
3936                         return $text;
3937                 }
3938                 $executing = true;
3939
3940                 wfProfileIn(__METHOD__);
3941                 $text = $this->preprocess( $text, $wgTitle, $options );
3942
3943                 $executing = false;
3944                 wfProfileOut(__METHOD__);
3945                 return $text;
3946         }
3947
3948         /**
3949          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3950          * The callback should have the following form:
3951          *    function myParserHook( $text, $params, &$parser ) { ... }
3952          *
3953          * Transform and return $text. Use $parser for any required context, e.g. use
3954          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3955          *
3956          * @public
3957          *
3958          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3959          * @param mixed $callback The callback function (and object) to use for the tag
3960          *
3961          * @return The old value of the mTagHooks array associated with the hook
3962          */
3963         function setHook( $tag, $callback ) {
3964                 $tag = strtolower( $tag );
3965                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3966                 $this->mTagHooks[$tag] = $callback;
3967                 if( !in_array( $tag, $this->mStripList ) ) {
3968                         $this->mStripList[] = $tag;
3969                 }
3970
3971                 return $oldVal;
3972         }
3973
3974         function setTransparentTagHook( $tag, $callback ) {
3975                 $tag = strtolower( $tag );
3976                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3977                 $this->mTransparentTagHooks[$tag] = $callback;
3978
3979                 return $oldVal;
3980         }
3981
3982         /**
3983          * Remove all tag hooks
3984          */
3985         function clearTagHooks() {
3986                 $this->mTagHooks = array();
3987                 $this->mStripList = $this->mDefaultStripList;
3988         }
3989
3990         /**
3991          * Create a function, e.g. {{sum:1|2|3}}
3992          * The callback function should have the form:
3993          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3994          *
3995          * Or with SFH_OBJECT_ARGS:
3996          *    function myParserFunction( $parser, $frame, $args ) { ... }
3997          *
3998          * The callback may either return the text result of the function, or an array with the text
3999          * in element 0, and a number of flags in the other elements. The names of the flags are
4000          * specified in the keys. Valid flags are:
4001          *   found                     The text returned is valid, stop processing the template. This
4002          *                             is on by default.
4003          *   nowiki                    Wiki markup in the return value should be escaped
4004          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4005          *
4006          * @public
4007          *
4008          * @param string $id The magic word ID
4009          * @param mixed $callback The callback function (and object) to use
4010          * @param integer $flags a combination of the following flags:
4011          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4012          *
4013          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4014          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4015          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4016          *     the arguments, and to control the way they are expanded.
4017          *
4018          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4019          *     arguments, for instance:
4020          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4021          *
4022          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4023          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4024          *     working if/when this is changed.
4025          *
4026          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4027          *     expansion.
4028          *
4029          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4030          *     about the methods available in PPFrame and PPNode.
4031          *
4032          * @return The old callback function for this name, if any
4033          */
4034         function setFunctionHook( $id, $callback, $flags = 0 ) {
4035                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4036                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4037
4038                 # Add to function cache
4039                 $mw = MagicWord::get( $id );
4040                 if( !$mw )
4041                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4042
4043                 $synonyms = $mw->getSynonyms();
4044                 $sensitive = intval( $mw->isCaseSensitive() );
4045
4046                 foreach ( $synonyms as $syn ) {
4047                         # Case
4048                         if ( !$sensitive ) {
4049                                 $syn = strtolower( $syn );
4050                         }
4051                         # Add leading hash
4052                         if ( !( $flags & SFH_NO_HASH ) ) {
4053                                 $syn = '#' . $syn;
4054                         }
4055                         # Remove trailing colon
4056                         if ( substr( $syn, -1, 1 ) === ':' ) {
4057                                 $syn = substr( $syn, 0, -1 );
4058                         }
4059                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4060                 }
4061                 return $oldVal;
4062         }
4063
4064         /**
4065          * Get all registered function hook identifiers
4066          *
4067          * @return array
4068          */
4069         function getFunctionHooks() {
4070                 return array_keys( $this->mFunctionHooks );
4071         }
4072
4073         /**
4074          * FIXME: update documentation. makeLinkObj() is deprecated.
4075          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4076          * Placeholders created in Skin::makeLinkObj()
4077          * Returns an array of link CSS classes, indexed by PDBK.
