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, $mDefaultStripList, $mVarCache, $mConf;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mInterwikiLinkHolders, $mLinkHolders;
 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
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions,      // ParserOptions object
 110                 $mTitle,        // Title context, used for self-link rendering and similar things
 111                 $mOutputType,   // Output type, one of the OT_xxx constants
 112                 $ot,            // Shortcut alias, see setOutputType()
 113                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 114                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 115                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 116
 117         /**#@-*/
 118
 119         /**
 120          * Constructor
 121          *
 122          * @public
 123          */
 124         function __construct( $conf = array() ) {
 125                 $this->mConf = $conf;
 126                 $this->mTagHooks = array();
 127                 $this->mTransparentTagHooks = array();
 128                 $this->mFunctionHooks = array();
 129                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 130                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 131                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 132                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 133                 $this->mVarCache = array();
 134                 if ( isset( $conf['preprocessorClass'] ) ) {
 135                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 136                 } else {
 137                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 138                 }
 139                 $this->mMarkerIndex = 0;
 140                 $this->mFirstCall = true;
 141         }
 142
 143         /**
 144          * Do various kinds of initialisation on the first call of the parser
 145          */
 146         function firstCallInit() {
 147                 if ( !$this->mFirstCall ) {
 148                         return;
 149                 }
 150                 $this->mFirstCall = false;
 151
 152                 wfProfileIn( __METHOD__ );
 153
 154                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 155                 CoreParserFunctions::register( $this );
 156                 $this->initialiseVariables();
 157
 158                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 159                 wfProfileOut( __METHOD__ );
 160         }
 161
 162         /**
 163          * Clear Parser state
 164          *
 165          * @private
 166          */
 167         function clearState() {
 168                 wfProfileIn( __METHOD__ );
 169                 if ( $this->mFirstCall ) {
 170                         $this->firstCallInit();
 171                 }
 172                 $this->mOutput = new ParserOutput;
 173                 $this->mAutonumber = 0;
 174                 $this->mLastSection = '';
 175                 $this->mDTopen = false;
 176                 $this->mIncludeCount = array();
 177                 $this->mStripState = new StripState;
 178                 $this->mArgStack = false;
 179                 $this->mInPre = false;
 180                 $this->mInterwikiLinkHolders = array(
 181                         'texts' => array(),
 182                         'titles' => array()
 183                 );
 184                 $this->mLinkHolders = array(
 185                         'namespaces' => array(),
 186                         'dbkeys' => array(),
 187                         'queries' => array(),
 188                         'texts' => array(),
 189                         'titles' => array()
 190                 );
 191                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 192
 193                 /**
 194                  * Prefix for temporary replacement strings for the multipass parser.
 195                  * \x07 should never appear in input as it's disallowed in XML.
 196                  * Using it at the front also gives us a little extra robustness
 197                  * since it shouldn't match when butted up against identifier-like
 198                  * string constructs.
 199                  *
 200                  * Must not consist of all title characters, or else it will change
 201                  * the behaviour of <nowiki> in a link.
 202                  */
 203                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 204                 # Changed to \x7f to allow XML double-parsing -- TS
 205                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 206
 207
 208                 # Clear these on every parse, bug 4549
 209                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 210
 211                 $this->mShowToc = true;
 212                 $this->mForceTocPosition = false;
 213                 $this->mIncludeSizes = array(
 214                         'post-expand' => 0,
 215                         'arg' => 0,
 216                 );
 217                 $this->mPPNodeCount = 0;
 218                 $this->mDefaultSort = false;
 219                 $this->mHeadings = array();
 220                 $this->mDoubleUnderscores = array();
 221                 $this->mExpensiveFunctionCount = 0;
 222
 223                 # Fix cloning
 224                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 225                         $this->mPreprocessor = null;
 226                 }
 227
 228                 wfRunHooks( 'ParserClearState', array( &$this ) );
 229                 wfProfileOut( __METHOD__ );
 230         }
 231
 232         function setOutputType( $ot ) {
 233                 $this->mOutputType = $ot;
 234                 // Shortcut alias
 235                 $this->ot = array(
 236                         'html' => $ot == self::OT_HTML,
 237                         'wiki' => $ot == self::OT_WIKI,
 238                         'pre' => $ot == self::OT_PREPROCESS,
 239                 );
 240         }
 241
 242         /**
 243          * Set the context title
 244          */
 245         function setTitle( $t ) {
 246                 if ( !$t || $t instanceof FakeTitle ) {
 247                         $t = Title::newFromText( 'NO TITLE' );
 248                 }
 249                 if ( strval( $t->getFragment() ) !== '' ) {
 250                         # Strip the fragment to avoid various odd effects
 251                         $this->mTitle = clone $t;
 252                         $this->mTitle->setFragment( '' );
 253                 } else {
 254                         $this->mTitle = $t;
 255                 }
 256         }
 257
 258         /**
 259          * Accessor for mUniqPrefix.
 260          *
 261          * @public
 262          */
 263         function uniqPrefix() {
 264                 if( !isset( $this->mUniqPrefix ) ) {
 265                         // @fixme this is probably *horribly wrong*
 266                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 267                         // if this is called for a parser cache hit, the parser may not
 268                         // have ever been initialized in the first place.
 269                         // Not really sure what the heck is supposed to be going on here.
 270                         return '';
 271                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 272                 }
 273                 return $this->mUniqPrefix;
 274         }
 275
 276         /**
 277          * Convert wikitext to HTML
 278          * Do not call this function recursively.
 279          *
 280          * @param string $text Text we want to parse
 281          * @param Title &$title A title object
 282          * @param array $options
 283          * @param boolean $linestart
 284          * @param boolean $clearState
 285          * @param int $revid number to pass in {{REVISIONID}}
 286          * @return ParserOutput a ParserOutput
 287          */
 288         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 289                 /**
 290                  * First pass--just handle <nowiki> sections, pass the rest off
 291                  * to internalParse() which does all the real work.
 292                  */
 293
 294                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 295                 $fname = 'Parser::parse-' . wfGetCaller();
 296                 wfProfileIn( __METHOD__ );
 297                 wfProfileIn( $fname );
 298
 299                 if ( $clearState ) {
 300                         $this->clearState();
 301                 }
 302
 303                 $this->mOptions = $options;
 304                 $this->setTitle( $title );
 305                 $oldRevisionId = $this->mRevisionId;
 306                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 307                 if( $revid !== null ) {
 308                         $this->mRevisionId = $revid;
 309                         $this->mRevisionTimestamp = null;
 310                 }
 311                 $this->setOutputType( self::OT_HTML );
 312                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 313                 # No more strip!
 314                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 315                 $text = $this->internalParse( $text );
 316                 $text = $this->mStripState->unstripGeneral( $text );
 317
 318                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 319                 $fixtags = array(
 320                         # french spaces, last one Guillemet-left
 321                         # only if there is something before the space
 322                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 323                         # french spaces, Guillemet-right
 324                         '/(\\302\\253) /' => '\\1&nbsp;',
 325                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 326                 );
 327                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 328
 329                 # only once and last
 330                 $text = $this->doBlockLevels( $text, $linestart );
 331
 332                 $this->replaceLinkHolders( $text );
 333
 334                 # the position of the parserConvert() call should not be changed. it
 335                 # assumes that the links are all replaced and the only thing left
 336                 # is the <nowiki> mark.
 337                 # Side-effects: this calls $this->mOutput->setTitleText()
 338                 $text = $wgContLang->parserConvert( $text, $this );
 339
 340                 $text = $this->mStripState->unstripNoWiki( $text );
 341
 342                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 343
 344 //!JF Move to its own function
 345
 346                 $uniq_prefix = $this->mUniqPrefix;
 347                 $matches = array();
 348                 $elements = array_keys( $this->mTransparentTagHooks );
 349                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 350
 351                 foreach( $matches as $marker => $data ) {
 352                         list( $element, $content, $params, $tag ) = $data;
 353                         $tagName = strtolower( $element );
 354                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 355                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 356                                         array( $content, $params, $this ) );
 357                         } else {
 358                                 $output = $tag;
 359                         }
 360                         $this->mStripState->general->setPair( $marker, $output );
 361                 }
 362                 $text = $this->mStripState->unstripGeneral( $text );
 363
 364                 $text = Sanitizer::normalizeCharReferences( $text );
 365
 366                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 367                         $text = Parser::tidy($text);
 368                 } else {
 369                         # attempt to sanitize at least some nesting problems
 370                         # (bug #2702 and quite a few others)
 371                         $tidyregs = array(
 372                                 # ''Something [http://www.cool.com cool''] -->
 373                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 374                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 375                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 376                                 # fix up an anchor inside another anchor, only
 377                                 # at least for a single single nested link (bug 3695)
 378                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 379                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 380                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 381                                 # contains a div, so fix it up here; replace
 382                                 # div with escaped text
 383                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 384                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 385                                 # remove empty italic or bold tag pairs, some
 386                                 # introduced by rules above
 387                                 '/<([bi])><\/\\1>/' => '',
 388                         );
 389
 390                         $text = preg_replace(
 391                                 array_keys( $tidyregs ),
 392                                 array_values( $tidyregs ),
 393                                 $text );
 394                 }
 395                 global $wgExpensiveParserFunctionLimit;
 396                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 397                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 398                 }
 399
 400                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 401
 402                 # Information on include size limits, for the benefit of users who try to skirt them
 403                 if ( $this->mOptions->getEnableLimitReport() ) {
 404                         global $wgExpensiveParserFunctionLimit;
 405                         $max = $this->mOptions->getMaxIncludeSize();
 406                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 407                         $limitReport =
 408                                 "NewPP limit report\n" .
 409                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 410                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 411                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 412                                 $PFreport;
 413                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 414                         $text .= "\n<!-- \n$limitReport-->\n";
 415                 }
 416                 $this->mOutput->setText( $text );
 417                 $this->mRevisionId = $oldRevisionId;
 418                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 419                 wfProfileOut( $fname );
 420                 wfProfileOut( __METHOD__ );
 421
 422                 return $this->mOutput;
 423         }
 424
 425         /**
 426          * Recursive parser entry point that can be called from an extension tag
 427          * hook.
 428          */
 429         function recursiveTagParse( $text ) {
 430                 wfProfileIn( __METHOD__ );
 431                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 432                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 433                 $text = $this->internalParse( $text );
 434                 wfProfileOut( __METHOD__ );
 435                 return $text;
 436         }
 437
 438         /**
 439          * Expand templates and variables in the text, producing valid, static wikitext.
 440          * Also removes comments.
 441          */
 442         function preprocess( $text, $title, $options, $revid = null ) {
 443                 wfProfileIn( __METHOD__ );
 444                 $this->clearState();
 445                 $this->setOutputType( self::OT_PREPROCESS );
 446                 $this->mOptions = $options;
 447                 $this->setTitle( $title );
 448                 if( $revid !== null ) {
 449                         $this->mRevisionId = $revid;
 450                 }
 451                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 452                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 453                 $text = $this->replaceVariables( $text );
 454                 $text = $this->mStripState->unstripBoth( $text );
 455                 wfProfileOut( __METHOD__ );
 456                 return $text;
 457         }
 458
 459         /**
 460          * Get a random string
 461          *
 462          * @private
 463          * @static
 464          */
 465         function getRandomString() {
 466                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 467         }
 468
 469         function &getTitle() { return $this->mTitle; }
 470         function getOptions() { return $this->mOptions; }
 471         function getRevisionId() { return $this->mRevisionId; }
 472
 473         function getFunctionLang() {
 474                 global $wgLang, $wgContLang;
 475
 476                 $target = $this->mOptions->getTargetLanguage();
 477                 if ( $target !== null ) {
 478                         return $target;
 479                 } else {
 480                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 481                 }
 482         }
 483
 484         /**
 485          * Get a preprocessor object
 486          */
 487         function getPreprocessor() {
 488                 if ( !isset( $this->mPreprocessor ) ) {
 489                         $class = $this->mPreprocessorClass;
 490                         $this->mPreprocessor = new $class( $this );
 491                 }
 492                 return $this->mPreprocessor;
 493         }
 494
 495         /**
 496          * Replaces all occurrences of HTML-style comments and the given tags
 497          * in the text with a random marker and returns the next text. The output
 498          * parameter $matches will be an associative array filled with data in
 499          * the form:
 500          *   'UNIQ-xxxxx' => array(
 501          *     'element',
 502          *     'tag content',
 503          *     array( 'param' => 'x' ),
 504          *     '<element param="x">tag content</element>' ) )
 505          *
 506          * @param $elements list of element names. Comments are always extracted.
 507          * @param $text Source text string.
 508          * @param $uniq_prefix
 509          *
 510          * @public
 511          * @static
 512          */
 513         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 514                 static $n = 1;
 515                 $stripped = '';
 516                 $matches = array();
 517
 518                 $taglist = implode( '|', $elements );
 519                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 520
 521                 while ( '' != $text ) {
 522                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 523                         $stripped .= $p[0];
 524                         if( count( $p ) < 5 ) {
 525                                 break;
 526                         }
 527                         if( count( $p ) > 5 ) {
 528                                 // comment
 529                                 $element    = $p[4];
 530                                 $attributes = '';
 531                                 $close      = '';
 532                                 $inside     = $p[5];
 533                         } else {
 534                                 // tag
 535                                 $element    = $p[1];
 536                                 $attributes = $p[2];
 537                                 $close      = $p[3];
 538                                 $inside     = $p[4];
 539                         }
 540
 541                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 542                         $stripped .= $marker;
 543
 544                         if ( $close === '/>' ) {
 545                                 // Empty element tag, <tag />
 546                                 $content = null;
 547                                 $text = $inside;
 548                                 $tail = null;
 549                         } else {
 550                                 if( $element == '!--' ) {
 551                                         $end = '/(-->)/';
 552                                 } else {
 553                                         $end = "/(<\\/$element\\s*>)/i";
 554                                 }
 555                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 556                                 $content = $q[0];
 557                                 if( count( $q ) < 3 ) {
 558                                         # No end tag -- let it run out to the end of the text.
 559                                         $tail = '';
 560                                         $text = '';
 561                                 } else {
 562                                         $tail = $q[1];
 563                                         $text = $q[2];
 564                                 }
 565                         }
 566
 567                         $matches[$marker] = array( $element,
 568                                 $content,
 569                                 Sanitizer::decodeTagAttributes( $attributes ),
 570                                 "<$element$attributes$close$content$tail" );
 571                 }
 572                 return $stripped;
 573         }
 574
 575         /**
 576          * Get a list of strippable XML-like elements
 577          */
 578         function getStripList() {
 579                 global $wgRawHtml;
 580                 $elements = $this->mStripList;
 581                 if( $wgRawHtml ) {
 582                         $elements[] = 'html';
 583                 }
 584                 if( $this->mOptions->getUseTeX() ) {
 585                         $elements[] = 'math';
 586                 }
 587                 return $elements;
 588         }
 589
 590         /**
 591          * @deprecated use replaceVariables
 592          */
 593         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 594                 return $text;
 595         }
 596
 597         /**
 598          * Restores pre, math, and other extensions removed by strip()
 599          *
 600          * always call unstripNoWiki() after this one
 601          * @private
 602          * @deprecated use $this->mStripState->unstrip()
 603          */
 604         function unstrip( $text, $state ) {
 605                 return $state->unstripGeneral( $text );
 606         }
 607
 608         /**
 609          * Always call this after unstrip() to preserve the order
 610          *
 611          * @private
 612          * @deprecated use $this->mStripState->unstrip()
 613          */
 614         function unstripNoWiki( $text, $state ) {
 615                 return $state->unstripNoWiki( $text );
 616         }
 617
 618         /**
 619          * @deprecated use $this->mStripState->unstripBoth()
 620          */
 621         function unstripForHTML( $text ) {
 622                 return $this->mStripState->unstripBoth( $text );
 623         }
 624
 625         /**
 626          * Add an item to the strip state
 627          * Returns the unique tag which must be inserted into the stripped text
 628          * The tag will be replaced with the original text in unstrip()
 629          *
 630          * @private
 631          */
 632         function insertStripItem( $text ) {
 633                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 634                 $this->mMarkerIndex++;
 635                 $this->mStripState->general->setPair( $rnd, $text );
 636                 return $rnd;
 637         }
 638
 639         /**
 640          * Interface with html tidy, used if $wgUseTidy = true.
 641          * If tidy isn't able to correct the markup, the original will be
 642          * returned in all its glory with a warning comment appended.
 643          *
 644          * Either the external tidy program or the in-process tidy extension
 645          * will be used depending on availability. Override the default
 646          * $wgTidyInternal setting to disable the internal if it's not working.
 647          *
 648          * @param string $text Hideous HTML input
 649          * @return string Corrected HTML output
 650          * @public
 651          * @static
 652          */
 653         function tidy( $text ) {
 654                 global $wgTidyInternal;
 655                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 656 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 657 '<head><title>test</title></head><body>'.$text.'</body></html>';
 658                 if( $wgTidyInternal ) {
 659                         $correctedtext = Parser::internalTidy( $wrappedtext );
 660                 } else {
 661                         $correctedtext = Parser::externalTidy( $wrappedtext );
 662                 }
 663                 if( is_null( $correctedtext ) ) {
 664                         wfDebug( "Tidy error detected!\n" );
 665                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 666                 }
 667                 return $correctedtext;
 668         }
 669
 670         /**
 671          * Spawn an external HTML tidy process and get corrected markup back from it.
 672          *
 673          * @private
 674          * @static
 675          */
 676         function externalTidy( $text ) {
 677                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 678                 $fname = 'Parser::externalTidy';
 679                 wfProfileIn( $fname );
 680
 681                 $cleansource = '';
 682                 $opts = ' -utf8';
 683
 684                 $descriptorspec = array(
 685                         0 => array('pipe', 'r'),
 686                         1 => array('pipe', 'w'),
 687                         2 => array('file', wfGetNull(), 'a')
 688                 );
 689                 $pipes = array();
 690                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 691                 if (is_resource($process)) {
 692                         // Theoretically, this style of communication could cause a deadlock
 693                         // here. If the stdout buffer fills up, then writes to stdin could
 694                         // block. This doesn't appear to happen with tidy, because tidy only
 695                         // writes to stdout after it's finished reading from stdin. Search
 696                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 697                         fwrite($pipes[0], $text);
 698                         fclose($pipes[0]);
 699                         while (!feof($pipes[1])) {
 700                                 $cleansource .= fgets($pipes[1], 1024);
 701                         }
 702                         fclose($pipes[1]);
 703                         proc_close($process);
 704                 }
 705
 706                 wfProfileOut( $fname );
 707
 708                 if( $cleansource == '' && $text != '') {
 709                         // Some kind of error happened, so we couldn't get the corrected text.
 710                         // Just give up; we'll use the source text and append a warning.
 711                         return null;
 712                 } else {
 713                         return $cleansource;
 714                 }
 715         }
 716
 717         /**
 718          * Use the HTML tidy PECL extension to use the tidy library in-process,
 719          * saving the overhead of spawning a new process.
 720          *
 721          * 'pear install tidy' should be able to compile the extension module.
 722          *
 723          * @private
 724          * @static
 725          */
 726         function internalTidy( $text ) {
 727                 global $wgTidyConf, $IP, $wgDebugTidy;
 728                 $fname = 'Parser::internalTidy';
 729                 wfProfileIn( $fname );
 730
 731                 $tidy = new tidy;
 732                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 733                 $tidy->cleanRepair();
 734                 if( $tidy->getStatus() == 2 ) {
 735                         // 2 is magic number for fatal error
 736                         // http://www.php.net/manual/en/function.tidy-get-status.php
 737                         $cleansource = null;
 738                 } else {
 739                         $cleansource = tidy_get_output( $tidy );
 740                 }
 741                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 742                         $cleansource .= "<!--\nTidy reports:\n" .
