includes/Parser.php

   1 <?php
   2
   3 /**
   4  *
   5  * File for Parser and related classes
   6  *
   7  * @addtogroup Parser
   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  *  $wgMaxArticleSize*
  40  *
  41  *  * only within ParserOptions
  42  * </pre>
  43  *
  44  * @addtogroup Parser
  45  */
  46 class Parser
  47 {
  48         /**
  49          * Update this version number when the ParserOutput format
  50          * changes in an incompatible way, so the parser cache
  51          * can automatically discard old data.
  52          */
  53         const VERSION = '1.6.4';
  54
  55         # Flags for Parser::setFunctionHook
  56         # Also available as global constants from Defines.php
  57         const SFH_NO_HASH = 1;
  58         const SFH_OBJECT_ARGS = 2;
  59
  60         # Constants needed for external link processing
  61         # Everything except bracket, space, or control characters
  62         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  63         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  64                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/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_DOM';
 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                         if ( is_callable( array( $this->mOutput, 'addWarning' ) ) ) {
 398                                 $warning = wfMsg( 'expensive-parserfunction-warning', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 399                                 $this->mOutput->addWarning( $warning );
 400                                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'expensive-parserfunction-category' ) );
 401                                 if ( $cat ) {
 402                                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
 403                                 }
 404                         }
 405                 }
 406
 407                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 408
 409                 # Information on include size limits, for the benefit of users who try to skirt them
 410                 if ( $this->mOptions->getEnableLimitReport() ) {
 411                         global $wgExpensiveParserFunctionLimit;
 412                         $max = $this->mOptions->getMaxIncludeSize();
 413                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 414                         $limitReport =
 415                                 "NewPP limit report\n" .
 416                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 417                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 418                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 419                                 $PFreport;
 420                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 421                         $text .= "\n<!-- \n$limitReport-->\n";
 422                 }
 423                 $this->mOutput->setText( $text );
 424                 $this->mRevisionId = $oldRevisionId;
 425                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 426                 wfProfileOut( $fname );
 427                 wfProfileOut( __METHOD__ );
 428
 429                 return $this->mOutput;
 430         }
 431
 432         /**
 433          * Recursive parser entry point that can be called from an extension tag
 434          * hook.
 435          */
 436         function recursiveTagParse( $text ) {
 437                 wfProfileIn( __METHOD__ );
 438                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 439                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 440                 $text = $this->internalParse( $text );
 441                 wfProfileOut( __METHOD__ );
 442                 return $text;
 443         }
 444
 445         /**
 446          * Expand templates and variables in the text, producing valid, static wikitext.
 447          * Also removes comments.
 448          */
 449         function preprocess( $text, $title, $options, $revid = null ) {
 450                 wfProfileIn( __METHOD__ );
 451                 $this->clearState();
 452                 $this->setOutputType( self::OT_PREPROCESS );
 453                 $this->mOptions = $options;
 454                 $this->setTitle( $title );
 455                 if( $revid !== null ) {
 456                         $this->mRevisionId = $revid;
 457                 }
 458                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 459                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 460                 $text = $this->replaceVariables( $text );
 461                 $text = $this->mStripState->unstripBoth( $text );
 462                 wfProfileOut( __METHOD__ );
 463                 return $text;
 464         }
 465
 466         /**
 467          * Get a random string
 468          *
 469          * @private
 470          * @static
 471          */
 472         function getRandomString() {
 473                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 474         }
 475
 476         function &getTitle() { return $this->mTitle; }
 477         function getOptions() { return $this->mOptions; }
 478
 479         function getFunctionLang() {
 480                 global $wgLang, $wgContLang;
 481
 482                 $target = $this->mOptions->getTargetLanguage();
 483                 if ( $target !== null ) {
 484                         return $target;
 485                 } else {
 486                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 487                 }
 488         }
 489
 490         /**
 491          * Get a preprocessor object
 492          */
 493         function getPreprocessor() {
 494                 if ( !isset( $this->mPreprocessor ) ) {
 495                         $class = $this->mPreprocessorClass;
 496                         $this->mPreprocessor = new $class( $this );
 497                 }
 498                 return $this->mPreprocessor;
 499         }
 500
 501         /**
 502          * Replaces all occurrences of HTML-style comments and the given tags
 503          * in the text with a random marker and returns the next text. The output
 504          * parameter $matches will be an associative array filled with data in
 505          * the form:
 506          *   'UNIQ-xxxxx' => array(
 507          *     'element',
 508          *     'tag content',
 509          *     array( 'param' => 'x' ),
 510          *     '<element param="x">tag content</element>' ) )
 511          *
 512          * @param $elements list of element names. Comments are always extracted.
 513          * @param $text Source text string.
 514          * @param $uniq_prefix
 515          *
 516          * @public
 517          * @static
 518          */
 519         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 520                 static $n = 1;
 521                 $stripped = '';
 522                 $matches = array();
 523
 524                 $taglist = implode( '|', $elements );
 525                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 526
 527                 while ( '' != $text ) {
 528                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 529                         $stripped .= $p[0];
 530                         if( count( $p ) < 5 ) {
 531                                 break;
 532                         }
 533                         if( count( $p ) > 5 ) {
 534                                 // comment
 535                                 $element    = $p[4];
 536                                 $attributes = '';
 537                                 $close      = '';
 538                                 $inside     = $p[5];
 539                         } else {
 540                                 // tag
 541                                 $element    = $p[1];
 542                                 $attributes = $p[2];
 543                                 $close      = $p[3];
 544                                 $inside     = $p[4];
 545                         }
 546
 547                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 548                         $stripped .= $marker;
 549
 550                         if ( $close === '/>' ) {
 551                                 // Empty element tag, <tag />
 552                                 $content = null;
 553                                 $text = $inside;
 554                                 $tail = null;
 555                         } else {
 556                                 if( $element == '!--' ) {
 557                                         $end = '/(-->)/';
 558                                 } else {
 559                                         $end = "/(<\\/$element\\s*>)/i";
 560                                 }
 561                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 562                                 $content = $q[0];
 563                                 if( count( $q ) < 3 ) {
 564                                         # No end tag -- let it run out to the end of the text.
 565                                         $tail = '';
 566                                         $text = '';
 567                                 } else {
 568                                         $tail = $q[1];
 569                                         $text = $q[2];
 570                                 }
 571                         }
 572
 573                         $matches[$marker] = array( $element,
 574                                 $content,
 575                                 Sanitizer::decodeTagAttributes( $attributes ),
 576                                 "<$element$attributes$close$content$tail" );
 577                 }
 578                 return $stripped;
 579         }
 580
 581         /**
 582          * Get a list of strippable XML-like elements
 583          */
 584         function getStripList() {
 585                 global $wgRawHtml;
 586                 $elements = $this->mStripList;
 587                 if( $wgRawHtml ) {
 588                         $elements[] = 'html';
 589                 }
 590                 if( $this->mOptions->getUseTeX() ) {
 591                         $elements[] = 'math';
 592                 }
 593                 return $elements;
 594         }
 595
 596         /**
 597          * @deprecated use replaceVariables
 598          */
 599         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 600                 return $text;
 601         }
 602
 603         /**
 604          * Restores pre, math, and other extensions removed by strip()
 605          *
 606          * always call unstripNoWiki() after this one
 607          * @private
 608          * @deprecated use $this->mStripState->unstrip()
 609          */
 610         function unstrip( $text, $state ) {
 611                 return $state->unstripGeneral( $text );
 612         }
 613
 614         /**
 615          * Always call this after unstrip() to preserve the order
 616          *
 617          * @private
 618          * @deprecated use $this->mStripState->unstrip()
 619          */
 620         function unstripNoWiki( $text, $state ) {
 621                 return $state->unstripNoWiki( $text );
 622         }
 623
 624         /**
 625          * @deprecated use $this->mStripState->unstripBoth()
 626          */
 627         function unstripForHTML( $text ) {
 628                 return $this->mStripState->unstripBoth( $text );
 629         }
 630
 631         /**
 632          * Add an item to the strip state
 633          * Returns the unique tag which must be inserted into the stripped text
 634          * The tag will be replaced with the original text in unstrip()
 635          *
 636          * @private
 637          */
 638         function insertStripItem( $text ) {
 639                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 640                 $this->mMarkerIndex++;
 641                 $this->mStripState->general->setPair( $rnd, $text );
 642                 return $rnd;
 643         }
 644
 645         /**
 646          * Interface with html tidy, used if $wgUseTidy = true.
 647          * If tidy isn't able to correct the markup, the original will be
 648          * returned in all its glory with a warning comment appended.
 649          *
 650          * Either the external tidy program or the in-process tidy extension
 651          * will be used depending on availability. Override the default
 652          * $wgTidyInternal setting to disable the internal if it's not working.
 653          *
 654          * @param string $text Hideous HTML input
 655          * @return string Corrected HTML output
 656          * @public
 657          * @static
 658          */
 659         function tidy( $text ) {
 660                 global $wgTidyInternal;
 661                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 662 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 663 '<head><title>test</title></head><body>'.$text.'</body></html>';
 664                 if( $wgTidyInternal ) {
 665                         $correctedtext = Parser::internalTidy( $wrappedtext );
 666                 } else {
 667                         $correctedtext = Parser::externalTidy( $wrappedtext );
 668                 }
 669                 if( is_null( $correctedtext ) ) {
 670                         wfDebug( "Tidy error detected!\n" );
 671                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 672                 }
 673                 return $correctedtext;
 674         }
 675
 676         /**
 677          * Spawn an external HTML tidy process and get corrected markup back from it.
 678          *
 679          * @private
 680          * @static
 681          */
 682         function externalTidy( $text ) {
 683                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 684                 $fname = 'Parser::externalTidy';
 685                 wfProfileIn( $fname );
 686
 687                 $cleansource = '';
 688                 $opts = ' -utf8';
 689
 690                 $descriptorspec = array(
 691                         0 => array('pipe', 'r'),
 692                         1 => array('pipe', 'w'),
 693                         2 => array('file', wfGetNull(), 'a')
 694                 );
 695                 $pipes = array();
 696                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 697                 if (is_resource($process)) {
 698                         // Theoretically, this style of communication could cause a deadlock
 699                         // here. If the stdout buffer fills up, then writes to stdin could
 700                         // block. This doesn't appear to happen with tidy, because tidy only
 701                         // writes to stdout after it's finished reading from stdin. Search
 702                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 703                         fwrite($pipes[0], $text);
 704                         fclose($pipes[0]);
 705                         while (!feof($pipes[1])) {
 706                                 $cleansource .= fgets($pipes[1], 1024);
 707                         }
 708                         fclose($pipes[1]);
 709                         proc_close($process);
 710                 }
 711
 712                 wfProfileOut( $fname );
 713
 714                 if( $cleansource == '' && $text != '') {
 715                         // Some kind of error happened, so we couldn't get the corrected text.
 716                         // Just give up; we'll use the source text and append a warning.
 717                         return null;
 718                 } else {
 719                         return $cleansource;
 720                 }
 721         }
 722
 723         /**
 724          * Use the HTML tidy PECL extension to use the tidy library in-process,
 725          * saving the overhead of spawning a new process.
 726          *
 727          * 'pear install tidy' should be able to compile the extension module.
 728          *
 729          * @private
 730          * @static
 731          */
 732         function internalTidy( $text ) {
 733                 global $wgTidyConf, $IP, $wgDebugTidy;
 734                 $fname = 'Parser::internalTidy';
 735                 wfProfileIn( $fname );
 736
 737                 $tidy = new tidy;
 738                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 739                 $tidy->cleanRepair();
 740                 if( $tidy->getStatus() == 2 ) {
 741                         // 2 is magic number for fatal error
 742                         // http://www.php.net/manual/en/function.tidy-get-status.php
 743                         $cleansource = null;
 744                 } else {
 745                         $cleansource = tidy_get_output( $tidy );
 746                 }
 747                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 748                         $cleansource .= "<!--\nTidy reports:\n" .
 749                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 750                                 "\n-->";
 751                 }
 752
 753                 wfProfileOut( $fname );
 754                 return $cleansource;
 755         }
 756
 757         /**
 758          * parse the wiki syntax used to render tables
 759          *
 760          * @private
 761          */
 762         function doTableStuff ( $text ) {
 763                 $fname = 'Parser::doTableStuff';
 764                 wfProfileIn( $fname );
 765
 766                 $lines = explode ( "\n" , $text );
 767                 $td_history = array (); // Is currently a td tag open?
 768                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 769                 $tr_history = array (); // Is currently a tr tag open?
 770                 $tr_attributes = array (); // history of tr attributes
 771                 $has_opened_tr = array(); // Did this table open a <tr> element?
 772                 $indent_level = 0; // indent level of the table
 773                 foreach ( $lines as $key => $line )
 774                 {
 775                         $line = trim ( $line );
 776
 777                         if( $line == '' ) { // empty line, go to next line
 778                                 continue;
 779                         }
 780                         $first_character = $line{0};
 781                         $matches = array();
 782
 783                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 784                                 // First check if we are starting a new table
 785                                 $indent_level = strlen( $matches[1] );
 786
 787                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 788                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 789
 790                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 791                                 array_push ( $td_history , false );
 792                                 array_push ( $last_tag_history , '' );
 793                                 array_push ( $tr_history , false );
 794                                 array_push ( $tr_attributes , '' );
 795                                 array_push ( $has_opened_tr , false );
 796                         } else if ( count ( $td_history ) == 0 ) {
 797                                 // Don't do any of the following
 798                                 continue;
 799                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 800                                 // We are ending a table
 801                                 $line = '</table>' . substr ( $line , 2 );
 802                                 $last_tag = array_pop ( $last_tag_history );
 803
 804                                 if ( !array_pop ( $has_opened_tr ) ) {
 805                                         $line = "<tr><td></td></tr>{$line}";
 806                                 }
 807
 808                                 if ( array_pop ( $tr_history ) ) {
 809                                         $line = "</tr>{$line}";
 810                                 }
 811
 812                                 if ( array_pop ( $td_history ) ) {
 813                                         $line = "</{$last_tag}>{$line}";
 814                                 }
 815                                 array_pop ( $tr_attributes );
 816                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 817                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 818                                 // Now we have a table row
 819                                 $line = preg_replace( '#^\|-+#', '', $line );
 820
 821                                 // Whats after the tag is now only attributes
 822                                 $attributes = $this->mStripState->unstripBoth( $line );
 823                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 824                                 array_pop ( $tr_attributes );
 825                                 array_push ( $tr_attributes , $attributes );
 826
 827                                 $line = '';
 828                                 $last_tag = array_pop ( $last_tag_history );
 829                                 array_pop ( $has_opened_tr );
 830                                 array_push ( $has_opened_tr , true );
 831
 832                                 if ( array_pop ( $tr_history ) ) {
 833                                         $line = '</tr>';
 834                                 }
 835
 836                                 if ( array_pop ( $td_history ) ) {
 837                                         $line = "</{$last_tag}>{$line}";
 838                                 }
 839
 840                                 $lines[$key] = $line;
 841                                 array_push ( $tr_history , false );
 842                                 array_push ( $td_history , false );
 843                                 array_push ( $last_tag_history , '' );
 844                         }
 845                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 846                                 // This might be cell elements, td, th or captions
 847                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 848                                         $first_character = '+';
 849                                         $line = substr ( $line , 1 );
 850                                 }
 851
 852                                 $line = substr ( $line , 1 );
 853
 854                                 if ( $first_character == '!' ) {
 855                                         $line = str_replace ( '!!' , '||' , $line );
 856                                 }
 857
 858                                 // Split up multiple cells on the same line.
 859                                 // FIXME : This can result in improper nesting of tags processed
 860                                 // by earlier parser steps, but should avoid splitting up eg
 861                                 // attribute values containing literal "||".
 862                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 863
 864                                 $lines[$key] = '';
 865
 866                                 // Loop through each table cell
 867                                 foreach ( $cells as $cell )
 868                                 {
 869                                         $previous = '';
 870                                         if ( $first_character != '+' )
 871                                         {
 872                                                 $tr_after = array_pop ( $tr_attributes );
 873                                                 if ( !array_pop ( $tr_history ) ) {
 874                                                         $previous = "<tr{$tr_after}>\n";
 875                                                 }
 876                                                 array_push ( $tr_history , true );
 877                                                 array_push ( $tr_attributes , '' );
 878                                                 array_pop ( $has_opened_tr );
 879                                                 array_push ( $has_opened_tr , true );
 880                                         }
 881
 882                                         $last_tag = array_pop ( $last_tag_history );
 883
 884                                         if ( array_pop ( $td_history ) ) {
 885                                                 $previous = "</{$last_tag}>{$previous}";
 886                                         }
 887
 888                                         if ( $first_character == '|' ) {
 889                                                 $last_tag = 'td';
 890                                         } else if ( $first_character == '!' ) {
 891                                                 $last_tag = 'th';
 892                                         } else if ( $first_character == '+' ) {
 893                                                 $last_tag = 'caption';
 894                                         } else {
 895                                                 $last_tag = '';
 896                                         }
 897
 898                                         array_push ( $last_tag_history , $last_tag );
 899
 900                                         // A cell could contain both parameters and data
 901                                         $cell_data = explode ( '|' , $cell , 2 );
 902
 903                                         // Bug 553: Note that a '|' inside an invalid link should not
 904                                         // be mistaken as delimiting cell parameters
 905                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 906                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 907                                         } else if ( count ( $cell_data ) == 1 )
 908                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 909                                         else {
 910                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 911                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 912                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 913                                         }
 914
 915                                         $lines[$key] .= $cell;
 916                                         array_push ( $td_history , true );
 917                                 }
 918                         }
 919                 }
 920
 921                 // Closing open td, tr && table
 922                 while ( count ( $td_history ) > 0 )
 923                 {
 924                         if ( array_pop ( $td_history ) ) {
 925                                 $lines[] = '</td>' ;
 926                         }
 927                         if ( array_pop ( $tr_history ) ) {
 928                                 $lines[] = '</tr>' ;
 929                         }
 930                         if ( !array_pop ( $has_opened_tr ) ) {
 931                                 $lines[] = "<tr><td></td></tr>" ;
 932                         }
 933
 934                         $lines[] = '</table>' ;
 935                 }
 936
 937                 $output = implode ( "\n" , $lines ) ;
 938
 939                 // special case: don't return empty table
 940                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 941                         $output = '';
 942                 }
 943
 944                 wfProfileOut( $fname );
 945
 946                 return $output;
 947         }
 948
 949         /**
 950          * Helper function for parse() that transforms wiki markup into
 951          * HTML. Only called for $mOutputType == self::OT_HTML.
