includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * preprocess()
  24  *   removes HTML comments and expands templates
  25  * cleanSig()
  26  *   Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *   Extracts sections from an article for section editing
  29  *
  30  * Globals used:
  31  *    objects:   $wgLang, $wgContLang
  32  *
  33  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  34  *
  35  * settings:
  36  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  37  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  38  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  39  *  $wgMaxArticleSize*
  40  *
  41  *  * only within ParserOptions
  42  * </pre>
  43  *
  44  * @ingroup Parser
  45  */
  46 class Parser
  47 {
  48         /**
  49          * Update this version number when the ParserOutput format
  50          * changes in an incompatible way, so the parser cache
  51          * can automatically discard old data.
  52          */
  53         const VERSION = '1.6.4';
  54
  55         # Flags for Parser::setFunctionHook
  56         # Also available as global constants from Defines.php
  57         const SFH_NO_HASH = 1;
  58         const SFH_OBJECT_ARGS = 2;
  59
  60         # Constants needed for external link processing
  61         # Everything except bracket, space, or control characters
  62         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  63         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  64                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  65
  66         // State constants for the definition list colon extraction
  67         const COLON_STATE_TEXT = 0;
  68         const COLON_STATE_TAG = 1;
  69         const COLON_STATE_TAGSTART = 2;
  70         const COLON_STATE_CLOSETAG = 3;
  71         const COLON_STATE_TAGSLASH = 4;
  72         const COLON_STATE_COMMENT = 5;
  73         const COLON_STATE_COMMENTDASH = 6;
  74         const COLON_STATE_COMMENTDASHDASH = 7;
  75
  76         // Flags for preprocessToDom
  77         const PTD_FOR_INCLUSION = 1;
  78
  79         // Allowed values for $this->mOutputType
  80         // Parameter to startExternalParse().
  81         const OT_HTML = 1;
  82         const OT_WIKI = 2;
  83         const OT_PREPROCESS = 3;
  84         const OT_MSG = 3;
  85
  86         // Marker Suffix needs to be accessible staticly.
  87         const MARKER_SUFFIX = "-QINU\x7f";
  88
  89         /**#@+
  90          * @private
  91          */
  92         # Persistent:
  93         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  94                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex, $mPreprocessor,
  95                 $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList, $mVarCache, $mConf;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mLinkHolders, $mLinkID;
 102         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 103         var $mTplExpandCache; // empty-frame expansion cache
 104         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 105         var $mExpensiveFunctionCount; // number of expensive parser function calls
 106         var $mFileCache;
 107
 108         # Temporary
 109         # These are variables reset at least once per parse regardless of $clearState
 110         var $mOptions,      // ParserOptions object
 111                 $mTitle,        // Title context, used for self-link rendering and similar things
 112                 $mOutputType,   // Output type, one of the OT_xxx constants
 113                 $ot,            // Shortcut alias, see setOutputType()
 114                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 115                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 116                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 117
 118         /**#@-*/
 119
 120         /**
 121          * Constructor
 122          *
 123          * @public
 124          */
 125         function __construct( $conf = array() ) {
 126                 $this->mConf = $conf;
 127                 $this->mTagHooks = array();
 128                 $this->mTransparentTagHooks = array();
 129                 $this->mFunctionHooks = array();
 130                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 131                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 132                 $this->mUrlProtocols = wfUrlProtocols();
 133                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 134                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 135                 $this->mVarCache = array();
 136                 if ( isset( $conf['preprocessorClass'] ) ) {
 137                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 138                 } elseif ( extension_loaded( 'domxml' ) ) {
 139                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 140                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 141                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 142                 } elseif ( extension_loaded( 'dom' ) ) {
 143                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 144                 } else {
 145                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 146                 }
 147                 $this->mMarkerIndex = 0;
 148                 $this->mFirstCall = true;
 149         }
 150
 151         /**
 152          * Reduce memory usage to reduce the impact of circular references
 153          */
 154         function __destruct() {
 155                 if ( isset( $this->mLinkHolders ) ) {
 156                         $this->mLinkHolders->__destruct();
 157                 }
 158                 foreach ( $this as $name => $value ) {
 159                         unset( $this->$name );
 160                 }
 161         }
 162
 163         /**
 164          * Do various kinds of initialisation on the first call of the parser
 165          */
 166         function firstCallInit() {
 167                 if ( !$this->mFirstCall ) {
 168                         return;
 169                 }
 170                 $this->mFirstCall = false;
 171
 172                 wfProfileIn( __METHOD__ );
 173
 174                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 175                 CoreParserFunctions::register( $this );
 176                 $this->initialiseVariables();
 177
 178                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 179                 wfProfileOut( __METHOD__ );
 180         }
 181
 182         /**
 183          * Clear Parser state
 184          *
 185          * @private
 186          */
 187         function clearState() {
 188                 wfProfileIn( __METHOD__ );
 189                 if ( $this->mFirstCall ) {
 190                         $this->firstCallInit();
 191                 }
 192                 $this->mOutput = new ParserOutput;
 193                 $this->mAutonumber = 0;
 194                 $this->mLastSection = '';
 195                 $this->mDTopen = false;
 196                 $this->mIncludeCount = array();
 197                 $this->mStripState = new StripState;
 198                 $this->mArgStack = false;
 199                 $this->mInPre = false;
 200                 $this->mLinkHolders = new LinkHolderArray( $this );
 201                 $this->mLinkID = 0;
 202                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 203
 204                 /**
 205                  * Prefix for temporary replacement strings for the multipass parser.
 206                  * \x07 should never appear in input as it's disallowed in XML.
 207                  * Using it at the front also gives us a little extra robustness
 208                  * since it shouldn't match when butted up against identifier-like
 209                  * string constructs.
 210                  *
 211                  * Must not consist of all title characters, or else it will change
 212                  * the behaviour of <nowiki> in a link.
 213                  */
 214                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 215                 # Changed to \x7f to allow XML double-parsing -- TS
 216                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 217
 218
 219                 # Clear these on every parse, bug 4549
 220                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 221
 222                 $this->mShowToc = true;
 223                 $this->mForceTocPosition = false;
 224                 $this->mIncludeSizes = array(
 225                         'post-expand' => 0,
 226                         'arg' => 0,
 227                 );
 228                 $this->mPPNodeCount = 0;
 229                 $this->mDefaultSort = false;
 230                 $this->mHeadings = array();
 231                 $this->mDoubleUnderscores = array();
 232                 $this->mExpensiveFunctionCount = 0;
 233                 $this->mFileCache = array();
 234
 235                 # Fix cloning
 236                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 237                         $this->mPreprocessor = null;
 238                 }
 239
 240                 wfRunHooks( 'ParserClearState', array( &$this ) );
 241                 wfProfileOut( __METHOD__ );
 242         }
 243
 244         function setOutputType( $ot ) {
 245                 $this->mOutputType = $ot;
 246                 // Shortcut alias
 247                 $this->ot = array(
 248                         'html' => $ot == self::OT_HTML,
 249                         'wiki' => $ot == self::OT_WIKI,
 250                         'pre' => $ot == self::OT_PREPROCESS,
 251                 );
 252         }
 253
 254         /**
 255          * Set the context title
 256          */
 257         function setTitle( $t ) {
 258                 if ( !$t || $t instanceof FakeTitle ) {
 259                         $t = Title::newFromText( 'NO TITLE' );
 260                 }
 261                 if ( strval( $t->getFragment() ) !== '' ) {
 262                         # Strip the fragment to avoid various odd effects
 263                         $this->mTitle = clone $t;
 264                         $this->mTitle->setFragment( '' );
 265                 } else {
 266                         $this->mTitle = $t;
 267                 }
 268         }
 269
 270         /**
 271          * Accessor for mUniqPrefix.
 272          *
 273          * @public
 274          */
 275         function uniqPrefix() {
 276                 if( !isset( $this->mUniqPrefix ) ) {
 277                         // @fixme this is probably *horribly wrong*
 278                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 279                         // if this is called for a parser cache hit, the parser may not
 280                         // have ever been initialized in the first place.
 281                         // Not really sure what the heck is supposed to be going on here.
 282                         return '';
 283                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 284                 }
 285                 return $this->mUniqPrefix;
 286         }
 287
 288         /**
 289          * Convert wikitext to HTML
 290          * Do not call this function recursively.
 291          *
 292          * @param $text String: text we want to parse
 293          * @param $title A title object
 294          * @param $options ParserOptions
 295          * @param $linestart boolean
 296          * @param $clearState boolean
 297          * @param $revid Int: number to pass in {{REVISIONID}}
 298          * @return ParserOutput a ParserOutput
 299          */
 300         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 301                 /**
 302                  * First pass--just handle <nowiki> sections, pass the rest off
 303                  * to internalParse() which does all the real work.
 304                  */
 305
 306                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 307                 $fname = __METHOD__.'-' . wfGetCaller();
 308                 wfProfileIn( __METHOD__ );
 309                 wfProfileIn( $fname );
 310
 311                 if ( $clearState ) {
 312                         $this->clearState();
 313                 }
 314
 315                 $this->mOptions = $options;
 316                 $this->setTitle( $title );
 317                 $oldRevisionId = $this->mRevisionId;
 318                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 319                 if( $revid !== null ) {
 320                         $this->mRevisionId = $revid;
 321                         $this->mRevisionTimestamp = null;
 322                 }
 323                 $this->setOutputType( self::OT_HTML );
 324                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 325                 # No more strip!
 326                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 327                 $text = $this->internalParse( $text );
 328                 $text = $this->mStripState->unstripGeneral( $text );
 329
 330                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 331                 $fixtags = array(
 332                         # french spaces, last one Guillemet-left
 333                         # only if there is something before the space
 334                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 335                         # french spaces, Guillemet-right
 336                         '/(\\302\\253) /' => '\\1&nbsp;',
 337                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 338                 );
 339                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 340
 341                 $text = $this->doBlockLevels( $text, $linestart );
 342
 343                 $this->replaceLinkHolders( $text );
 344
 345                 # the position of the parserConvert() call should not be changed. it
 346                 # assumes that the links are all replaced and the only thing left
 347                 # is the <nowiki> mark.
 348                 # Side-effects: this calls $this->mOutput->setTitleText()
 349                 $text = $wgContLang->parserConvert( $text, $this );
 350
 351                 $text = $this->mStripState->unstripNoWiki( $text );
 352
 353                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 354
 355 //!JF Move to its own function
 356
 357                 $uniq_prefix = $this->mUniqPrefix;
 358                 $matches = array();
 359                 $elements = array_keys( $this->mTransparentTagHooks );
 360                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 361
 362                 foreach( $matches as $marker => $data ) {
 363                         list( $element, $content, $params, $tag ) = $data;
 364                         $tagName = strtolower( $element );
 365                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 366                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 367                                         array( $content, $params, $this ) );
 368                         } else {
 369                                 $output = $tag;
 370                         }
 371                         $this->mStripState->general->setPair( $marker, $output );
 372                 }
 373                 $text = $this->mStripState->unstripGeneral( $text );
 374
 375                 $text = Sanitizer::normalizeCharReferences( $text );
 376
 377                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 378                         $text = self::tidy($text);
 379                 } else {
 380                         # attempt to sanitize at least some nesting problems
 381                         # (bug #2702 and quite a few others)
 382                         $tidyregs = array(
 383                                 # ''Something [http://www.cool.com cool''] -->
 384                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 385                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 386                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 387                                 # fix up an anchor inside another anchor, only
 388                                 # at least for a single single nested link (bug 3695)
 389                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 390                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 391                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 392                                 # contains a div, so fix it up here; replace
 393                                 # div with escaped text
 394                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 395                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 396                                 # remove empty italic or bold tag pairs, some
 397                                 # introduced by rules above
 398                                 '/<([bi])><\/\\1>/' => '',
 399                         );
 400
 401                         $text = preg_replace(
 402                                 array_keys( $tidyregs ),
 403                                 array_values( $tidyregs ),
 404                                 $text );
 405                 }
 406                 global $wgExpensiveParserFunctionLimit;
 407                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 408                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 409                 }
 410
 411                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 412
 413                 # Information on include size limits, for the benefit of users who try to skirt them
 414                 if ( $this->mOptions->getEnableLimitReport() ) {
 415                         global $wgExpensiveParserFunctionLimit;
 416                         $max = $this->mOptions->getMaxIncludeSize();
 417                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 418                         $limitReport =
 419                                 "NewPP limit report\n" .
 420                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 421                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 422                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 423                                 $PFreport;
 424                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 425                         $text .= "\n<!-- \n$limitReport-->\n";
 426                 }
 427                 $this->mOutput->setText( $text );
 428                 $this->mRevisionId = $oldRevisionId;
 429                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 430                 wfProfileOut( $fname );
 431                 wfProfileOut( __METHOD__ );
 432
 433                 return $this->mOutput;
 434         }
 435
 436         /**
 437          * Recursive parser entry point that can be called from an extension tag
 438          * hook.
 439          */
 440         function recursiveTagParse( $text ) {
 441                 wfProfileIn( __METHOD__ );
 442                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 443                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 444                 $text = $this->internalParse( $text );
 445                 wfProfileOut( __METHOD__ );
 446                 return $text;
 447         }
 448
 449         /**
 450          * Expand templates and variables in the text, producing valid, static wikitext.
 451          * Also removes comments.
 452          */
 453         function preprocess( $text, $title, $options, $revid = null ) {
 454                 wfProfileIn( __METHOD__ );
 455                 $this->clearState();
 456                 $this->setOutputType( self::OT_PREPROCESS );
 457                 $this->mOptions = $options;
 458                 $this->setTitle( $title );
 459                 if( $revid !== null ) {
 460                         $this->mRevisionId = $revid;
 461                 }
 462                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 463                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 464                 $text = $this->replaceVariables( $text );
 465                 $text = $this->mStripState->unstripBoth( $text );
 466                 wfProfileOut( __METHOD__ );
 467                 return $text;
 468         }
 469
 470         /**
 471          * Get a random string
 472          *
 473          * @private
 474          * @static
 475          */
 476         function getRandomString() {
 477                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 478         }
 479
 480         function &getTitle() { return $this->mTitle; }
 481         function getOptions() { return $this->mOptions; }
 482         function getRevisionId() { return $this->mRevisionId; }
 483         function getOutput() { return $this->mOutput; }
 484         function nextLinkID() { return $this->mLinkID++; }
 485
 486         function getFunctionLang() {
 487                 global $wgLang, $wgContLang;
 488
 489                 $target = $this->mOptions->getTargetLanguage();
 490                 if ( $target !== null ) {
 491                         return $target;
 492                 } else {
 493                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 494                 }
 495         }
 496
 497         /**
 498          * Get a preprocessor object
 499          */
 500         function getPreprocessor() {
 501                 if ( !isset( $this->mPreprocessor ) ) {
 502                         $class = $this->mPreprocessorClass;
 503                         $this->mPreprocessor = new $class( $this );
 504                 }
 505                 return $this->mPreprocessor;
 506         }
 507
 508         /**
 509          * Replaces all occurrences of HTML-style comments and the given tags
 510          * in the text with a random marker and returns the next text. The output
 511          * parameter $matches will be an associative array filled with data in
 512          * the form:
 513          *   'UNIQ-xxxxx' => array(
 514          *     'element',
 515          *     'tag content',
 516          *     array( 'param' => 'x' ),
 517          *     '<element param="x">tag content</element>' ) )
 518          *
 519          * @param $elements list of element names. Comments are always extracted.
 520          * @param $text Source text string.
 521          * @param $uniq_prefix
 522          *
 523          * @public
 524          * @static
 525          */
 526         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 527                 static $n = 1;
 528                 $stripped = '';
 529                 $matches = array();
 530
 531                 $taglist = implode( '|', $elements );
 532                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 533
 534                 while ( '' != $text ) {
 535                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 536                         $stripped .= $p[0];
 537                         if( count( $p ) < 5 ) {
 538                                 break;
 539                         }
 540                         if( count( $p ) > 5 ) {
 541                                 // comment
 542                                 $element    = $p[4];
 543                                 $attributes = '';
 544                                 $close      = '';
 545                                 $inside     = $p[5];
 546                         } else {
 547                                 // tag
 548                                 $element    = $p[1];
 549                                 $attributes = $p[2];
 550                                 $close      = $p[3];
 551                                 $inside     = $p[4];
 552                         }
 553
 554                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 555                         $stripped .= $marker;
 556
 557                         if ( $close === '/>' ) {
 558                                 // Empty element tag, <tag />
 559                                 $content = null;
 560                                 $text = $inside;
 561                                 $tail = null;
 562                         } else {
 563                                 if( $element === '!--' ) {
 564                                         $end = '/(-->)/';
 565                                 } else {
 566                                         $end = "/(<\\/$element\\s*>)/i";
 567                                 }
 568                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 569                                 $content = $q[0];
 570                                 if( count( $q ) < 3 ) {
 571                                         # No end tag -- let it run out to the end of the text.
 572                                         $tail = '';
 573                                         $text = '';
 574                                 } else {
 575                                         $tail = $q[1];
 576                                         $text = $q[2];
 577                                 }
 578                         }
 579
 580                         $matches[$marker] = array( $element,
 581                                 $content,
 582                                 Sanitizer::decodeTagAttributes( $attributes ),
 583                                 "<$element$attributes$close$content$tail" );
 584                 }
 585                 return $stripped;
 586         }
 587
 588         /**
 589          * Get a list of strippable XML-like elements
 590          */
 591         function getStripList() {
 592                 global $wgRawHtml;
 593                 $elements = $this->mStripList;
 594                 if( $wgRawHtml ) {
 595                         $elements[] = 'html';
 596                 }
 597                 if( $this->mOptions->getUseTeX() ) {
 598                         $elements[] = 'math';
 599                 }
 600                 return $elements;
 601         }
 602
 603         /**
 604          * @deprecated use replaceVariables
 605          */
 606         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 607                 return $text;
 608         }
 609
 610         /**
 611          * Restores pre, math, and other extensions removed by strip()
 612          *
 613          * always call unstripNoWiki() after this one
 614          * @private
 615          * @deprecated use $this->mStripState->unstrip()
 616          */
 617         function unstrip( $text, $state ) {
 618                 return $state->unstripGeneral( $text );
 619         }
 620
 621         /**
 622          * Always call this after unstrip() to preserve the order
 623          *
 624          * @private
 625          * @deprecated use $this->mStripState->unstrip()
 626          */
 627         function unstripNoWiki( $text, $state ) {
 628                 return $state->unstripNoWiki( $text );
 629         }
 630
 631         /**
 632          * @deprecated use $this->mStripState->unstripBoth()
 633          */
 634         function unstripForHTML( $text ) {
 635                 return $this->mStripState->unstripBoth( $text );
 636         }
 637
 638         /**
 639          * Add an item to the strip state
 640          * Returns the unique tag which must be inserted into the stripped text
 641          * The tag will be replaced with the original text in unstrip()
 642          *
 643          * @private
 644          */
 645         function insertStripItem( $text ) {
 646                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 647                 $this->mMarkerIndex++;
 648                 $this->mStripState->general->setPair( $rnd, $text );
 649                 return $rnd;
 650         }
 651
 652         /**
 653          * Interface with html tidy, used if $wgUseTidy = true.
 654          * If tidy isn't able to correct the markup, the original will be
 655          * returned in all its glory with a warning comment appended.
 656          *
 657          * Either the external tidy program or the in-process tidy extension
 658          * will be used depending on availability. Override the default
 659          * $wgTidyInternal setting to disable the internal if it's not working.