 743                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 744                                 "\n-->";
 745                 }
 746
 747                 wfProfileOut( $fname );
 748                 return $cleansource;
 749         }
 750
 751         /**
 752          * parse the wiki syntax used to render tables
 753          *
 754          * @private
 755          */
 756         function doTableStuff ( $text ) {
 757                 $fname = 'Parser::doTableStuff';
 758                 wfProfileIn( $fname );
 759
 760                 $lines = explode ( "\n" , $text );
 761                 $td_history = array (); // Is currently a td tag open?
 762                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 763                 $tr_history = array (); // Is currently a tr tag open?
 764                 $tr_attributes = array (); // history of tr attributes
 765                 $has_opened_tr = array(); // Did this table open a <tr> element?
 766                 $indent_level = 0; // indent level of the table
 767                 foreach ( $lines as $key => $line )
 768                 {
 769                         $line = trim ( $line );
 770
 771                         if( $line == '' ) { // empty line, go to next line
 772                                 continue;
 773                         }
 774                         $first_character = $line{0};
 775                         $matches = array();
 776
 777                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 778                                 // First check if we are starting a new table
 779                                 $indent_level = strlen( $matches[1] );
 780
 781                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 782                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 783
 784                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 785                                 array_push ( $td_history , false );
 786                                 array_push ( $last_tag_history , '' );
 787                                 array_push ( $tr_history , false );
 788                                 array_push ( $tr_attributes , '' );
 789                                 array_push ( $has_opened_tr , false );
 790                         } else if ( count ( $td_history ) == 0 ) {
 791                                 // Don't do any of the following
 792                                 continue;
 793                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 794                                 // We are ending a table
 795                                 $line = '</table>' . substr ( $line , 2 );
 796                                 $last_tag = array_pop ( $last_tag_history );
 797
 798                                 if ( !array_pop ( $has_opened_tr ) ) {
 799                                         $line = "<tr><td></td></tr>{$line}";
 800                                 }
 801
 802                                 if ( array_pop ( $tr_history ) ) {
 803                                         $line = "</tr>{$line}";
 804                                 }
 805
 806                                 if ( array_pop ( $td_history ) ) {
 807                                         $line = "</{$last_tag}>{$line}";
 808                                 }
 809                                 array_pop ( $tr_attributes );
 810                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 811                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 812                                 // Now we have a table row
 813                                 $line = preg_replace( '#^\|-+#', '', $line );
 814
 815                                 // Whats after the tag is now only attributes
 816                                 $attributes = $this->mStripState->unstripBoth( $line );
 817                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 818                                 array_pop ( $tr_attributes );
 819                                 array_push ( $tr_attributes , $attributes );
 820
 821                                 $line = '';
 822                                 $last_tag = array_pop ( $last_tag_history );
 823                                 array_pop ( $has_opened_tr );
 824                                 array_push ( $has_opened_tr , true );
 825
 826                                 if ( array_pop ( $tr_history ) ) {
 827                                         $line = '</tr>';
 828                                 }
 829
 830                                 if ( array_pop ( $td_history ) ) {
 831                                         $line = "</{$last_tag}>{$line}";
 832                                 }
 833
 834                                 $lines[$key] = $line;
 835                                 array_push ( $tr_history , false );
 836                                 array_push ( $td_history , false );
 837                                 array_push ( $last_tag_history , '' );
 838                         }
 839                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 840                                 // This might be cell elements, td, th or captions
 841                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 842                                         $first_character = '+';
 843                                         $line = substr ( $line , 1 );
 844                                 }
 845
 846                                 $line = substr ( $line , 1 );
 847
 848                                 if ( $first_character == '!' ) {
 849                                         $line = str_replace ( '!!' , '||' , $line );
 850                                 }
 851
 852                                 // Split up multiple cells on the same line.
 853                                 // FIXME : This can result in improper nesting of tags processed
 854                                 // by earlier parser steps, but should avoid splitting up eg
 855                                 // attribute values containing literal "||".
 856                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 857
 858                                 $lines[$key] = '';
 859
 860                                 // Loop through each table cell
 861                                 foreach ( $cells as $cell )
 862                                 {
 863                                         $previous = '';
 864                                         if ( $first_character != '+' )
 865                                         {
 866                                                 $tr_after = array_pop ( $tr_attributes );
 867                                                 if ( !array_pop ( $tr_history ) ) {
 868                                                         $previous = "<tr{$tr_after}>\n";
 869                                                 }
 870                                                 array_push ( $tr_history , true );
 871                                                 array_push ( $tr_attributes , '' );
 872                                                 array_pop ( $has_opened_tr );
 873                                                 array_push ( $has_opened_tr , true );
 874                                         }
 875
 876                                         $last_tag = array_pop ( $last_tag_history );
 877
 878                                         if ( array_pop ( $td_history ) ) {
 879                                                 $previous = "</{$last_tag}>{$previous}";
 880                                         }
 881
 882                                         if ( $first_character == '|' ) {
 883                                                 $last_tag = 'td';
 884                                         } else if ( $first_character == '!' ) {
 885                                                 $last_tag = 'th';
 886                                         } else if ( $first_character == '+' ) {
 887                                                 $last_tag = 'caption';
 888                                         } else {
 889                                                 $last_tag = '';
 890                                         }
 891
 892                                         array_push ( $last_tag_history , $last_tag );
 893
 894                                         // A cell could contain both parameters and data
 895                                         $cell_data = explode ( '|' , $cell , 2 );
 896
 897                                         // Bug 553: Note that a '|' inside an invalid link should not
 898                                         // be mistaken as delimiting cell parameters
 899                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 900                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 901                                         } else if ( count ( $cell_data ) == 1 )
 902                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 903                                         else {
 904                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 905                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 906                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 907                                         }
 908
 909                                         $lines[$key] .= $cell;
 910                                         array_push ( $td_history , true );
 911                                 }
 912                         }
 913                 }
 914
 915                 // Closing open td, tr && table
 916                 while ( count ( $td_history ) > 0 )
 917                 {
 918                         if ( array_pop ( $td_history ) ) {
 919                                 $lines[] = '</td>' ;
 920                         }
 921                         if ( array_pop ( $tr_history ) ) {
 922                                 $lines[] = '</tr>' ;
 923                         }
 924                         if ( !array_pop ( $has_opened_tr ) ) {
 925                                 $lines[] = "<tr><td></td></tr>" ;
 926                         }
 927
 928                         $lines[] = '</table>' ;
 929                 }
 930
 931                 $output = implode ( "\n" , $lines ) ;
 932
 933                 // special case: don't return empty table
 934                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 935                         $output = '';
 936                 }
 937
 938                 wfProfileOut( $fname );
 939
 940                 return $output;
 941         }
 942
 943         /**
 944          * Helper function for parse() that transforms wiki markup into
 945          * HTML. Only called for $mOutputType == self::OT_HTML.
 946          *
 947          * @private
 948          */
 949         function internalParse( $text ) {
 950                 $isMain = true;
 951                 $fname = 'Parser::internalParse';
 952                 wfProfileIn( $fname );
 953
 954                 # Hook to suspend the parser in this state
 955                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 956                         wfProfileOut( $fname );
 957                         return $text ;
 958                 }
 959
 960                 $text = $this->replaceVariables( $text );
 961                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 962                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 963
 964                 // Tables need to come after variable replacement for things to work
 965                 // properly; putting them before other transformations should keep
 966                 // exciting things like link expansions from showing up in surprising
 967                 // places.
 968                 $text = $this->doTableStuff( $text );
 969
 970                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 971
 972                 $text = $this->doDoubleUnderscore( $text );
 973                 $text = $this->doHeadings( $text );
 974                 if($this->mOptions->getUseDynamicDates()) {
 975                         $df = DateFormatter::getInstance();
 976                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 977                 }
 978                 $text = $this->doAllQuotes( $text );
 979                 $text = $this->replaceInternalLinks( $text );
 980                 $text = $this->replaceExternalLinks( $text );
 981
 982                 # replaceInternalLinks may sometimes leave behind
 983                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 984                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 985
 986                 $text = $this->doMagicLinks( $text );
 987                 $text = $this->formatHeadings( $text, $isMain );
 988
 989                 wfProfileOut( $fname );
 990                 return $text;
 991         }
 992
 993         /**
 994          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 995          * magic external links.
 996          *
 997          * @private
 998          */
 999         function doMagicLinks( $text ) {
1000                 wfProfileIn( __METHOD__ );
1001                 $text = preg_replace_callback(
1002                         '!(?:                           # Start cases
1003                             <a.*?</a> |                 # Skip link text
1004                             <.*?> |                     # Skip stuff inside HTML elements
1005                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1006                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1007                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1008                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1009                                      [0-9Xx]                 # check digit
1010                                    \b)
1011                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1012                 wfProfileOut( __METHOD__ );
1013                 return $text;
1014         }
1015
1016         function magicLinkCallback( $m ) {
1017                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1018                         # Skip HTML element
1019                         return $m[0];
1020                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1021                         $isbn = $m[2];
1022                         $num = strtr( $isbn, array(
1023                                 '-' => '',
1024                                 ' ' => '',
1025                                 'x' => 'X',
1026                         ));
1027                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1028                         $text = '<a href="' .
1029                                 $titleObj->escapeLocalUrl() .
1030                                 "\" class=\"internal\">ISBN $isbn</a>";
1031                 } else {
1032                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1033                                 $keyword = 'RFC';
1034                                 $urlmsg = 'rfcurl';
1035                                 $id = $m[1];
1036                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1037                                 $keyword = 'PMID';
1038                                 $urlmsg = 'pubmedurl';
1039                                 $id = $m[1];
1040                         } else {
1041                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1042                                         substr($m[0], 0, 20 ) . '"' );
1043                         }
1044
1045                         $url = wfMsg( $urlmsg, $id);
1046                         $sk = $this->mOptions->getSkin();
1047                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1048                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1049                 }
1050                 return $text;
1051         }
1052
1053         /**
1054          * Parse headers and return html
1055          *
1056          * @private
1057          */
1058         function doHeadings( $text ) {
1059                 $fname = 'Parser::doHeadings';
1060                 wfProfileIn( $fname );
1061                 for ( $i = 6; $i >= 1; --$i ) {
1062                         $h = str_repeat( '=', $i );
1063                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1064                           "<h$i>\\1</h$i>", $text );
1065                 }
1066                 wfProfileOut( $fname );
1067                 return $text;
1068         }
1069
1070         /**
1071          * Replace single quotes with HTML markup
1072          * @private
1073          * @return string the altered text
1074          */
1075         function doAllQuotes( $text ) {
1076                 $fname = 'Parser::doAllQuotes';
1077                 wfProfileIn( $fname );
1078                 $outtext = '';
1079                 $lines = explode( "\n", $text );
1080                 foreach ( $lines as $line ) {
1081                         $outtext .= $this->doQuotes ( $line ) . "\n";
1082                 }
1083                 $outtext = substr($outtext, 0,-1);
1084                 wfProfileOut( $fname );
1085                 return $outtext;
1086         }
1087
1088         /**
1089          * Helper function for doAllQuotes()
1090          */
1091         public function doQuotes( $text ) {
1092                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1093                 if ( count( $arr ) == 1 )
1094                         return $text;
1095                 else
1096                 {
1097                         # First, do some preliminary work. This may shift some apostrophes from
1098                         # being mark-up to being text. It also counts the number of occurrences
1099                         # of bold and italics mark-ups.
1100                         $i = 0;
1101                         $numbold = 0;
1102                         $numitalics = 0;
1103                         foreach ( $arr as $r )
1104                         {
1105                                 if ( ( $i % 2 ) == 1 )
1106                                 {
1107                                         # If there are ever four apostrophes, assume the first is supposed to
1108                                         # be text, and the remaining three constitute mark-up for bold text.
1109                                         if ( strlen( $arr[$i] ) == 4 )
1110                                         {
1111                                                 $arr[$i-1] .= "'";
1112                                                 $arr[$i] = "'''";
1113                                         }
1114                                         # If there are more than 5 apostrophes in a row, assume they're all
1115                                         # text except for the last 5.
1116                                         else if ( strlen( $arr[$i] ) > 5 )
1117                                         {
1118                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1119                                                 $arr[$i] = "'''''";
1120                                         }
1121                                         # Count the number of occurrences of bold and italics mark-ups.
1122                                         # We are not counting sequences of five apostrophes.
1123                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1124                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1125                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1126                                 }
1127                                 $i++;
1128                         }
1129
1130                         # If there is an odd number of both bold and italics, it is likely
1131                         # that one of the bold ones was meant to be an apostrophe followed
1132                         # by italics. Which one we cannot know for certain, but it is more
1133                         # likely to be one that has a single-letter word before it.
1134                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1135                         {
1136                                 $i = 0;
1137                                 $firstsingleletterword = -1;
1138                                 $firstmultiletterword = -1;
1139                                 $firstspace = -1;
1140                                 foreach ( $arr as $r )
1141                                 {
1142                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1143                                         {
1144                                                 $x1 = substr ($arr[$i-1], -1);
1145                                                 $x2 = substr ($arr[$i-1], -2, 1);
1146                                                 if ($x1 == ' ') {
1147                                                         if ($firstspace == -1) $firstspace = $i;
1148                                                 } else if ($x2 == ' ') {
1149                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1150                                                 } else {
1151                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1152                                                 }
1153                                         }
1154                                         $i++;
1155                                 }
1156
1157                                 # If there is a single-letter word, use it!
1158                                 if ($firstsingleletterword > -1)
1159                                 {
1160                                         $arr [ $firstsingleletterword ] = "''";
1161                                         $arr [ $firstsingleletterword-1 ] .= "'";
1162                                 }
1163                                 # If not, but there's a multi-letter word, use that one.
1164                                 else if ($firstmultiletterword > -1)
1165                                 {
1166                                         $arr [ $firstmultiletterword ] = "''";
1167                                         $arr [ $firstmultiletterword-1 ] .= "'";
1168                                 }
1169                                 # ... otherwise use the first one that has neither.
1170                                 # (notice that it is possible for all three to be -1 if, for example,
1171                                 # there is only one pentuple-apostrophe in the line)
1172                                 else if ($firstspace > -1)
1173                                 {
1174                                         $arr [ $firstspace ] = "''";
1175                                         $arr [ $firstspace-1 ] .= "'";
1176                                 }
1177                         }
1178
1179                         # Now let's actually convert our apostrophic mush to HTML!
1180                         $output = '';
1181                         $buffer = '';
1182                         $state = '';
1183                         $i = 0;
1184                         foreach ($arr as $r)
1185                         {
1186                                 if (($i % 2) == 0)
1187                                 {
1188                                         if ($state == 'both')
1189                                                 $buffer .= $r;
1190                                         else
1191                                                 $output .= $r;
1192                                 }
1193                                 else
1194                                 {
1195                                         if (strlen ($r) == 2)
1196                                         {
1197                                                 if ($state == 'i')
1198                                                 { $output .= '</i>'; $state = ''; }
1199                                                 else if ($state == 'bi')
1200                                                 { $output .= '</i>'; $state = 'b'; }
1201                                                 else if ($state == 'ib')
1202                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1203                                                 else if ($state == 'both')
1204                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1205                                                 else # $state can be 'b' or ''
1206                                                 { $output .= '<i>'; $state .= 'i'; }
1207                                         }
1208                                         else if (strlen ($r) == 3)
1209                                         {
1210                                                 if ($state == 'b')
1211                                                 { $output .= '</b>'; $state = ''; }
1212                                                 else if ($state == 'bi')
1213                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1214                                                 else if ($state == 'ib')
1215                                                 { $output .= '</b>'; $state = 'i'; }
1216                                                 else if ($state == 'both')
1217                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1218                                                 else # $state can be 'i' or ''
1219                                                 { $output .= '<b>'; $state .= 'b'; }
1220                                         }
1221                                         else if (strlen ($r) == 5)
1222                                         {
1223                                                 if ($state == 'b')
1224                                                 { $output .= '</b><i>'; $state = 'i'; }
1225                                                 else if ($state == 'i')
1226                                                 { $output .= '</i><b>'; $state = 'b'; }
1227                                                 else if ($state == 'bi')
1228                                                 { $output .= '</i></b>'; $state = ''; }
1229                                                 else if ($state == 'ib')
1230                                                 { $output .= '</b></i>'; $state = ''; }
1231                                                 else if ($state == 'both')
1232                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1233                                                 else # ($state == '')
1234                                                 { $buffer = ''; $state = 'both'; }
1235                                         }
1236                                 }
1237                                 $i++;
1238                         }
1239                         # Now close all remaining tags.  Notice that the order is important.
1240                         if ($state == 'b' || $state == 'ib')
1241                                 $output .= '</b>';
1242                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1243                                 $output .= '</i>';
1244                         if ($state == 'bi')
1245                                 $output .= '</b>';
1246                         # There might be lonely ''''', so make sure we have a buffer
1247                         if ($state == 'both' && $buffer)
1248                                 $output .= '<b><i>'.$buffer.'</i></b>';
1249                         return $output;
1250                 }
1251         }
1252
1253         /**
1254          * Replace external links
1255          *
1256          * Note: this is all very hackish and the order of execution matters a lot.
1257          * Make sure to run maintenance/parserTests.php if you change this code.
1258          *
1259          * @private
1260          */
1261         function replaceExternalLinks( $text ) {
1262                 global $wgContLang;
1263                 $fname = 'Parser::replaceExternalLinks';
1264                 wfProfileIn( $fname );
1265
1266                 $sk = $this->mOptions->getSkin();
1267
1268                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1269
1270                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1271
1272                 $i = 0;
1273                 while ( $i<count( $bits ) ) {
1274                         $url = $bits[$i++];
1275                         $protocol = $bits[$i++];
1276                         $text = $bits[$i++];
1277                         $trail = $bits[$i++];
1278
1279                         # The characters '<' and '>' (which were escaped by
1280                         # removeHTMLtags()) should not be included in
1281                         # URLs, per RFC 2396.
1282                         $m2 = array();
1283                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1284                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1285                                 $url = substr($url, 0, $m2[0][1]);
1286                         }
1287
1288                         # If the link text is an image URL, replace it with an <img> tag
1289                         # This happened by accident in the original parser, but some people used it extensively
1290                         $img = $this->maybeMakeExternalImage( $text );
1291                         if ( $img !== false ) {
1292                                 $text = $img;
1293                         }
1294
1295                         $dtrail = '';
1296
1297                         # Set linktype for CSS - if URL==text, link is essentially free
1298                         $linktype = ($text == $url) ? 'free' : 'text';
1299
1300                         # No link text, e.g. [http://domain.tld/some.link]
1301                         if ( $text == '' ) {
1302                                 # Autonumber if allowed. See bug #5918
1303                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1304                                         $text = '[' . ++$this->mAutonumber . ']';
1305                                         $linktype = 'autonumber';
1306                                 } else {
1307                                         # Otherwise just use the URL
1308                                         $text = htmlspecialchars( $url );
1309                                         $linktype = 'free';
1310                                 }
1311                         } else {
1312                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1313                                 # Check for trail
1314                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1315                         }
1316
1317                         $text = $wgContLang->markNoConversion($text);
1318
1319                         $url = Sanitizer::cleanUrl( $url );
1320
1321                         # Process the trail (i.e. everything after this link up until start of the next link),
1322                         # replacing any non-bracketed links
1323                         $trail = $this->replaceFreeExternalLinks( $trail );
1324
1325                         # Use the encoded URL
1326                         # This means that users can paste URLs directly into the text
1327                         # Funny characters like &ouml; aren't valid in URLs anyway
1328                         # This was changed in August 2004
1329                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1330
1331                         # Register link in the output object.