 952          *
 953          * @private
 954          */
 955         function internalParse( $text ) {
 956                 $isMain = true;
 957                 $fname = 'Parser::internalParse';
 958                 wfProfileIn( $fname );
 959
 960                 # Hook to suspend the parser in this state
 961                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 962                         wfProfileOut( $fname );
 963                         return $text ;
 964                 }
 965
 966                 $text = $this->replaceVariables( $text );
 967                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 968                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 969
 970                 // Tables need to come after variable replacement for things to work
 971                 // properly; putting them before other transformations should keep
 972                 // exciting things like link expansions from showing up in surprising
 973                 // places.
 974                 $text = $this->doTableStuff( $text );
 975
 976                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 977
 978                 $text = $this->doDoubleUnderscore( $text );
 979                 $text = $this->doHeadings( $text );
 980                 if($this->mOptions->getUseDynamicDates()) {
 981                         $df = DateFormatter::getInstance();
 982                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 983                 }
 984                 $text = $this->doAllQuotes( $text );
 985                 $text = $this->replaceInternalLinks( $text );
 986                 $text = $this->replaceExternalLinks( $text );
 987
 988                 # replaceInternalLinks may sometimes leave behind
 989                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 990                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 991
 992                 $text = $this->doMagicLinks( $text );
 993                 $text = $this->formatHeadings( $text, $isMain );
 994
 995                 wfProfileOut( $fname );
 996                 return $text;
 997         }
 998
 999         /**
1000          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1001          * magic external links.
1002          *
1003          * @private
1004          */
1005         function doMagicLinks( $text ) {
1006                 wfProfileIn( __METHOD__ );
1007                 $text = preg_replace_callback(
1008                         '!(?:                           # Start cases
1009                             <a.*?</a> |                 # Skip link text
1010                             <.*?> |                     # Skip stuff inside HTML elements
1011                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1012                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1013                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1014                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1015                                      [0-9Xx]                 # check digit
1016                                    \b)
1017                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1018                 wfProfileOut( __METHOD__ );
1019                 return $text;
1020         }
1021
1022         function magicLinkCallback( $m ) {
1023                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1024                         # Skip HTML element
1025                         return $m[0];
1026                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1027                         $isbn = $m[2];
1028                         $num = strtr( $isbn, array(
1029                                 '-' => '',
1030                                 ' ' => '',
1031                                 'x' => 'X',
1032                         ));
1033                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1034                         $text = '<a href="' .
1035                                 $titleObj->escapeLocalUrl() .
1036                                 "\" class=\"internal\">ISBN $isbn</a>";
1037                 } else {
1038                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1039                                 $keyword = 'RFC';
1040                                 $urlmsg = 'rfcurl';
1041                                 $id = $m[1];
1042                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1043                                 $keyword = 'PMID';
1044                                 $urlmsg = 'pubmedurl';
1045                                 $id = $m[1];
1046                         } else {
1047                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1048                                         substr($m[0], 0, 20 ) . '"' );
1049                         }
1050
1051                         $url = wfMsg( $urlmsg, $id);
1052                         $sk = $this->mOptions->getSkin();
1053                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1054                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1055                 }
1056                 return $text;
1057         }
1058
1059         /**
1060          * Parse headers and return html
1061          *
1062          * @private
1063          */
1064         function doHeadings( $text ) {
1065                 $fname = 'Parser::doHeadings';
1066                 wfProfileIn( $fname );
1067                 for ( $i = 6; $i >= 1; --$i ) {
1068                         $h = str_repeat( '=', $i );
1069                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1070                           "<h$i>\\1</h$i>", $text );
1071                 }
1072                 wfProfileOut( $fname );
1073                 return $text;
1074         }
1075
1076         /**
1077          * Replace single quotes with HTML markup
1078          * @private
1079          * @return string the altered text
1080          */
1081         function doAllQuotes( $text ) {
1082                 $fname = 'Parser::doAllQuotes';
1083                 wfProfileIn( $fname );
1084                 $outtext = '';
1085                 $lines = explode( "\n", $text );
1086                 foreach ( $lines as $line ) {
1087                         $outtext .= $this->doQuotes ( $line ) . "\n";
1088                 }
1089                 $outtext = substr($outtext, 0,-1);
1090                 wfProfileOut( $fname );
1091                 return $outtext;
1092         }
1093
1094         /**
1095          * Helper function for doAllQuotes()
1096          */
1097         public function doQuotes( $text ) {
1098                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1099                 if ( count( $arr ) == 1 )
1100                         return $text;
1101                 else
1102                 {
1103                         # First, do some preliminary work. This may shift some apostrophes from
1104                         # being mark-up to being text. It also counts the number of occurrences
1105                         # of bold and italics mark-ups.
1106                         $i = 0;
1107                         $numbold = 0;
1108                         $numitalics = 0;
1109                         foreach ( $arr as $r )
1110                         {
1111                                 if ( ( $i % 2 ) == 1 )
1112                                 {
1113                                         # If there are ever four apostrophes, assume the first is supposed to
1114                                         # be text, and the remaining three constitute mark-up for bold text.
1115                                         if ( strlen( $arr[$i] ) == 4 )
1116                                         {
1117                                                 $arr[$i-1] .= "'";
1118                                                 $arr[$i] = "'''";
1119                                         }
1120                                         # If there are more than 5 apostrophes in a row, assume they're all
1121                                         # text except for the last 5.
1122                                         else if ( strlen( $arr[$i] ) > 5 )
1123                                         {
1124                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1125                                                 $arr[$i] = "'''''";
1126                                         }
1127                                         # Count the number of occurrences of bold and italics mark-ups.
1128                                         # We are not counting sequences of five apostrophes.
1129                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1130                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1131                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1132                                 }
1133                                 $i++;
1134                         }
1135
1136                         # If there is an odd number of both bold and italics, it is likely
1137                         # that one of the bold ones was meant to be an apostrophe followed
1138                         # by italics. Which one we cannot know for certain, but it is more
1139                         # likely to be one that has a single-letter word before it.
1140                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1141                         {
1142                                 $i = 0;
1143                                 $firstsingleletterword = -1;
1144                                 $firstmultiletterword = -1;
1145                                 $firstspace = -1;
1146                                 foreach ( $arr as $r )
1147                                 {
1148                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1149                                         {
1150                                                 $x1 = substr ($arr[$i-1], -1);
1151                                                 $x2 = substr ($arr[$i-1], -2, 1);
1152                                                 if ($x1 == ' ') {
1153                                                         if ($firstspace == -1) $firstspace = $i;
1154                                                 } else if ($x2 == ' ') {
1155                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1156                                                 } else {
1157                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1158                                                 }
1159                                         }
1160                                         $i++;
1161                                 }
1162
1163                                 # If there is a single-letter word, use it!
1164                                 if ($firstsingleletterword > -1)
1165                                 {
1166                                         $arr [ $firstsingleletterword ] = "''";
1167                                         $arr [ $firstsingleletterword-1 ] .= "'";
1168                                 }
1169                                 # If not, but there's a multi-letter word, use that one.
1170                                 else if ($firstmultiletterword > -1)
1171                                 {
1172                                         $arr [ $firstmultiletterword ] = "''";
1173                                         $arr [ $firstmultiletterword-1 ] .= "'";
1174                                 }
1175                                 # ... otherwise use the first one that has neither.
1176                                 # (notice that it is possible for all three to be -1 if, for example,
1177                                 # there is only one pentuple-apostrophe in the line)
1178                                 else if ($firstspace > -1)
1179                                 {
1180                                         $arr [ $firstspace ] = "''";
1181                                         $arr [ $firstspace-1 ] .= "'";
1182                                 }
1183                         }
1184
1185                         # Now let's actually convert our apostrophic mush to HTML!
1186                         $output = '';
1187                         $buffer = '';
1188                         $state = '';
1189                         $i = 0;
1190                         foreach ($arr as $r)
1191                         {
1192                                 if (($i % 2) == 0)
1193                                 {
1194                                         if ($state == 'both')
1195                                                 $buffer .= $r;
1196                                         else
1197                                                 $output .= $r;
1198                                 }
1199                                 else
1200                                 {
1201                                         if (strlen ($r) == 2)
1202                                         {
1203                                                 if ($state == 'i')
1204                                                 { $output .= '</i>'; $state = ''; }
1205                                                 else if ($state == 'bi')
1206                                                 { $output .= '</i>'; $state = 'b'; }
1207                                                 else if ($state == 'ib')
1208                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1209                                                 else if ($state == 'both')
1210                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1211                                                 else # $state can be 'b' or ''
1212                                                 { $output .= '<i>'; $state .= 'i'; }
1213                                         }
1214                                         else if (strlen ($r) == 3)
1215                                         {
1216                                                 if ($state == 'b')
1217                                                 { $output .= '</b>'; $state = ''; }
1218                                                 else if ($state == 'bi')
1219                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1220                                                 else if ($state == 'ib')
1221                                                 { $output .= '</b>'; $state = 'i'; }
1222                                                 else if ($state == 'both')
1223                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1224                                                 else # $state can be 'i' or ''
1225                                                 { $output .= '<b>'; $state .= 'b'; }
1226                                         }
1227                                         else if (strlen ($r) == 5)
1228                                         {
1229                                                 if ($state == 'b')
1230                                                 { $output .= '</b><i>'; $state = 'i'; }
1231                                                 else if ($state == 'i')
1232                                                 { $output .= '</i><b>'; $state = 'b'; }
1233                                                 else if ($state == 'bi')
1234                                                 { $output .= '</i></b>'; $state = ''; }
1235                                                 else if ($state == 'ib')
1236                                                 { $output .= '</b></i>'; $state = ''; }
1237                                                 else if ($state == 'both')
1238                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1239                                                 else # ($state == '')
1240                                                 { $buffer = ''; $state = 'both'; }
1241                                         }
1242                                 }
1243                                 $i++;
1244                         }
1245                         # Now close all remaining tags.  Notice that the order is important.
1246                         if ($state == 'b' || $state == 'ib')
1247                                 $output .= '</b>';
1248                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1249                                 $output .= '</i>';
1250                         if ($state == 'bi')
1251                                 $output .= '</b>';
1252                         # There might be lonely ''''', so make sure we have a buffer
1253                         if ($state == 'both' && $buffer)
1254                                 $output .= '<b><i>'.$buffer.'</i></b>';
1255                         return $output;
1256                 }
1257         }
1258
1259         /**
1260          * Replace external links
1261          *
1262          * Note: this is all very hackish and the order of execution matters a lot.
1263          * Make sure to run maintenance/parserTests.php if you change this code.
1264          *
1265          * @private
1266          */
1267         function replaceExternalLinks( $text ) {
1268                 global $wgContLang;
1269                 $fname = 'Parser::replaceExternalLinks';
1270                 wfProfileIn( $fname );
1271
1272                 $sk = $this->mOptions->getSkin();
1273
1274                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1275
1276                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1277
1278                 $i = 0;
1279                 while ( $i<count( $bits ) ) {
1280                         $url = $bits[$i++];
1281                         $protocol = $bits[$i++];
1282                         $text = $bits[$i++];
1283                         $trail = $bits[$i++];
1284
1285                         # The characters '<' and '>' (which were escaped by
1286                         # removeHTMLtags()) should not be included in
1287                         # URLs, per RFC 2396.
1288                         $m2 = array();
1289                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1290                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1291                                 $url = substr($url, 0, $m2[0][1]);
1292                         }
1293
1294                         # If the link text is an image URL, replace it with an <img> tag
1295                         # This happened by accident in the original parser, but some people used it extensively
1296                         $img = $this->maybeMakeExternalImage( $text );
1297                         if ( $img !== false ) {
1298                                 $text = $img;
1299                         }
1300
1301                         $dtrail = '';
1302
1303                         # Set linktype for CSS - if URL==text, link is essentially free
1304                         $linktype = ($text == $url) ? 'free' : 'text';
1305
1306                         # No link text, e.g. [http://domain.tld/some.link]
1307                         if ( $text == '' ) {
1308                                 # Autonumber if allowed. See bug #5918
1309                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1310                                         $text = '[' . ++$this->mAutonumber . ']';
1311                                         $linktype = 'autonumber';
1312                                 } else {
1313                                         # Otherwise just use the URL
1314                                         $text = htmlspecialchars( $url );
1315                                         $linktype = 'free';
1316                                 }
1317                         } else {
1318                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1319                                 # Check for trail
1320                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1321                         }
1322
1323                         $text = $wgContLang->markNoConversion($text);
1324
1325                         $url = Sanitizer::cleanUrl( $url );
1326
1327                         # Process the trail (i.e. everything after this link up until start of the next link),
1328                         # replacing any non-bracketed links
1329                         $trail = $this->replaceFreeExternalLinks( $trail );
1330
1331                         # Use the encoded URL
1332                         # This means that users can paste URLs directly into the text
1333                         # Funny characters like &ouml; aren't valid in URLs anyway
1334                         # This was changed in August 2004
1335                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1336
1337                         # Register link in the output object.
1338                         # Replace unnecessary URL escape codes with the referenced character
1339                         # This prevents spammers from hiding links from the filters
1340                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1341                         $this->mOutput->addExternalLink( $pasteurized );
1342                 }
1343
1344                 wfProfileOut( $fname );
1345                 return $s;
1346         }
1347
1348         /**
1349          * Replace anything that looks like a URL with a link
1350          * @private
1351          */
1352         function replaceFreeExternalLinks( $text ) {
1353                 global $wgContLang;
1354                 $fname = 'Parser::replaceFreeExternalLinks';
1355                 wfProfileIn( $fname );
1356
1357                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1358                 $s = array_shift( $bits );
1359                 $i = 0;
1360
1361                 $sk = $this->mOptions->getSkin();
1362
1363                 while ( $i < count( $bits ) ){
1364                         $protocol = $bits[$i++];
1365                         $remainder = $bits[$i++];
1366
1367                         $m = array();
1368                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1369                                 # Found some characters after the protocol that look promising
1370                                 $url = $protocol . $m[1];
1371                                 $trail = $m[2];
1372
1373                                 # special case: handle urls as url args:
1374                                 # http://www.example.com/foo?=http://www.example.com/bar
1375                                 if(strlen($trail) == 0 &&
1376                                         isset($bits[$i]) &&
1377                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1378                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1379                                 {
1380                                         # add protocol, arg
1381                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1382                                         $i += 2;
1383                                         $trail = $m[2];
1384                                 }
1385
1386                                 # The characters '<' and '>' (which were escaped by
1387                                 # removeHTMLtags()) should not be included in
1388                                 # URLs, per RFC 2396.
1389                                 $m2 = array();
1390                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1391                                         $trail = substr($url, $m2[0][1]) . $trail;
1392                                         $url = substr($url, 0, $m2[0][1]);
1393                                 }
1394
1395                                 # Move trailing punctuation to $trail
1396                                 $sep = ',;\.:!?';
1397                                 # If there is no left bracket, then consider right brackets fair game too
1398                                 if ( strpos( $url, '(' ) === false ) {
1399                                         $sep .= ')';
1400                                 }
1401
1402                                 $numSepChars = strspn( strrev( $url ), $sep );
1403                                 if ( $numSepChars ) {
1404                                         $trail = substr( $url, -$numSepChars ) . $trail;
1405                                         $url = substr( $url, 0, -$numSepChars );
1406                                 }
1407
1408                                 $url = Sanitizer::cleanUrl( $url );
1409
1410                                 # Is this an external image?
1411                                 $text = $this->maybeMakeExternalImage( $url );
1412                                 if ( $text === false ) {
1413                                         # Not an image, make a link
1414                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1415                                         # Register it in the output object...
1416                                         # Replace unnecessary URL escape codes with their equivalent characters
1417                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1418                                         $this->mOutput->addExternalLink( $pasteurized );
1419                                 }
1420                                 $s .= $text . $trail;
1421                         } else {
1422                                 $s .= $protocol . $remainder;
1423                         }
1424                 }
1425                 wfProfileOut( $fname );
1426                 return $s;
1427         }
1428
1429         /**
1430          * Replace unusual URL escape codes with their equivalent characters
1431          * @param string
1432          * @return string
1433          * @static
1434          * @todo  This can merge genuinely required bits in the path or query string,
1435          *        breaking legit URLs. A proper fix would treat the various parts of
1436          *        the URL differently; as a workaround, just use the output for
1437          *        statistical records, not for actual linking/output.
1438          */
1439         static function replaceUnusualEscapes( $url ) {
1440                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1441                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1442         }
1443
1444         /**
1445          * Callback function used in replaceUnusualEscapes().