 660          *
 661          * @param string $text Hideous HTML input
 662          * @return string Corrected HTML output
 663          * @public
 664          * @static
 665          */
 666         function tidy( $text ) {
 667                 global $wgTidyInternal;
 668
 669                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 670 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 671 '<head><title>test</title></head><body>'.$text.'</body></html>';
 672
 673                 # Tidy is known to clobber tabs; convert 'em to entities
 674                 $wrappedtext = str_replace("\t", '&#9;', $wrappedtext);
 675
 676                 if( $wgTidyInternal ) {
 677                         $correctedtext = self::internalTidy( $wrappedtext );
 678                 } else {
 679                         $correctedtext = self::externalTidy( $wrappedtext );
 680                 }
 681                 if( is_null( $correctedtext ) ) {
 682                         wfDebug( "Tidy error detected!\n" );
 683                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 684                 }
 685
 686                 # Convert the tabs back from entities
 687                 $correctedtext = str_replace('&#9;', "\t", $correctedtext);
 688
 689                 return $correctedtext;
 690         }
 691
 692         /**
 693          * Spawn an external HTML tidy process and get corrected markup back from it.
 694          *
 695          * @private
 696          * @static
 697          */
 698         function externalTidy( $text ) {
 699                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 700                 wfProfileIn( __METHOD__ );
 701
 702                 $cleansource = '';
 703                 $opts = ' -utf8';
 704
 705                 $descriptorspec = array(
 706                         0 => array('pipe', 'r'),
 707                         1 => array('pipe', 'w'),
 708                         2 => array('file', wfGetNull(), 'a')
 709                 );
 710                 $pipes = array();
 711                 if( function_exists('proc_open') ) {
 712                         $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 713                         if (is_resource($process)) {
 714                                 // Theoretically, this style of communication could cause a deadlock
 715                                 // here. If the stdout buffer fills up, then writes to stdin could
 716                                 // block. This doesn't appear to happen with tidy, because tidy only
 717                                 // writes to stdout after it's finished reading from stdin. Search
 718                                 // for tidyParseStdin and tidySaveStdout in console/tidy.c
 719                                 fwrite($pipes[0], $text);
 720                                 fclose($pipes[0]);
 721                                 while (!feof($pipes[1])) {
 722                                         $cleansource .= fgets($pipes[1], 1024);
 723                                 }
 724                                 fclose($pipes[1]);
 725                                 proc_close($process);
 726                         }
 727                 }
 728
 729                 wfProfileOut( __METHOD__ );
 730
 731                 if( $cleansource == '' && $text != '') {
 732                         // Some kind of error happened, so we couldn't get the corrected text.
 733                         // Just give up; we'll use the source text and append a warning.
 734                         return null;
 735                 } else {
 736                         return $cleansource;
 737                 }
 738         }
 739
 740         /**
 741          * Use the HTML tidy PECL extension to use the tidy library in-process,
 742          * saving the overhead of spawning a new process.
 743          *
 744          * 'pear install tidy' should be able to compile the extension module.
 745          *
 746          * @private
 747          * @static
 748          */
 749         function internalTidy( $text ) {
 750                 global $wgTidyConf, $IP, $wgDebugTidy;
 751                 wfProfileIn( __METHOD__ );
 752
 753                 $tidy = new tidy;
 754                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 755                 $tidy->cleanRepair();
 756                 if( $tidy->getStatus() == 2 ) {
 757                         // 2 is magic number for fatal error
 758                         // http://www.php.net/manual/en/function.tidy-get-status.php
 759                         $cleansource = null;
 760                 } else {
 761                         $cleansource = tidy_get_output( $tidy );
 762                 }
 763                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 764                         $cleansource .= "<!--\nTidy reports:\n" .
 765                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 766                                 "\n-->";
 767                 }
 768
 769                 wfProfileOut( __METHOD__ );
 770                 return $cleansource;
 771         }
 772
 773         /**
 774          * parse the wiki syntax used to render tables
 775          *
 776          * @private
 777          */
 778         function doTableStuff ( $text ) {
 779                 wfProfileIn( __METHOD__ );
 780
 781                 $lines = StringUtils::explode( "\n", $text );
 782                 $out = '';
 783                 $td_history = array (); // Is currently a td tag open?
 784                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 785                 $tr_history = array (); // Is currently a tr tag open?
 786                 $tr_attributes = array (); // history of tr attributes
 787                 $has_opened_tr = array(); // Did this table open a <tr> element?
 788                 $indent_level = 0; // indent level of the table
 789
 790                 foreach ( $lines as $outLine ) {
 791                         $line = trim( $outLine );
 792
 793                         if( $line == '' ) { // empty line, go to next line
 794                                 $out .= $outLine."\n";
 795                                 continue;
 796                         }
 797                         $first_character = $line[0];
 798                         $matches = array();
 799
 800                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 801                                 // First check if we are starting a new table
 802                                 $indent_level = strlen( $matches[1] );
 803
 804                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 805                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 806
 807                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 808                                 array_push ( $td_history , false );
 809                                 array_push ( $last_tag_history , '' );
 810                                 array_push ( $tr_history , false );
 811                                 array_push ( $tr_attributes , '' );
 812                                 array_push ( $has_opened_tr , false );
 813                         } else if ( count ( $td_history ) == 0 ) {
 814                                 // Don't do any of the following
 815                                 $out .= $outLine."\n";
 816                                 continue;
 817                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 818                                 // We are ending a table
 819                                 $line = '</table>' . substr ( $line , 2 );
 820                                 $last_tag = array_pop ( $last_tag_history );
 821
 822                                 if ( !array_pop ( $has_opened_tr ) ) {
 823                                         $line = "<tr><td></td></tr>{$line}";
 824                                 }
 825
 826                                 if ( array_pop ( $tr_history ) ) {
 827                                         $line = "</tr>{$line}";
 828                                 }
 829
 830                                 if ( array_pop ( $td_history ) ) {
 831                                         $line = "</{$last_tag}>{$line}";
 832                                 }
 833                                 array_pop ( $tr_attributes );
 834                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 835                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 836                                 // Now we have a table row
 837                                 $line = preg_replace( '#^\|-+#', '', $line );
 838
 839                                 // Whats after the tag is now only attributes
 840                                 $attributes = $this->mStripState->unstripBoth( $line );
 841                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 842                                 array_pop ( $tr_attributes );
 843                                 array_push ( $tr_attributes , $attributes );
 844
 845                                 $line = '';
 846                                 $last_tag = array_pop ( $last_tag_history );
 847                                 array_pop ( $has_opened_tr );
 848                                 array_push ( $has_opened_tr , true );
 849
 850                                 if ( array_pop ( $tr_history ) ) {
 851                                         $line = '</tr>';
 852                                 }
 853
 854                                 if ( array_pop ( $td_history ) ) {
 855                                         $line = "</{$last_tag}>{$line}";
 856                                 }
 857
 858                                 $outLine = $line;
 859                                 array_push ( $tr_history , false );
 860                                 array_push ( $td_history , false );
 861                                 array_push ( $last_tag_history , '' );
 862                         }
 863                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 864                                 // This might be cell elements, td, th or captions
 865                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 866                                         $first_character = '+';
 867                                         $line = substr ( $line , 1 );
 868                                 }
 869
 870                                 $line = substr ( $line , 1 );
 871
 872                                 if ( $first_character === '!' ) {
 873                                         $line = str_replace ( '!!' , '||' , $line );
 874                                 }
 875
 876                                 // Split up multiple cells on the same line.
 877                                 // FIXME : This can result in improper nesting of tags processed
 878                                 // by earlier parser steps, but should avoid splitting up eg
 879                                 // attribute values containing literal "||".
 880                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 881
 882                                 $outLine = '';
 883
 884                                 // Loop through each table cell
 885                                 foreach ( $cells as $cell )
 886                                 {
 887                                         $previous = '';
 888                                         if ( $first_character !== '+' )
 889                                         {
 890                                                 $tr_after = array_pop ( $tr_attributes );
 891                                                 if ( !array_pop ( $tr_history ) ) {
 892                                                         $previous = "<tr{$tr_after}>\n";
 893                                                 }
 894                                                 array_push ( $tr_history , true );
 895                                                 array_push ( $tr_attributes , '' );
 896                                                 array_pop ( $has_opened_tr );
 897                                                 array_push ( $has_opened_tr , true );
 898                                         }
 899
 900                                         $last_tag = array_pop ( $last_tag_history );
 901
 902                                         if ( array_pop ( $td_history ) ) {
 903                                                 $previous = "</{$last_tag}>{$previous}";
 904                                         }
 905
 906                                         if ( $first_character === '|' ) {
 907                                                 $last_tag = 'td';
 908                                         } else if ( $first_character === '!' ) {
 909                                                 $last_tag = 'th';
 910                                         } else if ( $first_character === '+' ) {
 911                                                 $last_tag = 'caption';
 912                                         } else {
 913                                                 $last_tag = '';
 914                                         }
 915
 916                                         array_push ( $last_tag_history , $last_tag );
 917
 918                                         // A cell could contain both parameters and data
 919                                         $cell_data = explode ( '|' , $cell , 2 );
 920
 921                                         // Bug 553: Note that a '|' inside an invalid link should not
 922                                         // be mistaken as delimiting cell parameters
 923                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 924                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 925                                         } else if ( count ( $cell_data ) == 1 )
 926                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 927                                         else {
 928                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 929                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 930                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 931                                         }
 932
 933                                         $outLine .= $cell;
 934                                         array_push ( $td_history , true );
 935                                 }
 936                         }
 937                         $out .= $outLine . "\n";
 938                 }
 939
 940                 // Closing open td, tr && table
 941                 while ( count ( $td_history ) > 0 )
 942                 {
 943                         if ( array_pop ( $td_history ) ) {
 944                                 $out .= "</td>\n";
 945                         }
 946                         if ( array_pop ( $tr_history ) ) {
 947                                 $out .= "</tr>\n";
 948                         }
 949                         if ( !array_pop ( $has_opened_tr ) ) {
 950                                 $out .= "<tr><td></td></tr>\n" ;
 951                         }
 952
 953                         $out .= "</table>\n";
 954                 }
 955
 956                 // Remove trailing line-ending (b/c)
 957                 if ( substr( $out, -1 ) === "\n" ) {
 958                         $out = substr( $out, 0, -1 );
 959                 }
 960
 961                 // special case: don't return empty table
 962                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 963                         $out = '';
 964                 }
 965
 966                 wfProfileOut( __METHOD__ );
 967
 968                 return $out;
 969         }
 970
 971         /**
 972          * Helper function for parse() that transforms wiki markup into
 973          * HTML. Only called for $mOutputType == self::OT_HTML.
 974          *
 975          * @private
 976          */
 977         function internalParse( $text ) {
 978                 $isMain = true;
 979                 wfProfileIn( __METHOD__ );
 980
 981                 # Hook to suspend the parser in this state
 982                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 983                         wfProfileOut( __METHOD__ );
 984                         return $text ;
 985                 }
 986
 987                 $text = $this->replaceVariables( $text );
 988                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 989                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 990
 991                 // Tables need to come after variable replacement for things to work
 992                 // properly; putting them before other transformations should keep
 993                 // exciting things like link expansions from showing up in surprising
 994                 // places.
 995                 $text = $this->doTableStuff( $text );
 996
 997                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 998
 999                 $text = $this->doDoubleUnderscore( $text );
1000                 $text = $this->doHeadings( $text );
1001                 if($this->mOptions->getUseDynamicDates()) {
1002                         $df = DateFormatter::getInstance();
1003                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1004                 }
1005                 $text = $this->doAllQuotes( $text );
1006                 $text = $this->replaceInternalLinks( $text );
1007                 $text = $this->replaceExternalLinks( $text );
1008
1009                 # replaceInternalLinks may sometimes leave behind
1010                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1011                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1012
1013                 $text = $this->doMagicLinks( $text );
1014                 $text = $this->formatHeadings( $text, $isMain );
1015
1016                 wfProfileOut( __METHOD__ );
1017                 return $text;
1018         }
1019
1020         /**
1021          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1022          * magic external links.
1023          *
1024          * DML
1025          * @private
1026          */
1027         function doMagicLinks( $text ) {
1028                 wfProfileIn( __METHOD__ );
1029                 $prots = $this->mUrlProtocols;
1030                 $urlChar = self::EXT_LINK_URL_CLASS;
1031                 $text = preg_replace_callback(
1032                         '!(?:                           # Start cases
1033                                 (<a.*?</a>) |               # m[1]: Skip link text
1034                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1035                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1036                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1037                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1038                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1039                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1040                                     [0-9Xx]                 # check digit
1041                                     \b)
1042                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1043                 wfProfileOut( __METHOD__ );
1044                 return $text;
1045         }
1046
1047         function magicLinkCallback( $m ) {
1048                 if ( isset( $m[1] ) && strval( $m[1] ) !== '' ) {
1049                         # Skip anchor
1050                         return $m[0];
1051                 } elseif ( isset( $m[2] ) && strval( $m[2] ) !== '' ) {
1052                         # Skip HTML element
1053                         return $m[0];
1054                 } elseif ( isset( $m[3] ) && strval( $m[3] ) !== '' ) {
1055                         # Free external link
1056                         return $this->makeFreeExternalLink( $m[0] );
1057                 } elseif ( isset( $m[4] ) && strval( $m[4] ) !== '' ) {
1058                         # RFC or PMID
1059                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1060                                 $keyword = 'RFC';
1061                                 $urlmsg = 'rfcurl';
1062                                 $id = $m[4];
1063                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1064                                 $keyword = 'PMID';
1065                                 $urlmsg = 'pubmedurl';
1066                                 $id = $m[4];
1067                         } else {
1068                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1069                                         substr($m[0], 0, 20 ) . '"' );
1070                         }
1071                         $url = wfMsg( $urlmsg, $id);
1072                         $sk = $this->mOptions->getSkin();
1073                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1074                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1075                 } elseif ( isset( $m[5] ) && strval( $m[5] ) !== '' ) {
1076                         # ISBN
1077                         $isbn = $m[5];
1078                         $num = strtr( $isbn, array(
1079                                 '-' => '',
1080                                 ' ' => '',
1081                                 'x' => 'X',
1082                         ));
1083                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1084                         return'<a href="' .
1085                                 $titleObj->escapeLocalUrl() .
1086                                 "\" class=\"internal\">ISBN $isbn</a>";
1087                 } else {
1088                         return $m[0];
1089                 }
1090         }
1091
1092         /**
1093          * Make a free external link, given a user-supplied URL
1094          * @return HTML
1095          * @private
1096          */
1097         function makeFreeExternalLink( $url ) {
1098                 global $wgContLang;
1099                 wfProfileIn( __METHOD__ );
1100
1101                 $sk = $this->mOptions->getSkin();
1102                 $trail = '';
1103
1104                 # The characters '<' and '>' (which were escaped by
1105                 # removeHTMLtags()) should not be included in
1106                 # URLs, per RFC 2396.
1107                 $m2 = array();
1108                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1109                         $trail = substr($url, $m2[0][1]) . $trail;
1110                         $url = substr($url, 0, $m2[0][1]);
1111                 }
1112
1113                 # Move trailing punctuation to $trail
1114                 $sep = ',;\.:!?';
1115                 # If there is no left bracket, then consider right brackets fair game too
1116                 if ( strpos( $url, '(' ) === false ) {
1117                         $sep .= ')';
1118                 }
1119
1120                 $numSepChars = strspn( strrev( $url ), $sep );
1121                 if ( $numSepChars ) {
1122                         $trail = substr( $url, -$numSepChars ) . $trail;
1123                         $url = substr( $url, 0, -$numSepChars );
1124                 }
1125
1126                 $url = Sanitizer::cleanUrl( $url );
1127
1128                 # Is this an external image?
1129                 $text = $this->maybeMakeExternalImage( $url );
1130                 if ( $text === false ) {
1131                         # Not an image, make a link
1132                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1133                                 $this->getExternalLinkAttribs() );
1134                         # Register it in the output object...
1135                         # Replace unnecessary URL escape codes with their equivalent characters
1136                         $pasteurized = self::replaceUnusualEscapes( $url );
1137                         $this->mOutput->addExternalLink( $pasteurized );
1138                 }
1139                 wfProfileOut( __METHOD__ );
1140                 return $text . $trail;
1141         }
1142
1143
1144         /**
1145          * Parse headers and return html
1146          *
1147          * @private
1148          */
1149         function doHeadings( $text ) {
1150                 wfProfileIn( __METHOD__ );
1151                 for ( $i = 6; $i >= 1; --$i ) {
1152                         $h = str_repeat( '=', $i );
1153                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1154                           "<h$i>\\1</h$i>", $text );
1155                 }
1156                 wfProfileOut( __METHOD__ );
1157                 return $text;
1158         }
1159
1160         /**
1161          * Replace single quotes with HTML markup
1162          * @private
1163          * @return string the altered text
1164          */
1165         function doAllQuotes( $text ) {
1166                 wfProfileIn( __METHOD__ );
1167                 $outtext = '';
1168                 $lines = StringUtils::explode( "\n", $text );
1169                 foreach ( $lines as $line ) {
1170                         $outtext .= $this->doQuotes( $line ) . "\n";
1171                 }
1172                 $outtext = substr($outtext, 0,-1);
1173                 wfProfileOut( __METHOD__ );
1174                 return $outtext;
1175         }
1176
1177         /**
1178          * Helper function for doAllQuotes()
1179          */
1180         public function doQuotes( $text ) {
1181                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1182                 if ( count( $arr ) == 1 )
1183                         return $text;
1184                 else
1185                 {
1186                         # First, do some preliminary work. This may shift some apostrophes from
1187                         # being mark-up to being text. It also counts the number of occurrences
1188                         # of bold and italics mark-ups.
1189                         $i = 0;
1190                         $numbold = 0;
1191                         $numitalics = 0;
1192                         foreach ( $arr as $r )
1193                         {
1194                                 if ( ( $i % 2 ) == 1 )
1195                                 {
1196                                         # If there are ever four apostrophes, assume the first is supposed to
1197                                         # be text, and the remaining three constitute mark-up for bold text.
1198                                         if ( strlen( $arr[$i] ) == 4 )
1199                                         {
1200                                                 $arr[$i-1] .= "'";
1201                                                 $arr[$i] = "'''";
1202                                         }
1203                                         # If there are more than 5 apostrophes in a row, assume they're all
1204                                         # text except for the last 5.
1205                                         else if ( strlen( $arr[$i] ) > 5 )
1206                                         {
1207                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1208                                                 $arr[$i] = "'''''";
1209                                         }
1210                                         # Count the number of occurrences of bold and italics mark-ups.
1211                                         # We are not counting sequences of five apostrophes.
1212                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1213                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1214                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1215                                 }
1216                                 $i++;
1217                         }
1218
1219                         # If there is an odd number of both bold and italics, it is likely
1220                         # that one of the bold ones was meant to be an apostrophe followed
1221                         # by italics. Which one we cannot know for certain, but it is more
1222                         # likely to be one that has a single-letter word before it.
1223                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1224                         {
1225                                 $i = 0;
1226                                 $firstsingleletterword = -1;
1227                                 $firstmultiletterword = -1;
1228                                 $firstspace = -1;
1229                                 foreach ( $arr as $r )
1230                                 {
1231                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1232                                         {
1233                                                 $x1 = substr ($arr[$i-1], -1);
1234                                                 $x2 = substr ($arr[$i-1], -2, 1);
1235                                                 if ($x1 === ' ') {
1236                                                         if ($firstspace == -1) $firstspace = $i;
1237                                                 } else if ($x2 === ' ') {
1238                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1239                                                 } else {
1240                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1241                                                 }
1242                                         }
1243                                         $i++;
1244                                 }
1245
1246                                 # If there is a single-letter word, use it!
1247                                 if ($firstsingleletterword > -1)
1248                                 {
1249                                         $arr [ $firstsingleletterword ] = "''";
1250                                         $arr [ $firstsingleletterword-1 ] .= "'";
1251                                 }
1252                                 # If not, but there's a multi-letter word, use that one.