4078          */
4079         function replaceLinkHolders( &$text, $options = 0 ) {
4080                 return $this->mLinkHolders->replace( $text );
4081         }
4082
4083         /**
4084          * Replace <!--LINK--> link placeholders with plain text of links
4085          * (not HTML-formatted).
4086          * @param string $text
4087          * @return string
4088          */
4089         function replaceLinkHoldersText( $text ) {
4090                 return $this->mLinkHolders->replaceText( $text );
4091         }
4092
4093         /**
4094          * Tag hook handler for 'pre'.
4095          */
4096         function renderPreTag( $text, $attribs ) {
4097                 // Backwards-compatibility hack
4098                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4099
4100                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4101                 return Xml::openElement( 'pre', $attribs ) .
4102                         Xml::escapeTagsOnly( $content ) .
4103                         '</pre>';
4104         }
4105
4106         /**
4107          * Renders an image gallery from a text with one line per image.
4108          * text labels may be given by using |-style alternative text. E.g.
4109          *   Image:one.jpg|The number "1"
4110          *   Image:tree.jpg|A tree
4111          * given as text will return the HTML of a gallery with two images,
4112          * labeled 'The number "1"' and
4113          * 'A tree'.
4114          */
4115         function renderImageGallery( $text, $params ) {
4116                 $ig = new ImageGallery();
4117                 $ig->setContextTitle( $this->mTitle );
4118                 $ig->setShowBytes( false );
4119                 $ig->setShowFilename( false );
4120                 $ig->setParser( $this );
4121                 $ig->setHideBadImages();
4122                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4123                 $ig->useSkin( $this->mOptions->getSkin() );
4124                 $ig->mRevisionId = $this->mRevisionId;
4125
4126                 if( isset( $params['caption'] ) ) {
4127                         $caption = $params['caption'];
4128                         $caption = htmlspecialchars( $caption );
4129                         $caption = $this->replaceInternalLinks( $caption );
4130                         $ig->setCaptionHtml( $caption );
4131                 }
4132                 if( isset( $params['perrow'] ) ) {
4133                         $ig->setPerRow( $params['perrow'] );
4134                 }
4135                 if( isset( $params['widths'] ) ) {
4136                         $ig->setWidths( $params['widths'] );
4137                 }
4138                 if( isset( $params['heights'] ) ) {
4139                         $ig->setHeights( $params['heights'] );
4140                 }
4141
4142                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4143
4144                 $lines = StringUtils::explode( "\n", $text );
4145                 foreach ( $lines as $line ) {
4146                         # match lines like these:
4147                         # Image:someimage.jpg|This is some image
4148                         $matches = array();
4149                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4150                         # Skip empty lines
4151                         if ( count( $matches ) == 0 ) {
4152                                 continue;
4153                         }
4154
4155                         if ( strpos( $matches[0], '%' ) !== false )
4156                                 $matches[1] = urldecode( $matches[1] );
4157                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4158                         $nt =& $tp;
4159                         if( is_null( $nt ) ) {
4160                                 # Bogus title. Ignore these so we don't bomb out later.