1446          * Replaces unusual URL escape codes with their equivalent character
1447          * @static
1448          * @private
1449          */
1450         private static function replaceUnusualEscapesCallback( $matches ) {
1451                 $char = urldecode( $matches[0] );
1452                 $ord = ord( $char );
1453                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1454                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1455                         // No, shouldn't be escaped
1456                         return $char;
1457                 } else {
1458                         // Yes, leave it escaped
1459                         return $matches[0];
1460                 }
1461         }
1462
1463         /**
1464          * make an image if it's allowed, either through the global
1465          * option or through the exception
1466          * @private
1467          */
1468         function maybeMakeExternalImage( $url ) {
1469                 $sk = $this->mOptions->getSkin();
1470                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1471                 $imagesexception = !empty($imagesfrom);
1472                 $text = false;
1473                 if ( $this->mOptions->getAllowExternalImages()
1474                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1475                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1476                                 # Image found
1477                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1478                         }
1479                 }
1480                 return $text;
1481         }
1482
1483         /**
1484          * Process [[ ]] wikilinks
1485          *
1486          * @private
1487          */
1488         function replaceInternalLinks( $s ) {
1489                 global $wgContLang;
1490                 static $fname = 'Parser::replaceInternalLinks' ;
1491
1492                 wfProfileIn( $fname );
1493
1494                 wfProfileIn( $fname.'-setup' );
1495                 static $tc = FALSE;
1496                 # the % is needed to support urlencoded titles as well
1497                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1498
1499                 $sk = $this->mOptions->getSkin();
1500
1501                 #split the entire text string on occurences of [[
1502                 $a = explode( '[[', ' ' . $s );
1503                 #get the first element (all text up to first [[), and remove the space we added
1504                 $s = array_shift( $a );
1505                 $s = substr( $s, 1 );
1506
1507                 # Match a link having the form [[namespace:link|alternate]]trail
1508                 static $e1 = FALSE;
1509                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1510                 # Match cases where there is no "]]", which might still be images
1511                 static $e1_img = FALSE;
1512                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1513
1514                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1515                 $e2 = null;
1516                 if ( $useLinkPrefixExtension ) {
1517                         # Match the end of a line for a word that's not followed by whitespace,
1518                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1519                         $e2 = wfMsgForContent( 'linkprefix' );
1520                 }
1521
1522                 if( is_null( $this->mTitle ) ) {
1523                         wfProfileOut( $fname );
1524                         wfProfileOut( $fname.'-setup' );
1525                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1526                 }
1527                 $nottalk = !$this->mTitle->isTalkPage();
1528
1529                 if ( $useLinkPrefixExtension ) {
1530                         $m = array();
1531                         if ( preg_match( $e2, $s, $m ) ) {
1532                                 $first_prefix = $m[2];
1533                         } else {
1534                                 $first_prefix = false;
1535                         }
1536                 } else {
1537                         $prefix = '';
1538                 }
1539
1540                 if($wgContLang->hasVariants()) {
1541                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1542                 } else {
1543                         $selflink = array($this->mTitle->getPrefixedText());
1544                 }
1545                 $useSubpages = $this->areSubpagesAllowed();
1546                 wfProfileOut( $fname.'-setup' );
1547
1548                 # Loop for each link
1549                 for ($k = 0; isset( $a[$k] ); $k++) {
1550                         $line = $a[$k];
1551                         if ( $useLinkPrefixExtension ) {
1552                                 wfProfileIn( $fname.'-prefixhandling' );
1553                                 if ( preg_match( $e2, $s, $m ) ) {
1554                                         $prefix = $m[2];
1555                                         $s = $m[1];
1556                                 } else {
1557                                         $prefix='';
1558                                 }
1559                                 # first link
1560                                 if($first_prefix) {
1561                                         $prefix = $first_prefix;
1562                                         $first_prefix = false;
1563                                 }
1564                                 wfProfileOut( $fname.'-prefixhandling' );
1565                         }
1566
1567                         $might_be_img = false;
1568
1569                         wfProfileIn( "$fname-e1" );
1570                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1571                                 $text = $m[2];
1572                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1573                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1574                                 # the real problem is with the $e1 regex
1575                                 # See bug 1300.
1576                                 #
1577                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1578                                 # and no image is in sight. See bug 2095.
1579                                 #
1580                                 if( $text !== '' &&
1581                                         substr( $m[3], 0, 1 ) === ']' &&
1582                                         strpos($text, '[') !== false
1583                                 )
1584                                 {
1585                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1586                                         $m[3] = substr( $m[3], 1 );
1587                                 }
1588                                 # fix up urlencoded title texts
1589                                 if( strpos( $m[1], '%' ) !== false ) {
1590                                         # Should anchors '#' also be rejected?
1591                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1592                                 }
1593                                 $trail = $m[3];
1594                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1595                                 $might_be_img = true;
1596                                 $text = $m[2];
1597                                 if ( strpos( $m[1], '%' ) !== false ) {
1598                                         $m[1] = urldecode($m[1]);
1599                                 }
1600                                 $trail = "";
1601                         } else { # Invalid form; output directly
1602                                 $s .= $prefix . '[[' . $line ;
1603                                 wfProfileOut( "$fname-e1" );
1604                                 continue;
1605                         }
1606                         wfProfileOut( "$fname-e1" );
1607                         wfProfileIn( "$fname-misc" );
1608
1609                         # Don't allow internal links to pages containing
1610                         # PROTO: where PROTO is a valid URL protocol; these
1611                         # should be external links.
1612                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1613                                 $s .= $prefix . '[[' . $line ;
1614                                 wfProfileOut( "$fname-misc" );
1615                                 continue;
1616                         }
1617
1618                         # Make subpage if necessary
1619                         if( $useSubpages ) {
1620                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1621                         } else {
1622                                 $link = $m[1];
1623                         }
1624
1625                         $noforce = (substr($m[1], 0, 1) != ':');
1626                         if (!$noforce) {
1627                                 # Strip off leading ':'
1628                                 $link = substr($link, 1);
1629                         }
1630
1631                         wfProfileOut( "$fname-misc" );
1632                         wfProfileIn( "$fname-title" );
1633                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1634                         if( !$nt ) {
1635                                 $s .= $prefix . '[[' . $line;
1636                                 wfProfileOut( "$fname-title" );
1637                                 continue;
1638                         }
1639
1640                         $ns = $nt->getNamespace();
1641                         $iw = $nt->getInterWiki();
1642                         wfProfileOut( "$fname-title" );
1643
1644                         if ($might_be_img) { # if this is actually an invalid link
1645                                 wfProfileIn( "$fname-might_be_img" );
1646                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1647                                         $found = false;
1648                                         while (isset ($a[$k+1]) ) {
1649                                                 #look at the next 'line' to see if we can close it there
1650                                                 $spliced = array_splice( $a, $k + 1, 1 );
1651                                                 $next_line = array_shift( $spliced );
1652                                                 $m = explode( ']]', $next_line, 3 );
1653                                                 if ( count( $m ) == 3 ) {
1654                                                         # the first ]] closes the inner link, the second the image
1655                                                         $found = true;
1656                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1657                                                         $trail = $m[2];
1658                                                         break;
1659                                                 } elseif ( count( $m ) == 2 ) {
1660                                                         #if there's exactly one ]] that's fine, we'll keep looking
1661                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1662                                                 } else {
1663                                                         #if $next_line is invalid too, we need look no further
1664                                                         $text .= '[[' . $next_line;
1665                                                         break;
1666                                                 }
1667                                         }
1668                                         if ( !$found ) {
1669                                                 # we couldn't find the end of this imageLink, so output it raw
1670                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1671                                                 $text = $this->replaceInternalLinks($text);
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                                 } else { #it's not an image, so output it raw
1678                                         $s .= "{$prefix}[[$link|$text";
1679                                         # note: no $trail, because without an end, there *is* no trail
1680                                         wfProfileOut( "$fname-might_be_img" );
1681                                         continue;
1682                                 }
1683                                 wfProfileOut( "$fname-might_be_img" );
1684                         }
1685
1686                         $wasblank = ( '' == $text );
1687                         if( $wasblank ) $text = $link;
1688
1689                         # Link not escaped by : , create the various objects
1690                         if( $noforce ) {
1691
1692                                 # Interwikis
1693                                 wfProfileIn( "$fname-interwiki" );
1694                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1695                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1696                                         $s = rtrim($s . $prefix);
1697                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1698                                         wfProfileOut( "$fname-interwiki" );
1699                                         continue;
1700                                 }
1701                                 wfProfileOut( "$fname-interwiki" );
1702
1703                                 if ( $ns == NS_IMAGE ) {
1704                                         wfProfileIn( "$fname-image" );
1705                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1706                                                 # recursively parse links inside the image caption
1707                                                 # actually, this will parse them in any other parameters, too,
1708                                                 # but it might be hard to fix that, and it doesn't matter ATM
1709                                                 $text = $this->replaceExternalLinks($text);
1710                                                 $text = $this->replaceInternalLinks($text);
1711
1712                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1713                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1714                                                 $this->mOutput->addImage( $nt->getDBkey() );
1715
1716                                                 wfProfileOut( "$fname-image" );
1717                                                 continue;
1718                                         } else {
1719                                                 # We still need to record the image's presence on the page
1720                                                 $this->mOutput->addImage( $nt->getDBkey() );
1721                                         }
1722                                         wfProfileOut( "$fname-image" );
1723
1724                                 }
1725
1726                                 if ( $ns == NS_CATEGORY ) {
1727                                         wfProfileIn( "$fname-category" );
1728                                         $s = rtrim($s . "\n"); # bug 87
1729
1730                                         if ( $wasblank ) {
1731                                                 $sortkey = $this->getDefaultSort();
1732                                         } else {
1733                                                 $sortkey = $text;
1734                                         }
1735                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1736                                         $sortkey = str_replace( "\n", '', $sortkey );
1737                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1738                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1739
1740                                         /**
1741                                          * Strip the whitespace Category links produce, see bug 87
1742                                          * @todo We might want to use trim($tmp, "\n") here.
1743                                          */
1744                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1745
1746                                         wfProfileOut( "$fname-category" );
1747                                         continue;
1748                                 }
1749                         }
1750
1751                         # Self-link checking
1752                         if( $nt->getFragment() === '' ) {
1753                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1754                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1755                                         continue;
1756                                 }
1757                         }
1758
1759                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1760                         if( $ns == NS_MEDIA ) {
1761                                 # Give extensions a chance to select the file revision for us
1762                                 $skip = $time = false;
1763                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1764                                 if ( $skip ) {
1765                                         $link = $sk->makeLinkObj( $nt );
1766                                 } else {
1767                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1768                                 }
1769                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1770                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1771                                 $this->mOutput->addImage( $nt->getDBkey() );
1772                                 continue;
1773                         } elseif( $ns == NS_SPECIAL ) {
1774                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1775                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1776                                 } else {
1777                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1778                                 }
1779                                 continue;
1780                         } elseif( $ns == NS_IMAGE ) {
1781                                 $img = wfFindFile( $nt );
1782                                 if( $img ) {
1783                                         // Force a blue link if the file exists; may be a remote
1784                                         // upload on the shared repository, and we want to see its
1785                                         // auto-generated page.
1786                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1787                                         $this->mOutput->addLink( $nt );
1788                                         continue;
1789                                 }
1790                         }
1791                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1792                 }
1793                 wfProfileOut( $fname );
1794                 return $s;
1795         }
1796
1797         /**
1798          * Make a link placeholder. The text returned can be later resolved to a real link with
1799          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1800          * parsing of interwiki links, and secondly to allow all existence checks and
1801          * article length checks (for stub links) to be bundled into a single query.
1802          *
1803          */
1804         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1805                 wfProfileIn( __METHOD__ );
1806                 if ( ! is_object($nt) ) {
1807                         # Fail gracefully
1808                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1809                 } else {
1810                         # Separate the link trail from the rest of the link
1811                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1812
1813                         if ( $nt->isExternal() ) {
1814                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1815                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1816                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1817                         } else {
1818                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1819                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1820                                 $this->mLinkHolders['queries'][] = $query;
1821                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1822                                 $this->mLinkHolders['titles'][] = $nt;
1823
1824                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1825                         }
1826                 }
1827                 wfProfileOut( __METHOD__ );
1828                 return $retVal;
1829         }
1830
1831         /**
1832          * Render a forced-blue link inline; protect against double expansion of
1833          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1834          * Since this little disaster has to split off the trail text to avoid
1835          * breaking URLs in the following text without breaking trails on the
1836          * wiki links, it's been made into a horrible function.
1837          *
1838          * @param Title $nt
1839          * @param string $text
1840          * @param string $query
1841          * @param string $trail
1842          * @param string $prefix
1843          * @return string HTML-wikitext mix oh yuck
1844          */
1845         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1846                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1847                 $sk = $this->mOptions->getSkin();
1848                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1849                 return $this->armorLinks( $link ) . $trail;
1850         }
1851
1852         /**
1853          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1854          * going to go through further parsing steps before inline URL expansion.
1855          *
1856          * In particular this is important when using action=render, which causes
1857          * full URLs to be included.
1858          *
1859          * Oh man I hate our multi-layer parser!
1860          *
1861          * @param string more-or-less HTML
1862          * @return string less-or-more HTML with NOPARSE bits
1863          */
1864         function armorLinks( $text ) {
1865                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1866                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1867         }
1868
1869         /**
1870          * Return true if subpage links should be expanded on this page.
1871          * @return bool
1872          */
1873         function areSubpagesAllowed() {
1874                 # Some namespaces don't allow subpages
1875                 global $wgNamespacesWithSubpages;
1876                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1877         }
1878
1879         /**
1880          * Handle link to subpage if necessary
1881          * @param string $target the source of the link
1882          * @param string &$text the link text, modified as necessary
1883          * @return string the full name of the link
1884          * @private
1885          */
1886         function maybeDoSubpageLink($target, &$text) {
1887                 # Valid link forms:
1888                 # Foobar -- normal
1889                 # :Foobar -- override special treatment of prefix (images, language links)
1890                 # /Foobar -- convert to CurrentPage/Foobar
1891                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1892                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1893                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1894
1895                 $fname = 'Parser::maybeDoSubpageLink';
1896                 wfProfileIn( $fname );
1897                 $ret = $target; # default return value is no change
1898
1899                 # Some namespaces don't allow subpages,
1900                 # so only perform processing if subpages are allowed
1901                 if( $this->areSubpagesAllowed() ) {
1902                         $hash = strpos( $target, '#' );
1903                         if( $hash !== false ) {
1904                                 $suffix = substr( $target, $hash );
1905                                 $target = substr( $target, 0, $hash );
1906                         } else {
1907                                 $suffix = '';
1908                         }
1909                         # bug 7425
1910                         $target = trim( $target );
1911                         # Look at the first character
1912                         if( $target != '' && $target{0} == '/' ) {
1913                                 # / at end means we don't want the slash to be shown
1914                                 $m = array();
1915                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1916                                 if( $trailingSlashes ) {
1917                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1918                                 } else {
1919                                         $noslash = substr( $target, 1 );
1920                                 }
1921
1922                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1923                                 if( '' === $text ) {
1924                                         $text = $target . $suffix;
1925                                 } # this might be changed for ugliness reasons
1926                         } else {
1927                                 # check for .. subpage backlinks
1928                                 $dotdotcount = 0;
1929                                 $nodotdot = $target;
1930                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1931                                         ++$dotdotcount;
1932                                         $nodotdot = substr( $nodotdot, 3 );
1933                                 }
1934                                 if($dotdotcount > 0) {
1935                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1936                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1937                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1938                                                 # / at the end means don't show full path
1939                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1940                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1941                                                         if( '' === $text ) {
1942                                                                 $text = $nodotdot . $suffix;
1943                                                         }
1944                                                 }
1945                                                 $nodotdot = trim( $nodotdot );
1946                                                 if( $nodotdot != '' ) {
1947                                                         $ret .= '/' . $nodotdot;
1948                                                 }
1949                                                 $ret .= $suffix;
1950                                         }
1951                                 }
1952                         }
1953                 }
1954
1955                 wfProfileOut( $fname );
1956                 return $ret;
1957         }
1958
1959         /**#@+
1960          * Used by doBlockLevels()
1961          * @private
1962          */
1963         /* private */ function closeParagraph() {
1964                 $result = '';
1965                 if ( '' != $this->mLastSection ) {
1966                         $result = '</' . $this->mLastSection  . ">\n";
1967                 }
1968                 $this->mInPre = false;
1969                 $this->mLastSection = '';
1970                 return $result;
1971         }
1972         # getCommon() returns the length of the longest common substring
1973         # of both arguments, starting at the beginning of both.
1974         #
1975         /* private */ function getCommon( $st1, $st2 ) {
1976                 $fl = strlen( $st1 );
1977                 $shorter = strlen( $st2 );
1978                 if ( $fl < $shorter ) { $shorter = $fl; }
1979
1980                 for ( $i = 0; $i < $shorter; ++$i ) {
1981                         if ( $st1{$i} != $st2{$i} ) { break; }
1982                 }
1983                 return $i;
1984         }
1985         # These next three functions open, continue, and close the list
1986         # element appropriate to the prefix character passed into them.
1987         #
1988         /* private */ function openList( $char ) {
1989                 $result = $this->closeParagraph();
1990
1991                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1992                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1993                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1994                 else if ( ';' == $char ) {
1995                         $result .= '<dl><dt>';
1996                         $this->mDTopen = true;
1997                 }
1998                 else { $result = '<!-- ERR 1 -->'; }
1999
2000                 return $result;
2001         }
2002
2003         /* private */ function nextItem( $char ) {
2004                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2005                 else if ( ':' == $char || ';' == $char ) {
2006                         $close = '</dd>';
2007                         if ( $this->mDTopen ) { $close = '</dt>'; }
2008                         if ( ';' == $char ) {
2009                                 $this->mDTopen = true;
2010                                 return $close . '<dt>';
2011                         } else {
2012                                 $this->mDTopen = false;
2013                                 return $close . '<dd>';
2014                         }
2015                 }
2016                 return '<!-- ERR 2 -->';
2017         }
2018
2019         /* private */ function closeList( $char ) {
2020                 if ( '*' == $char ) { $text = '</li></ul>'; }
2021                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2022                 else if ( ':' == $char ) {
2023                         if ( $this->mDTopen ) {
2024                                 $this->mDTopen = false;
2025                                 $text = '</dt></dl>';
2026                         } else {
2027                                 $text = '</dd></dl>';
2028                         }
2029                 }
2030                 else {  return '<!-- ERR 3 -->'; }
2031                 return $text."\n";
2032         }
2033         /**#@-*/
2034
2035         /**
2036          * Make lists from lines starting with ':', '*', '#', etc.