1332                         # Replace unnecessary URL escape codes with the referenced character
1333                         # This prevents spammers from hiding links from the filters
1334                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1335                         $this->mOutput->addExternalLink( $pasteurized );
1336                 }
1337
1338                 wfProfileOut( $fname );
1339                 return $s;
1340         }
1341
1342         /**
1343          * Replace anything that looks like a URL with a link
1344          * @private
1345          */
1346         function replaceFreeExternalLinks( $text ) {
1347                 global $wgContLang;
1348                 $fname = 'Parser::replaceFreeExternalLinks';
1349                 wfProfileIn( $fname );
1350
1351                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1352                 $s = array_shift( $bits );
1353                 $i = 0;
1354
1355                 $sk = $this->mOptions->getSkin();
1356
1357                 while ( $i < count( $bits ) ){
1358                         $protocol = $bits[$i++];
1359                         $remainder = $bits[$i++];
1360
1361                         $m = array();
1362                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1363                                 # Found some characters after the protocol that look promising
1364                                 $url = $protocol . $m[1];
1365                                 $trail = $m[2];
1366
1367                                 # special case: handle urls as url args:
1368                                 # http://www.example.com/foo?=http://www.example.com/bar
1369                                 if(strlen($trail) == 0 &&
1370                                         isset($bits[$i]) &&
1371                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1372                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1373                                 {
1374                                         # add protocol, arg
1375                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1376                                         $i += 2;
1377                                         $trail = $m[2];
1378                                 }
1379
1380                                 # The characters '<' and '>' (which were escaped by
1381                                 # removeHTMLtags()) should not be included in
1382                                 # URLs, per RFC 2396.
1383                                 $m2 = array();
1384                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1385                                         $trail = substr($url, $m2[0][1]) . $trail;
1386                                         $url = substr($url, 0, $m2[0][1]);
1387                                 }
1388
1389                                 # Move trailing punctuation to $trail
1390                                 $sep = ',;\.:!?';
1391                                 # If there is no left bracket, then consider right brackets fair game too
1392                                 if ( strpos( $url, '(' ) === false ) {
1393                                         $sep .= ')';
1394                                 }
1395
1396                                 $numSepChars = strspn( strrev( $url ), $sep );
1397                                 if ( $numSepChars ) {
1398                                         $trail = substr( $url, -$numSepChars ) . $trail;
1399                                         $url = substr( $url, 0, -$numSepChars );
1400                                 }
1401
1402                                 $url = Sanitizer::cleanUrl( $url );
1403
1404                                 # Is this an external image?
1405                                 $text = $this->maybeMakeExternalImage( $url );
1406                                 if ( $text === false ) {
1407                                         # Not an image, make a link
1408                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1409                                         # Register it in the output object...
1410                                         # Replace unnecessary URL escape codes with their equivalent characters
1411                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1412                                         $this->mOutput->addExternalLink( $pasteurized );
1413                                 }
1414                                 $s .= $text . $trail;
1415                         } else {
1416                                 $s .= $protocol . $remainder;
1417                         }
1418                 }
1419                 wfProfileOut( $fname );
1420                 return $s;
1421         }
1422
1423         /**
1424          * Replace unusual URL escape codes with their equivalent characters
1425          * @param string
1426          * @return string
1427          * @static
1428          * @todo  This can merge genuinely required bits in the path or query string,
1429          *        breaking legit URLs. A proper fix would treat the various parts of
1430          *        the URL differently; as a workaround, just use the output for
1431          *        statistical records, not for actual linking/output.
1432          */
1433         static function replaceUnusualEscapes( $url ) {
1434                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1435                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1436         }
1437
1438         /**
1439          * Callback function used in replaceUnusualEscapes().
1440          * Replaces unusual URL escape codes with their equivalent character
1441          * @static
1442          * @private
1443          */
1444         private static function replaceUnusualEscapesCallback( $matches ) {
1445                 $char = urldecode( $matches[0] );
1446                 $ord = ord( $char );
1447                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1448                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1449                         // No, shouldn't be escaped
1450                         return $char;
1451                 } else {
1452                         // Yes, leave it escaped
1453                         return $matches[0];
1454                 }
1455         }
1456
1457         /**
1458          * make an image if it's allowed, either through the global
1459          * option or through the exception
1460          * @private
1461          */
1462         function maybeMakeExternalImage( $url ) {
1463                 $sk = $this->mOptions->getSkin();
1464                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1465                 $imagesexception = !empty($imagesfrom);
1466                 $text = false;
1467                 if ( $this->mOptions->getAllowExternalImages()
1468                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1469                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1470                                 # Image found
1471                                 $text = $sk->makeExternalImage( $url );
1472                         }
1473                 }
1474                 return $text;
1475         }
1476
1477         /**
1478          * Process [[ ]] wikilinks
1479          *
1480          * @private
1481          */
1482         function replaceInternalLinks( $s ) {
1483                 global $wgContLang;
1484                 static $fname = 'Parser::replaceInternalLinks' ;
1485
1486                 wfProfileIn( $fname );
1487
1488                 wfProfileIn( $fname.'-setup' );
1489                 static $tc = FALSE;
1490                 # the % is needed to support urlencoded titles as well
1491                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1492
1493                 $sk = $this->mOptions->getSkin();
1494
1495                 #split the entire text string on occurences of [[
1496                 $a = explode( '[[', ' ' . $s );
1497                 #get the first element (all text up to first [[), and remove the space we added
1498                 $s = array_shift( $a );
1499                 $s = substr( $s, 1 );
1500
1501                 # Match a link having the form [[namespace:link|alternate]]trail
1502                 static $e1 = FALSE;
1503                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1504                 # Match cases where there is no "]]", which might still be images
1505                 static $e1_img = FALSE;
1506                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1507
1508                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1509                 $e2 = null;
1510                 if ( $useLinkPrefixExtension ) {
1511                         # Match the end of a line for a word that's not followed by whitespace,
1512                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1513                         $e2 = wfMsgForContent( 'linkprefix' );
1514                 }
1515
1516                 if( is_null( $this->mTitle ) ) {
1517                         wfProfileOut( $fname );
1518                         wfProfileOut( $fname.'-setup' );
1519                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1520                 }
1521                 $nottalk = !$this->mTitle->isTalkPage();
1522
1523                 if ( $useLinkPrefixExtension ) {
1524                         $m = array();
1525                         if ( preg_match( $e2, $s, $m ) ) {
1526                                 $first_prefix = $m[2];
1527                         } else {
1528                                 $first_prefix = false;
1529                         }
1530                 } else {
1531                         $prefix = '';
1532                 }
1533
1534                 if($wgContLang->hasVariants()) {
1535                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1536                 } else {
1537                         $selflink = array($this->mTitle->getPrefixedText());
1538                 }
1539                 $useSubpages = $this->areSubpagesAllowed();
1540                 wfProfileOut( $fname.'-setup' );
1541
1542                 # Loop for each link
1543                 for ($k = 0; isset( $a[$k] ); $k++) {
1544                         $line = $a[$k];
1545                         if ( $useLinkPrefixExtension ) {
1546                                 wfProfileIn( $fname.'-prefixhandling' );
1547                                 if ( preg_match( $e2, $s, $m ) ) {
1548                                         $prefix = $m[2];
1549                                         $s = $m[1];
1550                                 } else {
1551                                         $prefix='';
1552                                 }
1553                                 # first link
1554                                 if($first_prefix) {
1555                                         $prefix = $first_prefix;
1556                                         $first_prefix = false;
1557                                 }
1558                                 wfProfileOut( $fname.'-prefixhandling' );
1559                         }
1560
1561                         $might_be_img = false;
1562
1563                         wfProfileIn( "$fname-e1" );
1564                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1565                                 $text = $m[2];
1566                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1567                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1568                                 # the real problem is with the $e1 regex
1569                                 # See bug 1300.
1570                                 #
1571                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1572                                 # and no image is in sight. See bug 2095.
1573                                 #
1574                                 if( $text !== '' &&
1575                                         substr( $m[3], 0, 1 ) === ']' &&
1576                                         strpos($text, '[') !== false
1577                                 )
1578                                 {
1579                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1580                                         $m[3] = substr( $m[3], 1 );
1581                                 }
1582                                 # fix up urlencoded title texts
1583                                 if( strpos( $m[1], '%' ) !== false ) {
1584                                         # Should anchors '#' also be rejected?
1585                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1586                                 }
1587                                 $trail = $m[3];
1588                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1589                                 $might_be_img = true;
1590                                 $text = $m[2];
1591                                 if ( strpos( $m[1], '%' ) !== false ) {
1592                                         $m[1] = urldecode($m[1]);
1593                                 }
1594                                 $trail = "";
1595                         } else { # Invalid form; output directly
1596                                 $s .= $prefix . '[[' . $line ;
1597                                 wfProfileOut( "$fname-e1" );
1598                                 continue;
1599                         }
1600                         wfProfileOut( "$fname-e1" );
1601                         wfProfileIn( "$fname-misc" );
1602
1603                         # Don't allow internal links to pages containing
1604                         # PROTO: where PROTO is a valid URL protocol; these
1605                         # should be external links.
1606                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1607                                 $s .= $prefix . '[[' . $line ;
1608                                 wfProfileOut( "$fname-misc" );
1609                                 continue;
1610                         }
1611
1612                         # Make subpage if necessary
1613                         if( $useSubpages ) {
1614                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1615                         } else {
1616                                 $link = $m[1];
1617                         }
1618
1619                         $noforce = (substr($m[1], 0, 1) != ':');
1620                         if (!$noforce) {
1621                                 # Strip off leading ':'
1622                                 $link = substr($link, 1);
1623                         }
1624
1625                         wfProfileOut( "$fname-misc" );
1626                         wfProfileIn( "$fname-title" );
1627                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1628                         if( !$nt ) {
1629                                 $s .= $prefix . '[[' . $line;
1630                                 wfProfileOut( "$fname-title" );
1631                                 continue;
1632                         }
1633
1634                         $ns = $nt->getNamespace();
1635                         $iw = $nt->getInterWiki();
1636                         wfProfileOut( "$fname-title" );
1637
1638                         if ($might_be_img) { # if this is actually an invalid link
1639                                 wfProfileIn( "$fname-might_be_img" );
1640                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1641                                         $found = false;
1642                                         while (isset ($a[$k+1]) ) {
1643                                                 #look at the next 'line' to see if we can close it there
1644                                                 $spliced = array_splice( $a, $k + 1, 1 );
1645                                                 $next_line = array_shift( $spliced );
1646                                                 $m = explode( ']]', $next_line, 3 );
1647                                                 if ( count( $m ) == 3 ) {
1648                                                         # the first ]] closes the inner link, the second the image
1649                                                         $found = true;
1650                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1651                                                         $trail = $m[2];
1652                                                         break;
1653                                                 } elseif ( count( $m ) == 2 ) {
1654                                                         #if there's exactly one ]] that's fine, we'll keep looking
1655                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1656                                                 } else {
1657                                                         #if $next_line is invalid too, we need look no further
1658                                                         $text .= '[[' . $next_line;
1659                                                         break;
1660                                                 }
1661                                         }
1662                                         if ( !$found ) {
1663                                                 # we couldn't find the end of this imageLink, so output it raw
1664                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1665                                                 $text = $this->replaceInternalLinks($text);
1666                                                 $s .= "{$prefix}[[$link|$text";
1667                                                 # note: no $trail, because without an end, there *is* no trail
1668                                                 wfProfileOut( "$fname-might_be_img" );
1669                                                 continue;
1670                                         }
1671                                 } else { #it's not an image, so output it raw
1672                                         $s .= "{$prefix}[[$link|$text";
1673                                         # note: no $trail, because without an end, there *is* no trail
1674                                         wfProfileOut( "$fname-might_be_img" );
1675                                         continue;
1676                                 }
1677                                 wfProfileOut( "$fname-might_be_img" );
1678                         }
1679
1680                         $wasblank = ( '' == $text );
1681                         if( $wasblank ) $text = $link;
1682
1683                         # Link not escaped by : , create the various objects
1684                         if( $noforce ) {
1685
1686                                 # Interwikis
1687                                 wfProfileIn( "$fname-interwiki" );
1688                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1689                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1690                                         $s = rtrim($s . $prefix);
1691                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1692                                         wfProfileOut( "$fname-interwiki" );
1693                                         continue;
1694                                 }
1695                                 wfProfileOut( "$fname-interwiki" );
1696
1697                                 if ( $ns == NS_IMAGE ) {
1698                                         wfProfileIn( "$fname-image" );
1699                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1700                                                 # recursively parse links inside the image caption
1701                                                 # actually, this will parse them in any other parameters, too,
1702                                                 # but it might be hard to fix that, and it doesn't matter ATM
1703                                                 $text = $this->replaceExternalLinks($text);
1704                                                 $text = $this->replaceInternalLinks($text);
1705
1706                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1707                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1708                                                 $this->mOutput->addImage( $nt->getDBkey() );
1709
1710                                                 wfProfileOut( "$fname-image" );
1711                                                 continue;
1712                                         } else {
1713                                                 # We still need to record the image's presence on the page
1714                                                 $this->mOutput->addImage( $nt->getDBkey() );
1715                                         }
1716                                         wfProfileOut( "$fname-image" );
1717
1718                                 }
1719
1720                                 if ( $ns == NS_CATEGORY ) {
1721                                         wfProfileIn( "$fname-category" );
1722                                         $s = rtrim($s . "\n"); # bug 87
1723
1724                                         if ( $wasblank ) {
1725                                                 $sortkey = $this->getDefaultSort();
1726                                         } else {
1727                                                 $sortkey = $text;
1728                                         }
1729                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1730                                         $sortkey = str_replace( "\n", '', $sortkey );
1731                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1732                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1733
1734                                         /**
1735                                          * Strip the whitespace Category links produce, see bug 87
1736                                          * @todo We might want to use trim($tmp, "\n") here.
1737                                          */
1738                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1739
1740                                         wfProfileOut( "$fname-category" );
1741                                         continue;
1742                                 }
1743                         }
1744
1745                         # Self-link checking
1746                         if( $nt->getFragment() === '' ) {
1747                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1748                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1749                                         continue;
1750                                 }
1751                         }
1752
1753                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1754                         if( $ns == NS_MEDIA ) {
1755                                 # Give extensions a chance to select the file revision for us
1756                                 $skip = $time = false;
1757                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1758                                 if ( $skip ) {
1759                                         $link = $sk->makeLinkObj( $nt );
1760                                 } else {
1761                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1762                                 }
1763                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1764                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1765                                 $this->mOutput->addImage( $nt->getDBkey() );
1766                                 continue;
1767                         } elseif( $ns == NS_SPECIAL ) {
1768                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1769                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1770                                 } else {
1771                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1772                                 }
1773                                 continue;
1774                         } elseif( $ns == NS_IMAGE ) {
1775                                 $img = wfFindFile( $nt );
1776                                 if( $img ) {
1777                                         // Force a blue link if the file exists; may be a remote
1778                                         // upload on the shared repository, and we want to see its
1779                                         // auto-generated page.
1780                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1781                                         $this->mOutput->addLink( $nt );
1782                                         continue;
1783                                 }
1784                         }
1785                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1786                 }
1787                 wfProfileOut( $fname );
1788                 return $s;
1789         }
1790
1791         /**
1792          * Make a link placeholder. The text returned can be later resolved to a real link with
1793          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1794          * parsing of interwiki links, and secondly to allow all existence checks and
1795          * article length checks (for stub links) to be bundled into a single query.
1796          *
1797          */
1798         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1799                 wfProfileIn( __METHOD__ );
1800                 if ( ! is_object($nt) ) {
1801                         # Fail gracefully
1802                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1803                 } else {
1804                         # Separate the link trail from the rest of the link
1805                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1806
1807                         if ( $nt->isExternal() ) {
1808                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1809                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1810                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1811                         } else {
1812                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1813                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1814                                 $this->mLinkHolders['queries'][] = $query;
1815                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1816                                 $this->mLinkHolders['titles'][] = $nt;
1817
1818                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1819                         }
1820                 }
1821                 wfProfileOut( __METHOD__ );
1822                 return $retVal;
1823         }
1824
1825         /**
1826          * Render a forced-blue link inline; protect against double expansion of
1827          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1828          * Since this little disaster has to split off the trail text to avoid
1829          * breaking URLs in the following text without breaking trails on the
1830          * wiki links, it's been made into a horrible function.
1831          *
1832          * @param Title $nt
1833          * @param string $text
1834          * @param string $query
1835          * @param string $trail
1836          * @param string $prefix
1837          * @return string HTML-wikitext mix oh yuck
1838          */
1839         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1840                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1841                 $sk = $this->mOptions->getSkin();
1842                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1843                 return $this->armorLinks( $link ) . $trail;
1844         }
1845
1846         /**
1847          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1848          * going to go through further parsing steps before inline URL expansion.
1849          *
1850          * In particular this is important when using action=render, which causes
1851          * full URLs to be included.
1852          *
1853          * Oh man I hate our multi-layer parser!
1854          *
1855          * @param string more-or-less HTML
1856          * @return string less-or-more HTML with NOPARSE bits
1857          */
1858         function armorLinks( $text ) {
1859                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1860                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1861         }
1862
1863         /**
1864          * Return true if subpage links should be expanded on this page.
1865          * @return bool
1866          */
1867         function areSubpagesAllowed() {
1868                 # Some namespaces don't allow subpages
1869                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1870         }
1871
1872         /**
1873          * Handle link to subpage if necessary
1874          * @param string $target the source of the link
1875          * @param string &$text the link text, modified as necessary
1876          * @return string the full name of the link
1877          * @private
1878          */
1879         function maybeDoSubpageLink($target, &$text) {
1880                 # Valid link forms:
1881                 # Foobar -- normal
1882                 # :Foobar -- override special treatment of prefix (images, language links)
1883                 # /Foobar -- convert to CurrentPage/Foobar
1884                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1885                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1886                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1887
1888                 $fname = 'Parser::maybeDoSubpageLink';
1889                 wfProfileIn( $fname );
1890                 $ret = $target; # default return value is no change
1891
1892                 # Some namespaces don't allow subpages,
1893                 # so only perform processing if subpages are allowed
1894                 if( $this->areSubpagesAllowed() ) {
1895                         $hash = strpos( $target, '#' );
1896                         if( $hash !== false ) {
1897                                 $suffix = substr( $target, $hash );
1898                                 $target = substr( $target, 0, $hash );
1899                         } else {
1900                                 $suffix = '';
1901                         }
1902                         # bug 7425
1903                         $target = trim( $target );
1904                         # Look at the first character
1905                         if( $target != '' && $target{0} == '/' ) {
1906                                 # / at end means we don't want the slash to be shown
1907                                 $m = array();
1908                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1909                                 if( $trailingSlashes ) {
1910                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1911                                 } else {
1912                                         $noslash = substr( $target, 1 );
1913                                 }
1914
1915                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1916                                 if( '' === $text ) {
1917                                         $text = $target . $suffix;
1918                                 } # this might be changed for ugliness reasons
1919                         } else {
1920                                 # check for .. subpage backlinks
1921                                 $dotdotcount = 0;
1922                                 $nodotdot = $target;
1923                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1924                                         ++$dotdotcount;
1925                                         $nodotdot = substr( $nodotdot, 3 );
1926                                 }
1927                                 if($dotdotcount > 0) {
1928                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1929                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1930                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1931                                                 # / at the end means don't show full path
1932                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1933                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1934                                                         if( '' === $text ) {
1935                                                                 $text = $nodotdot . $suffix;
1936                                                         }
1937                                                 }
1938                                                 $nodotdot = trim( $nodotdot );
1939                                                 if( $nodotdot != '' ) {
1940                                                         $ret .= '/' . $nodotdot;
1941                                                 }
1942                                                 $ret .= $suffix;
1943                                         }
1944                                 }
1945                         }
1946                 }
1947
1948                 wfProfileOut( $fname );
1949                 return $ret;
1950         }
1951
1952         /**#@+
1953          * Used by doBlockLevels()
1954          * @private
1955          */
1956         /* private */ function closeParagraph() {
1957                 $result = '';
1958                 if ( '' != $this->mLastSection ) {
1959                         $result = '</' . $this->mLastSection  . ">\n";
1960                 }
1961                 $this->mInPre = false;
1962                 $this->mLastSection = '';
1963                 return $result;
1964         }
1965         # getCommon() returns the length of the longest common substring
1966         # of both arguments, starting at the beginning of both.