4161                                 continue;
4162                         }
4163                         if ( isset( $matches[3] ) ) {
4164                                 $label = $matches[3];
4165                         } else {
4166                                 $label = '';
4167                         }
4168
4169                         $html = $this->recursiveTagParse( trim( $label ) );
4170
4171                         $ig->add( $nt, $html );
4172
4173                         # Only add real images (bug #5586)
4174                         if ( $nt->getNamespace() == NS_FILE ) {
4175                                 $this->mOutput->addImage( $nt->getDBkey() );
4176                         }
4177                 }
4178                 return $ig->toHTML();
4179         }
4180
4181         function getImageParams( $handler ) {
4182                 if ( $handler ) {
4183                         $handlerClass = get_class( $handler );
4184                 } else {
4185                         $handlerClass = '';
4186                 }
4187                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4188                         // Initialise static lists
4189                         static $internalParamNames = array(
4190                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4191                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4192                                         'bottom', 'text-bottom' ),
4193                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4194                                         'upright', 'border', 'link', 'alt' ),
4195                         );
4196                         static $internalParamMap;
4197                         if ( !$internalParamMap ) {
4198                                 $internalParamMap = array();
4199                                 foreach ( $internalParamNames as $type => $names ) {
4200                                         foreach ( $names as $name ) {
4201                                                 $magicName = str_replace( '-', '_', "img_$name" );
4202                                                 $internalParamMap[$magicName] = array( $type, $name );
4203                                         }
4204                                 }
4205                         }
4206
4207                         // Add handler params
4208                         $paramMap = $internalParamMap;
4209                         if ( $handler ) {
4210                                 $handlerParamMap = $handler->getParamMap();
4211                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4212                                         $paramMap[$magic] = array( 'handler', $paramName );
4213                                 }
4214                         }
4215                         $this->mImageParams[$handlerClass] = $paramMap;
4216                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4217                 }
4218                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4219         }
4220
4221         /**
4222          * Parse image options text and use it to make an image
4223          * @param Title $title
4224          * @param string $options
4225          * @param LinkHolderArray $holders
4226          */
4227         function makeImage( $title, $options, $holders = false ) {
4228                 # Check if the options text is of the form "options|alt text"
4229                 # Options are:
4230                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4231                 #  * left       no resizing, just left align. label is used for alt= only
4232                 #  * right      same, but right aligned
4233                 #  * none       same, but not aligned
4234                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4235                 #  * center     center the image
4236                 #  * framed     Keep original image size, no magnify-button.
4237                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4238                 #  * upright    reduce width for upright images, rounded to full __0 px
4239                 #  * border     draw a 1px border around the image
4240                 #  * alt        Text for HTML alt attribute (defaults to empty)
4241                 # vertical-align values (no % or length right now):
4242                 #  * baseline
4243                 #  * sub
4244                 #  * super
4245                 #  * top
4246                 #  * text-top
4247                 #  * middle
4248                 #  * bottom
4249                 #  * text-bottom
4250
4251                 $parts = StringUtils::explode( "|", $options );
4252                 $sk = $this->mOptions->getSkin();
4253
4254                 # Give extensions a chance to select the file revision for us
4255                 $skip = $time = $descQuery = false;
4256                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4257
4258                 if ( $skip ) {
4259                         return $sk->link( $title );
4260                 }
4261
4262                 # Get the file
4263                 $imagename = $title->getDBkey();
4264                 if ( isset( $this->mFileCache[$imagename][$time] ) ) {
4265                         $file = $this->mFileCache[$imagename][$time];
4266                 } else {
4267                         $file = wfFindFile( $title, $time );
4268                         if ( count( $this->mFileCache ) > 1000 ) {
4269                                 $this->mFileCache = array();
4270                         }
4271                         $this->mFileCache[$imagename][$time] = $file;
4272                 }
4273                 # Get parameter map
4274                 $handler = $file ? $file->getHandler() : false;
4275
4276                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4277
4278                 # Process the input parameters
4279                 $caption = '';
4280                 $params = array( 'frame' => array(), 'handler' => array(),
4281                         'horizAlign' => array(), 'vertAlign' => array() );
4282                 foreach( $parts as $part ) {
4283                         $part = trim( $part );
4284                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4285                         $validated = false;
4286                         if( isset( $paramMap[$magicName] ) ) {
4287                                 list( $type, $paramName ) = $paramMap[$magicName];
4288
4289                                 // Special case; width and height come in one variable together
4290                                 if( $type === 'handler' && $paramName === 'width' ) {
4291                                         $m = array();
4292                                         # (bug 13500) In both cases (width/height and width only),
4293                                         # permit trailing "px" for backward compatibility.
4294                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4295                                                 $width = intval( $m[1] );
4296                                                 $height = intval( $m[2] );
4297                                                 if ( $handler->validateParam( 'width', $width ) ) {
4298                                                         $params[$type]['width'] = $width;
4299                                                         $validated = true;
4300                                                 }
4301                                                 if ( $handler->validateParam( 'height', $height ) ) {
4302                                                         $params[$type]['height'] = $height;
4303                                                         $validated = true;
4304                                                 }
4305                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4306                                                 $width = intval( $value );
4307                                                 if ( $handler->validateParam( 'width', $width ) ) {
4308                                                         $params[$type]['width'] = $width;
4309                                                         $validated = true;
4310                                                 }
4311                                         } // else no validation -- bug 13436
4312                                 } else {
4313                                         if ( $type === 'handler' ) {
4314                                                 # Validate handler parameter
4315                                                 $validated = $handler->validateParam( $paramName, $value );
4316                                         } else {
4317                                                 # Validate internal parameters
4318                                                 switch( $paramName ) {
4319                                                 case 'manualthumb':
4320                                                 case 'alt':
4321                                                         // @fixme - possibly check validity here for
4322                                                         // manualthumb? downstream behavior seems odd with
4323                                                         // missing manual thumbs.