2037          *
2038          * @private
2039          * @return string the lists rendered as HTML
2040          */
2041         function doBlockLevels( $text, $linestart ) {
2042                 $fname = 'Parser::doBlockLevels';
2043                 wfProfileIn( $fname );
2044
2045                 # Parsing through the text line by line.  The main thing
2046                 # happening here is handling of block-level elements p, pre,
2047                 # and making lists from lines starting with * # : etc.
2048                 #
2049                 $textLines = explode( "\n", $text );
2050
2051                 $lastPrefix = $output = '';
2052                 $this->mDTopen = $inBlockElem = false;
2053                 $prefixLength = 0;
2054                 $paragraphStack = false;
2055
2056                 if ( !$linestart ) {
2057                         $output .= array_shift( $textLines );
2058                 }
2059                 foreach ( $textLines as $oLine ) {
2060                         $lastPrefixLength = strlen( $lastPrefix );
2061                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2062                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2063                         if ( !$this->mInPre ) {
2064                                 # Multiple prefixes may abut each other for nested lists.
2065                                 $prefixLength = strspn( $oLine, '*#:;' );
2066                                 $pref = substr( $oLine, 0, $prefixLength );
2067
2068                                 # eh?
2069                                 $pref2 = str_replace( ';', ':', $pref );
2070                                 $t = substr( $oLine, $prefixLength );
2071                                 $this->mInPre = !empty($preOpenMatch);
2072                         } else {
2073                                 # Don't interpret any other prefixes in preformatted text
2074                                 $prefixLength = 0;
2075                                 $pref = $pref2 = '';
2076                                 $t = $oLine;
2077                         }
2078
2079                         # List generation
2080                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2081                                 # Same as the last item, so no need to deal with nesting or opening stuff
2082                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2083                                 $paragraphStack = false;
2084
2085                                 if ( substr( $pref, -1 ) == ';') {
2086                                         # The one nasty exception: definition lists work like this:
2087                                         # ; title : definition text
2088                                         # So we check for : in the remainder text to split up the
2089                                         # title and definition, without b0rking links.
2090                                         $term = $t2 = '';
2091                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2092                                                 $t = $t2;
2093                                                 $output .= $term . $this->nextItem( ':' );
2094                                         }
2095                                 }
2096                         } elseif( $prefixLength || $lastPrefixLength ) {
2097                                 # Either open or close a level...
2098                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2099                                 $paragraphStack = false;
2100
2101                                 while( $commonPrefixLength < $lastPrefixLength ) {
2102                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2103                                         --$lastPrefixLength;
2104                                 }
2105                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2106                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2107                                 }
2108                                 while ( $prefixLength > $commonPrefixLength ) {
2109                                         $char = substr( $pref, $commonPrefixLength, 1 );
2110                                         $output .= $this->openList( $char );
2111
2112                                         if ( ';' == $char ) {
2113                                                 # FIXME: This is dupe of code above
2114                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2115                                                         $t = $t2;
2116                                                         $output .= $term . $this->nextItem( ':' );
2117                                                 }
2118                                         }
2119                                         ++$commonPrefixLength;
2120                                 }
2121                                 $lastPrefix = $pref2;
2122                         }
2123                         if( 0 == $prefixLength ) {
2124                                 wfProfileIn( "$fname-paragraph" );
2125                                 # No prefix (not in list)--go to paragraph mode
2126                                 // XXX: use a stack for nestable elements like span, table and div
2127                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2128                                 $closematch = preg_match(
2129                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2130                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2131                                 if ( $openmatch or $closematch ) {
2132                                         $paragraphStack = false;
2133                                         # TODO bug 5718: paragraph closed
2134                                         $output .= $this->closeParagraph();
2135                                         if ( $preOpenMatch and !$preCloseMatch ) {
2136                                                 $this->mInPre = true;
2137                                         }
2138                                         if ( $closematch ) {
2139                                                 $inBlockElem = false;
2140                                         } else {
2141                                                 $inBlockElem = true;
2142                                         }
2143                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2144                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2145                                                 // pre
2146                                                 if ($this->mLastSection != 'pre') {
2147                                                         $paragraphStack = false;
2148                                                         $output .= $this->closeParagraph().'<pre>';
2149                                                         $this->mLastSection = 'pre';
2150                                                 }
2151                                                 $t = substr( $t, 1 );
2152                                         } else {
2153                                                 // paragraph
2154                                                 if ( '' == trim($t) ) {
2155                                                         if ( $paragraphStack ) {
2156                                                                 $output .= $paragraphStack.'<br />';
2157                                                                 $paragraphStack = false;
2158                                                                 $this->mLastSection = 'p';
2159                                                         } else {
2160                                                                 if ($this->mLastSection != 'p' ) {
2161                                                                         $output .= $this->closeParagraph();
2162                                                                         $this->mLastSection = '';
2163                                                                         $paragraphStack = '<p>';
2164                                                                 } else {
2165                                                                         $paragraphStack = '</p><p>';
2166                                                                 }
2167                                                         }
2168                                                 } else {
2169                                                         if ( $paragraphStack ) {
2170                                                                 $output .= $paragraphStack;
2171                                                                 $paragraphStack = false;
2172                                                                 $this->mLastSection = 'p';
2173                                                         } else if ($this->mLastSection != 'p') {
2174                                                                 $output .= $this->closeParagraph().'<p>';
2175                                                                 $this->mLastSection = 'p';
2176                                                         }
2177                                                 }
2178                                         }
2179                                 }
2180                                 wfProfileOut( "$fname-paragraph" );
2181                         }
2182                         // somewhere above we forget to get out of pre block (bug 785)
2183                         if($preCloseMatch && $this->mInPre) {
2184                                 $this->mInPre = false;
2185                         }
2186                         if ($paragraphStack === false) {
2187                                 $output .= $t."\n";
2188                         }
2189                 }
2190                 while ( $prefixLength ) {
2191                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2192                         --$prefixLength;
2193                 }
2194                 if ( '' != $this->mLastSection ) {
2195                         $output .= '</' . $this->mLastSection . '>';
2196                         $this->mLastSection = '';
2197                 }
2198
2199                 wfProfileOut( $fname );
2200                 return $output;
2201         }
2202
2203         /**
2204          * Split up a string on ':', ignoring any occurences inside tags
2205          * to prevent illegal overlapping.
2206          * @param string $str the string to split
2207          * @param string &$before set to everything before the ':'
2208          * @param string &$after set to everything after the ':'
2209          * return string the position of the ':', or false if none found
2210          */
2211         function findColonNoLinks($str, &$before, &$after) {
2212                 $fname = 'Parser::findColonNoLinks';
2213                 wfProfileIn( $fname );
2214
2215                 $pos = strpos( $str, ':' );
2216                 if( $pos === false ) {
2217                         // Nothing to find!
2218                         wfProfileOut( $fname );
2219                         return false;
2220                 }
2221
2222                 $lt = strpos( $str, '<' );
2223                 if( $lt === false || $lt > $pos ) {
2224                         // Easy; no tag nesting to worry about
2225                         $before = substr( $str, 0, $pos );
2226                         $after = substr( $str, $pos+1 );
2227                         wfProfileOut( $fname );
2228                         return $pos;
2229                 }
2230
2231                 // Ugly state machine to walk through avoiding tags.
2232                 $state = self::COLON_STATE_TEXT;
2233                 $stack = 0;
2234                 $len = strlen( $str );
2235                 for( $i = 0; $i < $len; $i++ ) {
2236                         $c = $str{$i};
2237
2238                         switch( $state ) {
2239                         // (Using the number is a performance hack for common cases)
2240                         case 0: // self::COLON_STATE_TEXT:
2241                                 switch( $c ) {
2242                                 case "<":
2243                                         // Could be either a <start> tag or an </end> tag
2244                                         $state = self::COLON_STATE_TAGSTART;
2245                                         break;
2246                                 case ":":
2247                                         if( $stack == 0 ) {
2248                                                 // We found it!
2249                                                 $before = substr( $str, 0, $i );
2250                                                 $after = substr( $str, $i + 1 );
2251                                                 wfProfileOut( $fname );
2252                                                 return $i;
2253                                         }
2254                                         // Embedded in a tag; don't break it.
2255                                         break;
2256                                 default:
2257                                         // Skip ahead looking for something interesting
2258                                         $colon = strpos( $str, ':', $i );
2259                                         if( $colon === false ) {
2260                                                 // Nothing else interesting
2261                                                 wfProfileOut( $fname );
2262                                                 return false;
2263                                         }
2264                                         $lt = strpos( $str, '<', $i );
2265                                         if( $stack === 0 ) {
2266                                                 if( $lt === false || $colon < $lt ) {
2267                                                         // We found it!
2268                                                         $before = substr( $str, 0, $colon );
2269                                                         $after = substr( $str, $colon + 1 );
2270                                                         wfProfileOut( $fname );
2271                                                         return $i;
2272                                                 }
2273                                         }
2274                                         if( $lt === false ) {
2275                                                 // Nothing else interesting to find; abort!
2276                                                 // We're nested, but there's no close tags left. Abort!
2277                                                 break 2;
2278                                         }
2279                                         // Skip ahead to next tag start
2280                                         $i = $lt;
2281                                         $state = self::COLON_STATE_TAGSTART;
2282                                 }
2283                                 break;
2284                         case 1: // self::COLON_STATE_TAG:
2285                                 // In a <tag>
2286                                 switch( $c ) {
2287                                 case ">":
2288                                         $stack++;
2289                                         $state = self::COLON_STATE_TEXT;
2290                                         break;
2291                                 case "/":
2292                                         // Slash may be followed by >?
2293                                         $state = self::COLON_STATE_TAGSLASH;
2294                                         break;
2295                                 default:
2296                                         // ignore
2297                                 }
2298                                 break;
2299                         case 2: // self::COLON_STATE_TAGSTART:
2300                                 switch( $c ) {
2301                                 case "/":
2302                                         $state = self::COLON_STATE_CLOSETAG;
2303                                         break;
2304                                 case "!":
2305                                         $state = self::COLON_STATE_COMMENT;
2306                                         break;
2307                                 case ">":
2308                                         // Illegal early close? This shouldn't happen D:
2309                                         $state = self::COLON_STATE_TEXT;
2310                                         break;
2311                                 default:
2312                                         $state = self::COLON_STATE_TAG;
2313                                 }
2314                                 break;
2315                         case 3: // self::COLON_STATE_CLOSETAG:
2316                                 // In a </tag>
2317                                 if( $c == ">" ) {
2318                                         $stack--;
2319                                         if( $stack < 0 ) {
2320                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2321                                                 wfProfileOut( $fname );
2322                                                 return false;
2323                                         }
2324                                         $state = self::COLON_STATE_TEXT;
2325                                 }
2326                                 break;
2327                         case self::COLON_STATE_TAGSLASH:
2328                                 if( $c == ">" ) {
2329                                         // Yes, a self-closed tag <blah/>
2330                                         $state = self::COLON_STATE_TEXT;
2331                                 } else {
2332                                         // Probably we're jumping the gun, and this is an attribute
2333                                         $state = self::COLON_STATE_TAG;
2334                                 }
2335                                 break;
2336                         case 5: // self::COLON_STATE_COMMENT:
2337                                 if( $c == "-" ) {
2338                                         $state = self::COLON_STATE_COMMENTDASH;
2339                                 }
2340                                 break;
2341                         case self::COLON_STATE_COMMENTDASH:
2342                                 if( $c == "-" ) {
2343                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2344                                 } else {
2345                                         $state = self::COLON_STATE_COMMENT;
2346                                 }
2347                                 break;
2348                         case self::COLON_STATE_COMMENTDASHDASH:
2349                                 if( $c == ">" ) {
2350                                         $state = self::COLON_STATE_TEXT;
2351                                 } else {
2352                                         $state = self::COLON_STATE_COMMENT;
2353                                 }
2354                                 break;
2355                         default:
2356                                 throw new MWException( "State machine error in $fname" );
2357                         }
2358                 }
2359                 if( $stack > 0 ) {
2360                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2361                         return false;
2362                 }
2363                 wfProfileOut( $fname );
2364                 return false;
2365         }
2366
2367         /**
2368          * Return value of a magic variable (like PAGENAME)
2369          *
2370          * @private
2371          */
2372         function getVariableValue( $index ) {
2373                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2374
2375                 /**
2376                  * Some of these require message or data lookups and can be
2377                  * expensive to check many times.
2378                  */
2379                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2380                         if ( isset( $this->mVarCache[$index] ) ) {
2381                                 return $this->mVarCache[$index];
2382                         }
2383                 }
2384
2385                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2386                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2387
2388                 # Use the time zone
2389                 global $wgLocaltimezone;
2390                 if ( isset( $wgLocaltimezone ) ) {
2391                         $oldtz = getenv( 'TZ' );
2392                         putenv( 'TZ='.$wgLocaltimezone );
2393                 }
2394
2395                 wfSuppressWarnings(); // E_STRICT system time bitching
2396                 $localTimestamp = date( 'YmdHis', $ts );
2397                 $localMonth = date( 'm', $ts );
2398                 $localMonthName = date( 'n', $ts );
2399                 $localDay = date( 'j', $ts );
2400                 $localDay2 = date( 'd', $ts );
2401                 $localDayOfWeek = date( 'w', $ts );
2402                 $localWeek = date( 'W', $ts );
2403                 $localYear = date( 'Y', $ts );
2404                 $localHour = date( 'H', $ts );
2405                 if ( isset( $wgLocaltimezone ) ) {
2406                         putenv( 'TZ='.$oldtz );
2407                 }
2408                 wfRestoreWarnings();
2409
2410                 switch ( $index ) {
2411                         case 'currentmonth':
2412                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2413                         case 'currentmonthname':
2414                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2415                         case 'currentmonthnamegen':
2416                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2417                         case 'currentmonthabbrev':
2418                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2419                         case 'currentday':
2420                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2421                         case 'currentday2':
2422                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2423                         case 'localmonth':
2424                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2425                         case 'localmonthname':
2426                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2427                         case 'localmonthnamegen':
2428                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2429                         case 'localmonthabbrev':
2430                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2431                         case 'localday':
2432                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2433                         case 'localday2':
2434                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2435                         case 'pagename':
2436                                 return wfEscapeWikiText( $this->mTitle->getText() );
2437                         case 'pagenamee':
2438                                 return $this->mTitle->getPartialURL();
2439                         case 'fullpagename':
2440                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2441                         case 'fullpagenamee':
2442                                 return $this->mTitle->getPrefixedURL();
2443                         case 'subpagename':
2444                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2445                         case 'subpagenamee':
2446                                 return $this->mTitle->getSubpageUrlForm();
2447                         case 'basepagename':
2448                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2449                         case 'basepagenamee':
2450                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2451                         case 'talkpagename':
2452                                 if( $this->mTitle->canTalk() ) {
2453                                         $talkPage = $this->mTitle->getTalkPage();
2454                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2455                                 } else {
2456                                         return '';
2457                                 }
2458                         case 'talkpagenamee':
2459                                 if( $this->mTitle->canTalk() ) {
2460                                         $talkPage = $this->mTitle->getTalkPage();
2461                                         return $talkPage->getPrefixedUrl();
2462                                 } else {
2463                                         return '';
2464                                 }
2465                         case 'subjectpagename':
2466                                 $subjPage = $this->mTitle->getSubjectPage();
2467                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2468                         case 'subjectpagenamee':
2469                                 $subjPage = $this->mTitle->getSubjectPage();
2470                                 return $subjPage->getPrefixedUrl();
2471                         case 'revisionid':
2472                                 // Let the edit saving system know we should parse the page
2473                                 // *after* a revision ID has been assigned.
2474                                 $this->mOutput->setFlag( 'vary-revision' );
2475                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2476                                 return $this->mRevisionId;
2477                         case 'revisionday':
2478                                 // Let the edit saving system know we should parse the page
2479                                 // *after* a revision ID has been assigned. This is for null edits.
2480                                 $this->mOutput->setFlag( 'vary-revision' );
2481                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2482                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2483                         case 'revisionday2':
2484                                 // Let the edit saving system know we should parse the page
2485                                 // *after* a revision ID has been assigned. This is for null edits.
2486                                 $this->mOutput->setFlag( 'vary-revision' );
2487                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2488                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2489                         case 'revisionmonth':
2490                                 // Let the edit saving system know we should parse the page
2491                                 // *after* a revision ID has been assigned. This is for null edits.
2492                                 $this->mOutput->setFlag( 'vary-revision' );
2493                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2494                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2495                         case 'revisionyear':
2496                                 // Let the edit saving system know we should parse the page
2497                                 // *after* a revision ID has been assigned. This is for null edits.
2498                                 $this->mOutput->setFlag( 'vary-revision' );
2499                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2500                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2501                         case 'revisiontimestamp':
2502                                 // Let the edit saving system know we should parse the page
2503                                 // *after* a revision ID has been assigned. This is for null edits.