1253                                 else if ($firstmultiletterword > -1)
1254                                 {
1255                                         $arr [ $firstmultiletterword ] = "''";
1256                                         $arr [ $firstmultiletterword-1 ] .= "'";
1257                                 }
1258                                 # ... otherwise use the first one that has neither.
1259                                 # (notice that it is possible for all three to be -1 if, for example,
1260                                 # there is only one pentuple-apostrophe in the line)
1261                                 else if ($firstspace > -1)
1262                                 {
1263                                         $arr [ $firstspace ] = "''";
1264                                         $arr [ $firstspace-1 ] .= "'";
1265                                 }
1266                         }
1267
1268                         # Now let's actually convert our apostrophic mush to HTML!
1269                         $output = '';
1270                         $buffer = '';
1271                         $state = '';
1272                         $i = 0;
1273                         foreach ($arr as $r)
1274                         {
1275                                 if (($i % 2) == 0)
1276                                 {
1277                                         if ($state === 'both')
1278                                                 $buffer .= $r;
1279                                         else
1280                                                 $output .= $r;
1281                                 }
1282                                 else
1283                                 {
1284                                         if (strlen ($r) == 2)
1285                                         {
1286                                                 if ($state === 'i')
1287                                                 { $output .= '</i>'; $state = ''; }
1288                                                 else if ($state === 'bi')
1289                                                 { $output .= '</i>'; $state = 'b'; }
1290                                                 else if ($state === 'ib')
1291                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1292                                                 else if ($state === 'both')
1293                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1294                                                 else # $state can be 'b' or ''
1295                                                 { $output .= '<i>'; $state .= 'i'; }
1296                                         }
1297                                         else if (strlen ($r) == 3)
1298                                         {
1299                                                 if ($state === 'b')
1300                                                 { $output .= '</b>'; $state = ''; }
1301                                                 else if ($state === 'bi')
1302                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1303                                                 else if ($state === 'ib')
1304                                                 { $output .= '</b>'; $state = 'i'; }
1305                                                 else if ($state === 'both')
1306                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1307                                                 else # $state can be 'i' or ''
1308                                                 { $output .= '<b>'; $state .= 'b'; }
1309                                         }
1310                                         else if (strlen ($r) == 5)
1311                                         {
1312                                                 if ($state === 'b')
1313                                                 { $output .= '</b><i>'; $state = 'i'; }
1314                                                 else if ($state === 'i')
1315                                                 { $output .= '</i><b>'; $state = 'b'; }
1316                                                 else if ($state === 'bi')
1317                                                 { $output .= '</i></b>'; $state = ''; }
1318                                                 else if ($state === 'ib')
1319                                                 { $output .= '</b></i>'; $state = ''; }
1320                                                 else if ($state === 'both')
1321                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1322                                                 else # ($state == '')
1323                                                 { $buffer = ''; $state = 'both'; }
1324                                         }
1325                                 }
1326                                 $i++;
1327                         }
1328                         # Now close all remaining tags.  Notice that the order is important.
1329                         if ($state === 'b' || $state === 'ib')
1330                                 $output .= '</b>';
1331                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1332                                 $output .= '</i>';
1333                         if ($state === 'bi')
1334                                 $output .= '</b>';
1335                         # There might be lonely ''''', so make sure we have a buffer
1336                         if ($state === 'both' && $buffer)
1337                                 $output .= '<b><i>'.$buffer.'</i></b>';
1338                         return $output;
1339                 }
1340         }
1341
1342         /**
1343          * Replace external links (REL)
1344          *
1345          * Note: this is all very hackish and the order of execution matters a lot.
1346          * Make sure to run maintenance/parserTests.php if you change this code.
1347          *
1348          * @private
1349          */
1350         function replaceExternalLinks( $text ) {
1351                 global $wgContLang;
1352                 wfProfileIn( __METHOD__ );
1353
1354                 $sk = $this->mOptions->getSkin();
1355
1356                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1357                 $s = array_shift( $bits );
1358
1359                 $i = 0;
1360                 while ( $i<count( $bits ) ) {
1361                         $url = $bits[$i++];
1362                         $protocol = $bits[$i++];
1363                         $text = $bits[$i++];
1364                         $trail = $bits[$i++];
1365
1366                         # The characters '<' and '>' (which were escaped by
1367                         # removeHTMLtags()) should not be included in
1368                         # URLs, per RFC 2396.
1369                         $m2 = array();
1370                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1371                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1372                                 $url = substr($url, 0, $m2[0][1]);
1373                         }
1374
1375                         # If the link text is an image URL, replace it with an <img> tag
1376                         # This happened by accident in the original parser, but some people used it extensively
1377                         $img = $this->maybeMakeExternalImage( $text );
1378                         if ( $img !== false ) {
1379                                 $text = $img;
1380                         }
1381
1382                         $dtrail = '';
1383
1384                         # Set linktype for CSS - if URL==text, link is essentially free
1385                         $linktype = ($text === $url) ? 'free' : 'text';
1386
1387                         # No link text, e.g. [http://domain.tld/some.link]
1388                         if ( $text == '' ) {
1389                                 # Autonumber if allowed. See bug #5918
1390                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1391                                         $langObj = $this->getFunctionLang();
1392                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1393                                         $linktype = 'autonumber';
1394                                 } else {
1395                                         # Otherwise just use the URL
1396                                         $text = htmlspecialchars( $url );
1397                                         $linktype = 'free';
1398                                 }
1399                         } else {
1400                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1401                                 # Check for trail
1402                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1403                         }
1404
1405                         $text = $wgContLang->markNoConversion($text);
1406
1407                         $url = Sanitizer::cleanUrl( $url );
1408
1409                         if ( $this->mOptions->mExternalLinkTarget ) {
1410                                 $attribs = array( 'target' => $this->mOptions->mExternalLinkTarget );
1411                         } else {
1412                                 $attribs = array();
1413                         }
1414
1415                         # Use the encoded URL
1416                         # This means that users can paste URLs directly into the text
1417                         # Funny characters like &ouml; aren't valid in URLs anyway
1418                         # This was changed in August 2004
1419                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->getExternalLinkAttribs() )
1420                                 . $dtrail . $trail;
1421
1422                         # Register link in the output object.
1423                         # Replace unnecessary URL escape codes with the referenced character
1424                         # This prevents spammers from hiding links from the filters
1425                         $pasteurized = self::replaceUnusualEscapes( $url );
1426                         $this->mOutput->addExternalLink( $pasteurized );
1427                 }
1428
1429                 wfProfileOut( __METHOD__ );
1430                 return $s;
1431         }
1432
1433         function getExternalLinkAttribs() {
1434                 $attribs = array();
1435                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1436                 $ns = $this->mTitle->getNamespace();
1437                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1438                         $attribs['rel'] = 'nofollow';
1439                 }
1440                 if ( $this->mOptions->getExternalLinkTarget() ) {
1441                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1442                 }
1443                 return $attribs;
1444         }
1445
1446
1447         /**
1448          * Replace unusual URL escape codes with their equivalent characters
1449          * @param string
1450          * @return string
1451          * @static
1452          * @todo  This can merge genuinely required bits in the path or query string,
1453          *        breaking legit URLs. A proper fix would treat the various parts of
1454          *        the URL differently; as a workaround, just use the output for
1455          *        statistical records, not for actual linking/output.
1456          */
1457         static function replaceUnusualEscapes( $url ) {
1458                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1459                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1460         }
1461
1462         /**
1463          * Callback function used in replaceUnusualEscapes().
1464          * Replaces unusual URL escape codes with their equivalent character
1465          * @static
1466          * @private
1467          */
1468         private static function replaceUnusualEscapesCallback( $matches ) {
1469                 $char = urldecode( $matches[0] );
1470                 $ord = ord( $char );
1471                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1472                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1473                         // No, shouldn't be escaped
1474                         return $char;
1475                 } else {
1476                         // Yes, leave it escaped
1477                         return $matches[0];
1478                 }
1479         }
1480
1481         /**
1482          * make an image if it's allowed, either through the global
1483          * option, through the exception, or through the on-wiki whitelist
1484          * @private
1485          */
1486         function maybeMakeExternalImage( $url ) {
1487                 $sk = $this->mOptions->getSkin();
1488                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1489                 $imagesexception = !empty($imagesfrom);
1490                 $text = false;
1491                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1492                 if( $imagesexception && is_array( $imagesfrom ) ) {
1493                         $imagematch = false;
1494                         foreach( $imagesfrom as $match ) {
1495                                 if( strpos( $url, $match ) === 0 ) {
1496                                         $imagematch = true;
1497                                         break;
1498                                 }
1499                         }
1500                 } elseif( $imagesexception ) {
1501                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1502                 } else {
1503                         $imagematch = false;
1504                 }
1505                 if ( $this->mOptions->getAllowExternalImages()
1506                      || ( $imagesexception && $imagematch ) ) {
1507                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1508                                 # Image found
1509                                 $text = $sk->makeExternalImage( $url );
1510                         }
1511                 }
1512                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1513                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1514                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1515                         foreach( $whitelist as $entry ) {
1516                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1517                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1518                                         continue;
1519                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1520                                         # Image matches a whitelist entry
1521                                         $text = $sk->makeExternalImage( $url );
1522                                         break;
1523                                 }
1524                         }
1525                 }
1526                 return $text;
1527         }
1528
1529         /**
1530          * Process [[ ]] wikilinks
1531          * @return processed text
1532          *
1533          * @private
1534          */
1535         function replaceInternalLinks( $s ) {
1536                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1537                 return $s;
1538         }
1539
1540         /**
1541          * Process [[ ]] wikilinks (RIL)
1542          * @return LinkHolderArray
1543          *
1544          * @private
1545          */
1546         function replaceInternalLinks2( &$s ) {
1547                 global $wgContLang;
1548
1549                 wfProfileIn( __METHOD__ );
1550
1551                 wfProfileIn( __METHOD__.'-setup' );
1552                 static $tc = FALSE, $e1, $e1_img;
1553                 # the % is needed to support urlencoded titles as well
1554                 if ( !$tc ) {
1555                         $tc = Title::legalChars() . '#%';
1556                         # Match a link having the form [[namespace:link|alternate]]trail
1557                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1558                         # Match cases where there is no "]]", which might still be images
1559                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1560                 }
1561
1562                 $sk = $this->mOptions->getSkin();
1563                 $holders = new LinkHolderArray( $this );
1564
1565                 #split the entire text string on occurences of [[
1566                 $a = StringUtils::explode( '[[', ' ' . $s );
1567                 #get the first element (all text up to first [[), and remove the space we added
1568                 $s = $a->current();
1569                 $a->next();
1570                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1571                 $s = substr( $s, 1 );
1572
1573                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1574                 $e2 = null;
1575                 if ( $useLinkPrefixExtension ) {
1576                         # Match the end of a line for a word that's not followed by whitespace,
1577                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1578                         $e2 = wfMsgForContent( 'linkprefix' );
1579                 }
1580
1581                 if( is_null( $this->mTitle ) ) {
1582                         wfProfileOut( __METHOD__.'-setup' );
1583                         wfProfileOut( __METHOD__ );
1584                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1585                 }
1586                 $nottalk = !$this->mTitle->isTalkPage();
1587
1588                 if ( $useLinkPrefixExtension ) {
1589                         $m = array();
1590                         if ( preg_match( $e2, $s, $m ) ) {
1591                                 $first_prefix = $m[2];
1592                         } else {
1593                                 $first_prefix = false;
1594                         }
1595                 } else {
1596                         $prefix = '';
1597                 }
1598
1599                 if($wgContLang->hasVariants()) {
1600                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1601                 } else {
1602                         $selflink = array($this->mTitle->getPrefixedText());
1603                 }
1604                 $useSubpages = $this->areSubpagesAllowed();
1605                 wfProfileOut( __METHOD__.'-setup' );
1606
1607                 # Loop for each link
1608                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1609                         # Check for excessive memory usage
1610                         if ( $holders->isBig() ) {
1611                                 # Too big
1612                                 # Do the existence check, replace the link holders and clear the array
1613                                 $holders->replace( $s );
1614                                 $holders->clear();
1615                         }
1616
1617                         if ( $useLinkPrefixExtension ) {
1618                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1619                                 if ( preg_match( $e2, $s, $m ) ) {
1620                                         $prefix = $m[2];
1621                                         $s = $m[1];
1622                                 } else {
1623                                         $prefix='';
1624                                 }
1625                                 # first link
1626                                 if($first_prefix) {
1627                                         $prefix = $first_prefix;
1628                                         $first_prefix = false;
1629                                 }
1630                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1631                         }
1632
1633                         $might_be_img = false;
1634
1635                         wfProfileIn( __METHOD__."-e1" );
1636                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1637                                 $text = $m[2];
1638                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1639                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1640                                 # the real problem is with the $e1 regex
1641                                 # See bug 1300.
1642                                 #
1643                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1644                                 # and no image is in sight. See bug 2095.
1645                                 #
1646                                 if( $text !== '' &&
1647                                         substr( $m[3], 0, 1 ) === ']' &&
1648                                         strpos($text, '[') !== false
1649                                 )
1650                                 {
1651                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1652                                         $m[3] = substr( $m[3], 1 );
1653                                 }
1654                                 # fix up urlencoded title texts
1655                                 if( strpos( $m[1], '%' ) !== false ) {
1656                                         # Should anchors '#' also be rejected?
1657                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1658                                 }
1659                                 $trail = $m[3];
1660                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1661                                 $might_be_img = true;
1662                                 $text = $m[2];
1663                                 if ( strpos( $m[1], '%' ) !== false ) {
1664                                         $m[1] = urldecode($m[1]);
1665                                 }
1666                                 $trail = "";
1667                         } else { # Invalid form; output directly
1668                                 $s .= $prefix . '[[' . $line ;
1669                                 wfProfileOut( __METHOD__."-e1" );
1670                                 continue;
1671                         }
1672                         wfProfileOut( __METHOD__."-e1" );
1673                         wfProfileIn( __METHOD__."-misc" );
1674
1675                         # Don't allow internal links to pages containing
1676                         # PROTO: where PROTO is a valid URL protocol; these
1677                         # should be external links.
1678                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1679                                 $s .= $prefix . '[[' . $line ;
1680                                 wfProfileOut( __METHOD__."-misc" );
1681                                 continue;
1682                         }
1683
1684                         # Make subpage if necessary
1685                         if( $useSubpages ) {
1686                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1687                         } else {
1688                                 $link = $m[1];
1689                         }
1690
1691                         $noforce = (substr($m[1], 0, 1) !== ':');
1692                         if (!$noforce) {
1693                                 # Strip off leading ':'
1694                                 $link = substr($link, 1);
1695                         }
1696
1697                         wfProfileOut( __METHOD__."-misc" );
1698                         wfProfileIn( __METHOD__."-title" );
1699                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1700                         if( !$nt ) {
1701                                 $s .= $prefix . '[[' . $line;
1702                                 wfProfileOut( __METHOD__."-title" );
1703                                 continue;
1704                         }
1705
1706                         $ns = $nt->getNamespace();
1707                         $iw = $nt->getInterWiki();
1708                         wfProfileOut( __METHOD__."-title" );
1709
1710                         if ($might_be_img) { # if this is actually an invalid link
1711                                 wfProfileIn( __METHOD__."-might_be_img" );
1712                                 if ($ns == NS_FILE && $noforce) { #but might be an image
1713                                         $found = false;
1714                                         while ( true ) {
1715                                                 #look at the next 'line' to see if we can close it there
1716                                                 $a->next();
1717                                                 $next_line = $a->current();
1718                                                 if ( $next_line === false || $next_line === null ) {
1719                                                         break;
1720                                                 }
1721                                                 $m = explode( ']]', $next_line, 3 );
1722                                                 if ( count( $m ) == 3 ) {
1723                                                         # the first ]] closes the inner link, the second the image
1724                                                         $found = true;
1725                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1726                                                         $trail = $m[2];
1727                                                         break;
1728                                                 } elseif ( count( $m ) == 2 ) {
1729                                                         #if there's exactly one ]] that's fine, we'll keep looking
1730                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1731                                                 } else {
1732                                                         #if $next_line is invalid too, we need look no further
1733                                                         $text .= '[[' . $next_line;
1734                                                         break;
1735                                                 }
1736                                         }
1737                                         if ( !$found ) {
1738                                                 # we couldn't find the end of this imageLink, so output it raw
1739                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1740                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1741                                                 $s .= "{$prefix}[[$link|$text";
1742                                                 # note: no $trail, because without an end, there *is* no trail
1743                                                 wfProfileOut( __METHOD__."-might_be_img" );
1744                                                 continue;
1745                                         }
1746                                 } else { #it's not an image, so output it raw
1747                                         $s .= "{$prefix}[[$link|$text";
1748                                         # note: no $trail, because without an end, there *is* no trail
1749                                         wfProfileOut( __METHOD__."-might_be_img" );
1750                                         continue;
1751                                 }
1752                                 wfProfileOut( __METHOD__."-might_be_img" );
1753                         }
1754
1755                         $wasblank = ( '' == $text );
1756                         if( $wasblank ) $text = $link;
1757
1758                         # Link not escaped by : , create the various objects
1759                         if( $noforce ) {
1760
1761                                 # Interwikis
1762                                 wfProfileIn( __METHOD__."-interwiki" );
1763                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1764                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1765                                         $s = rtrim($s . $prefix);
1766                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1767                                         wfProfileOut( __METHOD__."-interwiki" );
1768                                         continue;
1769                                 }
1770                                 wfProfileOut( __METHOD__."-interwiki" );
1771
1772                                 if ( $ns == NS_FILE ) {
1773                                         wfProfileIn( __METHOD__."-image" );
1774                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1775                                                 # recursively parse links inside the image caption
1776                                                 # actually, this will parse them in any other parameters, too,
1777                                                 # but it might be hard to fix that, and it doesn't matter ATM
1778                                                 $text = $this->replaceExternalLinks($text);
1779                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1780
1781                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1782                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1783                                         }
1784                                         $this->mOutput->addImage( $nt->getDBkey() );
1785                                         wfProfileOut( __METHOD__."-image" );
1786                                         continue;
1787
1788                                 }
1789
1790                                 if ( $ns == NS_CATEGORY ) {
1791                                         wfProfileIn( __METHOD__."-category" );
1792                                         $s = rtrim($s . "\n"); # bug 87
1793
1794                                         if ( $wasblank ) {
1795                                                 $sortkey = $this->getDefaultSort();
1796                                         } else {
1797                                                 $sortkey = $text;
1798                                         }
1799                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1800                                         $sortkey = str_replace( "\n", '', $sortkey );
1801                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1802                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1803
1804                                         /**
1805                                          * Strip the whitespace Category links produce, see bug 87
1806                                          * @todo We might want to use trim($tmp, "\n") here.
1807                                          */
1808                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1809
1810                                         wfProfileOut( __METHOD__."-category" );
1811                                         continue;
1812                                 }
1813                         }
1814
1815                         # Self-link checking
1816                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1817                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1818                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1819                                         continue;
1820                                 }
1821                         }
1822
1823                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1824                         # FIXME: Should do batch file existence checks, see comment below
1825                         if( $ns == NS_MEDIA ) {
1826                                 # Give extensions a chance to select the file revision for us
1827                                 $skip = $time = false;
1828                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1829                                 if ( $skip ) {
1830                                         $link = $sk->link( $nt );
1831                                 } else {
1832                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1833                                 }
1834                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1835                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1836                                 $this->mOutput->addImage( $nt->getDBkey() );
1837                                 continue;
1838                         }
1839
1840                         # Some titles, such as valid special pages or files in foreign repos, should
1841                         # be shown as bluelinks even though they're not included in the page table
1842                         #
1843                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1844                         # batch file existence checks for NS_FILE and NS_MEDIA
1845                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1846                                 $this->mOutput->addLink( $nt );
1847                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1848                         } else {
1849                                 # Links will be added to the output link list after checking
1850                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1851                         }
1852                 }
1853                 wfProfileOut( __METHOD__ );
1854                 return $holders;
1855         }
1856
1857         /**
1858          * Make a link placeholder. The text returned can be later resolved to a real link with
1859          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1860          * parsing of interwiki links, and secondly to allow all existence checks and
1861          * article length checks (for stub links) to be bundled into a single query.