4324                                                         $validated = true;
4325                                                         $value = $this->stripAltText( $value, $holders );
4326                                                         break;
4327                                                 case 'link':
4328                                                         $chars = self::EXT_LINK_URL_CLASS;
4329                                                         $prots = $this->mUrlProtocols;
4330                                                         if ( $value === '' ) {
4331                                                                 $paramName = 'no-link';
4332                                                                 $value = true;
4333                                                                 $validated = true;
4334                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4335                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4336                                                                         $paramName = 'link-url';
4337                                                                         $this->mOutput->addExternalLink( $value );
4338                                                                         $validated = true;
4339                                                                 }
4340                                                         } else {
4341                                                                 $linkTitle = Title::newFromText( $value );
4342                                                                 if ( $linkTitle ) {
4343                                                                         $paramName = 'link-title';
4344                                                                         $value = $linkTitle;
4345                                                                         $this->mOutput->addLink( $linkTitle );
4346                                                                         $validated = true;
4347                                                                 }
4348                                                         }
4349                                                         break;
4350                                                 default:
4351                                                         // Most other things appear to be empty or numeric...
4352                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4353                                                 }
4354                                         }
4355
4356                                         if ( $validated ) {
4357                                                 $params[$type][$paramName] = $value;
4358                                         }
4359                                 }
4360                         }
4361                         if ( !$validated ) {
4362                                 $caption = $part;
4363                         }
4364                 }
4365
4366                 # Process alignment parameters
4367                 if ( $params['horizAlign'] ) {
4368                         $params['frame']['align'] = key( $params['horizAlign'] );
4369                 }
4370                 if ( $params['vertAlign'] ) {
4371                         $params['frame']['valign'] = key( $params['vertAlign'] );
4372                 }
4373
4374                 $params['frame']['caption'] = $caption;
4375
4376                 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4377
4378                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4379                 # came to also set the caption, ordinary text after the image -- which
4380                 # makes no sense, because that just repeats the text multiple times in
4381                 # screen readers.  It *also* came to set the title attribute.
4382                 #
4383                 # Now that we have an alt attribute, we should not set the alt text to
4384                 # equal the caption: that's worse than useless, it just repeats the
4385                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4386                 # use the unnamed parameter for alt text as well, just for the time be-
4387                 # ing, if the unnamed param is set and the alt param is not.
4388                 #
4389                 # For the future, we need to figure out if we want to tweak this more,
4390                 # e.g., introducing a title= parameter for the title; ignoring the un-
4391                 # named parameter entirely for images without a caption; adding an ex-
4392                 # plicit caption= parameter and preserving the old magic unnamed para-
4393                 # meter for BC; ...
4394                 if( $caption !== '' && !isset( $params['frame']['alt'] )
4395                 && !isset( $params['frame']['framed'] )
4396                 && !isset( $params['frame']['thumbnail'] )
4397                 && !isset( $params['frame']['manualthumb'] ) ) {
4398                         $params['frame']['alt'] = $params['frame']['title'];
4399                 }
4400
4401                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4402
4403                 # Linker does the rest
4404                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4405
4406                 # Give the handler a chance to modify the parser object
4407                 if ( $handler ) {
4408                         $handler->parserTransformHook( $this, $file );
4409                 }
4410
4411                 return $ret;
4412         }
4413
4414         protected function stripAltText( $caption, $holders ) {
4415                 # Strip bad stuff out of the title (tooltip).  We can't just use
4416                 # replaceLinkHoldersText() here, because if this function is called
4417                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4418                 if ( $holders ) {
4419                         $tooltip = $holders->replaceText( $caption );
4420                 } else {
4421                         $tooltip = $this->replaceLinkHoldersText( $caption );
4422                 }
4423
4424                 # make sure there are no placeholders in thumbnail attributes
4425                 # that are later expanded to html- so expand them now and
4426                 # remove the tags
4427                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4428                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4429
4430                 return $tooltip;
4431         }
4432
4433         /**
4434          * Set a flag in the output object indicating that the content is dynamic and
4435          * shouldn't be cached.