2504                                 $this->mOutput->setFlag( 'vary-revision' );
2505                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2506                                 return $this->getRevisionTimestamp();
2507                         case 'namespace':
2508                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2509                         case 'namespacee':
2510                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2511                         case 'talkspace':
2512                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2513                         case 'talkspacee':
2514                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2515                         case 'subjectspace':
2516                                 return $this->mTitle->getSubjectNsText();
2517                         case 'subjectspacee':
2518                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2519                         case 'currentdayname':
2520                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2521                         case 'currentyear':
2522                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2523                         case 'currenttime':
2524                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2525                         case 'currenthour':
2526                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2527                         case 'currentweek':
2528                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2529                                 // int to remove the padding
2530                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2531                         case 'currentdow':
2532                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2533                         case 'localdayname':
2534                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2535                         case 'localyear':
2536                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2537                         case 'localtime':
2538                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2539                         case 'localhour':
2540                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2541                         case 'localweek':
2542                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2543                                 // int to remove the padding
2544                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2545                         case 'localdow':
2546                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2547                         case 'numberofarticles':
2548                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2549                         case 'numberoffiles':
2550                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2551                         case 'numberofusers':
2552                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2553                         case 'numberofpages':
2554                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2555                         case 'numberofadmins':
2556                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2557                         case 'numberofedits':
2558                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2559                         case 'currenttimestamp':
2560                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2561                         case 'localtimestamp':
2562                                 return $this->mVarCache[$index] = $localTimestamp;
2563                         case 'currentversion':
2564                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2565                         case 'sitename':
2566                                 return $wgSitename;
2567                         case 'server':
2568                                 return $wgServer;
2569                         case 'servername':
2570                                 return $wgServerName;
2571                         case 'scriptpath':
2572                                 return $wgScriptPath;
2573                         case 'directionmark':
2574                                 return $wgContLang->getDirMark();
2575                         case 'contentlanguage':
2576                                 global $wgContLanguageCode;
2577                                 return $wgContLanguageCode;
2578                         default:
2579                                 $ret = null;
2580                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2581                                         return $ret;
2582                                 else
2583                                         return null;
2584                 }
2585         }
2586
2587         /**
2588          * initialise the magic variables (like CURRENTMONTHNAME)
2589          *
2590          * @private
2591          */
2592         function initialiseVariables() {
2593                 $fname = 'Parser::initialiseVariables';
2594                 wfProfileIn( $fname );
2595                 $variableIDs = MagicWord::getVariableIDs();
2596
2597                 $this->mVariables = new MagicWordArray( $variableIDs );
2598                 wfProfileOut( $fname );
2599         }
2600
2601         /**
2602          * Preprocess some wikitext and return the document tree.
2603          * This is the ghost of replace_variables().
2604          *
2605          * @param string $text The text to parse
2606          * @param integer flags Bitwise combination of:
2607          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2608          *                                     included. Default is to assume a direct page view.
2609          *
2610          * The generated DOM tree must depend only on the input text and the flags.
2611          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2612          *
2613          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2614          * change in the DOM tree for a given text, must be passed through the section identifier
2615          * in the section edit link and thus back to extractSections().
2616          *
2617          * The output of this function is currently only cached in process memory, but a persistent
2618          * cache may be implemented at a later date which takes further advantage of these strict
2619          * dependency requirements.
2620          *
2621          * @private
2622          */
2623         function preprocessToDom ( $text, $flags = 0 ) {
2624                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2625                 return $dom;
2626         }
2627
2628         /*
2629          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2630          */
2631         public static function splitWhitespace( $s ) {
2632                 $ltrimmed = ltrim( $s );
2633                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2634                 $trimmed = rtrim( $ltrimmed );
2635                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2636                 if ( $diff > 0 ) {
2637                         $w2 = substr( $ltrimmed, -$diff );
2638                 } else {
2639                         $w2 = '';
2640                 }
2641                 return array( $w1, $trimmed, $w2 );
2642         }
2643
2644         /**
2645          * Replace magic variables, templates, and template arguments
2646          * with the appropriate text. Templates are substituted recursively,
2647          * taking care to avoid infinite loops.
2648          *
2649          * Note that the substitution depends on value of $mOutputType:
2650          *  self::OT_WIKI: only {{subst:}} templates
2651          *  self::OT_PREPROCESS: templates but not extension tags
2652          *  self::OT_HTML: all templates and extension tags
2653          *
2654          * @param string $tex The text to transform
2655          * @param PPFrame $frame Object describing the arguments passed to the template
2656          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2657          * @private
2658          */
2659         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2660                 # Prevent too big inclusions
2661                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2662                         return $text;
2663                 }
2664
2665                 $fname = __METHOD__;
2666                 wfProfileIn( $fname );
2667
2668                 if ( $frame === false ) {
2669                         $frame = $this->getPreprocessor()->newFrame();
2670                 } elseif ( !( $frame instanceof PPFrame ) ) {
2671                         throw new MWException( __METHOD__ . ' called using the old argument format' );
2672                 }
2673
2674                 $dom = $this->preprocessToDom( $text );
2675                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2676                 $text = $frame->expand( $dom, $flags );
2677
2678                 wfProfileOut( $fname );
2679                 return $text;
2680         }
2681
2682         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2683         static function createAssocArgs( $args ) {
2684                 $assocArgs = array();
2685                 $index = 1;
2686                 foreach( $args as $arg ) {
2687                         $eqpos = strpos( $arg, '=' );
2688                         if ( $eqpos === false ) {
2689                                 $assocArgs[$index++] = $arg;
2690                         } else {
2691                                 $name = trim( substr( $arg, 0, $eqpos ) );
2692                                 $value = trim( substr( $arg, $eqpos+1 ) );
2693                                 if ( $value === false ) {
2694                                         $value = '';
2695                                 }
2696                                 if ( $name !== false ) {
2697                                         $assocArgs[$name] = $value;
2698                                 }
2699                         }
2700                 }
2701
2702                 return $assocArgs;
2703         }
2704
2705         /**
2706          * Return the text of a template, after recursively
2707          * replacing any variables or templates within the template.
2708          *
2709          * @param array $piece The parts of the template
2710          *  $piece['title']: the title, i.e. the part before the |
2711          *  $piece['parts']: the parameter array
2712          *  $piece['lineStart']: whether the brace was at the start of a line
2713          * @param PPFrame The current frame, contains template arguments
2714          * @return string the text of the template
2715          * @private
2716          */
2717         function braceSubstitution( $piece, $frame ) {
2718                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2719                 $fname = __METHOD__;
2720                 wfProfileIn( $fname );
2721                 wfProfileIn( __METHOD__.'-setup' );
2722
2723                 # Flags
2724                 $found = false;             # $text has been filled
2725                 $nowiki = false;            # wiki markup in $text should be escaped
2726                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2727                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2728                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2729                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2730
2731                 # Title object, where $text came from
2732                 $title = NULL;
2733
2734                 # $part1 is the bit before the first |, and must contain only title characters.
2735                 # Various prefixes will be stripped from it later.
2736                 $titleWithSpaces = $frame->expand( $piece['title'] );
2737                 $part1 = trim( $titleWithSpaces );
2738                 $titleText = false;
2739
2740                 # Original title text preserved for various purposes
2741                 $originalTitle = $part1;
2742
2743                 # $args is a list of argument nodes, starting from index 0, not including $part1
2744                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2745                 wfProfileOut( __METHOD__.'-setup' );
2746
2747                 # SUBST
2748                 wfProfileIn( __METHOD__.'-modifiers' );
2749                 if ( !$found ) {
2750                         $mwSubst = MagicWord::get( 'subst' );
2751                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2752                                 # One of two possibilities is true:
2753                                 # 1) Found SUBST but not in the PST phase
2754                                 # 2) Didn't find SUBST and in the PST phase
2755                                 # In either case, return without further processing
2756                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2757                                 $isLocalObj = true;
2758                                 $found = true;
2759                         }
2760                 }
2761
2762                 # Variables
2763                 if ( !$found && $args->getLength() == 0 ) {
2764                         $id = $this->mVariables->matchStartToEnd( $part1 );
2765                         if ( $id !== false ) {
2766                                 $text = $this->getVariableValue( $id );
2767                                 if (MagicWord::getCacheTTL($id)>-1)
2768                                         $this->mOutput->mContainsOldMagic = true;
2769                                 $found = true;
2770                         }
2771                 }
2772
2773                 # MSG, MSGNW and RAW
2774                 if ( !$found ) {
2775                         # Check for MSGNW:
2776                         $mwMsgnw = MagicWord::get( 'msgnw' );
2777                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2778                                 $nowiki = true;
2779                         } else {
2780                                 # Remove obsolete MSG:
2781                                 $mwMsg = MagicWord::get( 'msg' );
2782                                 $mwMsg->matchStartAndRemove( $part1 );
2783                         }
2784
2785                         # Check for RAW:
2786                         $mwRaw = MagicWord::get( 'raw' );
2787                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2788                                 $forceRawInterwiki = true;
2789                         }
2790                 }
2791                 wfProfileOut( __METHOD__.'-modifiers' );
2792
2793                 # Parser functions
2794                 if ( !$found ) {
2795                         wfProfileIn( __METHOD__ . '-pfunc' );
2796
2797                         $colonPos = strpos( $part1, ':' );
2798                         if ( $colonPos !== false ) {
2799                                 # Case sensitive functions
2800                                 $function = substr( $part1, 0, $colonPos );
2801                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2802                                         $function = $this->mFunctionSynonyms[1][$function];
2803                                 } else {
2804                                         # Case insensitive functions
2805                                         $function = strtolower( $function );
2806                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2807                                                 $function = $this->mFunctionSynonyms[0][$function];
2808                                         } else {
2809                                                 $function = false;
2810                                         }
2811                                 }
2812                                 if ( $function ) {
2813                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2814                                         $initialArgs = array( &$this );
2815                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2816                                         if ( $flags & SFH_OBJECT_ARGS ) {
2817                                                 # Add a frame parameter, and pass the arguments as an array
2818                                                 $allArgs = $initialArgs;
2819                                                 $allArgs[] = $frame;
2820                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2821                                                         $funcArgs[] = $args->item( $i );
2822                                                 }
2823                                                 $allArgs[] = $funcArgs;
2824                                         } else {
2825                                                 # Convert arguments to plain text
2826                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2827                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2828                                                 }
2829                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2830                                         }
2831
2832                                         # Workaround for PHP bug 35229 and similar
2833                                         if ( !is_callable( $callback ) ) {
2834                                                 throw new MWException( "Tag hook for $name is not callable\n" );
2835                                         }
2836                                         $result = call_user_func_array( $callback, $allArgs );
2837                                         $found = true;
2838
2839                                         if ( is_array( $result ) ) {
2840                                                 if ( isset( $result[0] ) ) {
2841                                                         $text = $result[0];
2842                                                         unset( $result[0] );
2843                                                 }
2844
2845                                                 // Extract flags into the local scope
2846                                                 // This allows callers to set flags such as nowiki, found, etc.
2847                                                 extract( $result );
2848                                         } else {
2849                                                 $text = $result;
2850                                         }
2851                                 }
2852                         }
2853                         wfProfileOut( __METHOD__ . '-pfunc' );
2854                 }
2855
2856                 # Finish mangling title and then check for loops.
2857                 # Set $title to a Title object and $titleText to the PDBK
2858                 if ( !$found ) {
2859                         $ns = NS_TEMPLATE;
2860                         # Split the title into page and subpage
2861                         $subpage = '';
2862                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2863                         if ($subpage !== '') {
2864                                 $ns = $this->mTitle->getNamespace();
2865                         }
2866                         $title = Title::newFromText( $part1, $ns );
2867                         if ( $title ) {
2868                                 $titleText = $title->getPrefixedText();
2869                                 # Check for language variants if the template is not found
2870                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2871                                         $wgContLang->findVariantLink($part1, $title);
2872                                 }
2873                                 # Do infinite loop check
2874                                 if ( !$frame->loopCheck( $title ) ) {
2875                                         $found = true;
2876                                         $text = "<span class=\"error\">Template loop detected: [[$titleText]]</span>";
2877                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2878                                 }
2879                                 # Do recursion depth check
2880                                 $limit = $this->mOptions->getMaxTemplateDepth();
2881                                 if ( $frame->depth >= $limit ) {
2882                                         $found = true;
2883                                         $text = "<span class=\"error\">Template recursion depth limit exceeded ($limit)</span>";
2884                                 }
2885                         }
2886                 }
2887
2888                 # Load from database
2889                 if ( !$found && $title ) {
2890                         wfProfileIn( __METHOD__ . '-loadtpl' );
2891                         if ( !$title->isExternal() ) {
2892                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2893                                         $text = SpecialPage::capturePath( $title );
2894                                         if ( is_string( $text ) ) {
2895                                                 $found = true;
2896                                                 $isHTML = true;
2897                                                 $this->disableCache();
2898                                         }
2899                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2900                                         $found = false; //access denied
2901                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
2902                                 } else {
2903                                         list( $text, $title ) = $this->getTemplateDom( $title );
2904                                         if ( $text !== false ) {
2905                                                 $found = true;
2906                                                 $isChildObj = true;
2907                                         }
2908                                 }
2909
2910                                 # If the title is valid but undisplayable, make a link to it
2911                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2912                                         $text = "[[:$titleText]]";
2913                                         $found = true;
2914                                 }
2915                         } elseif ( $title->isTrans() ) {
2916                                 // Interwiki transclusion
2917                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2918                                         $text = $this->interwikiTransclude( $title, 'render' );
2919                                         $isHTML = true;
2920                                 } else {
2921                                         $text = $this->interwikiTransclude( $title, 'raw' );
2922                                         // Preprocess it like a template
2923                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2924                                         $isChildObj = true;
2925                                 }
2926                                 $found = true;
2927                         }
2928                         wfProfileOut( __METHOD__ . '-loadtpl' );
2929                 }
2930
2931                 # If we haven't found text to substitute by now, we're done
2932                 # Recover the source wikitext and return it
2933                 if ( !$found ) {
2934                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2935                         wfProfileOut( $fname );
2936                         return array( 'object' => $text );
2937                 }
2938
2939                 # Expand DOM-style return values in a child frame
2940                 if ( $isChildObj ) {
2941                         # Clean up argument array
2942                         $newFrame = $frame->newChild( $args, $title );
2943
2944                         if ( $nowiki ) {
2945                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2946                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2947                                 # Expansion is eligible for the empty-frame cache
2948                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2949                                         $text = $this->mTplExpandCache[$titleText];
2950                                 } else {
2951                                         $text = $newFrame->expand( $text );
2952                                         $this->mTplExpandCache[$titleText] = $text;
2953                                 }
2954                         } else {
2955                                 # Uncached expansion
2956                                 $text = $newFrame->expand( $text );
2957                         }
2958                 }
2959                 if ( $isLocalObj && $nowiki ) {
2960                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2961                         $isLocalObj = false;
2962                 }
2963
2964                 # Replace raw HTML by a placeholder
2965                 # Add a blank line preceding, to prevent it from mucking up
2966                 # immediately preceding headings
2967                 if ( $isHTML ) {
2968                         $text = "\n\n" . $this->insertStripItem( $text );
2969                 }
2970                 # Escape nowiki-style return values
2971                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2972                         $text = wfEscapeWikiText( $text );
2973                 }
2974                 # Bug 529: if the template begins with a table or block-level
2975                 # element, it should be treated as beginning a new line.
2976                 # This behaviour is somewhat controversial.
2977                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
2978                         $text = "\n" . $text;
2979                 }
2980
2981                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
2982                         # Error, oversize inclusion
2983                         $text = "[[$originalTitle]]" .
2984                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
2985                 }
2986
2987                 if ( $isLocalObj ) {
2988                         $ret = array( 'object' => $text );
2989                 } else {
2990                         $ret = array( 'text' => $text );
2991                 }
2992
2993                 wfProfileOut( $fname );
2994                 return $ret;
2995         }
2996
2997         /**
2998          * Get the semi-parsed DOM representation of a template with a given title,
2999          * and its redirect destination title. Cached.
3000          */
3001         function getTemplateDom( $title ) {
3002                 $cacheTitle = $title;
3003                 $titleText = $title->getPrefixedDBkey();
3004
3005                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3006                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3007                         $title = Title::makeTitle( $ns, $dbk );
3008                         $titleText = $title->getPrefixedDBkey();
3009                 }
3010                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3011                         return array( $this->mTplDomCache[$titleText], $title );
3012                 }
3013
3014                 // Cache miss, go to the database
3015                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3016
3017                 if ( $text === false ) {
3018                         $this->mTplDomCache[$titleText] = false;
3019                         return array( false, $title );
3020                 }
3021
3022                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3023                 $this->mTplDomCache[ $titleText ] = $dom;
3024
3025                 if (! $title->equals($cacheTitle)) {
3026                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3027                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3028                 }
3029
3030                 return array( $dom, $title );
3031         }
3032
3033         /**
3034          * Fetch the unparsed text of a template and register a reference to it.
3035          */
3036         function fetchTemplateAndTitle( $title ) {
3037                 $templateCb = $this->mOptions->getTemplateCallback();
3038                 $stuff = call_user_func( $templateCb, $title );
3039                 $text = $stuff['text'];
3040                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3041                 if ( isset( $stuff['deps'] ) ) {
3042                         foreach ( $stuff['deps'] as $dep ) {
3043                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3044                         }
3045                 }
3046                 return array($text,$finalTitle);
3047         }
3048
3049         function fetchTemplate( $title ) {
3050                 $rv = $this->fetchTemplateAndTitle($title);
3051                 return $rv[0];
3052         }
3053
3054         /**
3055          * Static function to get a template
3056          * Can be overridden via ParserOptions::setTemplateCallback().