1967         #
1968         /* private */ function getCommon( $st1, $st2 ) {
1969                 $fl = strlen( $st1 );
1970                 $shorter = strlen( $st2 );
1971                 if ( $fl < $shorter ) { $shorter = $fl; }
1972
1973                 for ( $i = 0; $i < $shorter; ++$i ) {
1974                         if ( $st1{$i} != $st2{$i} ) { break; }
1975                 }
1976                 return $i;
1977         }
1978         # These next three functions open, continue, and close the list
1979         # element appropriate to the prefix character passed into them.
1980         #
1981         /* private */ function openList( $char ) {
1982                 $result = $this->closeParagraph();
1983
1984                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1985                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1986                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1987                 else if ( ';' == $char ) {
1988                         $result .= '<dl><dt>';
1989                         $this->mDTopen = true;
1990                 }
1991                 else { $result = '<!-- ERR 1 -->'; }
1992
1993                 return $result;
1994         }
1995
1996         /* private */ function nextItem( $char ) {
1997                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1998                 else if ( ':' == $char || ';' == $char ) {
1999                         $close = '</dd>';
2000                         if ( $this->mDTopen ) { $close = '</dt>'; }
2001                         if ( ';' == $char ) {
2002                                 $this->mDTopen = true;
2003                                 return $close . '<dt>';
2004                         } else {
2005                                 $this->mDTopen = false;
2006                                 return $close . '<dd>';
2007                         }
2008                 }
2009                 return '<!-- ERR 2 -->';
2010         }
2011
2012         /* private */ function closeList( $char ) {
2013                 if ( '*' == $char ) { $text = '</li></ul>'; }
2014                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2015                 else if ( ':' == $char ) {
2016                         if ( $this->mDTopen ) {
2017                                 $this->mDTopen = false;
2018                                 $text = '</dt></dl>';
2019                         } else {
2020                                 $text = '</dd></dl>';
2021                         }
2022                 }
2023                 else {  return '<!-- ERR 3 -->'; }
2024                 return $text."\n";
2025         }
2026         /**#@-*/
2027
2028         /**
2029          * Make lists from lines starting with ':', '*', '#', etc.
2030          *
2031          * @private
2032          * @return string the lists rendered as HTML
2033          */
2034         function doBlockLevels( $text, $linestart ) {
2035                 $fname = 'Parser::doBlockLevels';
2036                 wfProfileIn( $fname );
2037
2038                 # Parsing through the text line by line.  The main thing
2039                 # happening here is handling of block-level elements p, pre,
2040                 # and making lists from lines starting with * # : etc.
2041                 #
2042                 $textLines = explode( "\n", $text );
2043
2044                 $lastPrefix = $output = '';
2045                 $this->mDTopen = $inBlockElem = false;
2046                 $prefixLength = 0;
2047                 $paragraphStack = false;
2048
2049                 if ( !$linestart ) {
2050                         $output .= array_shift( $textLines );
2051                 }
2052                 foreach ( $textLines as $oLine ) {
2053                         $lastPrefixLength = strlen( $lastPrefix );
2054                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2055                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2056                         if ( !$this->mInPre ) {
2057                                 # Multiple prefixes may abut each other for nested lists.
2058                                 $prefixLength = strspn( $oLine, '*#:;' );
2059                                 $pref = substr( $oLine, 0, $prefixLength );
2060
2061                                 # eh?
2062                                 $pref2 = str_replace( ';', ':', $pref );
2063                                 $t = substr( $oLine, $prefixLength );
2064                                 $this->mInPre = !empty($preOpenMatch);
2065                         } else {
2066                                 # Don't interpret any other prefixes in preformatted text
2067                                 $prefixLength = 0;
2068                                 $pref = $pref2 = '';
2069                                 $t = $oLine;
2070                         }
2071
2072                         # List generation
2073                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2074                                 # Same as the last item, so no need to deal with nesting or opening stuff
2075                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2076                                 $paragraphStack = false;
2077
2078                                 if ( substr( $pref, -1 ) == ';') {
2079                                         # The one nasty exception: definition lists work like this:
2080                                         # ; title : definition text
2081                                         # So we check for : in the remainder text to split up the
2082                                         # title and definition, without b0rking links.
2083                                         $term = $t2 = '';
2084                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2085                                                 $t = $t2;
2086                                                 $output .= $term . $this->nextItem( ':' );
2087                                         }
2088                                 }
2089                         } elseif( $prefixLength || $lastPrefixLength ) {
2090                                 # Either open or close a level...
2091                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2092                                 $paragraphStack = false;
2093
2094                                 while( $commonPrefixLength < $lastPrefixLength ) {
2095                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2096                                         --$lastPrefixLength;
2097                                 }
2098                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2099                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2100                                 }
2101                                 while ( $prefixLength > $commonPrefixLength ) {
2102                                         $char = substr( $pref, $commonPrefixLength, 1 );
2103                                         $output .= $this->openList( $char );
2104
2105                                         if ( ';' == $char ) {
2106                                                 # FIXME: This is dupe of code above
2107                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2108                                                         $t = $t2;
2109                                                         $output .= $term . $this->nextItem( ':' );
2110                                                 }
2111                                         }
2112                                         ++$commonPrefixLength;
2113                                 }
2114                                 $lastPrefix = $pref2;
2115                         }
2116                         if( 0 == $prefixLength ) {
2117                                 wfProfileIn( "$fname-paragraph" );
2118                                 # No prefix (not in list)--go to paragraph mode
2119                                 // XXX: use a stack for nestable elements like span, table and div
2120                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2121                                 $closematch = preg_match(
2122                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2123                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2124                                 if ( $openmatch or $closematch ) {
2125                                         $paragraphStack = false;
2126                                         # TODO bug 5718: paragraph closed
2127                                         $output .= $this->closeParagraph();
2128                                         if ( $preOpenMatch and !$preCloseMatch ) {
2129                                                 $this->mInPre = true;
2130                                         }
2131                                         if ( $closematch ) {
2132                                                 $inBlockElem = false;
2133                                         } else {
2134                                                 $inBlockElem = true;
2135                                         }
2136                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2137                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2138                                                 // pre
2139                                                 if ($this->mLastSection != 'pre') {
2140                                                         $paragraphStack = false;
2141                                                         $output .= $this->closeParagraph().'<pre>';
2142                                                         $this->mLastSection = 'pre';
2143                                                 }
2144                                                 $t = substr( $t, 1 );
2145                                         } else {
2146                                                 // paragraph
2147                                                 if ( '' == trim($t) ) {
2148                                                         if ( $paragraphStack ) {
2149                                                                 $output .= $paragraphStack.'<br />';
2150                                                                 $paragraphStack = false;
2151                                                                 $this->mLastSection = 'p';
2152                                                         } else {
2153                                                                 if ($this->mLastSection != 'p' ) {
2154                                                                         $output .= $this->closeParagraph();
2155                                                                         $this->mLastSection = '';
2156                                                                         $paragraphStack = '<p>';
2157                                                                 } else {
2158                                                                         $paragraphStack = '</p><p>';
2159                                                                 }
2160                                                         }
2161                                                 } else {
2162                                                         if ( $paragraphStack ) {
2163                                                                 $output .= $paragraphStack;
2164                                                                 $paragraphStack = false;
2165                                                                 $this->mLastSection = 'p';
2166                                                         } else if ($this->mLastSection != 'p') {
2167                                                                 $output .= $this->closeParagraph().'<p>';
2168                                                                 $this->mLastSection = 'p';
2169                                                         }
2170                                                 }
2171                                         }
2172                                 }
2173                                 wfProfileOut( "$fname-paragraph" );
2174                         }
2175                         // somewhere above we forget to get out of pre block (bug 785)
2176                         if($preCloseMatch && $this->mInPre) {
2177                                 $this->mInPre = false;
2178                         }
2179                         if ($paragraphStack === false) {
2180                                 $output .= $t."\n";
2181                         }
2182                 }
2183                 while ( $prefixLength ) {
2184                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2185                         --$prefixLength;
2186                 }
2187                 if ( '' != $this->mLastSection ) {
2188                         $output .= '</' . $this->mLastSection . '>';
2189                         $this->mLastSection = '';
2190                 }
2191
2192                 wfProfileOut( $fname );
2193                 return $output;
2194         }
2195
2196         /**
2197          * Split up a string on ':', ignoring any occurences inside tags
2198          * to prevent illegal overlapping.
2199          * @param string $str the string to split
2200          * @param string &$before set to everything before the ':'
2201          * @param string &$after set to everything after the ':'
2202          * return string the position of the ':', or false if none found
2203          */
2204         function findColonNoLinks($str, &$before, &$after) {
2205                 $fname = 'Parser::findColonNoLinks';
2206                 wfProfileIn( $fname );
2207
2208                 $pos = strpos( $str, ':' );
2209                 if( $pos === false ) {
2210                         // Nothing to find!
2211                         wfProfileOut( $fname );
2212                         return false;
2213                 }
2214
2215                 $lt = strpos( $str, '<' );
2216                 if( $lt === false || $lt > $pos ) {
2217                         // Easy; no tag nesting to worry about
2218                         $before = substr( $str, 0, $pos );
2219                         $after = substr( $str, $pos+1 );
2220                         wfProfileOut( $fname );
2221                         return $pos;
2222                 }
2223
2224                 // Ugly state machine to walk through avoiding tags.
2225                 $state = self::COLON_STATE_TEXT;
2226                 $stack = 0;
2227                 $len = strlen( $str );
2228                 for( $i = 0; $i < $len; $i++ ) {
2229                         $c = $str{$i};
2230
2231                         switch( $state ) {
2232                         // (Using the number is a performance hack for common cases)
2233                         case 0: // self::COLON_STATE_TEXT:
2234                                 switch( $c ) {
2235                                 case "<":
2236                                         // Could be either a <start> tag or an </end> tag
2237                                         $state = self::COLON_STATE_TAGSTART;
2238                                         break;
2239                                 case ":":
2240                                         if( $stack == 0 ) {
2241                                                 // We found it!
2242                                                 $before = substr( $str, 0, $i );
2243                                                 $after = substr( $str, $i + 1 );
2244                                                 wfProfileOut( $fname );
2245                                                 return $i;
2246                                         }
2247                                         // Embedded in a tag; don't break it.
2248                                         break;
2249                                 default:
2250                                         // Skip ahead looking for something interesting
2251                                         $colon = strpos( $str, ':', $i );
2252                                         if( $colon === false ) {
2253                                                 // Nothing else interesting
2254                                                 wfProfileOut( $fname );
2255                                                 return false;
2256                                         }
2257                                         $lt = strpos( $str, '<', $i );
2258                                         if( $stack === 0 ) {
2259                                                 if( $lt === false || $colon < $lt ) {
2260                                                         // We found it!
2261                                                         $before = substr( $str, 0, $colon );
2262                                                         $after = substr( $str, $colon + 1 );
2263                                                         wfProfileOut( $fname );
2264                                                         return $i;
2265                                                 }
2266                                         }
2267                                         if( $lt === false ) {
2268                                                 // Nothing else interesting to find; abort!
2269                                                 // We're nested, but there's no close tags left. Abort!
2270                                                 break 2;
2271                                         }
2272                                         // Skip ahead to next tag start
2273                                         $i = $lt;
2274                                         $state = self::COLON_STATE_TAGSTART;
2275                                 }
2276                                 break;
2277                         case 1: // self::COLON_STATE_TAG:
2278                                 // In a <tag>
2279                                 switch( $c ) {
2280                                 case ">":
2281                                         $stack++;
2282                                         $state = self::COLON_STATE_TEXT;
2283                                         break;
2284                                 case "/":
2285                                         // Slash may be followed by >?
2286                                         $state = self::COLON_STATE_TAGSLASH;
2287                                         break;
2288                                 default:
2289                                         // ignore
2290                                 }
2291                                 break;
2292                         case 2: // self::COLON_STATE_TAGSTART:
2293                                 switch( $c ) {
2294                                 case "/":
2295                                         $state = self::COLON_STATE_CLOSETAG;
2296                                         break;
2297                                 case "!":
2298                                         $state = self::COLON_STATE_COMMENT;
2299                                         break;
2300                                 case ">":
2301                                         // Illegal early close? This shouldn't happen D:
2302                                         $state = self::COLON_STATE_TEXT;
2303                                         break;
2304                                 default:
2305                                         $state = self::COLON_STATE_TAG;
2306                                 }
2307                                 break;
2308                         case 3: // self::COLON_STATE_CLOSETAG:
2309                                 // In a </tag>
2310                                 if( $c == ">" ) {
2311                                         $stack--;
2312                                         if( $stack < 0 ) {
2313                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2314                                                 wfProfileOut( $fname );
2315                                                 return false;
2316                                         }
2317                                         $state = self::COLON_STATE_TEXT;
2318                                 }
2319                                 break;
2320                         case self::COLON_STATE_TAGSLASH:
2321                                 if( $c == ">" ) {
2322                                         // Yes, a self-closed tag <blah/>
2323                                         $state = self::COLON_STATE_TEXT;
2324                                 } else {
2325                                         // Probably we're jumping the gun, and this is an attribute
2326                                         $state = self::COLON_STATE_TAG;
2327                                 }
2328                                 break;
2329                         case 5: // self::COLON_STATE_COMMENT:
2330                                 if( $c == "-" ) {
2331                                         $state = self::COLON_STATE_COMMENTDASH;
2332                                 }
2333                                 break;
2334                         case self::COLON_STATE_COMMENTDASH:
2335                                 if( $c == "-" ) {
2336                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2337                                 } else {
2338                                         $state = self::COLON_STATE_COMMENT;
2339                                 }
2340                                 break;
2341                         case self::COLON_STATE_COMMENTDASHDASH:
2342                                 if( $c == ">" ) {
2343                                         $state = self::COLON_STATE_TEXT;
2344                                 } else {
2345                                         $state = self::COLON_STATE_COMMENT;
2346                                 }
2347                                 break;
2348                         default:
2349                                 throw new MWException( "State machine error in $fname" );
2350                         }
2351                 }
2352                 if( $stack > 0 ) {
2353                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2354                         return false;
2355                 }
2356                 wfProfileOut( $fname );
2357                 return false;
2358         }
2359
2360         /**
2361          * Return value of a magic variable (like PAGENAME)
2362          *
2363          * @private
2364          */
2365         function getVariableValue( $index ) {
2366                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2367
2368                 /**
2369                  * Some of these require message or data lookups and can be
2370                  * expensive to check many times.
2371                  */
2372                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2373                         if ( isset( $this->mVarCache[$index] ) ) {
2374                                 return $this->mVarCache[$index];
2375                         }
2376                 }
2377
2378                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2379                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2380
2381                 # Use the time zone
2382                 global $wgLocaltimezone;
2383                 if ( isset( $wgLocaltimezone ) ) {
2384                         $oldtz = getenv( 'TZ' );
2385                         putenv( 'TZ='.$wgLocaltimezone );
2386                 }
2387
2388                 wfSuppressWarnings(); // E_STRICT system time bitching
2389                 $localTimestamp = date( 'YmdHis', $ts );
2390                 $localMonth = date( 'm', $ts );
2391                 $localMonthName = date( 'n', $ts );
2392                 $localDay = date( 'j', $ts );
2393                 $localDay2 = date( 'd', $ts );
2394                 $localDayOfWeek = date( 'w', $ts );
2395                 $localWeek = date( 'W', $ts );
2396                 $localYear = date( 'Y', $ts );
2397                 $localHour = date( 'H', $ts );
2398                 if ( isset( $wgLocaltimezone ) ) {
2399                         putenv( 'TZ='.$oldtz );
2400                 }
2401                 wfRestoreWarnings();
2402
2403                 switch ( $index ) {
2404                         case 'currentmonth':
2405                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2406                         case 'currentmonthname':
2407                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2408                         case 'currentmonthnamegen':
2409                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2410                         case 'currentmonthabbrev':
2411                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2412                         case 'currentday':
2413                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2414                         case 'currentday2':
2415                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2416                         case 'localmonth':
2417                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2418                         case 'localmonthname':
2419                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2420                         case 'localmonthnamegen':
2421                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2422                         case 'localmonthabbrev':
2423                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2424                         case 'localday':
2425                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2426                         case 'localday2':
2427                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2428                         case 'pagename':
2429                                 return wfEscapeWikiText( $this->mTitle->getText() );
2430                         case 'pagenamee':
2431                                 return $this->mTitle->getPartialURL();
2432                         case 'fullpagename':
2433                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2434                         case 'fullpagenamee':
2435                                 return $this->mTitle->getPrefixedURL();
2436                         case 'subpagename':
2437                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2438                         case 'subpagenamee':
2439                                 return $this->mTitle->getSubpageUrlForm();
2440                         case 'basepagename':
2441                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2442                         case 'basepagenamee':
2443                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2444                         case 'talkpagename':
2445                                 if( $this->mTitle->canTalk() ) {
2446                                         $talkPage = $this->mTitle->getTalkPage();
2447                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2448                                 } else {
2449                                         return '';
2450                                 }
2451                         case 'talkpagenamee':
2452                                 if( $this->mTitle->canTalk() ) {
2453                                         $talkPage = $this->mTitle->getTalkPage();
2454                                         return $talkPage->getPrefixedUrl();
2455                                 } else {
2456                                         return '';
2457                                 }
2458                         case 'subjectpagename':
2459                                 $subjPage = $this->mTitle->getSubjectPage();
2460                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2461                         case 'subjectpagenamee':
2462                                 $subjPage = $this->mTitle->getSubjectPage();
2463                                 return $subjPage->getPrefixedUrl();
2464                         case 'revisionid':
2465                                 // Let the edit saving system know we should parse the page
2466                                 // *after* a revision ID has been assigned.
2467                                 $this->mOutput->setFlag( 'vary-revision' );
2468                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2469                                 return $this->mRevisionId;
2470                         case 'revisionday':
2471                                 // Let the edit saving system know we should parse the page
2472                                 // *after* a revision ID has been assigned. This is for null edits.
2473                                 $this->mOutput->setFlag( 'vary-revision' );
2474                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2475                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2476                         case 'revisionday2':
2477                                 // Let the edit saving system know we should parse the page
2478                                 // *after* a revision ID has been assigned. This is for null edits.
2479                                 $this->mOutput->setFlag( 'vary-revision' );
2480                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2481                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2482                         case 'revisionmonth':
2483                                 // Let the edit saving system know we should parse the page
2484                                 // *after* a revision ID has been assigned. This is for null edits.
2485                                 $this->mOutput->setFlag( 'vary-revision' );
2486                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2487                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2488                         case 'revisionyear':
2489                                 // Let the edit saving system know we should parse the page
2490                                 // *after* a revision ID has been assigned. This is for null edits.
2491                                 $this->mOutput->setFlag( 'vary-revision' );
2492                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2493                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2494                         case 'revisiontimestamp':
2495                                 // Let the edit saving system know we should parse the page
2496                                 // *after* a revision ID has been assigned. This is for null edits.