1862          *
1863          * @deprecated
1864          */
1865         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1866                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1867         }
1868
1869         /**
1870          * Render a forced-blue link inline; protect against double expansion of
1871          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1872          * Since this little disaster has to split off the trail text to avoid
1873          * breaking URLs in the following text without breaking trails on the
1874          * wiki links, it's been made into a horrible function.
1875          *
1876          * @param Title $nt
1877          * @param string $text
1878          * @param string $query
1879          * @param string $trail
1880          * @param string $prefix
1881          * @return string HTML-wikitext mix oh yuck
1882          */
1883         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1884                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1885                 $sk = $this->mOptions->getSkin();
1886                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1887                 return $this->armorLinks( $link ) . $trail;
1888         }
1889
1890         /**
1891          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1892          * going to go through further parsing steps before inline URL expansion.
1893          *
1894          * Not needed quite as much as it used to be since free links are a bit
1895          * more sensible these days. But bracketed links are still an issue.
1896          *
1897          * @param string more-or-less HTML
1898          * @return string less-or-more HTML with NOPARSE bits
1899          */
1900         function armorLinks( $text ) {
1901                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1902                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1903         }
1904
1905         /**
1906          * Return true if subpage links should be expanded on this page.
1907          * @return bool
1908          */
1909         function areSubpagesAllowed() {
1910                 # Some namespaces don't allow subpages
1911                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1912         }
1913
1914         /**
1915          * Handle link to subpage if necessary
1916          * @param string $target the source of the link
1917          * @param string &$text the link text, modified as necessary
1918          * @return string the full name of the link
1919          * @private
1920          */
1921         function maybeDoSubpageLink($target, &$text) {
1922                 # Valid link forms:
1923                 # Foobar -- normal
1924                 # :Foobar -- override special treatment of prefix (images, language links)
1925                 # /Foobar -- convert to CurrentPage/Foobar
1926                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1927                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1928                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1929
1930                 wfProfileIn( __METHOD__ );
1931                 $ret = $target; # default return value is no change
1932
1933                 # Some namespaces don't allow subpages,
1934                 # so only perform processing if subpages are allowed
1935                 if( $this->areSubpagesAllowed() ) {
1936                         $hash = strpos( $target, '#' );
1937                         if( $hash !== false ) {
1938                                 $suffix = substr( $target, $hash );
1939                                 $target = substr( $target, 0, $hash );
1940                         } else {
1941                                 $suffix = '';
1942                         }
1943                         # bug 7425
1944                         $target = trim( $target );
1945                         # Look at the first character
1946                         if( $target != '' && $target{0} === '/' ) {
1947                                 # / at end means we don't want the slash to be shown
1948                                 $m = array();
1949                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1950                                 if( $trailingSlashes ) {
1951                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1952                                 } else {
1953                                         $noslash = substr( $target, 1 );
1954                                 }
1955
1956                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1957                                 if( '' === $text ) {
1958                                         $text = $target . $suffix;
1959                                 } # this might be changed for ugliness reasons
1960                         } else {
1961                                 # check for .. subpage backlinks
1962                                 $dotdotcount = 0;
1963                                 $nodotdot = $target;
1964                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1965                                         ++$dotdotcount;
1966                                         $nodotdot = substr( $nodotdot, 3 );
1967                                 }
1968                                 if($dotdotcount > 0) {
1969                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1970                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1971                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1972                                                 # / at the end means don't show full path
1973                                                 if( substr( $nodotdot, -1, 1 ) === '/' ) {
1974                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1975                                                         if( '' === $text ) {
1976                                                                 $text = $nodotdot . $suffix;
1977                                                         }
1978                                                 }
1979                                                 $nodotdot = trim( $nodotdot );
1980                                                 if( $nodotdot != '' ) {
1981                                                         $ret .= '/' . $nodotdot;
1982                                                 }
1983                                                 $ret .= $suffix;
1984                                         }
1985                                 }
1986                         }
1987                 }
1988
1989                 wfProfileOut( __METHOD__ );
1990                 return $ret;
1991         }
1992
1993         /**#@+
1994          * Used by doBlockLevels()
1995          * @private
1996          */
1997         /* private */ function closeParagraph() {
1998                 $result = '';
1999                 if ( '' != $this->mLastSection ) {
2000                         $result = '</' . $this->mLastSection  . ">\n";
2001                 }
2002                 $this->mInPre = false;
2003                 $this->mLastSection = '';
2004                 return $result;
2005         }
2006         # getCommon() returns the length of the longest common substring
2007         # of both arguments, starting at the beginning of both.
2008         #
2009         /* private */ function getCommon( $st1, $st2 ) {
2010                 $fl = strlen( $st1 );
2011                 $shorter = strlen( $st2 );
2012                 if ( $fl < $shorter ) { $shorter = $fl; }
2013
2014                 for ( $i = 0; $i < $shorter; ++$i ) {
2015                         if ( $st1{$i} != $st2{$i} ) { break; }
2016                 }
2017                 return $i;
2018         }
2019         # These next three functions open, continue, and close the list
2020         # element appropriate to the prefix character passed into them.
2021         #
2022         /* private */ function openList( $char ) {
2023                 $result = $this->closeParagraph();
2024
2025                 if ( '*' === $char ) { $result .= '<ul><li>'; }
2026                 else if ( '#' === $char ) { $result .= '<ol><li>'; }
2027                 else if ( ':' === $char ) { $result .= '<dl><dd>'; }
2028                 else if ( ';' === $char ) {
2029                         $result .= '<dl><dt>';
2030                         $this->mDTopen = true;
2031                 }
2032                 else { $result = '<!-- ERR 1 -->'; }
2033
2034                 return $result;
2035         }
2036
2037         /* private */ function nextItem( $char ) {
2038                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
2039                 else if ( ':' === $char || ';' === $char ) {
2040                         $close = '</dd>';
2041                         if ( $this->mDTopen ) { $close = '</dt>'; }
2042                         if ( ';' === $char ) {
2043                                 $this->mDTopen = true;
2044                                 return $close . '<dt>';
2045                         } else {
2046                                 $this->mDTopen = false;
2047                                 return $close . '<dd>';
2048                         }
2049                 }
2050                 return '<!-- ERR 2 -->';
2051         }
2052
2053         /* private */ function closeList( $char ) {
2054                 if ( '*' === $char ) { $text = '</li></ul>'; }
2055                 else if ( '#' === $char ) { $text = '</li></ol>'; }
2056                 else if ( ':' === $char ) {
2057                         if ( $this->mDTopen ) {
2058                                 $this->mDTopen = false;
2059                                 $text = '</dt></dl>';
2060                         } else {
2061                                 $text = '</dd></dl>';
2062                         }
2063                 }
2064                 else {  return '<!-- ERR 3 -->'; }
2065                 return $text."\n";
2066         }
2067         /**#@-*/
2068
2069         /**
2070          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2071          *
2072          * @private
2073          * @return string the lists rendered as HTML
2074          */
2075         function doBlockLevels( $text, $linestart ) {
2076                 wfProfileIn( __METHOD__ );
2077
2078                 # Parsing through the text line by line.  The main thing
2079                 # happening here is handling of block-level elements p, pre,
2080                 # and making lists from lines starting with * # : etc.
2081                 #
2082                 $textLines = StringUtils::explode( "\n", $text );
2083
2084                 $lastPrefix = $output = '';
2085                 $this->mDTopen = $inBlockElem = false;
2086                 $prefixLength = 0;
2087                 $paragraphStack = false;
2088
2089                 foreach ( $textLines as $oLine ) {
2090                         # Fix up $linestart
2091                         if ( !$linestart ) {
2092                                 $output .= $oLine;
2093                                 $linestart = true;
2094                                 continue;
2095                         }
2096
2097                         $lastPrefixLength = strlen( $lastPrefix );
2098                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2099                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2100                         if ( !$this->mInPre ) {
2101                                 # Multiple prefixes may abut each other for nested lists.
2102                                 $prefixLength = strspn( $oLine, '*#:;' );
2103                                 $prefix = substr( $oLine, 0, $prefixLength );
2104
2105                                 # eh?
2106                                 $prefix2 = str_replace( ';', ':', $prefix );
2107                                 $t = substr( $oLine, $prefixLength );
2108                                 $this->mInPre = (bool)$preOpenMatch;
2109                         } else {
2110                                 # Don't interpret any other prefixes in preformatted text
2111                                 $prefixLength = 0;
2112                                 $prefix = $prefix2 = '';
2113                                 $t = $oLine;
2114                         }
2115
2116                         # List generation
2117                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2118                                 # Same as the last item, so no need to deal with nesting or opening stuff
2119                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2120                                 $paragraphStack = false;
2121
2122                                 if ( substr( $prefix, -1 ) === ';') {
2123                                         # The one nasty exception: definition lists work like this:
2124                                         # ; title : definition text
2125                                         # So we check for : in the remainder text to split up the
2126                                         # title and definition, without b0rking links.
2127                                         $term = $t2 = '';
2128                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2129                                                 $t = $t2;
2130                                                 $output .= $term . $this->nextItem( ':' );
2131                                         }
2132                                 }
2133                         } elseif( $prefixLength || $lastPrefixLength ) {
2134                                 # Either open or close a level...
2135                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2136                                 $paragraphStack = false;
2137
2138                                 while( $commonPrefixLength < $lastPrefixLength ) {
2139                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2140                                         --$lastPrefixLength;
2141                                 }
2142                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2143                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2144                                 }
2145                                 while ( $prefixLength > $commonPrefixLength ) {
2146                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2147                                         $output .= $this->openList( $char );
2148
2149                                         if ( ';' === $char ) {
2150                                                 # FIXME: This is dupe of code above
2151                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2152                                                         $t = $t2;
2153                                                         $output .= $term . $this->nextItem( ':' );
2154                                                 }
2155                                         }
2156                                         ++$commonPrefixLength;
2157                                 }
2158                                 $lastPrefix = $prefix2;
2159                         }
2160                         if( 0 == $prefixLength ) {
2161                                 wfProfileIn( __METHOD__."-paragraph" );
2162                                 # No prefix (not in list)--go to paragraph mode
2163                                 // XXX: use a stack for nestable elements like span, table and div
2164                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2165                                 $closematch = preg_match(
2166                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2167                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2168                                 if ( $openmatch or $closematch ) {
2169                                         $paragraphStack = false;
2170                                         # TODO bug 5718: paragraph closed
2171                                         $output .= $this->closeParagraph();
2172                                         if ( $preOpenMatch and !$preCloseMatch ) {
2173                                                 $this->mInPre = true;
2174                                         }
2175                                         if ( $closematch ) {
2176                                                 $inBlockElem = false;
2177                                         } else {
2178                                                 $inBlockElem = true;
2179                                         }
2180                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2181                                         if ( ' ' == $t{0} and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2182                                                 // pre
2183                                                 if ($this->mLastSection !== 'pre') {
2184                                                         $paragraphStack = false;
2185                                                         $output .= $this->closeParagraph().'<pre>';
2186                                                         $this->mLastSection = 'pre';
2187                                                 }
2188                                                 $t = substr( $t, 1 );
2189                                         } else {
2190                                                 // paragraph
2191                                                 if ( '' == trim($t) ) {
2192                                                         if ( $paragraphStack ) {
2193                                                                 $output .= $paragraphStack.'<br />';
2194                                                                 $paragraphStack = false;
2195                                                                 $this->mLastSection = 'p';
2196                                                         } else {
2197                                                                 if ($this->mLastSection !== 'p' ) {
2198                                                                         $output .= $this->closeParagraph();
2199                                                                         $this->mLastSection = '';
2200                                                                         $paragraphStack = '<p>';
2201                                                                 } else {
2202                                                                         $paragraphStack = '</p><p>';
2203                                                                 }
2204                                                         }
2205                                                 } else {
2206                                                         if ( $paragraphStack ) {
2207                                                                 $output .= $paragraphStack;
2208                                                                 $paragraphStack = false;
2209                                                                 $this->mLastSection = 'p';
2210                                                         } else if ($this->mLastSection !== 'p') {
2211                                                                 $output .= $this->closeParagraph().'<p>';
2212                                                                 $this->mLastSection = 'p';
2213                                                         }
2214                                                 }
2215                                         }
2216                                 }
2217                                 wfProfileOut( __METHOD__."-paragraph" );
2218                         }
2219                         // somewhere above we forget to get out of pre block (bug 785)
2220                         if($preCloseMatch && $this->mInPre) {
2221                                 $this->mInPre = false;
2222                         }
2223                         if ($paragraphStack === false) {
2224                                 $output .= $t."\n";
2225                         }
2226                 }
2227                 while ( $prefixLength ) {
2228                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2229                         --$prefixLength;
2230                 }
2231                 if ( '' != $this->mLastSection ) {
2232                         $output .= '</' . $this->mLastSection . '>';
2233                         $this->mLastSection = '';
2234                 }
2235
2236                 wfProfileOut( __METHOD__ );
2237                 return $output;
2238         }
2239
2240         /**
2241          * Split up a string on ':', ignoring any occurences inside tags
2242          * to prevent illegal overlapping.
2243          * @param string $str the string to split
2244          * @param string &$before set to everything before the ':'
2245          * @param string &$after set to everything after the ':'
2246          * return string the position of the ':', or false if none found
2247          */
2248         function findColonNoLinks($str, &$before, &$after) {
2249                 wfProfileIn( __METHOD__ );
2250
2251                 $pos = strpos( $str, ':' );
2252                 if( $pos === false ) {
2253                         // Nothing to find!
2254                         wfProfileOut( __METHOD__ );
2255                         return false;
2256                 }
2257
2258                 $lt = strpos( $str, '<' );
2259                 if( $lt === false || $lt > $pos ) {
2260                         // Easy; no tag nesting to worry about
2261                         $before = substr( $str, 0, $pos );
2262                         $after = substr( $str, $pos+1 );
2263                         wfProfileOut( __METHOD__ );
2264                         return $pos;
2265                 }
2266
2267                 // Ugly state machine to walk through avoiding tags.
2268                 $state = self::COLON_STATE_TEXT;
2269                 $stack = 0;
2270                 $len = strlen( $str );
2271                 for( $i = 0; $i < $len; $i++ ) {
2272                         $c = $str{$i};
2273
2274                         switch( $state ) {
2275                         // (Using the number is a performance hack for common cases)
2276                         case 0: // self::COLON_STATE_TEXT:
2277                                 switch( $c ) {
2278                                 case "<":
2279                                         // Could be either a <start> tag or an </end> tag
2280                                         $state = self::COLON_STATE_TAGSTART;
2281                                         break;
2282                                 case ":":
2283                                         if( $stack == 0 ) {
2284                                                 // We found it!
2285                                                 $before = substr( $str, 0, $i );
2286                                                 $after = substr( $str, $i + 1 );
2287                                                 wfProfileOut( __METHOD__ );
2288                                                 return $i;
2289                                         }
2290                                         // Embedded in a tag; don't break it.
2291                                         break;
2292                                 default:
2293                                         // Skip ahead looking for something interesting
2294                                         $colon = strpos( $str, ':', $i );
2295                                         if( $colon === false ) {
2296                                                 // Nothing else interesting
2297                                                 wfProfileOut( __METHOD__ );
2298                                                 return false;
2299                                         }
2300                                         $lt = strpos( $str, '<', $i );
2301                                         if( $stack === 0 ) {
2302                                                 if( $lt === false || $colon < $lt ) {
2303                                                         // We found it!
2304                                                         $before = substr( $str, 0, $colon );
2305                                                         $after = substr( $str, $colon + 1 );
2306                                                         wfProfileOut( __METHOD__ );
2307                                                         return $i;
2308                                                 }
2309                                         }
2310                                         if( $lt === false ) {
2311                                                 // Nothing else interesting to find; abort!
2312                                                 // We're nested, but there's no close tags left. Abort!
2313                                                 break 2;
2314                                         }
2315                                         // Skip ahead to next tag start
2316                                         $i = $lt;
2317                                         $state = self::COLON_STATE_TAGSTART;
2318                                 }
2319                                 break;
2320                         case 1: // self::COLON_STATE_TAG:
2321                                 // In a <tag>
2322                                 switch( $c ) {
2323                                 case ">":
2324                                         $stack++;
2325                                         $state = self::COLON_STATE_TEXT;
2326                                         break;
2327                                 case "/":
2328                                         // Slash may be followed by >?
2329                                         $state = self::COLON_STATE_TAGSLASH;
2330                                         break;
2331                                 default:
2332                                         // ignore
2333                                 }
2334                                 break;
2335                         case 2: // self::COLON_STATE_TAGSTART:
2336                                 switch( $c ) {
2337                                 case "/":
2338                                         $state = self::COLON_STATE_CLOSETAG;
2339                                         break;
2340                                 case "!":
2341                                         $state = self::COLON_STATE_COMMENT;
2342                                         break;
2343                                 case ">":
2344                                         // Illegal early close? This shouldn't happen D:
2345                                         $state = self::COLON_STATE_TEXT;
2346                                         break;
2347                                 default:
2348                                         $state = self::COLON_STATE_TAG;
2349                                 }
2350                                 break;
2351                         case 3: // self::COLON_STATE_CLOSETAG:
2352                                 // In a </tag>
2353                                 if( $c === ">" ) {
2354                                         $stack--;
2355                                         if( $stack < 0 ) {
2356                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2357                                                 wfProfileOut( __METHOD__ );
2358                                                 return false;
2359                                         }
2360                                         $state = self::COLON_STATE_TEXT;
2361                                 }
2362                                 break;
2363                         case self::COLON_STATE_TAGSLASH:
2364                                 if( $c === ">" ) {
2365                                         // Yes, a self-closed tag <blah/>
2366                                         $state = self::COLON_STATE_TEXT;
2367                                 } else {
2368                                         // Probably we're jumping the gun, and this is an attribute
2369                                         $state = self::COLON_STATE_TAG;
2370                                 }
2371                                 break;
2372                         case 5: // self::COLON_STATE_COMMENT:
2373                                 if( $c === "-" ) {
2374                                         $state = self::COLON_STATE_COMMENTDASH;
2375                                 }
2376                                 break;
2377                         case self::COLON_STATE_COMMENTDASH:
2378                                 if( $c === "-" ) {
2379                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2380                                 } else {
2381                                         $state = self::COLON_STATE_COMMENT;
2382                                 }
2383                                 break;
2384                         case self::COLON_STATE_COMMENTDASHDASH:
2385                                 if( $c === ">" ) {
2386                                         $state = self::COLON_STATE_TEXT;
2387                                 } else {
2388                                         $state = self::COLON_STATE_COMMENT;
2389                                 }
2390                                 break;
2391                         default:
2392                                 throw new MWException( "State machine error in " . __METHOD__ );
2393                         }
2394                 }
2395                 if( $stack > 0 ) {
2396                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2397                         return false;
2398                 }
2399                 wfProfileOut( __METHOD__ );
2400                 return false;
2401         }
2402
2403         /**
2404          * Return value of a magic variable (like PAGENAME)
2405          *
2406          * @private
2407          */
2408         function getVariableValue( $index ) {
2409                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2410
2411                 /**
2412                  * Some of these require message or data lookups and can be
2413                  * expensive to check many times.