4436          */
4437         function disableCache() {
4438                 wfDebug( "Parser output marked as uncacheable.\n" );
4439                 $this->mOutput->mCacheTime = -1;
4440         }
4441
4442         /**#@+
4443          * Callback from the Sanitizer for expanding items found in HTML attribute
4444          * values, so they can be safely tested and escaped.
4445          * @param string $text
4446          * @param PPFrame $frame
4447          * @return string
4448          * @private
4449          */
4450         function attributeStripCallback( &$text, $frame = false ) {
4451                 $text = $this->replaceVariables( $text, $frame );
4452                 $text = $this->mStripState->unstripBoth( $text );
4453                 return $text;
4454         }
4455
4456         /**#@-*/
4457
4458         /**#@+
4459          * Accessor/mutator
4460          */
4461         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4462         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4463         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4464         /**#@-*/
4465
4466         /**#@+
4467          * Accessor
4468          */
4469         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4470         /**#@-*/
4471
4472
4473         /**
4474          * Break wikitext input into sections, and either pull or replace
4475          * some particular section's text.
4476          *
4477          * External callers should use the getSection and replaceSection methods.
4478          *
4479          * @param string $text Page wikitext
4480          * @param string $section A section identifier string of the form:
4481          *   <flag1> - <flag2> - ... - <section number>
4482          *
4483          * Currently the only recognised flag is "T", which means the target section number
4484          * was derived during a template inclusion parse, in other words this is a template
4485          * section edit link. If no flags are given, it was an ordinary section edit link.
4486          * This flag is required to avoid a section numbering mismatch when a section is
4487          * enclosed by <includeonly> (bug 6563).
4488          *
4489          * The section number 0 pulls the text before the first heading; other numbers will
4490          * pull the given section along with its lower-level subsections. If the section is
4491          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4492          *
4493          * @param string $mode One of "get" or "replace"
4494          * @param string $newText Replacement text for section data.
4495          * @return string for "get", the extracted section text.
4496          *                for "replace", the whole page with the section replaced.
4497          */
4498         private function extractSections( $text, $section, $mode, $newText='' ) {
4499                 global $wgTitle;
4500                 $this->clearState();
4501                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4502                 $this->mOptions = new ParserOptions;
4503                 $this->setOutputType( self::OT_WIKI );
4504                 $outText = '';
4505                 $frame = $this->getPreprocessor()->newFrame();
4506
4507                 // Process section extraction flags
4508                 $flags = 0;
4509                 $sectionParts = explode( '-', $section );
4510                 $sectionIndex = array_pop( $sectionParts );
4511                 foreach ( $sectionParts as $part ) {
4512                         if ( $part === 'T' ) {
4513                                 $flags |= self::PTD_FOR_INCLUSION;
4514                         }
4515                 }
4516                 // Preprocess the text
4517                 $root = $this->preprocessToDom( $text, $flags );
4518
4519                 // <h> nodes indicate section breaks
4520                 // They can only occur at the top level, so we can find them by iterating the root's children
4521                 $node = $root->getFirstChild();
4522
4523                 // Find the target section
4524                 if ( $sectionIndex == 0 ) {
4525                         // Section zero doesn't nest, level=big
4526                         $targetLevel = 1000;
4527                 } else {
4528             while ( $node ) {
4529                 if ( $node->getName() === 'h' ) {
4530                     $bits = $node->splitHeading();
4531                                         if ( $bits['i'] == $sectionIndex ) {
4532                                         $targetLevel = $bits['level'];
4533                                                 break;
4534                                         }
4535                                 }
4536                                 if ( $mode === 'replace' ) {
4537                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4538                                 }
4539                                 $node = $node->getNextSibling();
4540                         }
4541                 }
4542
4543                 if ( !$node ) {
4544                         // Not found
4545                         if ( $mode === 'get' ) {
4546                                 return $newText;
4547                         } else {
4548                                 return $text;
4549                         }
4550                 }
4551
4552                 // Find the end of the section, including nested sections
4553                 do {
4554                         if ( $node->getName() === 'h' ) {
4555                                 $bits = $node->splitHeading();
4556                                 $curLevel = $bits['level'];
4557                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4558                                         break;
4559                                 }
4560                         }
4561                         if ( $mode === 'get' ) {
4562                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4563                         }
4564                         $node = $node->getNextSibling();
4565                 } while ( $node );
4566
4567                 // Write out the remainder (in replace mode only)
4568                 if ( $mode === 'replace' ) {
4569                         // Output the replacement text
4570                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4571                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4572                         // Only add trailing whitespace if there is newText
4573                         if($newText != "") {
4574                                 $outText .= $newText . "\n\n";
4575                         }
4576
4577                         while ( $node ) {
4578                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4579                                 $node = $node->getNextSibling();
4580                         }
4581                 }
4582
4583                 if ( is_string( $outText ) ) {
4584                         // Re-insert stripped tags
4585                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4586                 }
4587
4588                 return $outText;
4589         }
4590
4591         /**
4592          * This function returns the text of a section, specified by a number ($section).