2497                                 $this->mOutput->setFlag( 'vary-revision' );
2498                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2499                                 return $this->getRevisionTimestamp();
2500                         case 'namespace':
2501                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2502                         case 'namespacee':
2503                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2504                         case 'talkspace':
2505                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2506                         case 'talkspacee':
2507                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2508                         case 'subjectspace':
2509                                 return $this->mTitle->getSubjectNsText();
2510                         case 'subjectspacee':
2511                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2512                         case 'currentdayname':
2513                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2514                         case 'currentyear':
2515                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2516                         case 'currenttime':
2517                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2518                         case 'currenthour':
2519                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2520                         case 'currentweek':
2521                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2522                                 // int to remove the padding
2523                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2524                         case 'currentdow':
2525                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2526                         case 'localdayname':
2527                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2528                         case 'localyear':
2529                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2530                         case 'localtime':
2531                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2532                         case 'localhour':
2533                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2534                         case 'localweek':
2535                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2536                                 // int to remove the padding
2537                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2538                         case 'localdow':
2539                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2540                         case 'numberofarticles':
2541                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2542                         case 'numberoffiles':
2543                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2544                         case 'numberofusers':
2545                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2546                         case 'numberofpages':
2547                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2548                         case 'numberofadmins':
2549                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2550                         case 'numberofedits':
2551                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2552                         case 'currenttimestamp':
2553                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2554                         case 'localtimestamp':
2555                                 return $this->mVarCache[$index] = $localTimestamp;
2556                         case 'currentversion':
2557                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2558                         case 'sitename':
2559                                 return $wgSitename;
2560                         case 'server':
2561                                 return $wgServer;
2562                         case 'servername':
2563                                 return $wgServerName;
2564                         case 'scriptpath':
2565                                 return $wgScriptPath;
2566                         case 'directionmark':
2567                                 return $wgContLang->getDirMark();
2568                         case 'contentlanguage':
2569                                 global $wgContLanguageCode;
2570                                 return $wgContLanguageCode;
2571                         default:
2572                                 $ret = null;
2573                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2574                                         return $ret;
2575                                 else
2576                                         return null;
2577                 }
2578         }
2579
2580         /**
2581          * initialise the magic variables (like CURRENTMONTHNAME)
2582          *
2583          * @private
2584          */
2585         function initialiseVariables() {
2586                 $fname = 'Parser::initialiseVariables';
2587                 wfProfileIn( $fname );
2588                 $variableIDs = MagicWord::getVariableIDs();
2589
2590                 $this->mVariables = new MagicWordArray( $variableIDs );
2591                 wfProfileOut( $fname );
2592         }
2593
2594         /**
2595          * Preprocess some wikitext and return the document tree.
2596          * This is the ghost of replace_variables().
2597          *
2598          * @param string $text The text to parse
2599          * @param integer flags Bitwise combination of:
2600          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2601          *                                     included. Default is to assume a direct page view.
2602          *
2603          * The generated DOM tree must depend only on the input text and the flags.
2604          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2605          *
2606          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2607          * change in the DOM tree for a given text, must be passed through the section identifier
2608          * in the section edit link and thus back to extractSections().
2609          *
2610          * The output of this function is currently only cached in process memory, but a persistent
2611          * cache may be implemented at a later date which takes further advantage of these strict
2612          * dependency requirements.
2613          *
2614          * @private
2615          */
2616         function preprocessToDom ( $text, $flags = 0 ) {
2617                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2618                 return $dom;
2619         }
2620
2621         /*
2622          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2623          */
2624         public static function splitWhitespace( $s ) {
2625                 $ltrimmed = ltrim( $s );
2626                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2627                 $trimmed = rtrim( $ltrimmed );
2628                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2629                 if ( $diff > 0 ) {
2630                         $w2 = substr( $ltrimmed, -$diff );
2631                 } else {
2632                         $w2 = '';
2633                 }
2634                 return array( $w1, $trimmed, $w2 );
2635         }
2636
2637         /**
2638          * Replace magic variables, templates, and template arguments
2639          * with the appropriate text. Templates are substituted recursively,
2640          * taking care to avoid infinite loops.
2641          *
2642          * Note that the substitution depends on value of $mOutputType:
2643          *  self::OT_WIKI: only {{subst:}} templates
2644          *  self::OT_PREPROCESS: templates but not extension tags
2645          *  self::OT_HTML: all templates and extension tags
2646          *
2647          * @param string $tex The text to transform
2648          * @param PPFrame $frame Object describing the arguments passed to the template.
2649          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2650          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2651          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2652          * @private
2653          */
2654         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2655                 # Prevent too big inclusions
2656                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2657                         return $text;
2658                 }
2659
2660                 $fname = __METHOD__;
2661                 wfProfileIn( $fname );
2662
2663                 if ( $frame === false ) {
2664                         $frame = $this->getPreprocessor()->newFrame();
2665                 } elseif ( !( $frame instanceof PPFrame ) ) {
2666                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2667                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2668                 }
2669
2670                 $dom = $this->preprocessToDom( $text );
2671                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2672                 $text = $frame->expand( $dom, $flags );
2673
2674                 wfProfileOut( $fname );
2675                 return $text;
2676         }
2677
2678         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2679         static function createAssocArgs( $args ) {
2680                 $assocArgs = array();
2681                 $index = 1;
2682                 foreach( $args as $arg ) {
2683                         $eqpos = strpos( $arg, '=' );
2684                         if ( $eqpos === false ) {
2685                                 $assocArgs[$index++] = $arg;
2686                         } else {
2687                                 $name = trim( substr( $arg, 0, $eqpos ) );
2688                                 $value = trim( substr( $arg, $eqpos+1 ) );
2689                                 if ( $value === false ) {
2690                                         $value = '';
2691                                 }
2692                                 if ( $name !== false ) {
2693                                         $assocArgs[$name] = $value;
2694                                 }
2695                         }
2696                 }
2697
2698                 return $assocArgs;
2699         }
2700
2701         /**
2702          * Warn the user when a parser limitation is reached
2703          * Will warn at most once the user per limitation type
2704          *
2705          * @param string $limitationType, should be one of:
2706          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2707          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2708          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2709          * @params int $current, $max When an explicit limit has been
2710          *       exceeded, provide the values (optional)
2711          */
2712         function limitationWarn( $limitationType, $current=null, $max=null) {
2713                 $msgName = $limitationType . '-warning';
2714                 //does no harm if $current and $max are present but are unnecessary for the message
2715                 $warning = wfMsg( $msgName, $current, $max);
2716                 $this->mOutput->addWarning( $warning );
2717                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2718                 if ( $cat ) {
2719                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2720                 }
2721         }
2722
2723         /**
2724          * Return the text of a template, after recursively
2725          * replacing any variables or templates within the template.
2726          *
2727          * @param array $piece The parts of the template
2728          *  $piece['title']: the title, i.e. the part before the |
2729          *  $piece['parts']: the parameter array
2730          *  $piece['lineStart']: whether the brace was at the start of a line
2731          * @param PPFrame The current frame, contains template arguments
2732          * @return string the text of the template
2733          * @private
2734          */
2735         function braceSubstitution( $piece, $frame ) {
2736                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2737                 $fname = __METHOD__;
2738                 wfProfileIn( $fname );
2739                 wfProfileIn( __METHOD__.'-setup' );
2740
2741                 # Flags
2742                 $found = false;             # $text has been filled
2743                 $nowiki = false;            # wiki markup in $text should be escaped
2744                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2745                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2746                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2747                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2748
2749                 # Title object, where $text came from
2750                 $title = NULL;
2751
2752                 # $part1 is the bit before the first |, and must contain only title characters.
2753                 # Various prefixes will be stripped from it later.
2754                 $titleWithSpaces = $frame->expand( $piece['title'] );
2755                 $part1 = trim( $titleWithSpaces );
2756                 $titleText = false;
2757
2758                 # Original title text preserved for various purposes
2759                 $originalTitle = $part1;
2760
2761                 # $args is a list of argument nodes, starting from index 0, not including $part1
2762                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2763                 wfProfileOut( __METHOD__.'-setup' );
2764
2765                 # SUBST
2766                 wfProfileIn( __METHOD__.'-modifiers' );
2767                 if ( !$found ) {
2768                         $mwSubst = MagicWord::get( 'subst' );
2769                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2770                                 # One of two possibilities is true:
2771                                 # 1) Found SUBST but not in the PST phase
2772                                 # 2) Didn't find SUBST and in the PST phase
2773                                 # In either case, return without further processing
2774                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2775                                 $isLocalObj = true;
2776                                 $found = true;
2777                         }
2778                 }
2779
2780                 # Variables
2781                 if ( !$found && $args->getLength() == 0 ) {
2782                         $id = $this->mVariables->matchStartToEnd( $part1 );
2783                         if ( $id !== false ) {
2784                                 $text = $this->getVariableValue( $id );
2785                                 if (MagicWord::getCacheTTL($id)>-1)
2786                                         $this->mOutput->mContainsOldMagic = true;
2787                                 $found = true;
2788                         }
2789                 }
2790
2791                 # MSG, MSGNW and RAW
2792                 if ( !$found ) {
2793                         # Check for MSGNW:
2794                         $mwMsgnw = MagicWord::get( 'msgnw' );
2795                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2796                                 $nowiki = true;
2797                         } else {
2798                                 # Remove obsolete MSG:
2799                                 $mwMsg = MagicWord::get( 'msg' );
2800                                 $mwMsg->matchStartAndRemove( $part1 );
2801                         }
2802
2803                         # Check for RAW:
2804                         $mwRaw = MagicWord::get( 'raw' );
2805                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2806                                 $forceRawInterwiki = true;
2807                         }
2808                 }
2809                 wfProfileOut( __METHOD__.'-modifiers' );
2810
2811                 # Parser functions
2812                 if ( !$found ) {
2813                         wfProfileIn( __METHOD__ . '-pfunc' );
2814
2815                         $colonPos = strpos( $part1, ':' );
2816                         if ( $colonPos !== false ) {
2817                                 # Case sensitive functions
2818                                 $function = substr( $part1, 0, $colonPos );
2819                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2820                                         $function = $this->mFunctionSynonyms[1][$function];
2821                                 } else {
2822                                         # Case insensitive functions
2823                                         $function = strtolower( $function );
2824                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2825                                                 $function = $this->mFunctionSynonyms[0][$function];
2826                                         } else {
2827                                                 $function = false;
2828                                         }
2829                                 }
2830                                 if ( $function ) {
2831                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2832                                         $initialArgs = array( &$this );
2833                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2834                                         if ( $flags & SFH_OBJECT_ARGS ) {
2835                                                 # Add a frame parameter, and pass the arguments as an array
2836                                                 $allArgs = $initialArgs;
2837                                                 $allArgs[] = $frame;
2838                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2839                                                         $funcArgs[] = $args->item( $i );
2840                                                 }
2841                                                 $allArgs[] = $funcArgs;
2842                                         } else {
2843                                                 # Convert arguments to plain text
2844                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2845                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2846                                                 }
2847                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2848                                         }
2849
2850                                         # Workaround for PHP bug 35229 and similar
2851                                         if ( !is_callable( $callback ) ) {
2852                                                 throw new MWException( "Tag hook for $name is not callable\n" );
2853                                         }
2854                                         $result = call_user_func_array( $callback, $allArgs );
2855                                         $found = true;
2856                                         $noparse = true;
2857                                         $preprocessFlags = 0;
2858
2859                                         if ( is_array( $result ) ) {
2860                                                 if ( isset( $result[0] ) ) {
2861                                                         $text = $result[0];
2862                                                         unset( $result[0] );
2863                                                 }
2864
2865                                                 // Extract flags into the local scope
2866                                                 // This allows callers to set flags such as nowiki, found, etc.
2867                                                 extract( $result );
2868                                         } else {
2869                                                 $text = $result;
2870                                         }
2871                                         if ( !$noparse ) {
2872                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2873                                                 $isChildObj = true;
2874                                         }
2875                                 }
2876                         }
2877                         wfProfileOut( __METHOD__ . '-pfunc' );
2878                 }
2879
2880                 # Finish mangling title and then check for loops.
2881                 # Set $title to a Title object and $titleText to the PDBK
2882                 if ( !$found ) {
2883                         $ns = NS_TEMPLATE;
2884                         # Split the title into page and subpage
2885                         $subpage = '';
2886                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2887                         if ($subpage !== '') {
2888                                 $ns = $this->mTitle->getNamespace();
2889                         }
2890                         $title = Title::newFromText( $part1, $ns );
2891                         if ( $title ) {
2892                                 $titleText = $title->getPrefixedText();
2893                                 # Check for language variants if the template is not found
2894                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2895                                         $wgContLang->findVariantLink($part1, $title);
2896                                 }
2897                                 # Do infinite loop check
2898                                 if ( !$frame->loopCheck( $title ) ) {
2899                                         $found = true;
2900                                         $text = "<span class=\"error\">Template loop detected: [[$titleText]]</span>";
2901                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2902                                 }
2903                                 # Do recursion depth check
2904                                 $limit = $this->mOptions->getMaxTemplateDepth();
2905                                 if ( $frame->depth >= $limit ) {
2906                                         $found = true;
2907                                         $text = "<span class=\"error\">Template recursion depth limit exceeded ($limit)</span>";
2908                                 }
2909                         }
2910                 }
2911
2912                 # Load from database
2913                 if ( !$found && $title ) {
2914                         wfProfileIn( __METHOD__ . '-loadtpl' );
2915                         if ( !$title->isExternal() ) {
2916                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2917                                         $text = SpecialPage::capturePath( $title );
2918                                         if ( is_string( $text ) ) {
2919                                                 $found = true;
2920                                                 $isHTML = true;
2921                                                 $this->disableCache();
2922                                         }
2923                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2924                                         $found = false; //access denied
2925                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
2926                                 } else {
2927                                         list( $text, $title ) = $this->getTemplateDom( $title );
2928                                         if ( $text !== false ) {
2929                                                 $found = true;
2930                                                 $isChildObj = true;
2931                                         }
2932                                 }
2933
2934                                 # If the title is valid but undisplayable, make a link to it
2935                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2936                                         $text = "[[:$titleText]]";
2937                                         $found = true;
2938                                 }
2939                         } elseif ( $title->isTrans() ) {
2940                                 // Interwiki transclusion
2941                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2942                                         $text = $this->interwikiTransclude( $title, 'render' );
2943                                         $isHTML = true;
2944                                 } else {
2945                                         $text = $this->interwikiTransclude( $title, 'raw' );
2946                                         // Preprocess it like a template
2947                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2948                                         $isChildObj = true;
2949                                 }
2950                                 $found = true;
2951                         }
2952                         wfProfileOut( __METHOD__ . '-loadtpl' );
2953                 }
2954
2955                 # If we haven't found text to substitute by now, we're done
2956                 # Recover the source wikitext and return it
2957                 if ( !$found ) {
2958                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2959                         wfProfileOut( $fname );
2960                         return array( 'object' => $text );
2961                 }
2962
2963                 # Expand DOM-style return values in a child frame
2964                 if ( $isChildObj ) {
2965                         # Clean up argument array
2966                         $newFrame = $frame->newChild( $args, $title );
2967
2968                         if ( $nowiki ) {
2969                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2970                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2971                                 # Expansion is eligible for the empty-frame cache
2972                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2973                                         $text = $this->mTplExpandCache[$titleText];
2974                                 } else {
2975                                         $text = $newFrame->expand( $text );
2976                                         $this->mTplExpandCache[$titleText] = $text;
2977                                 }
2978                         } else {
2979                                 # Uncached expansion
2980                                 $text = $newFrame->expand( $text );
2981                         }
2982                 }
2983                 if ( $isLocalObj && $nowiki ) {
2984                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2985                         $isLocalObj = false;
2986                 }
2987
2988                 # Replace raw HTML by a placeholder
2989                 # Add a blank line preceding, to prevent it from mucking up
2990                 # immediately preceding headings
2991                 if ( $isHTML ) {
2992                         $text = "\n\n" . $this->insertStripItem( $text );
2993                 }
2994                 # Escape nowiki-style return values
2995                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2996                         $text = wfEscapeWikiText( $text );
2997                 }
2998                 # Bug 529: if the template begins with a table or block-level
2999                 # element, it should be treated as beginning a new line.
3000                 # This behaviour is somewhat controversial.
3001                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3002                         $text = "\n" . $text;
3003                 }
3004
3005                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3006                         # Error, oversize inclusion
3007                         $text = "[[$originalTitle]]" .
3008                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3009                         $this->limitationWarn( 'post-expand-template-inclusion' );
3010                 }
3011
3012                 if ( $isLocalObj ) {
3013                         $ret = array( 'object' => $text );
3014                 } else {
3015                         $ret = array( 'text' => $text );
3016                 }
3017
3018                 wfProfileOut( $fname );
3019                 return $ret;
3020         }
3021
3022         /**
3023          * Get the semi-parsed DOM representation of a template with a given title,
3024          * and its redirect destination title. Cached.
3025          */
3026         function getTemplateDom( $title ) {
3027                 $cacheTitle = $title;
3028                 $titleText = $title->getPrefixedDBkey();
3029
3030                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3031                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3032                         $title = Title::makeTitle( $ns, $dbk );
3033                         $titleText = $title->getPrefixedDBkey();
3034                 }
3035                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3036                         return array( $this->mTplDomCache[$titleText], $title );
3037                 }
3038
3039                 // Cache miss, go to the database
3040                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3041
3042                 if ( $text === false ) {
3043                         $this->mTplDomCache[$titleText] = false;
3044                         return array( false, $title );
3045                 }
3046
3047                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3048                 $this->mTplDomCache[ $titleText ] = $dom;
3049
3050                 if (! $title->equals($cacheTitle)) {
3051                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3052                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3053                 }
3054
3055                 return array( $dom, $title );
3056         }
3057
3058         /**
3059          * Fetch the unparsed text of a template and register a reference to it.
3060          */
3061         function fetchTemplateAndTitle( $title ) {
3062                 $templateCb = $this->mOptions->getTemplateCallback();
3063                 $stuff = call_user_func( $templateCb, $title, $this );
3064                 $text = $stuff['text'];
3065                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3066                 if ( isset( $stuff['deps'] ) ) {
3067                         foreach ( $stuff['deps'] as $dep ) {
3068                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3069                         }
3070                 }
3071                 return array($text,$finalTitle);
3072         }
3073
3074         function fetchTemplate( $title ) {
3075                 $rv = $this->fetchTemplateAndTitle($title);
3076                 return $rv[0];
3077         }
3078
3079         /**
3080          * Static function to get a template
3081          * Can be overridden via ParserOptions::setTemplateCallback().
3082          */
3083         static function statelessFetchTemplate( $title, $parser=false ) {
3084                 $text = $skip = false;
3085                 $finalTitle = $title;
3086                 $deps = array();
3087
3088                 // Loop to fetch the article, with up to 1 redirect
3089                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3090                         # Give extensions a chance to select the revision instead
3091                         $id = false; // Assume current
3092                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3093
3094                         if( $skip ) {
3095                                 $text = false;
3096                                 $deps[] = array(
3097                                         'title' => $title,
3098                                         'page_id' => $title->getArticleID(),
3099                                         'rev_id' => null );
3100                                 break;
3101                         }
3102                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3103                         $rev_id = $rev ? $rev->getId() : 0;
3104                         // If there is no current revision, there is no page
3105                         if( $id === false && !$rev ) {
3106                                 $linkCache = LinkCache::singleton();
3107                                 $linkCache->addBadLinkObj( $title );
3108                         }
3109
3110                         $deps[] = array(
3111                                 'title' => $title,
3112                                 'page_id' => $title->getArticleID(),
3113                                 'rev_id' => $rev_id );
3114
3115                         if( $rev ) {
3116                                 $text = $rev->getText();
3117                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3118                                 global $wgLang;
3119                                 $message = $wgLang->lcfirst( $title->getText() );
3120                                 $text = wfMsgForContentNoTrans( $message );
3121                                 if( wfEmptyMsg( $message, $text ) ) {
3122                                         $text = false;
3123                                         break;
3124                                 }
3125                         } else {
3126                                 break;
3127                         }
3128                         if ( $text === false ) {
3129                                 break;
3130                         }
3131                         // Redirect?