3057          */
3058         static function statelessFetchTemplate( $title ) {
3059                 $text = $skip = false;
3060                 $finalTitle = $title;
3061                 $deps = array();
3062
3063                 // Loop to fetch the article, with up to 1 redirect
3064                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3065                         # Give extensions a chance to select the revision instead
3066                         $id = false; // Assume current
3067                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3068
3069                         if( $skip ) {
3070                                 $text = false;
3071                                 $deps[] = array(
3072                                         'title' => $title,
3073                                         'page_id' => $title->getArticleID(),
3074                                         'rev_id' => null );
3075                                 break;
3076                         }
3077                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3078                         $rev_id = $rev ? $rev->getId() : 0;
3079
3080                         $deps[] = array(
3081                                 'title' => $title,
3082                                 'page_id' => $title->getArticleID(),
3083                                 'rev_id' => $rev_id );
3084
3085                         if( $rev ) {
3086                                 $text = $rev->getText();
3087                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3088                                 global $wgLang;
3089                                 $message = $wgLang->lcfirst( $title->getText() );
3090                                 $text = wfMsgForContentNoTrans( $message );
3091                                 if( wfEmptyMsg( $message, $text ) ) {
3092                                         $text = false;
3093                                         break;
3094                                 }
3095                         } else {
3096                                 break;
3097                         }
3098                         if ( $text === false ) {
3099                                 break;
3100                         }
3101                         // Redirect?
3102                         $finalTitle = $title;
3103                         $title = Title::newFromRedirect( $text );
3104                 }
3105                 return array(
3106                         'text' => $text,
3107                         'finalTitle' => $finalTitle,
3108                         'deps' => $deps );
3109         }
3110
3111         /**
3112          * Transclude an interwiki link.
3113          */
3114         function interwikiTransclude( $title, $action ) {
3115                 global $wgEnableScaryTranscluding;
3116
3117                 if (!$wgEnableScaryTranscluding)
3118                         return wfMsg('scarytranscludedisabled');
3119
3120                 $url = $title->getFullUrl( "action=$action" );
3121
3122                 if (strlen($url) > 255)
3123                         return wfMsg('scarytranscludetoolong');
3124                 return $this->fetchScaryTemplateMaybeFromCache($url);
3125         }
3126
3127         function fetchScaryTemplateMaybeFromCache($url) {
3128                 global $wgTranscludeCacheExpiry;
3129                 $dbr = wfGetDB(DB_SLAVE);
3130                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3131                                 array('tc_url' => $url));
3132                 if ($obj) {
3133                         $time = $obj->tc_time;
3134                         $text = $obj->tc_contents;
3135                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3136                                 return $text;
3137                         }
3138                 }
3139
3140                 $text = Http::get($url);
3141                 if (!$text)
3142                         return wfMsg('scarytranscludefailed', $url);
3143
3144                 $dbw = wfGetDB(DB_MASTER);
3145                 $dbw->replace('transcache', array('tc_url'), array(
3146                         'tc_url' => $url,
3147                         'tc_time' => time(),
3148                         'tc_contents' => $text));
3149                 return $text;
3150         }
3151
3152
3153         /**
3154          * Triple brace replacement -- used for template arguments
3155          * @private
3156          */
3157         function argSubstitution( $piece, $frame ) {
3158                 wfProfileIn( __METHOD__ );
3159
3160                 $error = false;
3161                 $parts = $piece['parts'];
3162                 $nameWithSpaces = $frame->expand( $piece['title'] );
3163                 $argName = trim( $nameWithSpaces );
3164                 $object = false;
3165                 $text = $frame->getArgument( $argName );
3166                 if (  $text === false && $parts->getLength() > 0
3167                   && (
3168                     $this->ot['html']
3169                     || $this->ot['pre']
3170                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3171                   )
3172                 ) {
3173                         # No match in frame, use the supplied default
3174                         $object = $parts->item( 0 )->getChildren();
3175                 }
3176                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3177                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3178                 }
3179
3180                 if ( $text === false && $object === false ) {
3181                         # No match anywhere
3182                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3183                 }
3184                 if ( $error !== false ) {
3185                         $text .= $error;
3186                 }
3187                 if ( $object !== false ) {
3188                         $ret = array( 'object' => $object );
3189                 } else {
3190                         $ret = array( 'text' => $text );
3191                 }
3192
3193                 wfProfileOut( __METHOD__ );
3194                 return $ret;
3195         }
3196
3197         /**
3198          * Return the text to be used for a given extension tag.
3199          * This is the ghost of strip().
3200          *
3201          * @param array $params Associative array of parameters:
3202          *     name       PPNode for the tag name
3203          *     attr       PPNode for unparsed text where tag attributes are thought to be
3204          *     attributes Optional associative array of parsed attributes
3205          *     inner      Contents of extension element
3206          *     noClose    Original text did not have a close tag
3207          * @param PPFrame $frame
3208          */
3209         function extensionSubstitution( $params, $frame ) {
3210                 global $wgRawHtml, $wgContLang;
3211
3212                 $name = $frame->expand( $params['name'] );
3213                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3214                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3215
3216                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3217
3218                 if ( $this->ot['html'] ) {
3219                         $name = strtolower( $name );
3220
3221                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3222                         if ( isset( $params['attributes'] ) ) {
3223                                 $attributes = $attributes + $params['attributes'];
3224                         }
3225                         switch ( $name ) {
3226                                 case 'html':
3227                                         if( $wgRawHtml ) {
3228                                                 $output = $content;
3229                                                 break;
3230                                         } else {
3231                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3232                                         }
3233                                 case 'nowiki':
3234                                         $output = Xml::escapeTagsOnly( $content );
3235                                         break;
3236                                 case 'math':
3237                                         $output = $wgContLang->armourMath(
3238                                                 MathRenderer::renderMath( $content, $attributes ) );
3239                                         break;
3240                                 case 'gallery':
3241                                         $output = $this->renderImageGallery( $content, $attributes );
3242                                         break;
3243                                 default:
3244                                         if( isset( $this->mTagHooks[$name] ) ) {
3245                                                 # Workaround for PHP bug 35229 and similar
3246                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3247                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3248                                                 }
3249                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3250                                                         array( $content, $attributes, $this ) );
3251                                         } else {
3252                                                 throw new MWException( "Invalid call hook $name" );
3253                                         }
3254                         }
3255                 } else {
3256                         if ( is_null( $attrText ) ) {
3257                                 $attrText = '';
3258                         }
3259                         if ( isset( $params['attributes'] ) ) {
3260                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3261                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3262                                                 htmlspecialchars( $attrValue ) . '"';
3263                                 }
3264                         }
3265                         if ( $content === null ) {
3266                                 $output = "<$name$attrText/>";
3267                         } else {
3268                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3269                                 $output = "<$name$attrText>$content$close";
3270                         }
3271                 }
3272
3273                 if ( $name == 'html' || $name == 'nowiki' ) {
3274                         $this->mStripState->nowiki->setPair( $marker, $output );
3275                 } else {
3276                         $this->mStripState->general->setPair( $marker, $output );
3277                 }
3278                 return $marker;
3279         }
3280
3281         /**
3282          * Increment an include size counter
3283          *
3284          * @param string $type The type of expansion
3285          * @param integer $size The size of the text
3286          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3287          */
3288         function incrementIncludeSize( $type, $size ) {
3289                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3290                         return false;
3291                 } else {
3292                         $this->mIncludeSizes[$type] += $size;
3293                         return true;
3294                 }
3295         }
3296
3297         /**
3298          * Increment the expensive function count
3299          *
3300          * @return boolean False if the limit has been exceeded
3301          */
3302         function incrementExpensiveFunctionCount() {
3303                 global $wgExpensiveParserFunctionLimit;
3304                 $this->mExpensiveFunctionCount++;
3305                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3306                         return true;
3307                 }
3308                 return false;
3309         }
3310
3311         /**
3312          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3313          * Fills $this->mDoubleUnderscores, returns the modified text
3314          */
3315         function doDoubleUnderscore( $text ) {
3316                 // The position of __TOC__ needs to be recorded
3317                 $mw = MagicWord::get( 'toc' );
3318                 if( $mw->match( $text ) ) {
3319                         $this->mShowToc = true;
3320                         $this->mForceTocPosition = true;
3321
3322                         // Set a placeholder. At the end we'll fill it in with the TOC.
3323                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3324
3325                         // Only keep the first one.
3326                         $text = $mw->replace( '', $text );
3327                 }
3328
3329                 // Now match and remove the rest of them
3330                 $mwa = MagicWord::getDoubleUnderscoreArray();
3331                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3332
3333                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3334                         $this->mOutput->mNoGallery = true;
3335                 }
3336                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3337                         $this->mShowToc = false;
3338                 }
3339                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3340                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3341
3342                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3343                         if ( $containerCategory ) {
3344                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3345                         } else {
3346                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3347                         }
3348                 }
3349                 return $text;
3350         }
3351
3352         /**
3353          * This function accomplishes several tasks:
3354          * 1) Auto-number headings if that option is enabled
3355          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3356          * 3) Add a Table of contents on the top for users who have enabled the option
3357          * 4) Auto-anchor headings
3358          *
3359          * It loops through all headlines, collects the necessary data, then splits up the
3360          * string and re-inserts the newly formatted headlines.
3361          *
3362          * @param string $text
3363          * @param boolean $isMain
3364          * @private
3365          */
3366         function formatHeadings( $text, $isMain=true ) {
3367                 global $wgMaxTocLevel, $wgContLang;
3368
3369                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3370                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3371                         $showEditLink = 0;
3372                 } else {
3373                         $showEditLink = $this->mOptions->getEditSection();
3374                 }
3375
3376                 # Inhibit editsection links if requested in the page
3377                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3378                         $showEditLink = 0;
3379                 }
3380
3381                 # Get all headlines for numbering them and adding funky stuff like [edit]
3382                 # links - this is for later, but we need the number of headlines right now
3383                 $matches = array();
3384                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3385
3386                 # if there are fewer than 4 headlines in the article, do not show TOC
3387                 # unless it's been explicitly enabled.
3388                 $enoughToc = $this->mShowToc &&
3389                         (($numMatches >= 4) || $this->mForceTocPosition);
3390
3391                 # Allow user to stipulate that a page should have a "new section"
3392                 # link added via __NEWSECTIONLINK__
3393                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3394                         $this->mOutput->setNewSection( true );
3395                 }
3396
3397                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3398                 # override above conditions and always show TOC above first header
3399                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3400                         $this->mShowToc = true;
3401                         $enoughToc = true;
3402                 }
3403
3404                 # We need this to perform operations on the HTML
3405                 $sk = $this->mOptions->getSkin();
3406
3407                 # headline counter
3408                 $headlineCount = 0;
3409                 $numVisible = 0;
3410
3411                 # Ugh .. the TOC should have neat indentation levels which can be
3412                 # passed to the skin functions. These are determined here
3413                 $toc = '';
3414                 $full = '';
3415                 $head = array();
3416                 $sublevelCount = array();
3417                 $levelCount = array();
3418                 $toclevel = 0;
3419                 $level = 0;
3420                 $prevlevel = 0;
3421                 $toclevel = 0;
3422                 $prevtoclevel = 0;
3423                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3424                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3425                 $tocraw = array();
3426
3427                 foreach( $matches[3] as $headline ) {
3428                         $isTemplate = false;
3429                         $titleText = false;
3430                         $sectionIndex = false;
3431                         $numbering = '';
3432                         $markerMatches = array();
3433                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3434                                 $serial = $markerMatches[1];
3435                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3436                                 $isTemplate = ($titleText != $baseTitleText);
3437                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3438                         }
3439
3440                         if( $toclevel ) {
3441                                 $prevlevel = $level;
3442                                 $prevtoclevel = $toclevel;
3443                         }
3444                         $level = $matches[1][$headlineCount];
3445
3446                         if( $doNumberHeadings || $enoughToc ) {
3447
3448                                 if ( $level > $prevlevel ) {
3449                                         # Increase TOC level
3450                                         $toclevel++;
3451                                         $sublevelCount[$toclevel] = 0;
3452                                         if( $toclevel<$wgMaxTocLevel ) {
3453                                                 $prevtoclevel = $toclevel;
3454                                                 $toc .= $sk->tocIndent();
3455                                                 $numVisible++;
3456                                         }
3457                                 }
3458                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3459                                         # Decrease TOC level, find level to jump to
3460
3461                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3462                                                 # Can only go down to level 1
3463                                                 $toclevel = 1;
3464                                         } else {
3465                                                 for ($i = $toclevel; $i > 0; $i--) {
3466                                                         if ( $levelCount[$i] == $level ) {
3467                                                                 # Found last matching level
3468                                                                 $toclevel = $i;
3469                                                                 break;
3470                                                         }
3471                                                         elseif ( $levelCount[$i] < $level ) {
3472                                                                 # Found first matching level below current level
3473                                                                 $toclevel = $i + 1;
3474                                                                 break;
3475                                                         }
3476                                                 }
3477                                         }
3478                                         if( $toclevel<$wgMaxTocLevel ) {
3479                                                 if($prevtoclevel < $wgMaxTocLevel) {
3480                                                         # Unindent only if the previous toc level was shown :p
3481                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3482                                                         $prevtoclevel = $toclevel;
3483                                                 } else {
3484                                                         $toc .= $sk->tocLineEnd();
3485                                                 }
3486                                         }
3487                                 }
3488                                 else {
3489                                         # No change in level, end TOC line
3490                                         if( $toclevel<$wgMaxTocLevel ) {
3491                                                 $toc .= $sk->tocLineEnd();
3492                                         }
3493                                 }
3494
3495                                 $levelCount[$toclevel] = $level;
3496
3497                                 # count number of headlines for each level
3498                                 @$sublevelCount[$toclevel]++;
3499                                 $dot = 0;
3500                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3501                                         if( !empty( $sublevelCount[$i] ) ) {
3502                                                 if( $dot ) {
3503                                                         $numbering .= '.';
3504                                                 }
3505                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3506                                                 $dot = 1;
3507                                         }
3508                                 }
3509                         }
3510
3511                         # The safe header is a version of the header text safe to use for links
3512                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3513                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3514
3515                         # Remove link placeholders by the link text.
3516                         #     <!--LINK number-->
3517                         # turns into
3518                         #     link text with suffix
3519                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3520                                                             "\$this->mLinkHolders['texts'][\$1]",
3521                                                             $safeHeadline );
3522                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3523                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3524                                                             $safeHeadline );
3525
3526                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3527                         $tocline = preg_replace(
3528                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3529                                 array( '',                          '<$1>'),
3530                                 $safeHeadline
3531                         );
3532                         $tocline = trim( $tocline );
3533
3534                         # For the anchor, strip out HTML-y stuff period
3535                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3536                         $safeHeadline = trim( $safeHeadline );
3537
3538                         # Save headline for section edit hint before it's escaped
3539                         $headlineHint = $safeHeadline;
3540                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3541                         # HTML names must be case-insensitively unique (bug 10721)
3542                         $arrayKey = strtolower( $safeHeadline );
3543
3544                         # XXX : Is $refers[$headlineCount] ever accessed, actually ?
3545                         $refers[$headlineCount] = $safeHeadline;
3546
3547                         # count how many in assoc. array so we can track dupes in anchors
3548                         isset( $refers[$arrayKey] ) ? $refers[$arrayKey]++ : $refers[$arrayKey] = 1;
3549                         $refcount[$headlineCount] = $refers[$arrayKey];
3550
3551                         # Don't number the heading if it is the only one (looks silly)
3552                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3553                                 # the two are different if the line contains a link
3554                                 $headline=$numbering . ' ' . $headline;
3555                         }
3556
3557                         # Create the anchor for linking from the TOC to the section
3558                         $anchor = $safeHeadline;
3559                         if($refcount[$headlineCount] > 1 ) {
3560                                 $anchor .= '_' . $refcount[$headlineCount];
3561                         }
3562                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3563                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3564                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3565                         }
3566                         # give headline the correct <h#> tag
3567                         if( $showEditLink && $sectionIndex !== false ) {
3568                                 if( $isTemplate ) {
3569                                         # Put a T flag in the section identifier, to indicate to extractSections()
3570                                         # that sections inside <includeonly> should be counted.
3571                                         $editlink = $sk->editSectionLinkForOther($titleText, "T-$sectionIndex");
3572                                 } else {
3573                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3574                                 }
3575                         } else {
3576                                 $editlink = '';
3577                         }
3578                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3579
3580                         $headlineCount++;
3581                 }
3582
3583                 $this->mOutput->setSections( $tocraw );
3584
3585                 # Never ever show TOC if no headers
3586                 if( $numVisible < 1 ) {
3587                         $enoughToc = false;
3588                 }
3589
3590                 if( $enoughToc ) {
3591                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3592                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3593                         }
3594                         $toc = $sk->tocList( $toc );
3595                 }
3596
3597                 # split up and insert constructed headlines
3598
3599                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3600                 $i = 0;
3601
3602                 foreach( $blocks as $block ) {
3603                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3604                                 # This is the [edit] link that appears for the top block of text when
3605                                 # section editing is enabled
3606
3607                                 # Disabled because it broke block formatting
3608                                 # For example, a bullet point in the top line
3609                                 # $full .= $sk->editSectionLink(0);
3610                         }
3611                         $full .= $block;
3612                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3613                                 # Top anchor now in skin
3614                                 $full = $full.$toc;
3615                         }
3616
3617                         if( !empty( $head[$i] ) ) {
3618                                 $full .= $head[$i];
3619                         }
3620                         $i++;
3621                 }
3622                 if( $this->mForceTocPosition ) {
3623                         return str_replace( '<!--MWTOC-->', $toc, $full );
3624                 } else {
3625                         return $full;
3626                 }
3627         }
3628
3629         /**
3630          * Transform wiki markup when saving a page by doing \r\n -> \n
3631          * conversion, substitting signatures, {{subst:}} templates, etc.