2414                  */
2415                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2416                         if ( isset( $this->mVarCache[$index] ) ) {
2417                                 return $this->mVarCache[$index];
2418                         }
2419                 }
2420
2421                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2422                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2423
2424                 # Use the time zone
2425                 global $wgLocaltimezone;
2426                 if ( isset( $wgLocaltimezone ) ) {
2427                         $oldtz = getenv( 'TZ' );
2428                         putenv( 'TZ='.$wgLocaltimezone );
2429                 }
2430
2431                 wfSuppressWarnings(); // E_STRICT system time bitching
2432                 $localTimestamp = date( 'YmdHis', $ts );
2433                 $localMonth = date( 'm', $ts );
2434                 $localMonthName = date( 'n', $ts );
2435                 $localDay = date( 'j', $ts );
2436                 $localDay2 = date( 'd', $ts );
2437                 $localDayOfWeek = date( 'w', $ts );
2438                 $localWeek = date( 'W', $ts );
2439                 $localYear = date( 'Y', $ts );
2440                 $localHour = date( 'H', $ts );
2441                 if ( isset( $wgLocaltimezone ) ) {
2442                         putenv( 'TZ='.$oldtz );
2443                 }
2444                 wfRestoreWarnings();
2445
2446                 switch ( $index ) {
2447                         case 'currentmonth':
2448                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2449                         case 'currentmonthname':
2450                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2451                         case 'currentmonthnamegen':
2452                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2453                         case 'currentmonthabbrev':
2454                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2455                         case 'currentday':
2456                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2457                         case 'currentday2':
2458                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2459                         case 'localmonth':
2460                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2461                         case 'localmonthname':
2462                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2463                         case 'localmonthnamegen':
2464                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2465                         case 'localmonthabbrev':
2466                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2467                         case 'localday':
2468                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2469                         case 'localday2':
2470                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2471                         case 'pagename':
2472                                 return wfEscapeWikiText( $this->mTitle->getText() );
2473                         case 'pagenamee':
2474                                 return $this->mTitle->getPartialURL();
2475                         case 'fullpagename':
2476                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2477                         case 'fullpagenamee':
2478                                 return $this->mTitle->getPrefixedURL();
2479                         case 'subpagename':
2480                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2481                         case 'subpagenamee':
2482                                 return $this->mTitle->getSubpageUrlForm();
2483                         case 'basepagename':
2484                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2485                         case 'basepagenamee':
2486                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2487                         case 'talkpagename':
2488                                 if( $this->mTitle->canTalk() ) {
2489                                         $talkPage = $this->mTitle->getTalkPage();
2490                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2491                                 } else {
2492                                         return '';
2493                                 }
2494                         case 'talkpagenamee':
2495                                 if( $this->mTitle->canTalk() ) {
2496                                         $talkPage = $this->mTitle->getTalkPage();
2497                                         return $talkPage->getPrefixedUrl();
2498                                 } else {
2499                                         return '';
2500                                 }
2501                         case 'subjectpagename':
2502                                 $subjPage = $this->mTitle->getSubjectPage();
2503                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2504                         case 'subjectpagenamee':
2505                                 $subjPage = $this->mTitle->getSubjectPage();
2506                                 return $subjPage->getPrefixedUrl();
2507                         case 'revisionid':
2508                                 // Let the edit saving system know we should parse the page
2509                                 // *after* a revision ID has been assigned.
2510                                 $this->mOutput->setFlag( 'vary-revision' );
2511                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2512                                 return $this->mRevisionId;
2513                         case 'revisionday':
2514                                 // Let the edit saving system know we should parse the page
2515                                 // *after* a revision ID has been assigned. This is for null edits.
2516                                 $this->mOutput->setFlag( 'vary-revision' );
2517                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2518                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2519                         case 'revisionday2':
2520                                 // Let the edit saving system know we should parse the page
2521                                 // *after* a revision ID has been assigned. This is for null edits.
2522                                 $this->mOutput->setFlag( 'vary-revision' );
2523                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2524                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2525                         case 'revisionmonth':
2526                                 // Let the edit saving system know we should parse the page
2527                                 // *after* a revision ID has been assigned. This is for null edits.
2528                                 $this->mOutput->setFlag( 'vary-revision' );
2529                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2530                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2531                         case 'revisionyear':
2532                                 // Let the edit saving system know we should parse the page
2533                                 // *after* a revision ID has been assigned. This is for null edits.
2534                                 $this->mOutput->setFlag( 'vary-revision' );
2535                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2536                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2537                         case 'revisiontimestamp':
2538                                 // Let the edit saving system know we should parse the page
2539                                 // *after* a revision ID has been assigned. This is for null edits.
2540                                 $this->mOutput->setFlag( 'vary-revision' );
2541                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2542                                 return $this->getRevisionTimestamp();
2543                         case 'namespace':
2544                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2545                         case 'namespacee':
2546                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2547                         case 'talkspace':
2548                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2549                         case 'talkspacee':
2550                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2551                         case 'subjectspace':
2552                                 return $this->mTitle->getSubjectNsText();
2553                         case 'subjectspacee':
2554                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2555                         case 'currentdayname':
2556                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2557                         case 'currentyear':
2558                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2559                         case 'currenttime':
2560                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2561                         case 'currenthour':
2562                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2563                         case 'currentweek':
2564                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2565                                 // int to remove the padding
2566                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2567                         case 'currentdow':
2568                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2569                         case 'localdayname':
2570                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2571                         case 'localyear':
2572                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2573                         case 'localtime':
2574                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2575                         case 'localhour':
2576                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2577                         case 'localweek':
2578                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2579                                 // int to remove the padding
2580                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2581                         case 'localdow':
2582                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2583                         case 'numberofarticles':
2584                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2585                         case 'numberoffiles':
2586                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2587                         case 'numberofusers':
2588                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2589                         case 'numberofpages':
2590                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2591                         case 'numberofadmins':
2592                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2593                         case 'numberofedits':
2594                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2595                         case 'numberofviews':
2596                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::views() );
2597                         case 'currenttimestamp':
2598                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2599                         case 'localtimestamp':
2600                                 return $this->mVarCache[$index] = $localTimestamp;
2601                         case 'currentversion':
2602                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2603                         case 'sitename':
2604                                 return $wgSitename;
2605                         case 'server':
2606                                 return $wgServer;
2607                         case 'servername':
2608                                 return $wgServerName;
2609                         case 'scriptpath':
2610                                 return $wgScriptPath;
2611                         case 'directionmark':
2612                                 return $wgContLang->getDirMark();
2613                         case 'contentlanguage':
2614                                 global $wgContLanguageCode;
2615                                 return $wgContLanguageCode;
2616                         default:
2617                                 $ret = null;
2618                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2619                                         return $ret;
2620                                 else
2621                                         return null;
2622                 }
2623         }
2624
2625         /**
2626          * initialise the magic variables (like CURRENTMONTHNAME)
2627          *
2628          * @private
2629          */
2630         function initialiseVariables() {
2631                 wfProfileIn( __METHOD__ );
2632                 $variableIDs = MagicWord::getVariableIDs();
2633
2634                 $this->mVariables = new MagicWordArray( $variableIDs );
2635                 wfProfileOut( __METHOD__ );
2636         }
2637
2638         /**
2639          * Preprocess some wikitext and return the document tree.
2640          * This is the ghost of replace_variables().
2641          *
2642          * @param string $text The text to parse
2643          * @param integer flags Bitwise combination of:
2644          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2645          *                                     included. Default is to assume a direct page view.
2646          *
2647          * The generated DOM tree must depend only on the input text and the flags.
2648          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2649          *
2650          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2651          * change in the DOM tree for a given text, must be passed through the section identifier
2652          * in the section edit link and thus back to extractSections().
2653          *
2654          * The output of this function is currently only cached in process memory, but a persistent
2655          * cache may be implemented at a later date which takes further advantage of these strict
2656          * dependency requirements.
2657          *
2658          * @private
2659          */
2660         function preprocessToDom ( $text, $flags = 0 ) {
2661                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2662                 return $dom;
2663         }
2664
2665         /*
2666          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2667          */
2668         public static function splitWhitespace( $s ) {
2669                 $ltrimmed = ltrim( $s );
2670                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2671                 $trimmed = rtrim( $ltrimmed );
2672                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2673                 if ( $diff > 0 ) {
2674                         $w2 = substr( $ltrimmed, -$diff );
2675                 } else {
2676                         $w2 = '';
2677                 }
2678                 return array( $w1, $trimmed, $w2 );
2679         }
2680
2681         /**
2682          * Replace magic variables, templates, and template arguments
2683          * with the appropriate text. Templates are substituted recursively,
2684          * taking care to avoid infinite loops.
2685          *
2686          * Note that the substitution depends on value of $mOutputType:
2687          *  self::OT_WIKI: only {{subst:}} templates
2688          *  self::OT_PREPROCESS: templates but not extension tags
2689          *  self::OT_HTML: all templates and extension tags
2690          *
2691          * @param string $tex The text to transform
2692          * @param PPFrame $frame Object describing the arguments passed to the template.
2693          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2694          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2695          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2696          * @private
2697          */
2698         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2699                 # Prevent too big inclusions
2700                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2701                         return $text;
2702                 }
2703
2704                 wfProfileIn( __METHOD__ );
2705
2706                 if ( $frame === false ) {
2707                         $frame = $this->getPreprocessor()->newFrame();
2708                 } elseif ( !( $frame instanceof PPFrame ) ) {
2709                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2710                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2711                 }
2712
2713                 $dom = $this->preprocessToDom( $text );
2714                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2715                 $text = $frame->expand( $dom, $flags );
2716
2717                 wfProfileOut( __METHOD__ );
2718                 return $text;
2719         }
2720
2721         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2722         static function createAssocArgs( $args ) {
2723                 $assocArgs = array();
2724                 $index = 1;
2725                 foreach( $args as $arg ) {
2726                         $eqpos = strpos( $arg, '=' );
2727                         if ( $eqpos === false ) {
2728                                 $assocArgs[$index++] = $arg;
2729                         } else {
2730                                 $name = trim( substr( $arg, 0, $eqpos ) );
2731                                 $value = trim( substr( $arg, $eqpos+1 ) );
2732                                 if ( $value === false ) {
2733                                         $value = '';
2734                                 }
2735                                 if ( $name !== false ) {
2736                                         $assocArgs[$name] = $value;
2737                                 }
2738                         }
2739                 }
2740
2741                 return $assocArgs;
2742         }
2743
2744         /**
2745          * Warn the user when a parser limitation is reached
2746          * Will warn at most once the user per limitation type
2747          *
2748          * @param string $limitationType, should be one of:
2749          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2750          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2751          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2752          * @params int $current, $max When an explicit limit has been
2753          *       exceeded, provide the values (optional)
2754          */
2755         function limitationWarn( $limitationType, $current=null, $max=null) {
2756                 $msgName = $limitationType . '-warning';
2757                 //does no harm if $current and $max are present but are unnecessary for the message
2758                 $warning = wfMsgExt( $msgName, array( 'parsemag', 'escape' ), $current, $max );
2759                 $this->mOutput->addWarning( $warning );
2760                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2761                 if ( $cat ) {
2762                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2763                 }
2764         }
2765
2766         /**
2767          * Return the text of a template, after recursively
2768          * replacing any variables or templates within the template.
2769          *
2770          * @param array $piece The parts of the template
2771          *  $piece['title']: the title, i.e. the part before the |
2772          *  $piece['parts']: the parameter array
2773          *  $piece['lineStart']: whether the brace was at the start of a line
2774          * @param PPFrame The current frame, contains template arguments
2775          * @return string the text of the template
2776          * @private
2777          */
2778         function braceSubstitution( $piece, $frame ) {
2779                 global $wgContLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2780                 wfProfileIn( __METHOD__ );
2781                 wfProfileIn( __METHOD__.'-setup' );
2782
2783                 # Flags
2784                 $found = false;             # $text has been filled
2785                 $nowiki = false;            # wiki markup in $text should be escaped
2786                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2787                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2788                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2789                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2790
2791                 # Title object, where $text came from
2792                 $title = NULL;
2793
2794                 # $part1 is the bit before the first |, and must contain only title characters.
2795                 # Various prefixes will be stripped from it later.
2796                 $titleWithSpaces = $frame->expand( $piece['title'] );
2797                 $part1 = trim( $titleWithSpaces );
2798                 $titleText = false;
2799
2800                 # Original title text preserved for various purposes
2801                 $originalTitle = $part1;
2802
2803                 # $args is a list of argument nodes, starting from index 0, not including $part1
2804                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2805                 wfProfileOut( __METHOD__.'-setup' );
2806
2807                 # SUBST
2808                 wfProfileIn( __METHOD__.'-modifiers' );
2809                 if ( !$found ) {
2810                         $mwSubst = MagicWord::get( 'subst' );
2811                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2812                                 # One of two possibilities is true:
2813                                 # 1) Found SUBST but not in the PST phase
2814                                 # 2) Didn't find SUBST and in the PST phase
2815                                 # In either case, return without further processing
2816                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2817                                 $isLocalObj = true;
2818                                 $found = true;
2819                         }
2820                 }
2821
2822                 # Variables
2823                 if ( !$found && $args->getLength() == 0 ) {
2824                         $id = $this->mVariables->matchStartToEnd( $part1 );
2825                         if ( $id !== false ) {
2826                                 $text = $this->getVariableValue( $id );
2827                                 if (MagicWord::getCacheTTL($id)>-1)
2828                                         $this->mOutput->mContainsOldMagic = true;
2829                                 $found = true;
2830                         }
2831                 }
2832
2833                 # MSG, MSGNW and RAW
2834                 if ( !$found ) {
2835                         # Check for MSGNW:
2836                         $mwMsgnw = MagicWord::get( 'msgnw' );
2837                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2838                                 $nowiki = true;
2839                         } else {
2840                                 # Remove obsolete MSG:
2841                                 $mwMsg = MagicWord::get( 'msg' );
2842                                 $mwMsg->matchStartAndRemove( $part1 );
2843                         }
2844
2845                         # Check for RAW:
2846                         $mwRaw = MagicWord::get( 'raw' );
2847                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2848                                 $forceRawInterwiki = true;
2849                         }
2850                 }
2851                 wfProfileOut( __METHOD__.'-modifiers' );
2852
2853                 # Parser functions
2854                 if ( !$found ) {
2855                         wfProfileIn( __METHOD__ . '-pfunc' );
2856
2857                         $colonPos = strpos( $part1, ':' );
2858                         if ( $colonPos !== false ) {
2859                                 # Case sensitive functions
2860                                 $function = substr( $part1, 0, $colonPos );
2861                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2862                                         $function = $this->mFunctionSynonyms[1][$function];
2863                                 } else {
2864                                         # Case insensitive functions
2865                                         $function = strtolower( $function );
2866                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2867                                                 $function = $this->mFunctionSynonyms[0][$function];
2868                                         } else {
2869                                                 $function = false;
2870                                         }
2871                                 }
2872                                 if ( $function ) {
2873                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2874                                         $initialArgs = array( &$this );
2875                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2876                                         if ( $flags & SFH_OBJECT_ARGS ) {
2877                                                 # Add a frame parameter, and pass the arguments as an array
2878                                                 $allArgs = $initialArgs;
2879                                                 $allArgs[] = $frame;
2880                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2881                                                         $funcArgs[] = $args->item( $i );
2882                                                 }
2883                                                 $allArgs[] = $funcArgs;
2884                                         } else {
2885                                                 # Convert arguments to plain text
2886                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2887                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2888                                                 }
2889                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2890                                         }
2891
2892                                         # Workaround for PHP bug 35229 and similar
2893                                         if ( !is_callable( $callback ) ) {
2894                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2895                                         }
2896                                         $result = call_user_func_array( $callback, $allArgs );
2897                                         $found = true;
2898                                         $noparse = true;
2899                                         $preprocessFlags = 0;
2900
2901                                         if ( is_array( $result ) ) {
2902                                                 if ( isset( $result[0] ) ) {
2903                                                         $text = $result[0];
2904                                                         unset( $result[0] );
2905                                                 }
2906
2907                                                 // Extract flags into the local scope
2908                                                 // This allows callers to set flags such as nowiki, found, etc.
2909                                                 extract( $result );
2910                                         } else {
2911                                                 $text = $result;
2912                                         }
2913                                         if ( !$noparse ) {
2914                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2915                                                 $isChildObj = true;
2916                                         }
2917                                 }
2918                         }
2919                         wfProfileOut( __METHOD__ . '-pfunc' );
2920                 }
2921
2922                 # Finish mangling title and then check for loops.
2923                 # Set $title to a Title object and $titleText to the PDBK
2924                 if ( !$found ) {
2925                         $ns = NS_TEMPLATE;
2926                         # Split the title into page and subpage
2927                         $subpage = '';
2928                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2929                         if ($subpage !== '') {
2930                                 $ns = $this->mTitle->getNamespace();
2931                         }
2932                         $title = Title::newFromText( $part1, $ns );
2933                         if ( $title ) {
2934                                 $titleText = $title->getPrefixedText();
2935                                 # Check for language variants if the template is not found
2936                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2937                                         $wgContLang->findVariantLink( $part1, $title, true );
2938                                 }
2939                                 # Do infinite loop check
2940                                 if ( !$frame->loopCheck( $title ) ) {
2941                                         $found = true;
2942                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2943                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2944                                 }
2945                                 # Do recursion depth check
2946                                 $limit = $this->mOptions->getMaxTemplateDepth();
2947                                 if ( $frame->depth >= $limit ) {
2948                                         $found = true;
2949                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2950                                 }
2951                         }
2952                 }
2953
2954                 # Load from database
2955                 if ( !$found && $title ) {
2956                         wfProfileIn( __METHOD__ . '-loadtpl' );
2957                         if ( !$title->isExternal() ) {
2958                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2959                                         $text = SpecialPage::capturePath( $title );
2960                                         if ( is_string( $text ) ) {
2961                                                 $found = true;
2962                                                 $isHTML = true;
2963                                                 $this->disableCache();
2964                                         }
2965                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2966                                         $found = false; //access denied
2967                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2968                                 } else {
2969                                         list( $text, $title ) = $this->getTemplateDom( $title );
2970                                         if ( $text !== false ) {
2971                                                 $found = true;
2972                                                 $isChildObj = true;
2973                                         }
2974                                 }
2975
2976                                 # If the title is valid but undisplayable, make a link to it
2977                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2978                                         $text = "[[:$titleText]]";
2979                                         $found = true;
2980                                 }
2981                         } elseif ( $title->isTrans() ) {
2982                                 // Interwiki transclusion
2983                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2984                                         $text = $this->interwikiTransclude( $title, 'render' );
2985                                         $isHTML = true;
2986                                 } else {
2987                                         $text = $this->interwikiTransclude( $title, 'raw' );
2988                                         // Preprocess it like a template
2989                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2990                                         $isChildObj = true;
2991                                 }
2992                                 $found = true;
2993                         }
2994                         wfProfileOut( __METHOD__ . '-loadtpl' );
2995                 }
2996
2997                 # If we haven't found text to substitute by now, we're done
2998                 # Recover the source wikitext and return it
2999                 if ( !$found ) {
3000                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3001                         wfProfileOut( __METHOD__ );
3002                         return array( 'object' => $text );
3003                 }
3004
3005                 # Expand DOM-style return values in a child frame
3006                 if ( $isChildObj ) {
3007                         # Clean up argument array
3008                         $newFrame = $frame->newChild( $args, $title );
3009
3010                         if ( $nowiki ) {
3011                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3012                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3013                                 # Expansion is eligible for the empty-frame cache
3014                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3015                                         $text = $this->mTplExpandCache[$titleText];
3016                                 } else {
3017                                         $text = $newFrame->expand( $text );
3018                                         $this->mTplExpandCache[$titleText] = $text;
3019                                 }
3020                         } else {
3021                                 # Uncached expansion
3022                                 $text = $newFrame->expand( $text );
3023                         }
3024                 }
3025                 if ( $isLocalObj && $nowiki ) {
3026                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3027                         $isLocalObj = false;
3028                 }
3029
3030                 # Replace raw HTML by a placeholder
3031                 # Add a blank line preceding, to prevent it from mucking up
3032                 # immediately preceding headings
3033                 if ( $isHTML ) {
3034                         $text = "\n\n" . $this->insertStripItem( $text );
3035                 }
3036                 # Escape nowiki-style return values
3037                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3038                         $text = wfEscapeWikiText( $text );
3039                 }
3040                 # Bug 529: if the template begins with a table or block-level
3041                 # element, it should be treated as beginning a new line.