3132                         $finalTitle = $title;
3133                         $title = Title::newFromRedirect( $text );
3134                 }
3135                 return array(
3136                         'text' => $text,
3137                         'finalTitle' => $finalTitle,
3138                         'deps' => $deps );
3139         }
3140
3141         /**
3142          * Transclude an interwiki link.
3143          */
3144         function interwikiTransclude( $title, $action ) {
3145                 global $wgEnableScaryTranscluding;
3146
3147                 if (!$wgEnableScaryTranscluding)
3148                         return wfMsg('scarytranscludedisabled');
3149
3150                 $url = $title->getFullUrl( "action=$action" );
3151
3152                 if (strlen($url) > 255)
3153                         return wfMsg('scarytranscludetoolong');
3154                 return $this->fetchScaryTemplateMaybeFromCache($url);
3155         }
3156
3157         function fetchScaryTemplateMaybeFromCache($url) {
3158                 global $wgTranscludeCacheExpiry;
3159                 $dbr = wfGetDB(DB_SLAVE);
3160                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3161                                 array('tc_url' => $url));
3162                 if ($obj) {
3163                         $time = $obj->tc_time;
3164                         $text = $obj->tc_contents;
3165                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3166                                 return $text;
3167                         }
3168                 }
3169
3170                 $text = Http::get($url);
3171                 if (!$text)
3172                         return wfMsg('scarytranscludefailed', $url);
3173
3174                 $dbw = wfGetDB(DB_MASTER);
3175                 $dbw->replace('transcache', array('tc_url'), array(
3176                         'tc_url' => $url,
3177                         'tc_time' => time(),
3178                         'tc_contents' => $text));
3179                 return $text;
3180         }
3181
3182
3183         /**
3184          * Triple brace replacement -- used for template arguments
3185          * @private
3186          */
3187         function argSubstitution( $piece, $frame ) {
3188                 wfProfileIn( __METHOD__ );
3189
3190                 $error = false;
3191                 $parts = $piece['parts'];
3192                 $nameWithSpaces = $frame->expand( $piece['title'] );
3193                 $argName = trim( $nameWithSpaces );
3194                 $object = false;
3195                 $text = $frame->getArgument( $argName );
3196                 if (  $text === false && $parts->getLength() > 0
3197                   && (
3198                     $this->ot['html']
3199                     || $this->ot['pre']
3200                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3201                   )
3202                 ) {
3203                         # No match in frame, use the supplied default
3204                         $object = $parts->item( 0 )->getChildren();
3205                 }
3206                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3207                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3208                         $this->limitationWarn( 'post-expand-template-argument' );
3209                 }
3210
3211                 if ( $text === false && $object === false ) {
3212                         # No match anywhere
3213                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3214                 }
3215                 if ( $error !== false ) {
3216                         $text .= $error;
3217                 }
3218                 if ( $object !== false ) {
3219                         $ret = array( 'object' => $object );
3220                 } else {
3221                         $ret = array( 'text' => $text );
3222                 }
3223
3224                 wfProfileOut( __METHOD__ );
3225                 return $ret;
3226         }
3227
3228         /**
3229          * Return the text to be used for a given extension tag.
3230          * This is the ghost of strip().
3231          *
3232          * @param array $params Associative array of parameters:
3233          *     name       PPNode for the tag name
3234          *     attr       PPNode for unparsed text where tag attributes are thought to be
3235          *     attributes Optional associative array of parsed attributes
3236          *     inner      Contents of extension element
3237          *     noClose    Original text did not have a close tag
3238          * @param PPFrame $frame
3239          */
3240         function extensionSubstitution( $params, $frame ) {
3241                 global $wgRawHtml, $wgContLang;
3242
3243                 $name = $frame->expand( $params['name'] );
3244                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3245                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3246
3247                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3248
3249                 if ( $this->ot['html'] ) {
3250                         $name = strtolower( $name );
3251
3252                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3253                         if ( isset( $params['attributes'] ) ) {
3254                                 $attributes = $attributes + $params['attributes'];
3255                         }
3256                         switch ( $name ) {
3257                                 case 'html':
3258                                         if( $wgRawHtml ) {
3259                                                 $output = $content;
3260                                                 break;
3261                                         } else {
3262                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3263                                         }
3264                                 case 'nowiki':
3265                                         $output = Xml::escapeTagsOnly( $content );
3266                                         break;
3267                                 case 'math':
3268                                         $output = $wgContLang->armourMath(
3269                                                 MathRenderer::renderMath( $content, $attributes ) );
3270                                         break;
3271                                 case 'gallery':
3272                                         $output = $this->renderImageGallery( $content, $attributes );
3273                                         break;
3274                                 default:
3275                                         if( isset( $this->mTagHooks[$name] ) ) {
3276                                                 # Workaround for PHP bug 35229 and similar
3277                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3278                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3279                                                 }
3280                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3281                                                         array( $content, $attributes, $this ) );
3282                                         } else {
3283                                                 $output = '<span class="error">Invalid tag extension name: ' .
3284                                                         htmlspecialchars( $name ) . '</span>';
3285                                         }
3286                         }
3287                 } else {
3288                         if ( is_null( $attrText ) ) {
3289                                 $attrText = '';
3290                         }
3291                         if ( isset( $params['attributes'] ) ) {
3292                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3293                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3294                                                 htmlspecialchars( $attrValue ) . '"';
3295                                 }
3296                         }
3297                         if ( $content === null ) {
3298                                 $output = "<$name$attrText/>";
3299                         } else {
3300                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3301                                 $output = "<$name$attrText>$content$close";
3302                         }
3303                 }
3304
3305                 if ( $name == 'html' || $name == 'nowiki' ) {
3306                         $this->mStripState->nowiki->setPair( $marker, $output );
3307                 } else {
3308                         $this->mStripState->general->setPair( $marker, $output );
3309                 }
3310                 return $marker;
3311         }
3312
3313         /**
3314          * Increment an include size counter
3315          *
3316          * @param string $type The type of expansion
3317          * @param integer $size The size of the text
3318          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3319          */
3320         function incrementIncludeSize( $type, $size ) {
3321                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3322                         return false;
3323                 } else {
3324                         $this->mIncludeSizes[$type] += $size;
3325                         return true;
3326                 }
3327         }
3328
3329         /**
3330          * Increment the expensive function count
3331          *
3332          * @return boolean False if the limit has been exceeded
3333          */
3334         function incrementExpensiveFunctionCount() {
3335                 global $wgExpensiveParserFunctionLimit;
3336                 $this->mExpensiveFunctionCount++;
3337                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3338                         return true;
3339                 }
3340                 return false;
3341         }
3342
3343         /**
3344          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3345          * Fills $this->mDoubleUnderscores, returns the modified text
3346          */
3347         function doDoubleUnderscore( $text ) {
3348                 // The position of __TOC__ needs to be recorded
3349                 $mw = MagicWord::get( 'toc' );
3350                 if( $mw->match( $text ) ) {
3351                         $this->mShowToc = true;
3352                         $this->mForceTocPosition = true;
3353
3354                         // Set a placeholder. At the end we'll fill it in with the TOC.
3355                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3356
3357                         // Only keep the first one.
3358                         $text = $mw->replace( '', $text );
3359                 }
3360
3361                 // Now match and remove the rest of them
3362                 $mwa = MagicWord::getDoubleUnderscoreArray();
3363                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3364
3365                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3366                         $this->mOutput->mNoGallery = true;
3367                 }
3368                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3369                         $this->mShowToc = false;
3370                 }
3371                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3372                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3373
3374                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3375                         if ( $containerCategory ) {
3376                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3377                         } else {
3378                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3379                         }
3380                 }
3381                 return $text;
3382         }
3383
3384         /**
3385          * This function accomplishes several tasks:
3386          * 1) Auto-number headings if that option is enabled
3387          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3388          * 3) Add a Table of contents on the top for users who have enabled the option
3389          * 4) Auto-anchor headings
3390          *
3391          * It loops through all headlines, collects the necessary data, then splits up the
3392          * string and re-inserts the newly formatted headlines.
3393          *
3394          * @param string $text
3395          * @param boolean $isMain
3396          * @private
3397          */
3398         function formatHeadings( $text, $isMain=true ) {
3399                 global $wgMaxTocLevel, $wgContLang;
3400
3401                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3402                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3403                         $showEditLink = 0;
3404                 } else {
3405                         $showEditLink = $this->mOptions->getEditSection();
3406                 }
3407
3408                 # Inhibit editsection links if requested in the page
3409                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3410                         $showEditLink = 0;
3411                 }
3412
3413                 # Get all headlines for numbering them and adding funky stuff like [edit]
3414                 # links - this is for later, but we need the number of headlines right now
3415                 $matches = array();
3416                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3417
3418                 # if there are fewer than 4 headlines in the article, do not show TOC
3419                 # unless it's been explicitly enabled.
3420                 $enoughToc = $this->mShowToc &&
3421                         (($numMatches >= 4) || $this->mForceTocPosition);
3422
3423                 # Allow user to stipulate that a page should have a "new section"
3424                 # link added via __NEWSECTIONLINK__
3425                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3426                         $this->mOutput->setNewSection( true );
3427                 }
3428
3429                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3430                 # override above conditions and always show TOC above first header
3431                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3432                         $this->mShowToc = true;
3433                         $enoughToc = true;
3434                 }
3435
3436                 # We need this to perform operations on the HTML
3437                 $sk = $this->mOptions->getSkin();
3438
3439                 # headline counter
3440                 $headlineCount = 0;
3441                 $numVisible = 0;
3442
3443                 # Ugh .. the TOC should have neat indentation levels which can be
3444                 # passed to the skin functions. These are determined here
3445                 $toc = '';
3446                 $full = '';
3447                 $head = array();
3448                 $sublevelCount = array();
3449                 $levelCount = array();
3450                 $toclevel = 0;
3451                 $level = 0;
3452                 $prevlevel = 0;
3453                 $toclevel = 0;
3454                 $prevtoclevel = 0;
3455                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3456                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3457                 $tocraw = array();
3458
3459                 foreach( $matches[3] as $headline ) {
3460                         $isTemplate = false;
3461                         $titleText = false;
3462                         $sectionIndex = false;
3463                         $numbering = '';
3464                         $markerMatches = array();
3465                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3466                                 $serial = $markerMatches[1];
3467                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3468                                 $isTemplate = ($titleText != $baseTitleText);
3469                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3470                         }
3471
3472                         if( $toclevel ) {
3473                                 $prevlevel = $level;
3474                                 $prevtoclevel = $toclevel;
3475                         }
3476                         $level = $matches[1][$headlineCount];
3477
3478                         if( $doNumberHeadings || $enoughToc ) {
3479
3480                                 if ( $level > $prevlevel ) {
3481                                         # Increase TOC level
3482                                         $toclevel++;
3483                                         $sublevelCount[$toclevel] = 0;
3484                                         if( $toclevel<$wgMaxTocLevel ) {
3485                                                 $prevtoclevel = $toclevel;
3486                                                 $toc .= $sk->tocIndent();
3487                                                 $numVisible++;
3488                                         }
3489                                 }
3490                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3491                                         # Decrease TOC level, find level to jump to
3492
3493                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3494                                                 # Can only go down to level 1
3495                                                 $toclevel = 1;
3496                                         } else {
3497                                                 for ($i = $toclevel; $i > 0; $i--) {
3498                                                         if ( $levelCount[$i] == $level ) {
3499                                                                 # Found last matching level
3500                                                                 $toclevel = $i;
3501                                                                 break;
3502                                                         }
3503                                                         elseif ( $levelCount[$i] < $level ) {
3504                                                                 # Found first matching level below current level
3505                                                                 $toclevel = $i + 1;
3506                                                                 break;
3507                                                         }
3508                                                 }
3509                                         }
3510                                         if( $toclevel<$wgMaxTocLevel ) {
3511                                                 if($prevtoclevel < $wgMaxTocLevel) {
3512                                                         # Unindent only if the previous toc level was shown :p
3513                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3514                                                         $prevtoclevel = $toclevel;
3515                                                 } else {
3516                                                         $toc .= $sk->tocLineEnd();
3517                                                 }
3518                                         }
3519                                 }
3520                                 else {
3521                                         # No change in level, end TOC line
3522                                         if( $toclevel<$wgMaxTocLevel ) {
3523                                                 $toc .= $sk->tocLineEnd();
3524                                         }
3525                                 }
3526
3527                                 $levelCount[$toclevel] = $level;
3528
3529                                 # count number of headlines for each level
3530                                 @$sublevelCount[$toclevel]++;
3531                                 $dot = 0;
3532                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3533                                         if( !empty( $sublevelCount[$i] ) ) {
3534                                                 if( $dot ) {
3535                                                         $numbering .= '.';
3536                                                 }
3537                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3538                                                 $dot = 1;
3539                                         }
3540                                 }
3541                         }
3542
3543                         # The safe header is a version of the header text safe to use for links
3544                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3545                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3546
3547                         # Remove link placeholders by the link text.
3548                         #     <!--LINK number-->
3549                         # turns into
3550                         #     link text with suffix
3551                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3552                                                             "\$this->mLinkHolders['texts'][\$1]",
3553                                                             $safeHeadline );
3554                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3555                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3556                                                             $safeHeadline );
3557
3558                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3559                         $tocline = preg_replace(
3560                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3561                                 array( '',                          '<$1>'),
3562                                 $safeHeadline
3563                         );
3564                         $tocline = trim( $tocline );
3565
3566                         # For the anchor, strip out HTML-y stuff period
3567                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3568                         $safeHeadline = trim( $safeHeadline );
3569
3570                         # Save headline for section edit hint before it's escaped
3571                         $headlineHint = $safeHeadline;
3572                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3573                         # HTML names must be case-insensitively unique (bug 10721)
3574                         $arrayKey = strtolower( $safeHeadline );
3575
3576                         # count how many in assoc. array so we can track dupes in anchors
3577                         isset( $refers[$arrayKey] ) ? $refers[$arrayKey]++ : $refers[$arrayKey] = 1;
3578                         $refcount[$headlineCount] = $refers[$arrayKey];
3579
3580                         # Don't number the heading if it is the only one (looks silly)
3581                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3582                                 # the two are different if the line contains a link
3583                                 $headline=$numbering . ' ' . $headline;
3584                         }
3585
3586                         # Create the anchor for linking from the TOC to the section
3587                         $anchor = $safeHeadline;
3588                         if($refcount[$headlineCount] > 1 ) {
3589                                 $anchor .= '_' . $refcount[$headlineCount];
3590                         }
3591                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3592                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3593                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3594                         }
3595                         # give headline the correct <h#> tag
3596                         if( $showEditLink && $sectionIndex !== false ) {
3597                                 if( $isTemplate ) {
3598                                         # Put a T flag in the section identifier, to indicate to extractSections()
3599                                         # that sections inside <includeonly> should be counted.
3600                                         $editlink = $sk->editSectionLinkForOther($titleText, "T-$sectionIndex");
3601                                 } else {
3602                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3603                                 }
3604                         } else {
3605                                 $editlink = '';
3606                         }
3607                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3608
3609                         $headlineCount++;
3610                 }
3611
3612                 $this->mOutput->setSections( $tocraw );
3613
3614                 # Never ever show TOC if no headers
3615                 if( $numVisible < 1 ) {
3616                         $enoughToc = false;
3617                 }
3618
3619                 if( $enoughToc ) {
3620                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3621                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3622                         }
3623                         $toc = $sk->tocList( $toc );
3624                 }
3625
3626                 # split up and insert constructed headlines
3627
3628                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3629                 $i = 0;
3630
3631                 foreach( $blocks as $block ) {
3632                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3633                                 # This is the [edit] link that appears for the top block of text when
3634                                 # section editing is enabled
3635
3636                                 # Disabled because it broke block formatting
3637                                 # For example, a bullet point in the top line
3638                                 # $full .= $sk->editSectionLink(0);
3639                         }
3640                         $full .= $block;
3641                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3642                                 # Top anchor now in skin
3643                                 $full = $full.$toc;
3644                         }
3645
3646                         if( !empty( $head[$i] ) ) {
3647                                 $full .= $head[$i];
3648                         }
3649                         $i++;
3650                 }
3651                 if( $this->mForceTocPosition ) {
3652                         return str_replace( '<!--MWTOC-->', $toc, $full );
3653                 } else {
3654                         return $full;
3655                 }
3656         }
3657
3658         /**
3659          * Transform wiki markup when saving a page by doing \r\n -> \n
3660          * conversion, substitting signatures, {{subst:}} templates, etc.
3661          *
3662          * @param string $text the text to transform
3663          * @param Title &$title the Title object for the current article
3664          * @param User &$user the User object describing the current user
3665          * @param ParserOptions $options parsing options
3666          * @param bool $clearState whether to clear the parser state first
3667          * @return string the altered wiki markup
3668          * @public
3669          */
3670         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3671                 $this->mOptions = $options;
3672                 $this->setTitle( $title );
3673                 $this->setOutputType( self::OT_WIKI );
3674
3675                 if ( $clearState ) {
3676                         $this->clearState();
3677                 }
3678
3679                 $pairs = array(
3680                         "\r\n" => "\n",
3681                 );
3682                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3683                 $text = $this->pstPass2( $text, $user );
3684                 $text = $this->mStripState->unstripBoth( $text );
3685                 return $text;
3686         }
3687
3688         /**
3689          * Pre-save transform helper function
3690          * @private
3691          */
3692         function pstPass2( $text, $user ) {
3693                 global $wgContLang, $wgLocaltimezone;
3694
3695                 /* Note: This is the timestamp saved as hardcoded wikitext to
3696                  * the database, we use $wgContLang here in order to give
3697                  * everyone the same signature and use the default one rather
3698                  * than the one selected in each user's preferences.
3699                  *
3700                  * (see also bug 12815)
3701                  */
3702                 $ts = $this->mOptions->getTimestamp();
3703                 $tz = wfMsgForContent( 'timezone-utc' );
3704                 if ( isset( $wgLocaltimezone ) ) {
3705                         $unixts = wfTimestamp( TS_UNIX, $ts );
3706                         $oldtz = getenv( 'TZ' );
3707                         putenv( 'TZ='.$wgLocaltimezone );
3708                         $ts = date( 'YmdHis', $unixts );
3709                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3710                         putenv( 'TZ='.$oldtz );
3711                 }
3712
3713                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3714
3715                 # Variable replacement
3716                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3717                 $text = $this->replaceVariables( $text );
3718
3719                 # Signatures
3720                 $sigText = $this->getUserSig( $user );
3721                 $text = strtr( $text, array(
3722                         '~~~~~' => $d,
3723                         '~~~~' => "$sigText $d",
3724                         '~~~' => $sigText
3725                 ) );
3726
3727                 # Context links: [[|name]] and [[name (context)|]]
3728                 #
3729                 global $wgLegalTitleChars;
3730                 $tc = "[$wgLegalTitleChars]";
3731                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3732
3733                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3734                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3735                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3736
3737                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3738                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3739                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3740
3741                 $t = $this->mTitle->getText();
3742                 $m = array();
3743                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3744                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3745                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3746                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3747                 } else {
3748                         # if there's no context, don't bother duplicating the title
3749                         $text = preg_replace( $p2, '[[\\1]]', $text );
3750                 }
3751
3752                 # Trim trailing whitespace
3753                 $text = rtrim( $text );
3754
3755                 return $text;
3756         }
3757
3758         /**
3759          * Fetch the user's signature text, if any, and normalize to
3760          * validated, ready-to-insert wikitext.