3632          *
3633          * @param string $text the text to transform
3634          * @param Title &$title the Title object for the current article
3635          * @param User &$user the User object describing the current user
3636          * @param ParserOptions $options parsing options
3637          * @param bool $clearState whether to clear the parser state first
3638          * @return string the altered wiki markup
3639          * @public
3640          */
3641         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3642                 $this->mOptions = $options;
3643                 $this->setTitle( $title );
3644                 $this->setOutputType( self::OT_WIKI );
3645
3646                 if ( $clearState ) {
3647                         $this->clearState();
3648                 }
3649
3650                 $pairs = array(
3651                         "\r\n" => "\n",
3652                 );
3653                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3654                 $text = $this->pstPass2( $text, $user );
3655                 $text = $this->mStripState->unstripBoth( $text );
3656                 return $text;
3657         }
3658
3659         /**
3660          * Pre-save transform helper function
3661          * @private
3662          */
3663         function pstPass2( $text, $user ) {
3664                 global $wgContLang, $wgLocaltimezone;
3665
3666                 /* Note: This is the timestamp saved as hardcoded wikitext to
3667                  * the database, we use $wgContLang here in order to give
3668                  * everyone the same signature and use the default one rather
3669                  * than the one selected in each user's preferences.
3670                  *
3671                  * (see also bug 12815)
3672                  */
3673                 $ts = $this->mOptions->getTimestamp();
3674                 $tz = 'UTC';
3675                 if ( isset( $wgLocaltimezone ) ) {
3676                         $unixts = wfTimestamp( TS_UNIX, $ts );
3677                         $oldtz = getenv( 'TZ' );
3678                         putenv( 'TZ='.$wgLocaltimezone );
3679                         $ts = date( 'YmdHis', $unixts );
3680                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3681                         putenv( 'TZ='.$oldtz );
3682                 }
3683                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3684
3685                 # Variable replacement
3686                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3687                 $text = $this->replaceVariables( $text );
3688
3689                 # Signatures
3690                 $sigText = $this->getUserSig( $user );
3691                 $text = strtr( $text, array(
3692                         '~~~~~' => $d,
3693                         '~~~~' => "$sigText $d",
3694                         '~~~' => $sigText
3695                 ) );
3696
3697                 # Context links: [[|name]] and [[name (context)|]]
3698                 #
3699                 global $wgLegalTitleChars;
3700                 $tc = "[$wgLegalTitleChars]";
3701                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3702
3703                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3704                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3705                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3706
3707                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3708                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3709                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3710
3711                 $t = $this->mTitle->getText();
3712                 $m = array();
3713                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3714                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3715                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3716                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3717                 } else {
3718                         # if there's no context, don't bother duplicating the title
3719                         $text = preg_replace( $p2, '[[\\1]]', $text );
3720                 }
3721
3722                 # Trim trailing whitespace
3723                 $text = rtrim( $text );
3724
3725                 return $text;
3726         }
3727
3728         /**
3729          * Fetch the user's signature text, if any, and normalize to
3730          * validated, ready-to-insert wikitext.
3731          *
3732          * @param User $user
3733          * @return string
3734          * @private
3735          */
3736         function getUserSig( &$user ) {
3737                 global $wgMaxSigChars;
3738
3739                 $username = $user->getName();
3740                 $nickname = $user->getOption( 'nickname' );
3741                 $nickname = $nickname === '' ? $username : $nickname;
3742
3743                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3744                         $nickname = $username;
3745                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3746                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3747                         # Sig. might contain markup; validate this
3748                         if( $this->validateSig( $nickname ) !== false ) {
3749                                 # Validated; clean up (if needed) and return it
3750                                 return $this->cleanSig( $nickname, true );
3751                         } else {
3752                                 # Failed to validate; fall back to the default
3753                                 $nickname = $username;
3754                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3755                         }
3756                 }
3757
3758                 // Make sure nickname doesnt get a sig in a sig
3759                 $nickname = $this->cleanSigInSig( $nickname );
3760
3761                 # If we're still here, make it a link to the user page
3762                 $userText = wfEscapeWikiText( $username );
3763                 $nickText = wfEscapeWikiText( $nickname );
3764                 if ( $user->isAnon() )  {
3765                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3766                 } else {
3767                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3768                 }
3769         }
3770
3771         /**
3772          * Check that the user's signature contains no bad XML
3773          *
3774          * @param string $text
3775          * @return mixed An expanded string, or false if invalid.
3776          */
3777         function validateSig( $text ) {
3778                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3779         }
3780
3781         /**
3782          * Clean up signature text
3783          *
3784          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3785          * 2) Substitute all transclusions
3786          *
3787          * @param string $text
3788          * @param $parsing Whether we're cleaning (preferences save) or parsing
3789          * @return string Signature text
3790          */
3791         function cleanSig( $text, $parsing = false ) {
3792                 if ( !$parsing ) {
3793                         global $wgTitle;
3794                         $this->clearState();
3795                         $this->setTitle( $wgTitle );
3796                         $this->mOptions = new ParserOptions;
3797                         $this->setOutputType = self::OT_PREPROCESS;
3798                 }
3799
3800                 # FIXME: regex doesn't respect extension tags or nowiki
3801                 #  => Move this logic to braceSubstitution()
3802                 $substWord = MagicWord::get( 'subst' );
3803                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3804                 $substText = '{{' . $substWord->getSynonym( 0 );
3805
3806                 $text = preg_replace( $substRegex, $substText, $text );
3807                 $text = $this->cleanSigInSig( $text );
3808                 $dom = $this->preprocessToDom( $text );
3809                 $frame = $this->getPreprocessor()->newFrame();
3810                 $text = $frame->expand( $dom );
3811
3812                 if ( !$parsing ) {
3813                         $text = $this->mStripState->unstripBoth( $text );
3814                 }
3815
3816                 return $text;
3817         }
3818
3819         /**
3820          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3821          * @param string $text
3822          * @return string Signature text with /~{3,5}/ removed
3823          */
3824         function cleanSigInSig( $text ) {
3825                 $text = preg_replace( '/~{3,5}/', '', $text );
3826                 return $text;
3827         }
3828
3829         /**
3830          * Set up some variables which are usually set up in parse()
3831          * so that an external function can call some class members with confidence
3832          * @public
3833          */
3834         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3835                 $this->setTitle( $title );
3836                 $this->mOptions = $options;
3837                 $this->setOutputType( $outputType );
3838                 if ( $clearState ) {
3839                         $this->clearState();
3840                 }
3841         }
3842
3843         /**
3844          * Wrapper for preprocess()
3845          *
3846          * @param string $text the text to preprocess
3847          * @param ParserOptions $options  options
3848          * @return string
3849          * @public
3850          */
3851         function transformMsg( $text, $options ) {
3852                 global $wgTitle;
3853                 static $executing = false;
3854
3855                 $fname = "Parser::transformMsg";
3856
3857                 # Guard against infinite recursion
3858                 if ( $executing ) {
3859                         return $text;
3860                 }
3861                 $executing = true;
3862
3863                 wfProfileIn($fname);
3864                 $text = $this->preprocess( $text, $wgTitle, $options );
3865
3866                 $executing = false;
3867                 wfProfileOut($fname);
3868                 return $text;
3869         }
3870
3871         /**
3872          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3873          * The callback should have the following form:
3874          *    function myParserHook( $text, $params, &$parser ) { ... }
3875          *
3876          * Transform and return $text. Use $parser for any required context, e.g. use
3877          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3878          *
3879          * @public
3880          *
3881          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3882          * @param mixed $callback The callback function (and object) to use for the tag
3883          *
3884          * @return The old value of the mTagHooks array associated with the hook
3885          */
3886         function setHook( $tag, $callback ) {
3887                 $tag = strtolower( $tag );
3888                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3889                 $this->mTagHooks[$tag] = $callback;
3890                 if( !in_array( $tag, $this->mStripList ) ) {
3891                         $this->mStripList[] = $tag;
3892                 }
3893
3894                 return $oldVal;
3895         }
3896
3897         function setTransparentTagHook( $tag, $callback ) {
3898                 $tag = strtolower( $tag );
3899                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3900                 $this->mTransparentTagHooks[$tag] = $callback;
3901
3902                 return $oldVal;
3903         }
3904
3905         /**
3906          * Remove all tag hooks
3907          */
3908         function clearTagHooks() {
3909                 $this->mTagHooks = array();
3910                 $this->mStripList = $this->mDefaultStripList;
3911         }
3912
3913         /**
3914          * Create a function, e.g. {{sum:1|2|3}}
3915          * The callback function should have the form:
3916          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3917          *
3918          * The callback may either return the text result of the function, or an array with the text
3919          * in element 0, and a number of flags in the other elements. The names of the flags are
3920          * specified in the keys. Valid flags are:
3921          *   found                     The text returned is valid, stop processing the template. This
3922          *                             is on by default.
3923          *   nowiki                    Wiki markup in the return value should be escaped
3924          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3925          *
3926          * @public
3927          *
3928          * @param string $id The magic word ID
3929          * @param mixed $callback The callback function (and object) to use
3930          * @param integer $flags a combination of the following flags:
3931          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3932          *
3933          * @return The old callback function for this name, if any
3934          */
3935         function setFunctionHook( $id, $callback, $flags = 0 ) {
3936                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
3937                 $this->mFunctionHooks[$id] = array( $callback, $flags );
3938
3939                 # Add to function cache
3940                 $mw = MagicWord::get( $id );
3941                 if( !$mw )
3942                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3943
3944                 $synonyms = $mw->getSynonyms();
3945                 $sensitive = intval( $mw->isCaseSensitive() );
3946
3947                 foreach ( $synonyms as $syn ) {
3948                         # Case
3949                         if ( !$sensitive ) {
3950                                 $syn = strtolower( $syn );
3951                         }
3952                         # Add leading hash
3953                         if ( !( $flags & SFH_NO_HASH ) ) {
3954                                 $syn = '#' . $syn;
3955                         }
3956                         # Remove trailing colon
3957                         if ( substr( $syn, -1, 1 ) == ':' ) {
3958                                 $syn = substr( $syn, 0, -1 );
3959                         }
3960                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3961                 }
3962                 return $oldVal;
3963         }
3964
3965         /**
3966          * Get all registered function hook identifiers
3967          *
3968          * @return array
3969          */
3970         function getFunctionHooks() {
3971                 return array_keys( $this->mFunctionHooks );
3972         }
3973
3974         /**
3975          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3976          * Placeholders created in Skin::makeLinkObj()
3977          * Returns an array of link CSS classes, indexed by PDBK.
3978          * $options is a bit field, RLH_FOR_UPDATE to select for update
3979          */
3980         function replaceLinkHolders( &$text, $options = 0 ) {
3981                 global $wgUser;
3982                 global $wgContLang;
3983
3984                 $fname = 'Parser::replaceLinkHolders';
3985                 wfProfileIn( $fname );
3986
3987                 $pdbks = array();
3988                 $colours = array();
3989                 $linkcolour_ids = array();
3990                 $sk = $this->mOptions->getSkin();
3991                 $linkCache = LinkCache::singleton();
3992
3993                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3994                         wfProfileIn( $fname.'-check' );
3995                         $dbr = wfGetDB( DB_SLAVE );
3996                         $page = $dbr->tableName( 'page' );
3997                         $threshold = $wgUser->getOption('stubthreshold');
3998
3999                         # Sort by namespace
4000                         asort( $this->mLinkHolders['namespaces'] );
4001
4002                         # Generate query
4003                         $query = false;
4004                         $current = null;
4005                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4006                                 # Make title object
4007                                 $title = $this->mLinkHolders['titles'][$key];
4008
4009                                 # Skip invalid entries.
4010                                 # Result will be ugly, but prevents crash.
4011                                 if ( is_null( $title ) ) {
4012                                         continue;
4013                                 }
4014                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4015
4016                                 # Check if it's a static known link, e.g. interwiki
4017                                 if ( $title->isAlwaysKnown() ) {
4018                                         $colours[$pdbk] = '';
4019                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4020                                         $colours[$pdbk] = '';
4021                                         $this->mOutput->addLink( $title, $id );
4022                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4023                                         $colours[$pdbk] = 'new';
4024                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4025                                         $colours[$pdbk] = 'new';
4026                                 } else {
4027                                         # Not in the link cache, add it to the query
4028                                         if ( !isset( $current ) ) {
4029                                                 $current = $ns;
4030                                                 $query =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
4031                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4032                                         } elseif ( $current != $ns ) {
4033                                                 $current = $ns;
4034                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4035                                         } else {
4036                                                 $query .= ', ';
4037                                         }
4038
4039                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4040                                 }
4041                         }
4042                         if ( $query ) {
4043                                 $query .= '))';
4044                                 if ( $options & RLH_FOR_UPDATE ) {
4045                                         $query .= ' FOR UPDATE';
4046                                 }
4047
4048                                 $res = $dbr->query( $query, $fname );
4049
4050                                 # Fetch data and form into an associative array
4051                                 # non-existent = broken
4052                                 while ( $s = $dbr->fetchObject($res) ) {
4053                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4054                                         $pdbk = $title->getPrefixedDBkey();
4055                                         $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect );
4056                                         $this->mOutput->addLink( $title, $s->page_id );
4057                                         $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
4058                                         //add id to the extension todolist
4059                                         $linkcolour_ids[$s->page_id] = $pdbk;
4060                                 }
4061                                 //pass an array of page_ids to an extension
4062                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4063                         }
4064                         wfProfileOut( $fname.'-check' );
4065
4066                         # Do a second query for different language variants of links and categories
4067                         if($wgContLang->hasVariants()){
4068                                 $linkBatch = new LinkBatch();
4069                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4070                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4071                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4072
4073                                 $categories = $this->mOutput->getCategoryLinks();
4074
4075                                 // Add variants of links to link batch
4076                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4077                                         $title = $this->mLinkHolders['titles'][$key];
4078                                         if ( is_null( $title ) )
4079                                                 continue;
4080
4081                                         $pdbk = $title->getPrefixedDBkey();
4082                                         $titleText = $title->getText();
4083
4084                                         // generate all variants of the link title text
4085                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4086
4087                                         // if link was not found (in first query), add all variants to query
4088                                         if ( !isset($colours[$pdbk]) ){
4089                                                 foreach($allTextVariants as $textVariant){
4090                                                         if($textVariant != $titleText){
4091                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4092                                                                 if(is_null($variantTitle)) continue;
4093                                                                 $linkBatch->addObj( $variantTitle );
4094                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4095                                                         }
4096                                                 }
4097                                         }
4098                                 }
4099
4100                                 // process categories, check if a category exists in some variant
4101                                 foreach( $categories as $category ){
4102                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4103                                         foreach($variants as $variant){
4104                                                 if($variant != $category){
4105                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4106                                                         if(is_null($variantTitle)) continue;
4107                                                         $linkBatch->addObj( $variantTitle );
4108                                                         $categoryMap[$variant] = $category;
4109                                                 }
4110                                         }
4111                                 }
4112
4113
4114                                 if(!$linkBatch->isEmpty()){
4115                                         // construct query
4116                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4117
4118                                         $variantQuery =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
4119
4120                                         $variantQuery .= " FROM $page WHERE $titleClause";
4121                                         if ( $options & RLH_FOR_UPDATE ) {
4122                                                 $variantQuery .= ' FOR UPDATE';
4123                                         }
4124
4125                                         $varRes = $dbr->query( $variantQuery, $fname );
4126
4127                                         // for each found variants, figure out link holders and replace
4128                                         while ( $s = $dbr->fetchObject($varRes) ) {
4129
4130                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4131                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4132                                                 $vardbk = $variantTitle->getDBkey();
4133
4134                                                 $holderKeys = array();
4135                                                 if(isset($variantMap[$varPdbk])){
4136                                                         $holderKeys = $variantMap[$varPdbk];
4137                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
4138                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4139                                                 }
4140
4141                                                 // loop over link holders
4142                                                 foreach($holderKeys as $key){
4143                                                         $title = $this->mLinkHolders['titles'][$key];
4144                                                         if ( is_null( $title ) ) continue;
4145
4146                                                         $pdbk = $title->getPrefixedDBkey();
4147
4148                                                         if(!isset($colours[$pdbk])){
4149                                                                 // found link in some of the variants, replace the link holder data
4150                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4151                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4152
4153                                                                 // set pdbk and colour
4154                                                                 $pdbks[$key] = $varPdbk;
4155                                                                 $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );
4156                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4157                                                         }
4158                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4159                                                 }
4160
4161                                                 // check if the object is a variant of a category
4162                                                 if(isset($categoryMap[$vardbk])){
4163                                                         $oldkey = $categoryMap[$vardbk];
4164                                                         if($oldkey != $vardbk)
4165                                                                 $varCategories[$oldkey]=$vardbk;
4166                                                 }
4167                                         }
4168
4169                                         // rebuild the categories in original order (if there are replacements)
4170                                         if(count($varCategories)>0){
4171                                                 $newCats = array();
4172                                                 $originalCats = $this->mOutput->getCategories();
4173                                                 foreach($originalCats as $cat => $sortkey){
4174                                                         // make the replacement
4175                                                         if( array_key_exists($cat,$varCategories) )
4176                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4177                                                         else $newCats[$cat] = $sortkey;
4178                                                 }
4179                                                 $this->mOutput->setCategoryLinks($newCats);
4180                                         }
4181                                 }
4182                         }
4183
4184                         # Construct search and replace arrays
4185                         wfProfileIn( $fname.'-construct' );
4186                         $replacePairs = array();
4187                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4188                                 $pdbk = $pdbks[$key];
4189                                 $searchkey = "<!--LINK $key-->";
4190                                 $title = $this->mLinkHolders['titles'][$key];
4191                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4192                                         $linkCache->addBadLinkObj( $title );
4193                                         $colours[$pdbk] = 'new';
4194                                         $this->mOutput->addLink( $title, 0 );
4195                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4196                                                                         $this->mLinkHolders['texts'][$key],
4197                                                                         $this->mLinkHolders['queries'][$key] );
4198                                 } else {
4199                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4200                                                                         $this->mLinkHolders['texts'][$key],
4201                                                                         $this->mLinkHolders['queries'][$key] );
4202                                 }
4203                         }
4204                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4205                         wfProfileOut( $fname.'-construct' );
4206
4207                         # Do the thing
4208                         wfProfileIn( $fname.'-replace' );
4209                         $text = preg_replace_callback(
4210                                 '/(<!--LINK .*?-->)/',
4211                                 $replacer->cb(),
4212                                 $text);
4213
4214                         wfProfileOut( $fname.'-replace' );
4215                 }
4216
4217                 # Now process interwiki link holders
4218                 # This is quite a bit simpler than internal links
4219                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4220                         wfProfileIn( $fname.'-interwiki' );
4221                         # Make interwiki link HTML
4222                         $replacePairs = array();
4223                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4224                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4225                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4226                         }
4227                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4228
4229                         $text = preg_replace_callback(
4230                                 '/<!--IWLINK (.*?)-->/',
4231                                 $replacer->cb(),
4232                                 $text );
4233                         wfProfileOut( $fname.'-interwiki' );
4234                 }
4235
4236                 wfProfileOut( $fname );
4237                 return $colours;
4238         }
4239
4240         /**
4241          * Replace <!--LINK--> link placeholders with plain text of links
4242          * (not HTML-formatted).