3042                 # This behaviour is somewhat controversial.
3043                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3044                         $text = "\n" . $text;
3045                 }
3046
3047                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3048                         # Error, oversize inclusion
3049                         $text = "[[$originalTitle]]" .
3050                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3051                         $this->limitationWarn( 'post-expand-template-inclusion' );
3052                 }
3053
3054                 if ( $isLocalObj ) {
3055                         $ret = array( 'object' => $text );
3056                 } else {
3057                         $ret = array( 'text' => $text );
3058                 }
3059
3060                 wfProfileOut( __METHOD__ );
3061                 return $ret;
3062         }
3063
3064         /**
3065          * Get the semi-parsed DOM representation of a template with a given title,
3066          * and its redirect destination title. Cached.
3067          */
3068         function getTemplateDom( $title ) {
3069                 $cacheTitle = $title;
3070                 $titleText = $title->getPrefixedDBkey();
3071
3072                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3073                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3074                         $title = Title::makeTitle( $ns, $dbk );
3075                         $titleText = $title->getPrefixedDBkey();
3076                 }
3077                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3078                         return array( $this->mTplDomCache[$titleText], $title );
3079                 }
3080
3081                 // Cache miss, go to the database
3082                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3083
3084                 if ( $text === false ) {
3085                         $this->mTplDomCache[$titleText] = false;
3086                         return array( false, $title );
3087                 }
3088
3089                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3090                 $this->mTplDomCache[ $titleText ] = $dom;
3091
3092                 if (! $title->equals($cacheTitle)) {
3093                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3094                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3095                 }
3096
3097                 return array( $dom, $title );
3098         }
3099
3100         /**
3101          * Fetch the unparsed text of a template and register a reference to it.
3102          */
3103         function fetchTemplateAndTitle( $title ) {
3104                 $templateCb = $this->mOptions->getTemplateCallback();
3105                 $stuff = call_user_func( $templateCb, $title, $this );
3106                 $text = $stuff['text'];
3107                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3108                 if ( isset( $stuff['deps'] ) ) {
3109                         foreach ( $stuff['deps'] as $dep ) {
3110                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3111                         }
3112                 }
3113                 return array($text,$finalTitle);
3114         }
3115
3116         function fetchTemplate( $title ) {
3117                 $rv = $this->fetchTemplateAndTitle($title);
3118                 return $rv[0];
3119         }
3120
3121         /**
3122          * Static function to get a template
3123          * Can be overridden via ParserOptions::setTemplateCallback().
3124          */
3125         static function statelessFetchTemplate( $title, $parser=false ) {
3126                 $text = $skip = false;
3127                 $finalTitle = $title;
3128                 $deps = array();
3129
3130                 // Loop to fetch the article, with up to 1 redirect
3131                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3132                         # Give extensions a chance to select the revision instead
3133                         $id = false; // Assume current
3134                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3135
3136                         if( $skip ) {
3137                                 $text = false;
3138                                 $deps[] = array(
3139                                         'title' => $title,
3140                                         'page_id' => $title->getArticleID(),
3141                                         'rev_id' => null );
3142                                 break;
3143                         }
3144                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3145                         $rev_id = $rev ? $rev->getId() : 0;
3146                         // If there is no current revision, there is no page
3147                         if( $id === false && !$rev ) {
3148                                 $linkCache = LinkCache::singleton();
3149                                 $linkCache->addBadLinkObj( $title );
3150                         }
3151
3152                         $deps[] = array(
3153                                 'title' => $title,
3154                                 'page_id' => $title->getArticleID(),
3155                                 'rev_id' => $rev_id );
3156
3157                         if( $rev ) {
3158                                 $text = $rev->getText();
3159                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3160                                 global $wgContLang;
3161                                 $message = $wgContLang->lcfirst( $title->getText() );
3162                                 $text = wfMsgForContentNoTrans( $message );
3163                                 if( wfEmptyMsg( $message, $text ) ) {
3164                                         $text = false;
3165                                         break;
3166                                 }
3167                         } else {
3168                                 break;
3169                         }
3170                         if ( $text === false ) {
3171                                 break;
3172                         }
3173                         // Redirect?
3174                         $finalTitle = $title;
3175                         $title = Title::newFromRedirect( $text );
3176                 }
3177                 return array(
3178                         'text' => $text,
3179                         'finalTitle' => $finalTitle,
3180                         'deps' => $deps );
3181         }
3182
3183         /**
3184          * Transclude an interwiki link.
3185          */
3186         function interwikiTransclude( $title, $action ) {
3187                 global $wgEnableScaryTranscluding;
3188
3189                 if (!$wgEnableScaryTranscluding)
3190                         return wfMsg('scarytranscludedisabled');
3191
3192                 $url = $title->getFullUrl( "action=$action" );
3193
3194                 if (strlen($url) > 255)
3195                         return wfMsg('scarytranscludetoolong');
3196                 return $this->fetchScaryTemplateMaybeFromCache($url);
3197         }
3198
3199         function fetchScaryTemplateMaybeFromCache($url) {
3200                 global $wgTranscludeCacheExpiry;
3201                 $dbr = wfGetDB(DB_SLAVE);
3202                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3203                                 array('tc_url' => $url));
3204                 if ($obj) {
3205                         $time = $obj->tc_time;
3206                         $text = $obj->tc_contents;
3207                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3208                                 return $text;
3209                         }
3210                 }
3211
3212                 $text = Http::get($url);
3213                 if (!$text)
3214                         return wfMsg('scarytranscludefailed', $url);
3215
3216                 $dbw = wfGetDB(DB_MASTER);
3217                 $dbw->replace('transcache', array('tc_url'), array(
3218                         'tc_url' => $url,
3219                         'tc_time' => time(),
3220                         'tc_contents' => $text));
3221                 return $text;
3222         }
3223
3224
3225         /**
3226          * Triple brace replacement -- used for template arguments
3227          * @private
3228          */
3229         function argSubstitution( $piece, $frame ) {
3230                 wfProfileIn( __METHOD__ );
3231
3232                 $error = false;
3233                 $parts = $piece['parts'];
3234                 $nameWithSpaces = $frame->expand( $piece['title'] );
3235                 $argName = trim( $nameWithSpaces );
3236                 $object = false;
3237                 $text = $frame->getArgument( $argName );
3238                 if (  $text === false && $parts->getLength() > 0
3239                   && (
3240                     $this->ot['html']
3241                     || $this->ot['pre']
3242                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3243                   )
3244                 ) {
3245                         # No match in frame, use the supplied default
3246                         $object = $parts->item( 0 )->getChildren();
3247                 }
3248                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3249                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3250                         $this->limitationWarn( 'post-expand-template-argument' );
3251                 }
3252
3253                 if ( $text === false && $object === false ) {
3254                         # No match anywhere
3255                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3256                 }
3257                 if ( $error !== false ) {
3258                         $text .= $error;
3259                 }
3260                 if ( $object !== false ) {
3261                         $ret = array( 'object' => $object );
3262                 } else {
3263                         $ret = array( 'text' => $text );
3264                 }
3265
3266                 wfProfileOut( __METHOD__ );
3267                 return $ret;
3268         }
3269
3270         /**
3271          * Return the text to be used for a given extension tag.
3272          * This is the ghost of strip().
3273          *
3274          * @param array $params Associative array of parameters:
3275          *     name       PPNode for the tag name
3276          *     attr       PPNode for unparsed text where tag attributes are thought to be
3277          *     attributes Optional associative array of parsed attributes
3278          *     inner      Contents of extension element
3279          *     noClose    Original text did not have a close tag
3280          * @param PPFrame $frame
3281          */
3282         function extensionSubstitution( $params, $frame ) {
3283                 global $wgRawHtml, $wgContLang;
3284
3285                 $name = $frame->expand( $params['name'] );
3286                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3287                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3288
3289                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3290
3291                 if ( $this->ot['html'] ) {
3292                         $name = strtolower( $name );
3293
3294                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3295                         if ( isset( $params['attributes'] ) ) {
3296                                 $attributes = $attributes + $params['attributes'];
3297                         }
3298                         switch ( $name ) {
3299                                 case 'html':
3300                                         if( $wgRawHtml ) {
3301                                                 $output = $content;
3302                                                 break;
3303                                         } else {
3304                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3305                                         }
3306                                 case 'nowiki':
3307                                         $output = Xml::escapeTagsOnly( $content );
3308                                         break;
3309                                 case 'math':
3310                                         $output = $wgContLang->armourMath(
3311                                                 MathRenderer::renderMath( $content, $attributes ) );
3312                                         break;
3313                                 case 'gallery':
3314                                         $output = $this->renderImageGallery( $content, $attributes );
3315                                         break;
3316                                 default:
3317                                         if( isset( $this->mTagHooks[$name] ) ) {
3318                                                 # Workaround for PHP bug 35229 and similar
3319                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3320                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3321                                                 }
3322                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3323                                                         array( $content, $attributes, $this ) );
3324                                         } else {
3325                                                 $output = '<span class="error">Invalid tag extension name: ' .
3326                                                         htmlspecialchars( $name ) . '</span>';
3327                                         }
3328                         }
3329                 } else {
3330                         if ( is_null( $attrText ) ) {
3331                                 $attrText = '';
3332                         }
3333                         if ( isset( $params['attributes'] ) ) {
3334                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3335                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3336                                                 htmlspecialchars( $attrValue ) . '"';
3337                                 }
3338                         }
3339                         if ( $content === null ) {
3340                                 $output = "<$name$attrText/>";
3341                         } else {
3342                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3343                                 $output = "<$name$attrText>$content$close";
3344                         }
3345                 }
3346
3347                 if ( $name === 'html' || $name === 'nowiki' ) {
3348                         $this->mStripState->nowiki->setPair( $marker, $output );
3349                 } else {
3350                         $this->mStripState->general->setPair( $marker, $output );
3351                 }
3352                 return $marker;
3353         }
3354
3355         /**
3356          * Increment an include size counter
3357          *
3358          * @param string $type The type of expansion
3359          * @param integer $size The size of the text
3360          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3361          */
3362         function incrementIncludeSize( $type, $size ) {
3363                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3364                         return false;
3365                 } else {
3366                         $this->mIncludeSizes[$type] += $size;
3367                         return true;
3368                 }
3369         }
3370
3371         /**
3372          * Increment the expensive function count
3373          *
3374          * @return boolean False if the limit has been exceeded
3375          */
3376         function incrementExpensiveFunctionCount() {
3377                 global $wgExpensiveParserFunctionLimit;
3378                 $this->mExpensiveFunctionCount++;
3379                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3380                         return true;
3381                 }
3382                 return false;
3383         }
3384
3385         /**
3386          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3387          * Fills $this->mDoubleUnderscores, returns the modified text
3388          */
3389         function doDoubleUnderscore( $text ) {
3390                 // The position of __TOC__ needs to be recorded
3391                 $mw = MagicWord::get( 'toc' );
3392                 if( $mw->match( $text ) ) {
3393                         $this->mShowToc = true;
3394                         $this->mForceTocPosition = true;
3395
3396                         // Set a placeholder. At the end we'll fill it in with the TOC.
3397                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3398
3399                         // Only keep the first one.
3400                         $text = $mw->replace( '', $text );
3401                 }
3402
3403                 // Now match and remove the rest of them
3404                 $mwa = MagicWord::getDoubleUnderscoreArray();
3405                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3406
3407                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3408                         $this->mOutput->mNoGallery = true;
3409                 }
3410                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3411                         $this->mShowToc = false;
3412                 }
3413                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3414                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3415
3416                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3417                         if ( $containerCategory ) {
3418                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3419                         } else {
3420                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3421                         }
3422                 }
3423                 # (bug 8068) Allow control over whether robots index a page.
3424                 #
3425                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3426                 # is not desirable, the last one on the page should win.
3427                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3428                         $this->mOutput->setIndexPolicy( 'noindex' );
3429                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3430                         $this->mOutput->setIndexPolicy( 'index' );
3431                 }
3432
3433                 return $text;
3434         }
3435
3436         /**
3437          * This function accomplishes several tasks:
3438          * 1) Auto-number headings if that option is enabled
3439          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3440          * 3) Add a Table of contents on the top for users who have enabled the option
3441          * 4) Auto-anchor headings
3442          *
3443          * It loops through all headlines, collects the necessary data, then splits up the
3444          * string and re-inserts the newly formatted headlines.
3445          *
3446          * @param string $text
3447          * @param boolean $isMain
3448          * @private
3449          */
3450         function formatHeadings( $text, $isMain=true ) {
3451                 global $wgMaxTocLevel, $wgContLang, $wgEnforceHtmlIds;
3452
3453                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3454                 $showEditLink = $this->mOptions->getEditSection();
3455
3456                 // Do not call quickUserCan unless necessary
3457                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3458                         $showEditLink = 0;
3459                 }
3460
3461                 # Inhibit editsection links if requested in the page
3462                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3463                         $showEditLink = 0;
3464                 }
3465
3466                 # Get all headlines for numbering them and adding funky stuff like [edit]
3467                 # links - this is for later, but we need the number of headlines right now
3468                 $matches = array();
3469                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3470
3471                 # if there are fewer than 4 headlines in the article, do not show TOC
3472                 # unless it's been explicitly enabled.
3473                 $enoughToc = $this->mShowToc &&
3474                         (($numMatches >= 4) || $this->mForceTocPosition);
3475
3476                 # Allow user to stipulate that a page should have a "new section"
3477                 # link added via __NEWSECTIONLINK__
3478                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3479                         $this->mOutput->setNewSection( true );
3480                 }
3481
3482                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3483                 # override above conditions and always show TOC above first header
3484                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3485                         $this->mShowToc = true;
3486                         $enoughToc = true;
3487                 }
3488
3489                 # We need this to perform operations on the HTML
3490                 $sk = $this->mOptions->getSkin();
3491
3492                 # headline counter
3493                 $headlineCount = 0;
3494                 $numVisible = 0;
3495
3496                 # Ugh .. the TOC should have neat indentation levels which can be
3497                 # passed to the skin functions. These are determined here
3498                 $toc = '';
3499                 $full = '';
3500                 $head = array();
3501                 $sublevelCount = array();
3502                 $levelCount = array();
3503                 $toclevel = 0;
3504                 $level = 0;
3505                 $prevlevel = 0;
3506                 $toclevel = 0;
3507                 $prevtoclevel = 0;
3508                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3509                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3510                 $tocraw = array();
3511
3512                 foreach( $matches[3] as $headline ) {
3513                         $isTemplate = false;
3514                         $titleText = false;
3515                         $sectionIndex = false;
3516                         $numbering = '';
3517                         $markerMatches = array();
3518                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3519                                 $serial = $markerMatches[1];
3520                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3521                                 $isTemplate = ($titleText != $baseTitleText);
3522                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3523                         }
3524
3525                         if( $toclevel ) {
3526                                 $prevlevel = $level;
3527                                 $prevtoclevel = $toclevel;
3528                         }
3529                         $level = $matches[1][$headlineCount];
3530
3531                         if( $doNumberHeadings || $enoughToc ) {
3532
3533                                 if ( $level > $prevlevel ) {
3534                                         # Increase TOC level
3535                                         $toclevel++;
3536                                         $sublevelCount[$toclevel] = 0;
3537                                         if( $toclevel<$wgMaxTocLevel ) {
3538                                                 $prevtoclevel = $toclevel;
3539                                                 $toc .= $sk->tocIndent();
3540                                                 $numVisible++;
3541                                         }
3542                                 }
3543                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3544                                         # Decrease TOC level, find level to jump to
3545
3546                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3547                                                 # Can only go down to level 1
3548                                                 $toclevel = 1;
3549                                         } else {
3550                                                 for ($i = $toclevel; $i > 0; $i--) {
3551                                                         if ( $levelCount[$i] == $level ) {
3552                                                                 # Found last matching level
3553                                                                 $toclevel = $i;
3554                                                                 break;
3555                                                         }
3556                                                         elseif ( $levelCount[$i] < $level ) {
3557                                                                 # Found first matching level below current level
3558                                                                 $toclevel = $i + 1;
3559                                                                 break;
3560                                                         }
3561                                                 }
3562                                         }
3563                                         if( $toclevel<$wgMaxTocLevel ) {
3564                                                 if($prevtoclevel < $wgMaxTocLevel) {
3565                                                         # Unindent only if the previous toc level was shown :p
3566                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3567                                                         $prevtoclevel = $toclevel;
3568                                                 } else {
3569                                                         $toc .= $sk->tocLineEnd();
3570                                                 }
3571                                         }
3572                                 }
3573                                 else {
3574                                         # No change in level, end TOC line
3575                                         if( $toclevel<$wgMaxTocLevel ) {
3576                                                 $toc .= $sk->tocLineEnd();
3577                                         }
3578                                 }
3579
3580                                 $levelCount[$toclevel] = $level;
3581
3582                                 # count number of headlines for each level
3583                                 @$sublevelCount[$toclevel]++;
3584                                 $dot = 0;
3585                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3586                                         if( !empty( $sublevelCount[$i] ) ) {
3587                                                 if( $dot ) {
3588                                                         $numbering .= '.';
3589                                                 }
3590                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3591                                                 $dot = 1;
3592                                         }
3593                                 }
3594                         }
3595
3596                         # The safe header is a version of the header text safe to use for links
3597                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3598                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3599
3600                         # Remove link placeholders by the link text.
3601                         #     <!--LINK number-->
3602                         # turns into
3603                         #     link text with suffix
3604                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3605
3606                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3607                         $tocline = preg_replace(
3608                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3609                                 array( '',                          '<$1>'),
3610                                 $safeHeadline
3611                         );
3612                         $tocline = trim( $tocline );
3613
3614                         # For the anchor, strip out HTML-y stuff period
3615                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3616                         $safeHeadline = trim( $safeHeadline );
3617
3618                         # Save headline for section edit hint before it's escaped
3619                         $headlineHint = $safeHeadline;
3620
3621                         if ( $wgEnforceHtmlIds ) {
3622                                 $legacyHeadline = false;
3623                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3624                                         'noninitial' );
3625                         } else {
3626                                 # For reverse compatibility, provide an id that's
3627                                 # HTML4-compatible, like we used to.
3628                                 #
3629                                 # It may be worth noting, academically, that it's possible for
3630                                 # the legacy anchor to conflict with a non-legacy headline
3631                                 # anchor on the page.  In this case likely the "correct" thing
3632                                 # would be to either drop the legacy anchors or make sure
3633                                 # they're numbered first.  However, this would require people
3634                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3635                                 # manually, so let's not bother worrying about it.
3636                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3637                                         'noninitial' );
3638                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline, 'xml' );
3639
3640                                 if ( $legacyHeadline == $safeHeadline ) {
3641                                         # No reason to have both (in fact, we can't)
3642                                         $legacyHeadline = false;
3643                                 } elseif ( $legacyHeadline != Sanitizer::escapeId(
3644                                 $legacyHeadline, 'xml' ) ) {
3645                                         # The legacy id is invalid XML.  We used to allow this, but
3646                                         # there's no reason to do so anymore.  Backward
3647                                         # compatibility will fail slightly in this case, but it's
3648                                         # no big deal.
3649                                         $legacyHeadline = false;
3650                                 }
3651                         }
3652
3653                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3654                         # Does this apply to Unicode characters?  Because we aren't
3655                         # handling those here.