3761          *
3762          * @param User $user
3763          * @return string
3764          * @private
3765          */
3766         function getUserSig( &$user ) {
3767                 global $wgMaxSigChars;
3768
3769                 $username = $user->getName();
3770                 $nickname = $user->getOption( 'nickname' );
3771                 $nickname = $nickname === '' ? $username : $nickname;
3772
3773                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3774                         $nickname = $username;
3775                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3776                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3777                         # Sig. might contain markup; validate this
3778                         if( $this->validateSig( $nickname ) !== false ) {
3779                                 # Validated; clean up (if needed) and return it
3780                                 return $this->cleanSig( $nickname, true );
3781                         } else {
3782                                 # Failed to validate; fall back to the default
3783                                 $nickname = $username;
3784                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3785                         }
3786                 }
3787
3788                 // Make sure nickname doesnt get a sig in a sig
3789                 $nickname = $this->cleanSigInSig( $nickname );
3790
3791                 # If we're still here, make it a link to the user page
3792                 $userText = wfEscapeWikiText( $username );
3793                 $nickText = wfEscapeWikiText( $nickname );
3794                 if ( $user->isAnon() )  {
3795                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3796                 } else {
3797                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3798                 }
3799         }
3800
3801         /**
3802          * Check that the user's signature contains no bad XML
3803          *
3804          * @param string $text
3805          * @return mixed An expanded string, or false if invalid.
3806          */
3807         function validateSig( $text ) {
3808                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3809         }
3810
3811         /**
3812          * Clean up signature text
3813          *
3814          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3815          * 2) Substitute all transclusions
3816          *
3817          * @param string $text
3818          * @param $parsing Whether we're cleaning (preferences save) or parsing
3819          * @return string Signature text
3820          */
3821         function cleanSig( $text, $parsing = false ) {
3822                 if ( !$parsing ) {
3823                         global $wgTitle;
3824                         $this->clearState();
3825                         $this->setTitle( $wgTitle );
3826                         $this->mOptions = new ParserOptions;
3827                         $this->setOutputType = self::OT_PREPROCESS;
3828                 }
3829
3830                 # FIXME: regex doesn't respect extension tags or nowiki
3831                 #  => Move this logic to braceSubstitution()
3832                 $substWord = MagicWord::get( 'subst' );
3833                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3834                 $substText = '{{' . $substWord->getSynonym( 0 );
3835
3836                 $text = preg_replace( $substRegex, $substText, $text );
3837                 $text = $this->cleanSigInSig( $text );
3838                 $dom = $this->preprocessToDom( $text );
3839                 $frame = $this->getPreprocessor()->newFrame();
3840                 $text = $frame->expand( $dom );
3841
3842                 if ( !$parsing ) {
3843                         $text = $this->mStripState->unstripBoth( $text );
3844                 }
3845
3846                 return $text;
3847         }
3848
3849         /**
3850          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3851          * @param string $text
3852          * @return string Signature text with /~{3,5}/ removed
3853          */
3854         function cleanSigInSig( $text ) {
3855                 $text = preg_replace( '/~{3,5}/', '', $text );
3856                 return $text;
3857         }
3858
3859         /**
3860          * Set up some variables which are usually set up in parse()
3861          * so that an external function can call some class members with confidence
3862          * @public
3863          */
3864         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3865                 $this->setTitle( $title );
3866                 $this->mOptions = $options;
3867                 $this->setOutputType( $outputType );
3868                 if ( $clearState ) {
3869                         $this->clearState();
3870                 }
3871         }
3872
3873         /**
3874          * Wrapper for preprocess()
3875          *
3876          * @param string $text the text to preprocess
3877          * @param ParserOptions $options  options
3878          * @return string
3879          * @public
3880          */
3881         function transformMsg( $text, $options ) {
3882                 global $wgTitle;
3883                 static $executing = false;
3884
3885                 $fname = "Parser::transformMsg";
3886
3887                 # Guard against infinite recursion
3888                 if ( $executing ) {
3889                         return $text;
3890                 }
3891                 $executing = true;
3892
3893                 wfProfileIn($fname);
3894                 $text = $this->preprocess( $text, $wgTitle, $options );
3895
3896                 $executing = false;
3897                 wfProfileOut($fname);
3898                 return $text;
3899         }
3900
3901         /**
3902          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3903          * The callback should have the following form:
3904          *    function myParserHook( $text, $params, &$parser ) { ... }
3905          *
3906          * Transform and return $text. Use $parser for any required context, e.g. use
3907          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3908          *
3909          * @public
3910          *
3911          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3912          * @param mixed $callback The callback function (and object) to use for the tag
3913          *
3914          * @return The old value of the mTagHooks array associated with the hook
3915          */
3916         function setHook( $tag, $callback ) {
3917                 $tag = strtolower( $tag );
3918                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3919                 $this->mTagHooks[$tag] = $callback;
3920                 if( !in_array( $tag, $this->mStripList ) ) {
3921                         $this->mStripList[] = $tag;
3922                 }
3923
3924                 return $oldVal;
3925         }
3926
3927         function setTransparentTagHook( $tag, $callback ) {
3928                 $tag = strtolower( $tag );
3929                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3930                 $this->mTransparentTagHooks[$tag] = $callback;
3931
3932                 return $oldVal;
3933         }
3934
3935         /**
3936          * Remove all tag hooks
3937          */
3938         function clearTagHooks() {
3939                 $this->mTagHooks = array();
3940                 $this->mStripList = $this->mDefaultStripList;
3941         }
3942
3943         /**
3944          * Create a function, e.g. {{sum:1|2|3}}
3945          * The callback function should have the form:
3946          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3947          *
3948          * The callback may either return the text result of the function, or an array with the text
3949          * in element 0, and a number of flags in the other elements. The names of the flags are
3950          * specified in the keys. Valid flags are:
3951          *   found                     The text returned is valid, stop processing the template. This
3952          *                             is on by default.
3953          *   nowiki                    Wiki markup in the return value should be escaped
3954          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3955          *
3956          * @public
3957          *
3958          * @param string $id The magic word ID
3959          * @param mixed $callback The callback function (and object) to use
3960          * @param integer $flags a combination of the following flags:
3961          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3962          *
3963          * @return The old callback function for this name, if any
3964          */
3965         function setFunctionHook( $id, $callback, $flags = 0 ) {
3966                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
3967                 $this->mFunctionHooks[$id] = array( $callback, $flags );
3968
3969                 # Add to function cache
3970                 $mw = MagicWord::get( $id );
3971                 if( !$mw )
3972                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3973
3974                 $synonyms = $mw->getSynonyms();
3975                 $sensitive = intval( $mw->isCaseSensitive() );
3976
3977                 foreach ( $synonyms as $syn ) {
3978                         # Case
3979                         if ( !$sensitive ) {
3980                                 $syn = strtolower( $syn );
3981                         }
3982                         # Add leading hash
3983                         if ( !( $flags & SFH_NO_HASH ) ) {
3984                                 $syn = '#' . $syn;
3985                         }
3986                         # Remove trailing colon
3987                         if ( substr( $syn, -1, 1 ) == ':' ) {
3988                                 $syn = substr( $syn, 0, -1 );
3989                         }
3990                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3991                 }
3992                 return $oldVal;
3993         }
3994
3995         /**
3996          * Get all registered function hook identifiers
3997          *
3998          * @return array
3999          */
4000         function getFunctionHooks() {
4001                 return array_keys( $this->mFunctionHooks );
4002         }
4003
4004         /**
4005          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4006          * Placeholders created in Skin::makeLinkObj()
4007          * Returns an array of link CSS classes, indexed by PDBK.
4008          * $options is a bit field, RLH_FOR_UPDATE to select for update
4009          */
4010         function replaceLinkHolders( &$text, $options = 0 ) {
4011                 global $wgUser;
4012                 global $wgContLang;
4013
4014                 $fname = 'Parser::replaceLinkHolders';
4015                 wfProfileIn( $fname );
4016
4017                 $pdbks = array();
4018                 $colours = array();
4019                 $linkcolour_ids = array();
4020                 $sk = $this->mOptions->getSkin();
4021                 $linkCache = LinkCache::singleton();
4022
4023                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4024                         wfProfileIn( $fname.'-check' );
4025                         $dbr = wfGetDB( DB_SLAVE );
4026                         $page = $dbr->tableName( 'page' );
4027                         $threshold = $wgUser->getOption('stubthreshold');
4028
4029                         # Sort by namespace
4030                         asort( $this->mLinkHolders['namespaces'] );
4031
4032                         # Generate query
4033                         $query = false;
4034                         $current = null;
4035                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4036                                 # Make title object
4037                                 $title = $this->mLinkHolders['titles'][$key];
4038
4039                                 # Skip invalid entries.
4040                                 # Result will be ugly, but prevents crash.
4041                                 if ( is_null( $title ) ) {
4042                                         continue;
4043                                 }
4044                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4045
4046                                 # Check if it's a static known link, e.g. interwiki
4047                                 if ( $title->isAlwaysKnown() ) {
4048                                         $colours[$pdbk] = '';
4049                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4050                                         $colours[$pdbk] = '';
4051                                         $this->mOutput->addLink( $title, $id );
4052                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4053                                         $colours[$pdbk] = 'new';
4054                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4055                                         $colours[$pdbk] = 'new';
4056                                 } else {
4057                                         # Not in the link cache, add it to the query
4058                                         if ( !isset( $current ) ) {
4059                                                 $current = $ns;
4060                                                 $query =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
4061                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4062                                         } elseif ( $current != $ns ) {
4063                                                 $current = $ns;
4064                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4065                                         } else {
4066                                                 $query .= ', ';
4067                                         }
4068
4069                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4070                                 }
4071                         }
4072                         if ( $query ) {
4073                                 $query .= '))';
4074                                 if ( $options & RLH_FOR_UPDATE ) {
4075                                         $query .= ' FOR UPDATE';
4076                                 }
4077
4078                                 $res = $dbr->query( $query, $fname );
4079
4080                                 # Fetch data and form into an associative array
4081                                 # non-existent = broken
4082                                 while ( $s = $dbr->fetchObject($res) ) {
4083                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4084                                         $pdbk = $title->getPrefixedDBkey();
4085                                         $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect );
4086                                         $this->mOutput->addLink( $title, $s->page_id );
4087                                         $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
4088                                         //add id to the extension todolist
4089                                         $linkcolour_ids[$s->page_id] = $pdbk;
4090                                 }
4091                                 //pass an array of page_ids to an extension
4092                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4093                         }
4094                         wfProfileOut( $fname.'-check' );
4095
4096                         # Do a second query for different language variants of links and categories
4097                         if($wgContLang->hasVariants()){
4098                                 $linkBatch = new LinkBatch();
4099                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4100                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4101                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4102
4103                                 $categories = $this->mOutput->getCategoryLinks();
4104
4105                                 // Add variants of links to link batch
4106                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4107                                         $title = $this->mLinkHolders['titles'][$key];
4108                                         if ( is_null( $title ) )
4109                                                 continue;
4110
4111                                         $pdbk = $title->getPrefixedDBkey();
4112                                         $titleText = $title->getText();
4113
4114                                         // generate all variants of the link title text
4115                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4116
4117                                         // if link was not found (in first query), add all variants to query
4118                                         if ( !isset($colours[$pdbk]) ){
4119                                                 foreach($allTextVariants as $textVariant){
4120                                                         if($textVariant != $titleText){
4121                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4122                                                                 if(is_null($variantTitle)) continue;
4123                                                                 $linkBatch->addObj( $variantTitle );
4124                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4125                                                         }
4126                                                 }
4127                                         }
4128                                 }
4129
4130                                 // process categories, check if a category exists in some variant
4131                                 foreach( $categories as $category ){
4132                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4133                                         foreach($variants as $variant){
4134                                                 if($variant != $category){
4135                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4136                                                         if(is_null($variantTitle)) continue;
4137                                                         $linkBatch->addObj( $variantTitle );
4138                                                         $categoryMap[$variant] = $category;
4139                                                 }
4140                                         }
4141                                 }
4142
4143
4144                                 if(!$linkBatch->isEmpty()){
4145                                         // construct query
4146                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4147
4148                                         $variantQuery =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
4149
4150                                         $variantQuery .= " FROM $page WHERE $titleClause";
4151                                         if ( $options & RLH_FOR_UPDATE ) {
4152                                                 $variantQuery .= ' FOR UPDATE';
4153                                         }
4154
4155                                         $varRes = $dbr->query( $variantQuery, $fname );
4156
4157                                         // for each found variants, figure out link holders and replace
4158                                         while ( $s = $dbr->fetchObject($varRes) ) {
4159
4160                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4161                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4162                                                 $vardbk = $variantTitle->getDBkey();
4163
4164                                                 $holderKeys = array();
4165                                                 if(isset($variantMap[$varPdbk])){
4166                                                         $holderKeys = $variantMap[$varPdbk];
4167                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
4168                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4169                                                 }
4170
4171                                                 // loop over link holders
4172                                                 foreach($holderKeys as $key){
4173                                                         $title = $this->mLinkHolders['titles'][$key];
4174                                                         if ( is_null( $title ) ) continue;
4175
4176                                                         $pdbk = $title->getPrefixedDBkey();
4177
4178                                                         if(!isset($colours[$pdbk])){
4179                                                                 // found link in some of the variants, replace the link holder data
4180                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4181                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4182
4183                                                                 // set pdbk and colour
4184                                                                 $pdbks[$key] = $varPdbk;
4185                                                                 $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );
4186                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4187                                                         }
4188                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4189                                                 }
4190
4191                                                 // check if the object is a variant of a category
4192                                                 if(isset($categoryMap[$vardbk])){
4193                                                         $oldkey = $categoryMap[$vardbk];
4194                                                         if($oldkey != $vardbk)
4195                                                                 $varCategories[$oldkey]=$vardbk;
4196                                                 }
4197                                         }
4198
4199                                         // rebuild the categories in original order (if there are replacements)
4200                                         if(count($varCategories)>0){
4201                                                 $newCats = array();
4202                                                 $originalCats = $this->mOutput->getCategories();
4203                                                 foreach($originalCats as $cat => $sortkey){
4204                                                         // make the replacement
4205                                                         if( array_key_exists($cat,$varCategories) )
4206                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4207                                                         else $newCats[$cat] = $sortkey;
4208                                                 }
4209                                                 $this->mOutput->setCategoryLinks($newCats);
4210                                         }
4211                                 }
4212                         }
4213
4214                         # Construct search and replace arrays
4215                         wfProfileIn( $fname.'-construct' );
4216                         $replacePairs = array();
4217                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4218                                 $pdbk = $pdbks[$key];
4219                                 $searchkey = "<!--LINK $key-->";
4220                                 $title = $this->mLinkHolders['titles'][$key];
4221                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4222                                         $linkCache->addBadLinkObj( $title );
4223                                         $colours[$pdbk] = 'new';
4224                                         $this->mOutput->addLink( $title, 0 );
4225                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4226                                                                         $this->mLinkHolders['texts'][$key],
4227                                                                         $this->mLinkHolders['queries'][$key] );
4228                                 } else {
4229                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4230                                                                         $this->mLinkHolders['texts'][$key],
4231                                                                         $this->mLinkHolders['queries'][$key] );
4232                                 }
4233                         }
4234                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4235                         wfProfileOut( $fname.'-construct' );
4236
4237                         # Do the thing
4238                         wfProfileIn( $fname.'-replace' );
4239                         $text = preg_replace_callback(
4240                                 '/(<!--LINK .*?-->)/',
4241                                 $replacer->cb(),
4242                                 $text);
4243
4244                         wfProfileOut( $fname.'-replace' );
4245                 }
4246
4247                 # Now process interwiki link holders
4248                 # This is quite a bit simpler than internal links
4249                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4250                         wfProfileIn( $fname.'-interwiki' );
4251                         # Make interwiki link HTML
4252                         $replacePairs = array();
4253                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4254                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4255                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4256                         }
4257                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4258
4259                         $text = preg_replace_callback(
4260                                 '/<!--IWLINK (.*?)-->/',
4261                                 $replacer->cb(),
4262                                 $text );
4263                         wfProfileOut( $fname.'-interwiki' );
4264                 }
4265
4266                 wfProfileOut( $fname );
4267                 return $colours;
4268         }
4269
4270         /**
4271          * Replace <!--LINK--> link placeholders with plain text of links
4272          * (not HTML-formatted).
4273          * @param string $text
4274          * @return string
4275          */
4276         function replaceLinkHoldersText( $text ) {
4277                 $fname = 'Parser::replaceLinkHoldersText';
4278                 wfProfileIn( $fname );
4279
4280                 $text = preg_replace_callback(
4281                         '/<!--(LINK|IWLINK) (.*?)-->/',
4282                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4283                         $text );
4284
4285                 wfProfileOut( $fname );
4286                 return $text;
4287         }
4288
4289         /**
4290          * @param array $matches
4291          * @return string
4292          * @private
4293          */
4294         function replaceLinkHoldersTextCallback( $matches ) {
4295                 $type = $matches[1];
4296                 $key  = $matches[2];
4297                 if( $type == 'LINK' ) {
4298                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4299                                 return $this->mLinkHolders['texts'][$key];
4300                         }
4301                 } elseif( $type == 'IWLINK' ) {
4302                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4303                                 return $this->mInterwikiLinkHolders['texts'][$key];
4304                         }
4305                 }
4306                 return $matches[0];
4307         }
4308
4309         /**
4310          * Tag hook handler for 'pre'.
4311          */
4312         function renderPreTag( $text, $attribs ) {
4313                 // Backwards-compatibility hack
4314                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4315
4316                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4317                 return wfOpenElement( 'pre', $attribs ) .
4318                         Xml::escapeTagsOnly( $content ) .
4319                         '</pre>';
4320         }
4321
4322         /**
4323          * Renders an image gallery from a text with one line per image.
4324          * text labels may be given by using |-style alternative text. E.g.
4325          *   Image:one.jpg|The number "1"
4326          *   Image:tree.jpg|A tree
4327          * given as text will return the HTML of a gallery with two images,
4328          * labeled 'The number "1"' and
4329          * 'A tree'.
4330          */
4331         function renderImageGallery( $text, $params ) {
4332                 $ig = new ImageGallery();
4333                 $ig->setContextTitle( $this->mTitle );
4334                 $ig->setShowBytes( false );
4335                 $ig->setShowFilename( false );
4336                 $ig->setParser( $this );
4337                 $ig->setHideBadImages();
4338                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4339                 $ig->useSkin( $this->mOptions->getSkin() );
4340                 $ig->mRevisionId = $this->mRevisionId;
4341
4342                 if( isset( $params['caption'] ) ) {
4343                         $caption = $params['caption'];
4344                         $caption = htmlspecialchars( $caption );
4345                         $caption = $this->replaceInternalLinks( $caption );
4346                         $ig->setCaptionHtml( $caption );
4347                 }
4348                 if( isset( $params['perrow'] ) ) {
4349                         $ig->setPerRow( $params['perrow'] );
4350                 }
4351                 if( isset( $params['widths'] ) ) {
4352                         $ig->setWidths( $params['widths'] );
4353                 }
4354                 if( isset( $params['heights'] ) ) {
4355                         $ig->setHeights( $params['heights'] );
4356                 }
4357
4358                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4359
4360                 $lines = explode( "\n", $text );
4361                 foreach ( $lines as $line ) {
4362                         # match lines like these:
4363                         # Image:someimage.jpg|This is some image
4364                         $matches = array();
4365                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4366                         # Skip empty lines
4367                         if ( count( $matches ) == 0 ) {
4368                                 continue;
4369                         }
4370
4371                         if ( strpos( $matches[0], '%' ) !== false )
4372                                 $matches[1] = urldecode( $matches[1] );
4373                         $tp = Title::newFromText( $matches[1] );
4374                         $nt =& $tp;
4375                         if( is_null( $nt ) ) {
4376                                 # Bogus title. Ignore these so we don't bomb out later.