4593          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4594          * the first section before any such heading (section 0).
4595          *
4596          * If a section contains subsections, these are also returned.
4597          *
4598          * @param string $text text to look in
4599          * @param string $section section identifier
4600          * @param string $deftext default to return if section is not found
4601          * @return string text of the requested section
4602          */
4603         public function getSection( $text, $section, $deftext='' ) {
4604                 return $this->extractSections( $text, $section, "get", $deftext );
4605         }
4606
4607         public function replaceSection( $oldtext, $section, $text ) {
4608                 return $this->extractSections( $oldtext, $section, "replace", $text );
4609         }
4610
4611         /**
4612          * Get the timestamp associated with the current revision, adjusted for
4613          * the default server-local timestamp
4614          */
4615         function getRevisionTimestamp() {
4616                 if ( is_null( $this->mRevisionTimestamp ) ) {
4617                         wfProfileIn( __METHOD__ );
4618                         global $wgContLang;
4619                         $dbr = wfGetDB( DB_SLAVE );
4620                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4621                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4622
4623                         // Normalize timestamp to internal MW format for timezone processing.
4624                         // This has the added side-effect of replacing a null value with
4625                         // the current time, which gives us more sensible behavior for
4626                         // previews.
4627                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4628
4629                         // The cryptic '' timezone parameter tells to use the site-default
4630                         // timezone offset instead of the user settings.
4631                         //
4632                         // Since this value will be saved into the parser cache, served
4633                         // to other users, and potentially even used inside links and such,
4634                         // it needs to be consistent for all visitors.
4635                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4636
4637                         wfProfileOut( __METHOD__ );
4638                 }
4639                 return $this->mRevisionTimestamp;
4640         }
4641
4642         /**
4643          * Get the name of the user that edited the last revision
4644          */
4645         function getRevisionUser() {
4646                 // if this template is subst: the revision id will be blank,
4647                 // so just use the current user's name
4648                 if( $this->mRevisionId ) {
4649                         $revision = Revision::newFromId( $this->mRevisionId );
4650                         $revuser = $revision->getUserText();
4651                 } else {
4652                         global $wgUser;
4653                         $revuser = $wgUser->getName();
4654                 }
4655                 return $revuser;
4656         }
4657
4658         /**
4659          * Mutator for $mDefaultSort
4660          *
4661          * @param $sort New value
4662          */
4663         public function setDefaultSort( $sort ) {
4664                 $this->mDefaultSort = $sort;
4665         }
4666
4667         /**
4668          * Accessor for $mDefaultSort
4669          * Will use the title/prefixed title if none is set
4670          *
4671          * @return string
4672          */
4673         public function getDefaultSort() {
4674                 global $wgCategoryPrefixedDefaultSortkey;
4675                 if( $this->mDefaultSort !== false ) {
4676                         return $this->mDefaultSort;
4677                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4678                         !$wgCategoryPrefixedDefaultSortkey) {
4679                         return $this->mTitle->getText();
4680                 } else {
4681                         return $this->mTitle->getPrefixedText();
4682                 }
4683         }
4684
4685         /**
4686          * Accessor for $mDefaultSort
4687          * Unlike getDefaultSort(), will return false if none is set
4688          *
4689          * @return string or false
4690          */
4691         public function getCustomDefaultSort() {
4692                 return $this->mDefaultSort;
4693         }
4694
4695         /**
4696          * Try to guess the section anchor name based on a wikitext fragment
4697          * presumably extracted from a heading, for example "Header" from
4698          * "== Header ==".