3656                         $arrayKey = strtolower( $safeHeadline );
3657                         if ( $legacyHeadline === false ) {
3658                                 $legacyArrayKey = false;
3659                         } else {
3660                                 $legacyArrayKey = strtolower( $legacyHeadline );
3661                         }
3662
3663                         # count how many in assoc. array so we can track dupes in anchors
3664                         if ( isset( $refers[$arrayKey] ) ) {
3665                                 $refers[$arrayKey]++;
3666                         } else {
3667                                 $refers[$arrayKey] = 1;
3668                         }
3669                         if ( isset( $refers[$legacyArrayKey] ) ) {
3670                                 $refers[$legacyArrayKey]++;
3671                         } else {
3672                                 $refers[$legacyArrayKey] = 1;
3673                         }
3674
3675                         # Don't number the heading if it is the only one (looks silly)
3676                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3677                                 # the two are different if the line contains a link
3678                                 $headline=$numbering . ' ' . $headline;
3679                         }
3680
3681                         # Create the anchor for linking from the TOC to the section
3682                         $anchor = $safeHeadline;
3683                         $legacyAnchor = $legacyHeadline;
3684                         if ( $refers[$arrayKey] > 1 ) {
3685                                 $anchor .= '_' . $refers[$arrayKey];
3686                         }
3687                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3688                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3689                         }
3690                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3691                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3692                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3693                         }
3694                         # give headline the correct <h#> tag
3695                         if( $showEditLink && $sectionIndex !== false ) {
3696                                 if( $isTemplate ) {
3697                                         # Put a T flag in the section identifier, to indicate to extractSections()
3698                                         # that sections inside <includeonly> should be counted.
3699                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3700                                 } else {
3701                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3702                                 }
3703                         } else {
3704                                 $editlink = '';
3705                         }
3706                         $head[$headlineCount] = $sk->makeHeadline( $level,
3707                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3708                                 $editlink, $legacyAnchor );
3709
3710                         $headlineCount++;
3711                 }
3712
3713                 $this->mOutput->setSections( $tocraw );
3714
3715                 # Never ever show TOC if no headers
3716                 if( $numVisible < 1 ) {
3717                         $enoughToc = false;
3718                 }
3719
3720                 if( $enoughToc ) {
3721                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3722                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3723                         }
3724                         $toc = $sk->tocList( $toc );
3725                 }
3726
3727                 # split up and insert constructed headlines
3728
3729                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3730                 $i = 0;
3731
3732                 foreach( $blocks as $block ) {
3733                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3734                                 # This is the [edit] link that appears for the top block of text when
3735                                 # section editing is enabled
3736
3737                                 # Disabled because it broke block formatting
3738                                 # For example, a bullet point in the top line
3739                                 # $full .= $sk->editSectionLink(0);
3740                         }
3741                         $full .= $block;
3742                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3743                                 # Top anchor now in skin
3744                                 $full = $full.$toc;
3745                         }
3746
3747                         if( !empty( $head[$i] ) ) {
3748                                 $full .= $head[$i];
3749                         }
3750                         $i++;
3751                 }
3752                 if( $this->mForceTocPosition ) {
3753                         return str_replace( '<!--MWTOC-->', $toc, $full );
3754                 } else {
3755                         return $full;
3756                 }
3757         }
3758
3759         /**
3760          * Transform wiki markup when saving a page by doing \r\n -> \n
3761          * conversion, substitting signatures, {{subst:}} templates, etc.
3762          *
3763          * @param string $text the text to transform
3764          * @param Title &$title the Title object for the current article
3765          * @param User &$user the User object describing the current user
3766          * @param ParserOptions $options parsing options
3767          * @param bool $clearState whether to clear the parser state first
3768          * @return string the altered wiki markup
3769          * @public
3770          */
3771         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3772                 $this->mOptions = $options;
3773                 $this->setTitle( $title );
3774                 $this->setOutputType( self::OT_WIKI );
3775
3776                 if ( $clearState ) {
3777                         $this->clearState();
3778                 }
3779
3780                 $pairs = array(
3781                         "\r\n" => "\n",
3782                 );
3783                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3784                 $text = $this->pstPass2( $text, $user );
3785                 $text = $this->mStripState->unstripBoth( $text );
3786                 return $text;
3787         }
3788
3789         /**
3790          * Pre-save transform helper function
3791          * @private
3792          */
3793         function pstPass2( $text, $user ) {
3794                 global $wgContLang, $wgLocaltimezone;
3795
3796                 /* Note: This is the timestamp saved as hardcoded wikitext to
3797                  * the database, we use $wgContLang here in order to give
3798                  * everyone the same signature and use the default one rather
3799                  * than the one selected in each user's preferences.
3800                  *
3801                  * (see also bug 12815)
3802                  */
3803                 $ts = $this->mOptions->getTimestamp();
3804                 $tz = wfMsgForContent( 'timezone-utc' );
3805                 if ( isset( $wgLocaltimezone ) ) {
3806                         $unixts = wfTimestamp( TS_UNIX, $ts );
3807                         $oldtz = getenv( 'TZ' );
3808                         putenv( 'TZ='.$wgLocaltimezone );
3809                         $ts = date( 'YmdHis', $unixts );
3810                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3811                         putenv( 'TZ='.$oldtz );
3812                 }
3813
3814                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3815
3816                 # Variable replacement
3817                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3818                 $text = $this->replaceVariables( $text );
3819
3820                 # Signatures
3821                 $sigText = $this->getUserSig( $user );
3822                 $text = strtr( $text, array(
3823                         '~~~~~' => $d,
3824                         '~~~~' => "$sigText $d",
3825                         '~~~' => $sigText
3826                 ) );
3827
3828                 # Context links: [[|name]] and [[name (context)|]]
3829                 #
3830                 global $wgLegalTitleChars;
3831                 $tc = "[$wgLegalTitleChars]";
3832                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3833
3834                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3835                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3836                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3837                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3838
3839                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3840                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3841                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3842                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3843
3844                 $t = $this->mTitle->getText();
3845                 $m = array();
3846                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3847                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3848                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3849                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3850                 } else {
3851                         # if there's no context, don't bother duplicating the title
3852                         $text = preg_replace( $p2, '[[\\1]]', $text );
3853                 }
3854
3855                 # Trim trailing whitespace
3856                 $text = rtrim( $text );
3857
3858                 return $text;
3859         }
3860
3861         /**
3862          * Fetch the user's signature text, if any, and normalize to
3863          * validated, ready-to-insert wikitext.
3864          *
3865          * @param User $user
3866          * @return string
3867          * @private
3868          */
3869         function getUserSig( &$user ) {
3870                 global $wgMaxSigChars;
3871
3872                 $username = $user->getName();
3873                 $nickname = $user->getOption( 'nickname' );
3874                 $nickname = $nickname === '' ? $username : $nickname;
3875
3876                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3877                         $nickname = $username;
3878                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3879                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3880                         # Sig. might contain markup; validate this
3881                         if( $this->validateSig( $nickname ) !== false ) {
3882                                 # Validated; clean up (if needed) and return it
3883                                 return $this->cleanSig( $nickname, true );
3884                         } else {
3885                                 # Failed to validate; fall back to the default
3886                                 $nickname = $username;
3887                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3888                         }
3889                 }
3890
3891                 // Make sure nickname doesnt get a sig in a sig
3892                 $nickname = $this->cleanSigInSig( $nickname );
3893
3894                 # If we're still here, make it a link to the user page
3895                 $userText = wfEscapeWikiText( $username );
3896                 $nickText = wfEscapeWikiText( $nickname );
3897                 if ( $user->isAnon() )  {
3898                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3899                 } else {
3900                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3901                 }
3902         }
3903
3904         /**
3905          * Check that the user's signature contains no bad XML
3906          *
3907          * @param string $text
3908          * @return mixed An expanded string, or false if invalid.
3909          */
3910         function validateSig( $text ) {
3911                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3912         }
3913
3914         /**
3915          * Clean up signature text
3916          *
3917          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3918          * 2) Substitute all transclusions
3919          *
3920          * @param string $text
3921          * @param $parsing Whether we're cleaning (preferences save) or parsing
3922          * @return string Signature text
3923          */
3924         function cleanSig( $text, $parsing = false ) {
3925                 if ( !$parsing ) {
3926                         global $wgTitle;
3927                         $this->clearState();
3928                         $this->setTitle( $wgTitle );
3929                         $this->mOptions = new ParserOptions;
3930                         $this->setOutputType = self::OT_PREPROCESS;
3931                 }
3932
3933                 # Option to disable this feature
3934                 if ( !$this->mOptions->getCleanSignatures() ) {
3935                         return $text;
3936                 }
3937
3938                 # FIXME: regex doesn't respect extension tags or nowiki
3939                 #  => Move this logic to braceSubstitution()
3940                 $substWord = MagicWord::get( 'subst' );
3941                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3942                 $substText = '{{' . $substWord->getSynonym( 0 );
3943
3944                 $text = preg_replace( $substRegex, $substText, $text );
3945                 $text = $this->cleanSigInSig( $text );
3946                 $dom = $this->preprocessToDom( $text );
3947                 $frame = $this->getPreprocessor()->newFrame();
3948                 $text = $frame->expand( $dom );
3949
3950                 if ( !$parsing ) {
3951                         $text = $this->mStripState->unstripBoth( $text );
3952                 }
3953
3954                 return $text;
3955         }
3956
3957         /**
3958          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3959          * @param string $text
3960          * @return string Signature text with /~{3,5}/ removed
3961          */
3962         function cleanSigInSig( $text ) {
3963                 $text = preg_replace( '/~{3,5}/', '', $text );
3964                 return $text;
3965         }
3966
3967         /**
3968          * Set up some variables which are usually set up in parse()
3969          * so that an external function can call some class members with confidence
3970          * @public
3971          */
3972         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3973                 $this->setTitle( $title );
3974                 $this->mOptions = $options;
3975                 $this->setOutputType( $outputType );
3976                 if ( $clearState ) {
3977                         $this->clearState();
3978                 }
3979         }
3980
3981         /**
3982          * Wrapper for preprocess()
3983          *
3984          * @param string $text the text to preprocess
3985          * @param ParserOptions $options  options
3986          * @return string
3987          * @public
3988          */
3989         function transformMsg( $text, $options ) {
3990                 global $wgTitle;
3991                 static $executing = false;
3992
3993                 # Guard against infinite recursion
3994                 if ( $executing ) {
3995                         return $text;
3996                 }
3997                 $executing = true;
3998
3999                 wfProfileIn(__METHOD__);
4000                 $text = $this->preprocess( $text, $wgTitle, $options );
4001
4002                 $executing = false;
4003                 wfProfileOut(__METHOD__);
4004                 return $text;
4005         }
4006
4007         /**
4008          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4009          * The callback should have the following form:
4010          *    function myParserHook( $text, $params, &$parser ) { ... }
4011          *
4012          * Transform and return $text. Use $parser for any required context, e.g. use
4013          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4014          *
4015          * @public
4016          *
4017          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4018          * @param mixed $callback The callback function (and object) to use for the tag
4019          *
4020          * @return The old value of the mTagHooks array associated with the hook
4021          */
4022         function setHook( $tag, $callback ) {
4023                 $tag = strtolower( $tag );
4024                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4025                 $this->mTagHooks[$tag] = $callback;
4026                 if( !in_array( $tag, $this->mStripList ) ) {
4027                         $this->mStripList[] = $tag;
4028                 }
4029
4030                 return $oldVal;
4031         }
4032
4033         function setTransparentTagHook( $tag, $callback ) {
4034                 $tag = strtolower( $tag );
4035                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4036                 $this->mTransparentTagHooks[$tag] = $callback;
4037
4038                 return $oldVal;
4039         }
4040
4041         /**
4042          * Remove all tag hooks
4043          */
4044         function clearTagHooks() {
4045                 $this->mTagHooks = array();
4046                 $this->mStripList = $this->mDefaultStripList;
4047         }
4048
4049         /**
4050          * Create a function, e.g. {{sum:1|2|3}}
4051          * The callback function should have the form:
4052          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4053          *
4054          * Or with SFH_OBJECT_ARGS:
4055          *    function myParserFunction( $parser, $frame, $args ) { ... }
4056          *
4057          * The callback may either return the text result of the function, or an array with the text
4058          * in element 0, and a number of flags in the other elements. The names of the flags are
4059          * specified in the keys. Valid flags are:
4060          *   found                     The text returned is valid, stop processing the template. This
4061          *                             is on by default.
4062          *   nowiki                    Wiki markup in the return value should be escaped
4063          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4064          *
4065          * @public
4066          *
4067          * @param string $id The magic word ID
4068          * @param mixed $callback The callback function (and object) to use
4069          * @param integer $flags a combination of the following flags:
4070          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4071          *
4072          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4073          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4074          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4075          *     the arguments, and to control the way they are expanded.
4076          *
4077          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4078          *     arguments, for instance:
4079          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4080          *
4081          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4082          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4083          *     working if/when this is changed.
4084          *
4085          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4086          *     expansion.
4087          *
4088          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4089          *     about the methods available in PPFrame and PPNode.
4090          *
4091          * @return The old callback function for this name, if any
4092          */
4093         function setFunctionHook( $id, $callback, $flags = 0 ) {
4094                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4095                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4096
4097                 # Add to function cache
4098                 $mw = MagicWord::get( $id );
4099                 if( !$mw )
4100                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4101
4102                 $synonyms = $mw->getSynonyms();
4103                 $sensitive = intval( $mw->isCaseSensitive() );
4104
4105                 foreach ( $synonyms as $syn ) {
4106                         # Case
4107                         if ( !$sensitive ) {
4108                                 $syn = strtolower( $syn );
4109                         }
4110                         # Add leading hash
4111                         if ( !( $flags & SFH_NO_HASH ) ) {
4112                                 $syn = '#' . $syn;
4113                         }
4114                         # Remove trailing colon
4115                         if ( substr( $syn, -1, 1 ) === ':' ) {
4116                                 $syn = substr( $syn, 0, -1 );
4117                         }
4118                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4119                 }
4120                 return $oldVal;
4121         }
4122
4123         /**
4124          * Get all registered function hook identifiers
4125          *
4126          * @return array
4127          */
4128         function getFunctionHooks() {
4129                 return array_keys( $this->mFunctionHooks );
4130         }
4131
4132         /**
4133          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4134          * Placeholders created in Skin::makeLinkObj()
4135          * Returns an array of link CSS classes, indexed by PDBK.
4136          */
4137         function replaceLinkHolders( &$text, $options = 0 ) {
4138                 return $this->mLinkHolders->replace( $text );
4139         }
4140
4141         /**
4142          * Replace <!--LINK--> link placeholders with plain text of links
4143          * (not HTML-formatted).
4144          * @param string $text
4145          * @return string
4146          */
4147         function replaceLinkHoldersText( $text ) {
4148                 return $this->mLinkHolders->replaceText( $text );
4149         }
4150
4151         /**
4152          * Tag hook handler for 'pre'.
4153          */
4154         function renderPreTag( $text, $attribs ) {
4155                 // Backwards-compatibility hack
4156                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4157
4158                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4159                 return Xml::openElement( 'pre', $attribs ) .
4160                         Xml::escapeTagsOnly( $content ) .
4161                         '</pre>';
4162         }
4163
4164         /**
4165          * Renders an image gallery from a text with one line per image.
4166          * text labels may be given by using |-style alternative text. E.g.
4167          *   Image:one.jpg|The number "1"
4168          *   Image:tree.jpg|A tree
4169          * given as text will return the HTML of a gallery with two images,
4170          * labeled 'The number "1"' and
4171          * 'A tree'.
4172          */
4173         function renderImageGallery( $text, $params ) {
4174                 $ig = new ImageGallery();
4175                 $ig->setContextTitle( $this->mTitle );
4176                 $ig->setShowBytes( false );
4177                 $ig->setShowFilename( false );
4178                 $ig->setParser( $this );
4179                 $ig->setHideBadImages();
4180                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4181                 $ig->useSkin( $this->mOptions->getSkin() );
4182                 $ig->mRevisionId = $this->mRevisionId;
4183
4184                 if( isset( $params['caption'] ) ) {
4185                         $caption = $params['caption'];
4186                         $caption = htmlspecialchars( $caption );
4187                         $caption = $this->replaceInternalLinks( $caption );
4188                         $ig->setCaptionHtml( $caption );
4189                 }
4190                 if( isset( $params['perrow'] ) ) {
4191                         $ig->setPerRow( $params['perrow'] );
4192                 }
4193                 if( isset( $params['widths'] ) ) {
4194                         $ig->setWidths( $params['widths'] );
4195                 }
4196                 if( isset( $params['heights'] ) ) {
4197                         $ig->setHeights( $params['heights'] );
4198                 }
4199
4200                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4201
4202                 $lines = StringUtils::explode( "\n", $text );
4203                 foreach ( $lines as $line ) {
4204                         # match lines like these:
4205                         # Image:someimage.jpg|This is some image
4206                         $matches = array();
4207                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4208                         # Skip empty lines
4209                         if ( count( $matches ) == 0 ) {
4210                                 continue;
4211                         }
4212
4213                         if ( strpos( $matches[0], '%' ) !== false )
4214                                 $matches[1] = urldecode( $matches[1] );
4215                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4216                         $nt =& $tp;
4217                         if( is_null( $nt ) ) {
4218                                 # Bogus title. Ignore these so we don't bomb out later.
4219                                 continue;
4220                         }
4221                         if ( isset( $matches[3] ) ) {
4222                                 $label = $matches[3];
4223                         } else {
4224                                 $label = '';
4225                         }
4226
4227                         $html = $this->recursiveTagParse( trim( $label ) );
4228
4229                         $ig->add( $nt, $html );
4230
4231                         # Only add real images (bug #5586)
4232                         if ( $nt->getNamespace() == NS_FILE ) {
4233                                 $this->mOutput->addImage( $nt->getDBkey() );
4234                         }
4235                 }
4236                 return $ig->toHTML();
4237         }
4238
4239         function getImageParams( $handler ) {
4240                 if ( $handler ) {
4241                         $handlerClass = get_class( $handler );
4242                 } else {
4243                         $handlerClass = '';
4244                 }
4245                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4246                         // Initialise static lists
4247                         static $internalParamNames = array(
4248                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4249                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4250                                         'bottom', 'text-bottom' ),
4251                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4252                                         'upright', 'border', 'link', 'alt' ),
4253                         );
4254                         static $internalParamMap;
4255                         if ( !$internalParamMap ) {
4256                                 $internalParamMap = array();
4257                                 foreach ( $internalParamNames as $type => $names ) {
4258                                         foreach ( $names as $name ) {
4259                                                 $magicName = str_replace( '-', '_', "img_$name" );
4260                                                 $internalParamMap[$magicName] = array( $type, $name );
4261                                         }
4262                                 }
4263                         }
4264
4265                         // Add handler params
4266                         $paramMap = $internalParamMap;
4267                         if ( $handler ) {
4268                                 $handlerParamMap = $handler->getParamMap();
4269                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4270                                         $paramMap[$magic] = array( 'handler', $paramName );
4271                                 }
4272                         }
4273                         $this->mImageParams[$handlerClass] = $paramMap;
4274                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4275                 }
4276                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4277         }
4278
4279         /**
4280          * Parse image options text and use it to make an image
4281          * @param Title $title
4282          * @param string $options
4283          * @param LinkHolderArray $holders
4284          */
4285         function makeImage( $title, $options, $holders = false ) {
4286                 # Check if the options text is of the form "options|alt text"
4287                 # Options are:
4288                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4289                 #  * left       no resizing, just left align. label is used for alt= only
4290                 #  * right      same, but right aligned
4291                 #  * none       same, but not aligned
4292                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4293                 #  * center     center the image
4294                 #  * framed     Keep original image size, no magnify-button.