4377                                 continue;
4378                         }
4379                         if ( isset( $matches[3] ) ) {
4380                                 $label = $matches[3];
4381                         } else {
4382                                 $label = '';
4383                         }
4384
4385                         $html = $this->recursiveTagParse( trim( $label ) );
4386
4387                         $ig->add( $nt, $html );
4388
4389                         # Only add real images (bug #5586)
4390                         if ( $nt->getNamespace() == NS_IMAGE ) {
4391                                 $this->mOutput->addImage( $nt->getDBkey() );
4392                         }
4393                 }
4394                 return $ig->toHTML();
4395         }
4396
4397         function getImageParams( $handler ) {
4398                 if ( $handler ) {
4399                         $handlerClass = get_class( $handler );
4400                 } else {
4401                         $handlerClass = '';
4402                 }
4403                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4404                         // Initialise static lists
4405                         static $internalParamNames = array(
4406                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4407                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4408                                         'bottom', 'text-bottom' ),
4409                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4410                                         'upright', 'border' ),
4411                         );
4412                         static $internalParamMap;
4413                         if ( !$internalParamMap ) {
4414                                 $internalParamMap = array();
4415                                 foreach ( $internalParamNames as $type => $names ) {
4416                                         foreach ( $names as $name ) {
4417                                                 $magicName = str_replace( '-', '_', "img_$name" );
4418                                                 $internalParamMap[$magicName] = array( $type, $name );
4419                                         }
4420                                 }
4421                         }
4422
4423                         // Add handler params
4424                         $paramMap = $internalParamMap;
4425                         if ( $handler ) {
4426                                 $handlerParamMap = $handler->getParamMap();
4427                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4428                                         $paramMap[$magic] = array( 'handler', $paramName );
4429                                 }
4430                         }
4431                         $this->mImageParams[$handlerClass] = $paramMap;
4432                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4433                 }
4434                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4435         }
4436
4437         /**
4438          * Parse image options text and use it to make an image
4439          */
4440         function makeImage( $title, $options ) {
4441                 # Check if the options text is of the form "options|alt text"
4442                 # Options are:
4443                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4444                 #  * left               no resizing, just left align. label is used for alt= only
4445                 #  * right              same, but right aligned
4446                 #  * none               same, but not aligned
4447                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4448                 #  * center             center the image
4449                 #  * framed             Keep original image size, no magnify-button.
4450                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4451                 #  * upright            reduce width for upright images, rounded to full __0 px
4452                 #  * border             draw a 1px border around the image
4453                 # vertical-align values (no % or length right now):
4454                 #  * baseline
4455                 #  * sub
4456                 #  * super
4457                 #  * top
4458                 #  * text-top
4459                 #  * middle
4460                 #  * bottom
4461                 #  * text-bottom
4462
4463                 $parts = array_map( 'trim', explode( '|', $options) );
4464                 $sk = $this->mOptions->getSkin();
4465
4466                 # Give extensions a chance to select the file revision for us
4467                 $skip = $time = $descQuery = false;
4468                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4469
4470                 if ( $skip ) {
4471                         return $sk->makeLinkObj( $title );
4472                 }
4473
4474                 # Get parameter map
4475                 $file = wfFindFile( $title, $time );
4476                 $handler = $file ? $file->getHandler() : false;
4477
4478                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4479
4480                 # Process the input parameters
4481                 $caption = '';
4482                 $params = array( 'frame' => array(), 'handler' => array(),
4483                         'horizAlign' => array(), 'vertAlign' => array() );
4484                 foreach( $parts as $part ) {
4485                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4486                         $validated = false;
4487                         if( isset( $paramMap[$magicName] ) ) {
4488                                 list( $type, $paramName ) = $paramMap[$magicName];
4489
4490                                 // Special case; width and height come in one variable together
4491                                 if( $type == 'handler' && $paramName == 'width' ) {
4492                                         $m = array();
4493                                         # (bug 13500) In both cases (width/height and width only),
4494                                         # permit trailing "px" for backward compatibility.
4495                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4496                                                 $width = intval( $m[1] );
4497                                                 $height = intval( $m[2] );
4498                                                 if ( $handler->validateParam( 'width', $width ) ) {
4499                                                         $params[$type]['width'] = $width;
4500                                                         $validated = true;
4501                                                 }
4502                                                 if ( $handler->validateParam( 'height', $height ) ) {
4503                                                         $params[$type]['height'] = $height;
4504                                                         $validated = true;
4505                                                 }
4506                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4507                                                 $width = intval( $value );
4508                                                 if ( $handler->validateParam( 'width', $width ) ) {
4509                                                         $params[$type]['width'] = $width;
4510                                                         $validated = true;
4511                                                 }
4512                                         } // else no validation -- bug 13436
4513                                 } else {
4514                                         if ( $type == 'handler' ) {
4515                                                 # Validate handler parameter
4516                                                 $validated = $handler->validateParam( $paramName, $value );
4517                                         } else {
4518                                                 # Validate internal parameters
4519                                                 switch( $paramName ) {
4520                                                 case "manualthumb":
4521                                                         /// @fixme - possibly check validity here?
4522                                                         /// downstream behavior seems odd with missing manual thumbs.
4523                                                         $validated = true;
4524                                                         break;
4525                                                 default:
4526                                                         // Most other things appear to be empty or numeric...
4527                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4528                                                 }
4529                                         }
4530
4531                                         if ( $validated ) {
4532                                                 $params[$type][$paramName] = $value;
4533                                         }
4534                                 }
4535                         }
4536                         if ( !$validated ) {
4537                                 $caption = $part;
4538                         }
4539                 }
4540
4541                 # Process alignment parameters
4542                 if ( $params['horizAlign'] ) {
4543                         $params['frame']['align'] = key( $params['horizAlign'] );
4544                 }
4545                 if ( $params['vertAlign'] ) {
4546                         $params['frame']['valign'] = key( $params['vertAlign'] );
4547                 }
4548
4549                 # Strip bad stuff out of the alt text
4550                 $alt = $this->replaceLinkHoldersText( $caption );
4551
4552                 # make sure there are no placeholders in thumbnail attributes
4553                 # that are later expanded to html- so expand them now and
4554                 # remove the tags
4555                 $alt = $this->mStripState->unstripBoth( $alt );
4556                 $alt = Sanitizer::stripAllTags( $alt );
4557
4558                 $params['frame']['alt'] = $alt;
4559                 $params['frame']['caption'] = $caption;
4560
4561                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4562
4563                 # Linker does the rest
4564                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4565
4566                 # Give the handler a chance to modify the parser object
4567                 if ( $handler ) {
4568                         $handler->parserTransformHook( $this, $file );
4569                 }
4570
4571                 return $ret;
4572         }
4573
4574         /**
4575          * Set a flag in the output object indicating that the content is dynamic and
4576          * shouldn't be cached.
4577          */
4578         function disableCache() {
4579                 wfDebug( "Parser output marked as uncacheable.\n" );
4580                 $this->mOutput->mCacheTime = -1;
4581         }
4582
4583         /**#@+
4584          * Callback from the Sanitizer for expanding items found in HTML attribute
4585          * values, so they can be safely tested and escaped.
4586          * @param string $text
4587          * @param PPFrame $frame
4588          * @return string
4589          * @private
4590          */
4591         function attributeStripCallback( &$text, $frame = false ) {
4592                 $text = $this->replaceVariables( $text, $frame );
4593                 $text = $this->mStripState->unstripBoth( $text );
4594                 return $text;
4595         }
4596
4597         /**#@-*/
4598
4599         /**#@+
4600          * Accessor/mutator
4601          */
4602         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4603         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4604         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4605         /**#@-*/
4606
4607         /**#@+
4608          * Accessor
4609          */
4610         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4611         /**#@-*/
4612
4613
4614         /**
4615          * Break wikitext input into sections, and either pull or replace
4616          * some particular section's text.
4617          *
4618          * External callers should use the getSection and replaceSection methods.
4619          *
4620          * @param string $text Page wikitext
4621          * @param string $section A section identifier string of the form:
4622          *   <flag1> - <flag2> - ... - <section number>
4623          *
4624          * Currently the only recognised flag is "T", which means the target section number
4625          * was derived during a template inclusion parse, in other words this is a template
4626          * section edit link. If no flags are given, it was an ordinary section edit link.
4627          * This flag is required to avoid a section numbering mismatch when a section is
4628          * enclosed by <includeonly> (bug 6563).
4629          *
4630          * The section number 0 pulls the text before the first heading; other numbers will
4631          * pull the given section along with its lower-level subsections. If the section is
4632          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4633          *
4634          * @param string $mode One of "get" or "replace"
4635          * @param string $newText Replacement text for section data.
4636          * @return string for "get", the extracted section text.
4637          *                for "replace", the whole page with the section replaced.
4638          */
4639         private function extractSections( $text, $section, $mode, $newText='' ) {
4640                 global $wgTitle;
4641                 $this->clearState();
4642                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4643                 $this->mOptions = new ParserOptions;
4644                 $this->setOutputType( self::OT_WIKI );
4645                 $outText = '';
4646                 $frame = $this->getPreprocessor()->newFrame();
4647
4648                 // Process section extraction flags
4649                 $flags = 0;
4650                 $sectionParts = explode( '-', $section );
4651                 $sectionIndex = array_pop( $sectionParts );
4652                 foreach ( $sectionParts as $part ) {
4653                         if ( $part == 'T' ) {
4654                                 $flags |= self::PTD_FOR_INCLUSION;
4655                         }
4656                 }
4657                 // Preprocess the text
4658                 $root = $this->preprocessToDom( $text, $flags );
4659
4660                 // <h> nodes indicate section breaks
4661                 // They can only occur at the top level, so we can find them by iterating the root's children
4662                 $node = $root->getFirstChild();
4663
4664                 // Find the target section
4665                 if ( $sectionIndex == 0 ) {
4666                         // Section zero doesn't nest, level=big
4667                         $targetLevel = 1000;
4668                 } else {
4669             while ( $node ) {
4670                 if ( $node->getName() == 'h' ) {
4671                     $bits = $node->splitHeading();
4672                                         if ( $bits['i'] == $sectionIndex ) {
4673                                         $targetLevel = $bits['level'];
4674                                                 break;
4675                                         }
4676                                 }
4677                                 if ( $mode == 'replace' ) {
4678                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4679                                 }
4680                                 $node = $node->getNextSibling();
4681                         }
4682                 }
4683
4684                 if ( !$node ) {
4685                         // Not found
4686                         if ( $mode == 'get' ) {
4687                                 return $newText;
4688                         } else {
4689                                 return $text;
4690                         }
4691                 }
4692
4693                 // Find the end of the section, including nested sections
4694                 do {
4695                         if ( $node->getName() == 'h' ) {
4696                                 $bits = $node->splitHeading();
4697                                 $curLevel = $bits['level'];
4698                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4699                                         break;
4700                                 }
4701                         }
4702                         if ( $mode == 'get' ) {
4703                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4704                         }
4705                         $node = $node->getNextSibling();
4706                 } while ( $node );
4707
4708                 // Write out the remainder (in replace mode only)
4709                 if ( $mode == 'replace' ) {
4710                         // Output the replacement text
4711                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4712                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4713                         $outText .= $newText . "\n\n";
4714                         while ( $node ) {
4715                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4716                                 $node = $node->getNextSibling();
4717                         }
4718                 }
4719
4720                 if ( is_string( $outText ) ) {
4721                         // Re-insert stripped tags
4722                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4723                 }
4724
4725                 return $outText;
4726         }
4727
4728         /**
4729          * This function returns the text of a section, specified by a number ($section).
4730          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4731          * the first section before any such heading (section 0).
4732          *
4733          * If a section contains subsections, these are also returned.
4734          *
4735          * @param string $text text to look in
4736          * @param string $section section identifier
4737          * @param string $deftext default to return if section is not found
4738          * @return string text of the requested section
4739          */
4740         public function getSection( $text, $section, $deftext='' ) {
4741                 return $this->extractSections( $text, $section, "get", $deftext );
4742         }
4743
4744         public function replaceSection( $oldtext, $section, $text ) {
4745                 return $this->extractSections( $oldtext, $section, "replace", $text );
4746         }
4747
4748         /**
4749          * Get the timestamp associated with the current revision, adjusted for
4750          * the default server-local timestamp
4751          */
4752         function getRevisionTimestamp() {
4753                 if ( is_null( $this->mRevisionTimestamp ) ) {
4754                         wfProfileIn( __METHOD__ );
4755                         global $wgContLang;
4756                         $dbr = wfGetDB( DB_SLAVE );
4757                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4758                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4759
4760                         // Normalize timestamp to internal MW format for timezone processing.
4761                         // This has the added side-effect of replacing a null value with
4762                         // the current time, which gives us more sensible behavior for
4763                         // previews.
4764                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4765
4766                         // The cryptic '' timezone parameter tells to use the site-default
4767                         // timezone offset instead of the user settings.
4768                         //
4769                         // Since this value will be saved into the parser cache, served
4770                         // to other users, and potentially even used inside links and such,
4771                         // it needs to be consistent for all visitors.
4772                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4773
4774                         wfProfileOut( __METHOD__ );
4775                 }
4776                 return $this->mRevisionTimestamp;
4777         }
4778
4779         /**
4780          * Mutator for $mDefaultSort
4781          *
4782          * @param $sort New value
4783          */
4784         public function setDefaultSort( $sort ) {
4785                 $this->mDefaultSort = $sort;
4786         }
4787
4788         /**
4789          * Accessor for $mDefaultSort
4790          * Will use the title/prefixed title if none is set
4791          *
4792          * @return string
4793          */
4794         public function getDefaultSort() {
4795                 if( $this->mDefaultSort !== false ) {
4796                         return $this->mDefaultSort;
4797                 } else {
4798                         return $this->mTitle->getNamespace() == NS_CATEGORY
4799                                         ? $this->mTitle->getText()
4800                                         : $this->mTitle->getPrefixedText();
4801                 }
4802         }
4803
4804         /**
4805          * Try to guess the section anchor name based on a wikitext fragment
4806          * presumably extracted from a heading, for example "Header" from
4807          * "== Header ==".
4808          */
4809         public function guessSectionNameFromWikiText( $text ) {
4810                 # Strip out wikitext links(they break the anchor)
4811                 $text = $this->stripSectionName( $text );
4812                 $headline = Sanitizer::decodeCharReferences( $text );
4813                 # strip out HTML
4814                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4815                 $headline = trim( $headline );
4816                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4817                 $replacearray = array(
4818                         '%3A' => ':',
4819                         '%' => '.'
4820                 );
4821                 return str_replace(
4822                         array_keys( $replacearray ),
4823                         array_values( $replacearray ),
4824                         $sectionanchor );
4825         }
4826
4827         /**
4828          * Strips a text string of wikitext for use in a section anchor
4829          *
4830          * Accepts a text string and then removes all wikitext from the
4831          * string and leaves only the resultant text (i.e. the result of
4832          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4833          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4834          * to create valid section anchors by mimicing the output of the
4835          * parser when headings are parsed.
4836          *
4837          * @param $text string Text string to be stripped of wikitext
4838          * for use in a Section anchor
4839          * @return Filtered text string
4840          */
4841         public function stripSectionName( $text ) {
4842                 # Strip internal link markup
4843                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4844                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4845
4846                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4847                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4848                 # on how many empty links there are on the page - need to figure that out.
4849                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4850
4851                 # Parse wikitext quotes (italics & bold)
4852                 $text = $this->doQuotes($text);
4853
4854                 # Strip HTML tags
4855                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4856                 return $text;
4857         }
4858
4859         function srvus( $text ) {
4860                 return $this->testSrvus( $text, $this->mOutputType );
4861         }
4862
4863         /**
4864          * strip/replaceVariables/unstrip for preprocessor regression testing
4865          */
4866         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4867                 $this->clearState();
4868                 if ( ! ( $title instanceof Title ) ) {
4869                         $title = Title::newFromText( $title );
4870                 }
4871                 $this->mTitle = $title;
4872                 $this->mOptions = $options;
4873                 $this->setOutputType( $outputType );
4874                 $text = $this->replaceVariables( $text );
4875                 $text = $this->mStripState->unstripBoth( $text );
4876                 $text = Sanitizer::removeHTMLtags( $text );
4877                 return $text;
4878         }
4879
4880         function testPst( $text, $title, $options ) {
4881                 global $wgUser;
4882                 if ( ! ( $title instanceof Title ) ) {
4883                         $title = Title::newFromText( $title );
4884                 }
4885                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4886         }
4887
4888         function testPreprocess( $text, $title, $options ) {
4889                 if ( ! ( $title instanceof Title ) ) {
4890                         $title = Title::newFromText( $title );
4891                 }
4892                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4893         }
4894
4895         function markerSkipCallback( $s, $callback ) {
4896                 $i = 0;
4897                 $out = '';
4898                 while ( $i < strlen( $s ) ) {
4899                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4900                         if ( $markerStart === false ) {
4901                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4902                                 break;
4903                         } else {
4904                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4905                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4906                                 if ( $markerEnd === false ) {
4907                                         $out .= substr( $s, $markerStart );
4908                                         break;
4909                                 } else {
4910                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4911                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4912                                         $i = $markerEnd;
4913                                 }
4914                         }
4915                 }
4916                 return $out;
4917         }
4918 }
4919
4920 /**
4921  * @todo document, briefly.
4922  * @ingroup Parser
4923  */
4924 class StripState {
4925         var $general, $nowiki;
4926
4927         function __construct() {
4928                 $this->general = new ReplacementArray;
4929                 $this->nowiki = new ReplacementArray;
4930         }
4931
4932         function unstripGeneral( $text ) {
4933                 wfProfileIn( __METHOD__ );
4934                 do {
4935                         $oldText = $text;
4936                         $text = $this->general->replace( $text );
4937                 } while ( $text != $oldText );
4938                 wfProfileOut( __METHOD__ );
4939                 return $text;
4940         }
4941
4942         function unstripNoWiki( $text ) {
4943                 wfProfileIn( __METHOD__ );
4944                 do {
4945                         $oldText = $text;
4946                         $text = $this->nowiki->replace( $text );
4947                 } while ( $text != $oldText );
4948                 wfProfileOut( __METHOD__ );
4949                 return $text;
4950         }
4951
4952         function unstripBoth( $text ) {
4953                 wfProfileIn( __METHOD__ );
4954                 do {
4955                         $oldText = $text;
4956                         $text = $this->general->replace( $text );
4957                         $text = $this->nowiki->replace( $text );
4958                 } while ( $text != $oldText );
4959                 wfProfileOut( __METHOD__ );
4960                 return $text;
4961         }
4962 }
4963
4964 /**
4965  * @todo document, briefly.
4966  * @ingroup Parser
4967  */
4968 class OnlyIncludeReplacer {
4969         var $output = '';
4970
4971         function replace( $matches ) {
4972                 if ( substr( $matches[1], -1 ) == "\n" ) {
4973                         $this->output .= substr( $matches[1], 0, -1 );
4974                 } else {
4975                         $this->output .= $matches[1];
4976                 }
4977         }
4978 }