4699          */
4700         public function guessSectionNameFromWikiText( $text ) {
4701                 # Strip out wikitext links(they break the anchor)
4702                 $text = $this->stripSectionName( $text );
4703                 $headline = Sanitizer::decodeCharReferences( $text );
4704                 # strip out HTML
4705                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4706                 $headline = trim( $headline );
4707                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4708                 $replacearray = array(
4709                         '%3A' => ':',
4710                         '%' => '.'
4711                 );
4712                 return str_replace(
4713                         array_keys( $replacearray ),
4714                         array_values( $replacearray ),
4715                         $sectionanchor );
4716         }
4717
4718         /**
4719          * Strips a text string of wikitext for use in a section anchor
4720          *
4721          * Accepts a text string and then removes all wikitext from the
4722          * string and leaves only the resultant text (i.e. the result of
4723          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4724          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4725          * to create valid section anchors by mimicing the output of the
4726          * parser when headings are parsed.
4727          *
4728          * @param $text string Text string to be stripped of wikitext
4729          * for use in a Section anchor
4730          * @return Filtered text string
4731          */
4732         public function stripSectionName( $text ) {
4733                 # Strip internal link markup
4734                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4735                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4736
4737                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4738                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4739                 # on how many empty links there are on the page - need to figure that out.
4740                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4741
4742                 # Parse wikitext quotes (italics & bold)
4743                 $text = $this->doQuotes($text);
4744
4745                 # Strip HTML tags
4746                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4747                 return $text;
4748         }
4749
4750         function srvus( $text ) {
4751                 return $this->testSrvus( $text, $this->mOutputType );
4752         }
4753
4754         /**
4755          * strip/replaceVariables/unstrip for preprocessor regression testing
4756          */
4757         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4758                 $this->clearState();
4759                 if ( ! ( $title instanceof Title ) ) {
4760                         $title = Title::newFromText( $title );
4761                 }
4762                 $this->mTitle = $title;
4763                 $this->mOptions = $options;
4764                 $this->setOutputType( $outputType );
4765                 $text = $this->replaceVariables( $text );
4766                 $text = $this->mStripState->unstripBoth( $text );
4767                 $text = Sanitizer::removeHTMLtags( $text );
4768                 return $text;
4769         }
4770
4771         function testPst( $text, $title, $options ) {
4772                 global $wgUser;
4773                 if ( ! ( $title instanceof Title ) ) {
4774                         $title = Title::newFromText( $title );
4775                 }
4776                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4777         }
4778
4779         function testPreprocess( $text, $title, $options ) {
4780                 if ( ! ( $title instanceof Title ) ) {
4781                         $title = Title::newFromText( $title );
4782                 }
4783                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4784         }
4785
4786         function markerSkipCallback( $s, $callback ) {
4787                 $i = 0;
4788                 $out = '';
4789                 while ( $i < strlen( $s ) ) {
4790                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4791                         if ( $markerStart === false ) {
4792                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4793                                 break;
4794                         } else {
4795                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4796                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4797                                 if ( $markerEnd === false ) {
4798                                         $out .= substr( $s, $markerStart );
4799                                         break;
4800                                 } else {
4801                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4802                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4803                                         $i = $markerEnd;
4804                                 }
4805                         }
4806                 }
4807                 return $out;
4808         }
4809
4810         function serialiseHalfParsedText( $text ) {
4811                 $data = array();
4812                 $data['text'] = $text;
4813
4814                 // First, find all strip markers, and store their
4815                 //  data in an array.