4295                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4296                 #  * upright    reduce width for upright images, rounded to full __0 px
4297                 #  * border     draw a 1px border around the image
4298                 #  * alt        Text for HTML alt attribute (defaults to empty)
4299                 # vertical-align values (no % or length right now):
4300                 #  * baseline
4301                 #  * sub
4302                 #  * super
4303                 #  * top
4304                 #  * text-top
4305                 #  * middle
4306                 #  * bottom
4307                 #  * text-bottom
4308
4309                 $parts = StringUtils::explode( "|", $options );
4310                 $sk = $this->mOptions->getSkin();
4311
4312                 # Give extensions a chance to select the file revision for us
4313                 $skip = $time = $descQuery = false;
4314                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4315
4316                 if ( $skip ) {
4317                         return $sk->link( $title );
4318                 }
4319
4320                 # Get the file
4321                 $imagename = $title->getDBkey();
4322                 if ( isset( $this->mFileCache[$imagename][$time] ) ) {
4323                         $file = $this->mFileCache[$imagename][$time];
4324                 } else {
4325                         $file = wfFindFile( $title, $time );
4326                         if ( count( $this->mFileCache ) > 1000 ) {
4327                                 $this->mFileCache = array();
4328                         }
4329                         $this->mFileCache[$imagename][$time] = $file;
4330                 }
4331                 # Get parameter map
4332                 $handler = $file ? $file->getHandler() : false;
4333
4334                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4335
4336                 # Process the input parameters
4337                 $caption = '';
4338                 $params = array( 'frame' => array(), 'handler' => array(),
4339                         'horizAlign' => array(), 'vertAlign' => array() );
4340                 foreach( $parts as $part ) {
4341                         $part = trim( $part );
4342                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4343                         $validated = false;
4344                         if( isset( $paramMap[$magicName] ) ) {
4345                                 list( $type, $paramName ) = $paramMap[$magicName];
4346
4347                                 // Special case; width and height come in one variable together
4348                                 if( $type === 'handler' && $paramName === 'width' ) {
4349                                         $m = array();
4350                                         # (bug 13500) In both cases (width/height and width only),
4351                                         # permit trailing "px" for backward compatibility.
4352                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4353                                                 $width = intval( $m[1] );
4354                                                 $height = intval( $m[2] );
4355                                                 if ( $handler->validateParam( 'width', $width ) ) {
4356                                                         $params[$type]['width'] = $width;
4357                                                         $validated = true;
4358                                                 }
4359                                                 if ( $handler->validateParam( 'height', $height ) ) {
4360                                                         $params[$type]['height'] = $height;
4361                                                         $validated = true;
4362                                                 }
4363                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4364                                                 $width = intval( $value );
4365                                                 if ( $handler->validateParam( 'width', $width ) ) {
4366                                                         $params[$type]['width'] = $width;
4367                                                         $validated = true;
4368                                                 }
4369                                         } // else no validation -- bug 13436
4370                                 } else {
4371                                         if ( $type === 'handler' ) {
4372                                                 # Validate handler parameter
4373                                                 $validated = $handler->validateParam( $paramName, $value );
4374                                         } else {
4375                                                 # Validate internal parameters
4376                                                 switch( $paramName ) {
4377                                                 case 'manualthumb':
4378                                                 case 'alt':
4379                                                         // @fixme - possibly check validity here for
4380                                                         // manualthumb? downstream behavior seems odd with
4381                                                         // missing manual thumbs.
4382                                                         $validated = true;
4383                                                         $value = $this->stripAltText( $value, $holders );
4384                                                         break;
4385                                                 case 'link':
4386                                                         $chars = self::EXT_LINK_URL_CLASS;
4387                                                         $prots = $this->mUrlProtocols;
4388                                                         if ( $value === '' ) {
4389                                                                 $paramName = 'no-link';
4390                                                                 $value = true;
4391                                                                 $validated = true;
4392                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4393                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4394                                                                         $paramName = 'link-url';
4395                                                                         $this->mOutput->addExternalLink( $value );
4396                                                                         $validated = true;
4397                                                                 }
4398                                                         } else {
4399                                                                 $linkTitle = Title::newFromText( $value );
4400                                                                 if ( $linkTitle ) {
4401                                                                         $paramName = 'link-title';
4402                                                                         $value = $linkTitle;
4403                                                                         $this->mOutput->addLink( $linkTitle );
4404                                                                         $validated = true;
4405                                                                 }
4406                                                         }
4407                                                         break;
4408                                                 default:
4409                                                         // Most other things appear to be empty or numeric...
4410                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4411                                                 }
4412                                         }
4413
4414                                         if ( $validated ) {
4415                                                 $params[$type][$paramName] = $value;
4416                                         }
4417                                 }
4418                         }
4419                         if ( !$validated ) {
4420                                 $caption = $part;
4421                         }
4422                 }
4423
4424                 # Process alignment parameters
4425                 if ( $params['horizAlign'] ) {
4426                         $params['frame']['align'] = key( $params['horizAlign'] );
4427                 }
4428                 if ( $params['vertAlign'] ) {
4429                         $params['frame']['valign'] = key( $params['vertAlign'] );
4430                 }
4431
4432                 $params['frame']['caption'] = $caption;
4433
4434                 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4435
4436                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4437                 # came to also set the caption, ordinary text after the image -- which
4438                 # makes no sense, because that just repeats the text multiple times in
4439                 # screen readers.  It *also* came to set the title attribute.
4440                 #
4441                 # Now that we have an alt attribute, we should not set the alt text to
4442                 # equal the caption: that's worse than useless, it just repeats the
4443                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4444                 # use the unnamed parameter for alt text as well, just for the time be-
4445                 # ing, if the unnamed param is set and the alt param is not.
4446                 #
4447                 # For the future, we need to figure out if we want to tweak this more,
4448                 # e.g., introducing a title= parameter for the title; ignoring the un-
4449                 # named parameter entirely for images without a caption; adding an ex-
4450                 # plicit caption= parameter and preserving the old magic unnamed para-
4451                 # meter for BC; ...
4452                 if( $caption !== '' && !isset( $params['frame']['alt'] )
4453                 && !isset( $params['frame']['framed'] )
4454                 && !isset( $params['frame']['thumbnail'] )
4455                 && !isset( $params['frame']['manualthumb'] ) ) {
4456                         $params['frame']['alt'] = $params['frame']['title'];
4457                 }
4458
4459                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4460
4461                 # Linker does the rest
4462                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4463
4464                 # Give the handler a chance to modify the parser object
4465                 if ( $handler ) {
4466                         $handler->parserTransformHook( $this, $file );
4467                 }
4468
4469                 return $ret;
4470         }
4471
4472         protected function stripAltText( $caption, $holders ) {
4473                 # Strip bad stuff out of the title (tooltip).  We can't just use
4474                 # replaceLinkHoldersText() here, because if this function is called
4475                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4476                 if ( $holders ) {
4477                         $tooltip = $holders->replaceText( $caption );
4478                 } else {
4479                         $tooltip = $this->replaceLinkHoldersText( $caption );
4480                 }
4481
4482                 # make sure there are no placeholders in thumbnail attributes
4483                 # that are later expanded to html- so expand them now and
4484                 # remove the tags
4485                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4486                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4487
4488                 return $tooltip;
4489         }
4490
4491         /**
4492          * Set a flag in the output object indicating that the content is dynamic and
4493          * shouldn't be cached.
4494          */
4495         function disableCache() {
4496                 wfDebug( "Parser output marked as uncacheable.\n" );
4497                 $this->mOutput->mCacheTime = -1;
4498         }
4499
4500         /**#@+
4501          * Callback from the Sanitizer for expanding items found in HTML attribute
4502          * values, so they can be safely tested and escaped.
4503          * @param string $text
4504          * @param PPFrame $frame
4505          * @return string
4506          * @private
4507          */
4508         function attributeStripCallback( &$text, $frame = false ) {
4509                 $text = $this->replaceVariables( $text, $frame );
4510                 $text = $this->mStripState->unstripBoth( $text );
4511                 return $text;
4512         }
4513
4514         /**#@-*/
4515
4516         /**#@+
4517          * Accessor/mutator
4518          */
4519         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4520         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4521         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4522         /**#@-*/
4523
4524         /**#@+
4525          * Accessor
4526          */
4527         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4528         /**#@-*/
4529
4530
4531         /**
4532          * Break wikitext input into sections, and either pull or replace
4533          * some particular section's text.
4534          *
4535          * External callers should use the getSection and replaceSection methods.
4536          *
4537          * @param string $text Page wikitext
4538          * @param string $section A section identifier string of the form:
4539          *   <flag1> - <flag2> - ... - <section number>
4540          *
4541          * Currently the only recognised flag is "T", which means the target section number
4542          * was derived during a template inclusion parse, in other words this is a template
4543          * section edit link. If no flags are given, it was an ordinary section edit link.
4544          * This flag is required to avoid a section numbering mismatch when a section is
4545          * enclosed by <includeonly> (bug 6563).
4546          *
4547          * The section number 0 pulls the text before the first heading; other numbers will
4548          * pull the given section along with its lower-level subsections. If the section is
4549          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4550          *
4551          * @param string $mode One of "get" or "replace"
4552          * @param string $newText Replacement text for section data.
4553          * @return string for "get", the extracted section text.
4554          *                for "replace", the whole page with the section replaced.
4555          */
4556         private function extractSections( $text, $section, $mode, $newText='' ) {
4557                 global $wgTitle;
4558                 $this->clearState();
4559                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4560                 $this->mOptions = new ParserOptions;
4561                 $this->setOutputType( self::OT_WIKI );
4562                 $outText = '';
4563                 $frame = $this->getPreprocessor()->newFrame();
4564
4565                 // Process section extraction flags
4566                 $flags = 0;
4567                 $sectionParts = explode( '-', $section );
4568                 $sectionIndex = array_pop( $sectionParts );
4569                 foreach ( $sectionParts as $part ) {
4570                         if ( $part === 'T' ) {
4571                                 $flags |= self::PTD_FOR_INCLUSION;
4572                         }
4573                 }
4574                 // Preprocess the text
4575                 $root = $this->preprocessToDom( $text, $flags );
4576
4577                 // <h> nodes indicate section breaks
4578                 // They can only occur at the top level, so we can find them by iterating the root's children
4579                 $node = $root->getFirstChild();
4580
4581                 // Find the target section
4582                 if ( $sectionIndex == 0 ) {
4583                         // Section zero doesn't nest, level=big
4584                         $targetLevel = 1000;
4585                 } else {
4586             while ( $node ) {
4587                 if ( $node->getName() === 'h' ) {
4588                     $bits = $node->splitHeading();
4589                                         if ( $bits['i'] == $sectionIndex ) {
4590                                         $targetLevel = $bits['level'];
4591                                                 break;
4592                                         }
4593                                 }
4594                                 if ( $mode === 'replace' ) {
4595                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4596                                 }
4597                                 $node = $node->getNextSibling();
4598                         }
4599                 }
4600
4601                 if ( !$node ) {
4602                         // Not found
4603                         if ( $mode === 'get' ) {
4604                                 return $newText;
4605                         } else {
4606                                 return $text;
4607                         }
4608                 }
4609
4610                 // Find the end of the section, including nested sections
4611                 do {
4612                         if ( $node->getName() === 'h' ) {
4613                                 $bits = $node->splitHeading();
4614                                 $curLevel = $bits['level'];
4615                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4616                                         break;
4617                                 }
4618                         }
4619                         if ( $mode === 'get' ) {
4620                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4621                         }
4622                         $node = $node->getNextSibling();
4623                 } while ( $node );
4624
4625                 // Write out the remainder (in replace mode only)
4626                 if ( $mode === 'replace' ) {
4627                         // Output the replacement text
4628                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4629                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4630                         $outText .= $newText . "\n\n";
4631                         while ( $node ) {
4632                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4633                                 $node = $node->getNextSibling();
4634                         }
4635                 }
4636
4637                 if ( is_string( $outText ) ) {
4638                         // Re-insert stripped tags
4639                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4640                 }
4641
4642                 return $outText;
4643         }
4644
4645         /**
4646          * This function returns the text of a section, specified by a number ($section).
4647          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4648          * the first section before any such heading (section 0).
4649          *
4650          * If a section contains subsections, these are also returned.
4651          *
4652          * @param string $text text to look in
4653          * @param string $section section identifier
4654          * @param string $deftext default to return if section is not found
4655          * @return string text of the requested section
4656          */
4657         public function getSection( $text, $section, $deftext='' ) {
4658                 return $this->extractSections( $text, $section, "get", $deftext );
4659         }
4660
4661         public function replaceSection( $oldtext, $section, $text ) {
4662                 return $this->extractSections( $oldtext, $section, "replace", $text );
4663         }
4664
4665         /**
4666          * Get the timestamp associated with the current revision, adjusted for
4667          * the default server-local timestamp
4668          */
4669         function getRevisionTimestamp() {
4670                 if ( is_null( $this->mRevisionTimestamp ) ) {
4671                         wfProfileIn( __METHOD__ );
4672                         global $wgContLang;
4673                         $dbr = wfGetDB( DB_SLAVE );
4674                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4675                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4676
4677                         // Normalize timestamp to internal MW format for timezone processing.
4678                         // This has the added side-effect of replacing a null value with
4679                         // the current time, which gives us more sensible behavior for
4680                         // previews.
4681                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4682
4683                         // The cryptic '' timezone parameter tells to use the site-default
4684                         // timezone offset instead of the user settings.
4685                         //
4686                         // Since this value will be saved into the parser cache, served
4687                         // to other users, and potentially even used inside links and such,
4688                         // it needs to be consistent for all visitors.
4689                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4690
4691                         wfProfileOut( __METHOD__ );
4692                 }
4693                 return $this->mRevisionTimestamp;
4694         }
4695
4696         /**
4697          * Mutator for $mDefaultSort
4698          *
4699          * @param $sort New value
4700          */
4701         public function setDefaultSort( $sort ) {
4702                 $this->mDefaultSort = $sort;
4703         }
4704
4705         /**
4706          * Accessor for $mDefaultSort
4707          * Will use the title/prefixed title if none is set
4708          *
4709          * @return string
4710          */
4711         public function getDefaultSort() {
4712                 global $wgCategoryPrefixedDefaultSortkey;
4713                 if( $this->mDefaultSort !== false ) {
4714                         return $this->mDefaultSort;
4715                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4716                         !$wgCategoryPrefixedDefaultSortkey) {
4717                         return $this->mTitle->getText();
4718                 } else {
4719                         return $this->mTitle->getPrefixedText();
4720                 }
4721         }
4722
4723         /**
4724          * Accessor for $mDefaultSort
4725          * Unlike getDefaultSort(), will return false if none is set
4726          *
4727          * @return string or false
4728          */
4729         public function getCustomDefaultSort() {
4730                 return $this->mDefaultSort;
4731         }
4732
4733         /**
4734          * Try to guess the section anchor name based on a wikitext fragment
4735          * presumably extracted from a heading, for example "Header" from
4736          * "== Header ==".
4737          */
4738         public function guessSectionNameFromWikiText( $text ) {
4739                 # Strip out wikitext links(they break the anchor)
4740                 $text = $this->stripSectionName( $text );
4741                 $headline = Sanitizer::decodeCharReferences( $text );
4742                 # strip out HTML
4743                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4744                 $headline = trim( $headline );
4745                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4746                 $replacearray = array(
4747                         '%3A' => ':',
4748                         '%' => '.'
4749                 );
4750                 return str_replace(
4751                         array_keys( $replacearray ),
4752                         array_values( $replacearray ),
4753                         $sectionanchor );
4754         }
4755
4756         /**
4757          * Strips a text string of wikitext for use in a section anchor
4758          *
4759          * Accepts a text string and then removes all wikitext from the
4760          * string and leaves only the resultant text (i.e. the result of
4761          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4762          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4763          * to create valid section anchors by mimicing the output of the
4764          * parser when headings are parsed.
4765          *
4766          * @param $text string Text string to be stripped of wikitext
4767          * for use in a Section anchor
4768          * @return Filtered text string
4769          */
4770         public function stripSectionName( $text ) {
4771                 # Strip internal link markup
4772                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4773                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4774
4775                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4776                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4777                 # on how many empty links there are on the page - need to figure that out.
4778                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4779
4780                 # Parse wikitext quotes (italics & bold)
4781                 $text = $this->doQuotes($text);
4782
4783                 # Strip HTML tags
4784                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4785                 return $text;
4786         }
4787
4788         function srvus( $text ) {
4789                 return $this->testSrvus( $text, $this->mOutputType );
4790         }
4791
4792         /**
4793          * strip/replaceVariables/unstrip for preprocessor regression testing
4794          */
4795         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4796                 $this->clearState();
4797                 if ( ! ( $title instanceof Title ) ) {
4798                         $title = Title::newFromText( $title );
4799                 }
4800                 $this->mTitle = $title;
4801                 $this->mOptions = $options;
4802                 $this->setOutputType( $outputType );
4803                 $text = $this->replaceVariables( $text );
4804                 $text = $this->mStripState->unstripBoth( $text );
4805                 $text = Sanitizer::removeHTMLtags( $text );
4806                 return $text;
4807         }
4808
4809         function testPst( $text, $title, $options ) {
4810                 global $wgUser;
4811                 if ( ! ( $title instanceof Title ) ) {
4812                         $title = Title::newFromText( $title );
4813                 }
4814                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4815         }
4816
4817         function testPreprocess( $text, $title, $options ) {
4818                 if ( ! ( $title instanceof Title ) ) {
4819                         $title = Title::newFromText( $title );
4820                 }
4821                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4822         }
4823
4824         function markerSkipCallback( $s, $callback ) {
4825                 $i = 0;
4826                 $out = '';
4827                 while ( $i < strlen( $s ) ) {
4828                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4829                         if ( $markerStart === false ) {
4830                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4831                                 break;
4832                         } else {
4833                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4834                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4835                                 if ( $markerEnd === false ) {
4836                                         $out .= substr( $s, $markerStart );
4837                                         break;
4838                                 } else {
4839                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4840                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4841                                         $i = $markerEnd;
4842                                 }
4843                         }
4844                 }
4845                 return $out;
4846         }
4847 }
4848
4849 /**
4850  * @todo document, briefly.
4851  * @ingroup Parser
4852  */
4853 class StripState {
4854         var $general, $nowiki;
4855
4856         function __construct() {
4857                 $this->general = new ReplacementArray;
4858                 $this->nowiki = new ReplacementArray;
4859         }
4860
4861         function unstripGeneral( $text ) {
4862                 wfProfileIn( __METHOD__ );
4863                 do {
4864                         $oldText = $text;
4865                         $text = $this->general->replace( $text );
4866                 } while ( $text !== $oldText );
4867                 wfProfileOut( __METHOD__ );
4868                 return $text;
4869         }
4870
4871         function unstripNoWiki( $text ) {
4872                 wfProfileIn( __METHOD__ );
4873                 do {
4874                         $oldText = $text;
4875                         $text = $this->nowiki->replace( $text );
4876                 } while ( $text !== $oldText );
4877                 wfProfileOut( __METHOD__ );
4878                 return $text;
4879         }
4880
4881         function unstripBoth( $text ) {
4882                 wfProfileIn( __METHOD__ );
4883                 do {
4884                         $oldText = $text;
4885                         $text = $this->general->replace( $text );
4886                         $text = $this->nowiki->replace( $text );
4887                 } while ( $text !== $oldText );
4888                 wfProfileOut( __METHOD__ );
4889                 return $text;
4890         }
4891 }
4892
4893 /**
4894  * @todo document, briefly.
4895  * @ingroup Parser
4896  */
4897 class OnlyIncludeReplacer {
4898         var $output = '';
4899
4900         function replace( $matches ) {
4901                 if ( substr( $matches[1], -1 ) === "\n" ) {
4902                         $this->output .= substr( $matches[1], 0, -1 );
4903                 } else {
4904                         $this->output .= $matches[1];
4905                 }
4906         }
4907 }