4816                 $stripState = new StripState;
4817                 $pos = 0;
4818                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
4819                         $end_pos += strlen( self::MARKER_SUFFIX );
4820                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
4821
4822                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
4823                                 $replaceArray = $stripState->general;
4824                                 $stripText = $this->mStripState->general->data[$marker];
4825                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
4826                                 $replaceArray = $stripState->nowiki;
4827                                 $stripText = $this->mStripState->nowiki->data[$marker];
4828                         } else {
4829                                 throw new MWException( "Hanging strip marker: '$marker'." );
4830                         }
4831
4832                         $replaceArray->setPair( $marker, $stripText );
4833                         $pos = $end_pos;
4834                 }
4835                 $data['stripstate'] = $stripState;
4836
4837                 // Now, find all of our links, and store THEIR
4838                 //  data in an array! :)
4839                 $links = array( 'internal' => array(), 'interwiki' => array() );
4840                 $pos = 0;
4841
4842                 // Internal links
4843                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
4844                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
4845
4846                         $ns = trim($ns);
4847                         if (empty( $links['internal'][$ns] )) {
4848                                 $links['internal'][$ns] = array();
4849                         }
4850
4851                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
4852                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
4853                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
4854                 }
4855
4856                 $pos = 0;
4857
4858                 // Interwiki links
4859                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
4860                         $data = substr( $text, $start_pos );
4861                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
4862                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
4863                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
4864                 }
4865
4866                 $data['linkholder'] = $links;
4867
4868                 return $data;
4869         }
4870
4871         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
4872                 if (!$intPrefix)
4873                         $intPrefix = $this->getRandomString();
4874
4875                 // First, extract the strip state.
4876                 $stripState = $data['stripstate'];
4877                 $this->mStripState->general->merge( $stripState->general );
4878                 $this->mStripState->nowiki->merge( $stripState->nowiki );
4879
4880                 // Now, extract the text, and renumber links
4881                 $text = $data['text'];
4882                 $links = $data['linkholder'];
4883
4884                 // Internal...
4885                 foreach( $links['internal'] as $ns => $nsLinks ) {
4886                         foreach( $nsLinks as $key => $entry ) {
4887                                 $newKey = $intPrefix . '-' . $key;
4888                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
4889
4890                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
4891                         }
4892                 }
4893
4894                 // Interwiki...
4895                 foreach( $links['interwiki'] as $key => $entry ) {
4896                         $newKey = "$intPrefix-$key";
4897                         $this->mLinkHolders->interwikis[$newKey] = $entry;
4898
4899                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
4900                 }
4901
4902                 // Should be good to go.
4903                 return $text;
4904         }
4905 }
4906
4907 /**
4908  * @todo document, briefly.
4909  * @ingroup Parser
4910  */
4911 class StripState {
4912         var $general, $nowiki;
4913
4914         function __construct() {
4915                 $this->general = new ReplacementArray;
4916                 $this->nowiki = new ReplacementArray;
4917         }
4918
4919         function unstripGeneral( $text ) {
4920                 wfProfileIn( __METHOD__ );
4921                 do {
4922                         $oldText = $text;
4923                         $text = $this->general->replace( $text );
4924                 } while ( $text !== $oldText );
4925                 wfProfileOut( __METHOD__ );
4926                 return $text;
4927         }
4928
4929         function unstripNoWiki( $text ) {
4930                 wfProfileIn( __METHOD__ );
4931                 do {
4932                         $oldText = $text;
4933                         $text = $this->nowiki->replace( $text );
4934                 } while ( $text !== $oldText );
4935                 wfProfileOut( __METHOD__ );
4936                 return $text;
4937         }
4938
4939         function unstripBoth( $text ) {
4940                 wfProfileIn( __METHOD__ );
4941                 do {
4942                         $oldText = $text;
4943                         $text = $this->general->replace( $text );
4944                         $text = $this->nowiki->replace( $text );
4945                 } while ( $text !== $oldText );
4946                 wfProfileOut( __METHOD__ );
4947                 return $text;
4948         }
4949 }
4950
4951 /**
4952  * @todo document, briefly.
4953  * @ingroup Parser
4954  */
4955 class OnlyIncludeReplacer {
4956         var $output = '';
4957
4958         function replace( $matches ) {
4959                 if ( substr( $matches[1], -1 ) === "\n" ) {
4960                         $this->output .= substr( $matches[1], 0, -1 );
4961                 } else {
4962                         $this->output .= $matches[1];
4963                 }
4964         }
4965 }