4243          * @param string $text
4244          * @return string
4245          */
4246         function replaceLinkHoldersText( $text ) {
4247                 $fname = 'Parser::replaceLinkHoldersText';
4248                 wfProfileIn( $fname );
4249
4250                 $text = preg_replace_callback(
4251                         '/<!--(LINK|IWLINK) (.*?)-->/',
4252                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4253                         $text );
4254
4255                 wfProfileOut( $fname );
4256                 return $text;
4257         }
4258
4259         /**
4260          * @param array $matches
4261          * @return string
4262          * @private
4263          */
4264         function replaceLinkHoldersTextCallback( $matches ) {
4265                 $type = $matches[1];
4266                 $key  = $matches[2];
4267                 if( $type == 'LINK' ) {
4268                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4269                                 return $this->mLinkHolders['texts'][$key];
4270                         }
4271                 } elseif( $type == 'IWLINK' ) {
4272                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4273                                 return $this->mInterwikiLinkHolders['texts'][$key];
4274                         }
4275                 }
4276                 return $matches[0];
4277         }
4278
4279         /**
4280          * Tag hook handler for 'pre'.
4281          */
4282         function renderPreTag( $text, $attribs ) {
4283                 // Backwards-compatibility hack
4284                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4285
4286                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4287                 return wfOpenElement( 'pre', $attribs ) .
4288                         Xml::escapeTagsOnly( $content ) .
4289                         '</pre>';
4290         }
4291
4292         /**
4293          * Renders an image gallery from a text with one line per image.
4294          * text labels may be given by using |-style alternative text. E.g.
4295          *   Image:one.jpg|The number "1"
4296          *   Image:tree.jpg|A tree
4297          * given as text will return the HTML of a gallery with two images,
4298          * labeled 'The number "1"' and
4299          * 'A tree'.
4300          */
4301         function renderImageGallery( $text, $params ) {
4302                 $ig = new ImageGallery();
4303                 $ig->setContextTitle( $this->mTitle );
4304                 $ig->setShowBytes( false );
4305                 $ig->setShowFilename( false );
4306                 $ig->setParser( $this );
4307                 $ig->setHideBadImages();
4308                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4309                 $ig->useSkin( $this->mOptions->getSkin() );
4310                 $ig->mRevisionId = $this->mRevisionId;
4311
4312                 if( isset( $params['caption'] ) ) {
4313                         $caption = $params['caption'];
4314                         $caption = htmlspecialchars( $caption );
4315                         $caption = $this->replaceInternalLinks( $caption );
4316                         $ig->setCaptionHtml( $caption );
4317                 }
4318                 if( isset( $params['perrow'] ) ) {
4319                         $ig->setPerRow( $params['perrow'] );
4320                 }
4321                 if( isset( $params['widths'] ) ) {
4322                         $ig->setWidths( $params['widths'] );
4323                 }
4324                 if( isset( $params['heights'] ) ) {
4325                         $ig->setHeights( $params['heights'] );
4326                 }
4327
4328                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4329
4330                 $lines = explode( "\n", $text );
4331                 foreach ( $lines as $line ) {
4332                         # match lines like these:
4333                         # Image:someimage.jpg|This is some image
4334                         $matches = array();
4335                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4336                         # Skip empty lines
4337                         if ( count( $matches ) == 0 ) {
4338                                 continue;
4339                         }
4340                         $tp = Title::newFromText( $matches[1] );
4341                         $nt =& $tp;
4342                         if( is_null( $nt ) ) {
4343                                 # Bogus title. Ignore these so we don't bomb out later.
4344                                 continue;
4345                         }
4346                         if ( isset( $matches[3] ) ) {
4347                                 $label = $matches[3];
4348                         } else {
4349                                 $label = '';
4350                         }
4351
4352                         $html = $this->recursiveTagParse( trim( $label ) );
4353
4354                         $ig->add( $nt, $html );
4355
4356                         # Only add real images (bug #5586)
4357                         if ( $nt->getNamespace() == NS_IMAGE ) {
4358                                 $this->mOutput->addImage( $nt->getDBkey() );
4359                         }
4360                 }
4361                 return $ig->toHTML();
4362         }
4363
4364         function getImageParams( $handler ) {
4365                 if ( $handler ) {
4366                         $handlerClass = get_class( $handler );
4367                 } else {
4368                         $handlerClass = '';
4369                 }
4370                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4371                         // Initialise static lists
4372                         static $internalParamNames = array(
4373                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4374                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4375                                         'bottom', 'text-bottom' ),
4376                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4377                                         'upright', 'border' ),
4378                         );
4379                         static $internalParamMap;
4380                         if ( !$internalParamMap ) {
4381                                 $internalParamMap = array();
4382                                 foreach ( $internalParamNames as $type => $names ) {
4383                                         foreach ( $names as $name ) {
4384                                                 $magicName = str_replace( '-', '_', "img_$name" );
4385                                                 $internalParamMap[$magicName] = array( $type, $name );
4386                                         }
4387                                 }
4388                         }
4389
4390                         // Add handler params
4391                         $paramMap = $internalParamMap;
4392                         if ( $handler ) {
4393                                 $handlerParamMap = $handler->getParamMap();
4394                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4395                                         $paramMap[$magic] = array( 'handler', $paramName );
4396                                 }
4397                         }
4398                         $this->mImageParams[$handlerClass] = $paramMap;
4399                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4400                 }
4401                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4402         }
4403
4404         /**
4405          * Parse image options text and use it to make an image
4406          */
4407         function makeImage( $title, $options ) {
4408                 # Check if the options text is of the form "options|alt text"
4409                 # Options are:
4410                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4411                 #  * left               no resizing, just left align. label is used for alt= only
4412                 #  * right              same, but right aligned
4413                 #  * none               same, but not aligned
4414                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4415                 #  * center             center the image
4416                 #  * framed             Keep original image size, no magnify-button.
4417                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4418                 #  * upright            reduce width for upright images, rounded to full __0 px
4419                 #  * border             draw a 1px border around the image
4420                 # vertical-align values (no % or length right now):
4421                 #  * baseline
4422                 #  * sub
4423                 #  * super
4424                 #  * top
4425                 #  * text-top
4426                 #  * middle
4427                 #  * bottom
4428                 #  * text-bottom
4429
4430                 $parts = array_map( 'trim', explode( '|', $options) );
4431                 $sk = $this->mOptions->getSkin();
4432
4433                 # Give extensions a chance to select the file revision for us
4434                 $skip = $time = false;
4435                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4436
4437                 if ( $skip ) {
4438                         return $sk->makeLinkObj( $title );
4439                 }
4440
4441                 # Get parameter map
4442                 $file = wfFindFile( $title, $time );
4443                 $handler = $file ? $file->getHandler() : false;
4444
4445                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4446
4447                 # Process the input parameters
4448                 $caption = '';
4449                 $params = array( 'frame' => array(), 'handler' => array(),
4450                         'horizAlign' => array(), 'vertAlign' => array() );
4451                 foreach( $parts as $part ) {
4452                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4453                         $validated = false;
4454                         if( isset( $paramMap[$magicName] ) ) {
4455                                 list( $type, $paramName ) = $paramMap[$magicName];
4456
4457                                 // Special case; width and height come in one variable together
4458                                 if( $type == 'handler' && $paramName == 'width' ) {
4459                                         $m = array();
4460                                         # (bug 13500) In both cases (width/height and width only),
4461                                         # permit trailing "px" for backward compatibility.
4462                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4463                                                 $width = intval( $m[1] );
4464                                                 $height = intval( $m[2] );
4465                                                 if ( $handler->validateParam( 'width', $width ) ) {
4466                                                         $params[$type]['width'] = $width;
4467                                                         $validated = true;
4468                                                 }
4469                                                 if ( $handler->validateParam( 'height', $height ) ) {
4470                                                         $params[$type]['height'] = $height;
4471                                                         $validated = true;
4472                                                 }
4473                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4474                                                 $width = intval( $value );
4475                                                 if ( $handler->validateParam( 'width', $width ) ) {
4476                                                         $params[$type]['width'] = $width;
4477                                                         $validated = true;
4478                                                 }
4479                                         } // else no validation -- bug 13436
4480                                 } else {
4481                                         if ( $type == 'handler' ) {
4482                                                 # Validate handler parameter
4483                                                 $validated = $handler->validateParam( $paramName, $value );
4484                                         } else {
4485                                                 # Validate internal parameters
4486                                                 switch( $paramName ) {
4487                                                 case "manualthumb":
4488                                                         /// @fixme - possibly check validity here?
4489                                                         /// downstream behavior seems odd with missing manual thumbs.
4490                                                         $validated = true;
4491                                                         break;
4492                                                 default:
4493                                                         // Most other things appear to be empty or numeric...
4494                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4495                                                 }
4496                                         }
4497
4498                                         if ( $validated ) {
4499                                                 $params[$type][$paramName] = $value;
4500                                         }
4501                                 }
4502                         }
4503                         if ( !$validated ) {
4504                                 $caption = $part;
4505                         }
4506                 }
4507
4508                 # Process alignment parameters
4509                 if ( $params['horizAlign'] ) {
4510                         $params['frame']['align'] = key( $params['horizAlign'] );
4511                 }
4512                 if ( $params['vertAlign'] ) {
4513                         $params['frame']['valign'] = key( $params['vertAlign'] );
4514                 }
4515
4516                 # Strip bad stuff out of the alt text
4517                 $alt = $this->replaceLinkHoldersText( $caption );
4518
4519                 # make sure there are no placeholders in thumbnail attributes
4520                 # that are later expanded to html- so expand them now and
4521                 # remove the tags
4522                 $alt = $this->mStripState->unstripBoth( $alt );
4523                 $alt = Sanitizer::stripAllTags( $alt );
4524
4525                 $params['frame']['alt'] = $alt;
4526                 $params['frame']['caption'] = $caption;
4527
4528                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4529
4530                 # Linker does the rest
4531                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time );
4532
4533                 # Give the handler a chance to modify the parser object
4534                 if ( $handler ) {
4535                         $handler->parserTransformHook( $this, $file );
4536                 }
4537
4538                 return $ret;
4539         }
4540
4541         /**
4542          * Set a flag in the output object indicating that the content is dynamic and
4543          * shouldn't be cached.
4544          */
4545         function disableCache() {
4546                 wfDebug( "Parser output marked as uncacheable.\n" );
4547                 $this->mOutput->mCacheTime = -1;
4548         }
4549
4550         /**#@+
4551          * Callback from the Sanitizer for expanding items found in HTML attribute
4552          * values, so they can be safely tested and escaped.
4553          * @param string $text
4554          * @param PPFrame $frame
4555          * @return string
4556          * @private
4557          */
4558         function attributeStripCallback( &$text, $frame = false ) {
4559                 $text = $this->replaceVariables( $text, $frame );
4560                 $text = $this->mStripState->unstripBoth( $text );
4561                 return $text;
4562         }
4563
4564         /**#@-*/
4565
4566         /**#@+
4567          * Accessor/mutator
4568          */
4569         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4570         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4571         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4572         /**#@-*/
4573
4574         /**#@+
4575          * Accessor
4576          */
4577         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4578         /**#@-*/
4579
4580
4581         /**
4582          * Break wikitext input into sections, and either pull or replace
4583          * some particular section's text.
4584          *
4585          * External callers should use the getSection and replaceSection methods.
4586          *
4587          * @param string $text Page wikitext
4588          * @param string $section A section identifier string of the form:
4589          *   <flag1> - <flag2> - ... - <section number>
4590          *
4591          * Currently the only recognised flag is "T", which means the target section number
4592          * was derived during a template inclusion parse, in other words this is a template
4593          * section edit link. If no flags are given, it was an ordinary section edit link.
4594          * This flag is required to avoid a section numbering mismatch when a section is
4595          * enclosed by <includeonly> (bug 6563).
4596          *
4597          * The section number 0 pulls the text before the first heading; other numbers will
4598          * pull the given section along with its lower-level subsections. If the section is
4599          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4600          *
4601          * @param string $mode One of "get" or "replace"
4602          * @param string $newText Replacement text for section data.
4603          * @return string for "get", the extracted section text.
4604          *                for "replace", the whole page with the section replaced.
4605          */
4606         private function extractSections( $text, $section, $mode, $newText='' ) {
4607                 global $wgTitle;
4608                 $this->clearState();
4609                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4610                 $this->mOptions = new ParserOptions;
4611                 $this->setOutputType( self::OT_WIKI );
4612                 $outText = '';
4613                 $frame = $this->getPreprocessor()->newFrame();
4614
4615                 // Process section extraction flags
4616                 $flags = 0;
4617                 $sectionParts = explode( '-', $section );
4618                 $sectionIndex = array_pop( $sectionParts );
4619                 foreach ( $sectionParts as $part ) {
4620                         if ( $part == 'T' ) {
4621                                 $flags |= self::PTD_FOR_INCLUSION;
4622                         }
4623                 }
4624                 // Preprocess the text
4625                 $root = $this->preprocessToDom( $text, $flags );
4626
4627                 // <h> nodes indicate section breaks
4628                 // They can only occur at the top level, so we can find them by iterating the root's children
4629                 $node = $root->getFirstChild();
4630
4631                 // Find the target section
4632                 if ( $sectionIndex == 0 ) {
4633                         // Section zero doesn't nest, level=big
4634                         $targetLevel = 1000;
4635                 } else {
4636             while ( $node ) {
4637                 if ( $node->getName() == 'h' ) {
4638                     $bits = $node->splitHeading();
4639                                         if ( $bits['i'] == $sectionIndex ) {
4640                                         $targetLevel = $bits['level'];
4641                                                 break;
4642                                         }
4643                                 }
4644                                 if ( $mode == 'replace' ) {
4645                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4646                                 }
4647                                 $node = $node->getNextSibling();
4648                         }
4649                 }
4650
4651                 if ( !$node ) {
4652                         // Not found
4653                         if ( $mode == 'get' ) {
4654                                 return $newText;
4655                         } else {
4656                                 return $text;
4657                         }
4658                 }
4659
4660                 // Find the end of the section, including nested sections
4661                 do {
4662                         if ( $node->getName() == 'h' ) {
4663                                 $bits = $node->splitHeading();
4664                                 $curLevel = $bits['level'];
4665                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4666                                         break;
4667                                 }
4668                         }
4669                         if ( $mode == 'get' ) {
4670                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4671                         }
4672                         $node = $node->getNextSibling();
4673                 } while ( $node );
4674
4675                 // Write out the remainder (in replace mode only)
4676                 if ( $mode == 'replace' ) {
4677                         // Output the replacement text
4678                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4679                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4680                         $outText .= $newText . "\n\n";
4681                         while ( $node ) {
4682                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4683                                 $node = $node->getNextSibling();
4684                         }
4685                 }
4686
4687                 if ( is_string( $outText ) ) {
4688                         // Re-insert stripped tags
4689                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4690                 }
4691
4692                 return $outText;
4693         }
4694
4695         /**
4696          * This function returns the text of a section, specified by a number ($section).
4697          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4698          * the first section before any such heading (section 0).
4699          *
4700          * If a section contains subsections, these are also returned.
4701          *
4702          * @param string $text text to look in
4703          * @param string $section section identifier
4704          * @param string $deftext default to return if section is not found
4705          * @return string text of the requested section
4706          */
4707         public function getSection( $text, $section, $deftext='' ) {
4708                 return $this->extractSections( $text, $section, "get", $deftext );
4709         }
4710
4711         public function replaceSection( $oldtext, $section, $text ) {
4712                 return $this->extractSections( $oldtext, $section, "replace", $text );
4713         }
4714
4715         /**
4716          * Get the timestamp associated with the current revision, adjusted for
4717          * the default server-local timestamp
4718          */
4719         function getRevisionTimestamp() {
4720                 if ( is_null( $this->mRevisionTimestamp ) ) {
4721                         wfProfileIn( __METHOD__ );
4722                         global $wgContLang;
4723                         $dbr = wfGetDB( DB_SLAVE );
4724                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4725                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4726
4727                         // Normalize timestamp to internal MW format for timezone processing.
4728                         // This has the added side-effect of replacing a null value with
4729                         // the current time, which gives us more sensible behavior for
4730                         // previews.
4731                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4732
4733                         // The cryptic '' timezone parameter tells to use the site-default
4734                         // timezone offset instead of the user settings.
4735                         //
4736                         // Since this value will be saved into the parser cache, served
4737                         // to other users, and potentially even used inside links and such,
4738                         // it needs to be consistent for all visitors.
4739                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4740
4741                         wfProfileOut( __METHOD__ );
4742                 }
4743                 return $this->mRevisionTimestamp;
4744         }
4745
4746         /**
4747          * Mutator for $mDefaultSort
4748          *
4749          * @param $sort New value
4750          */
4751         public function setDefaultSort( $sort ) {
4752                 $this->mDefaultSort = $sort;
4753         }
4754
4755         /**
4756          * Accessor for $mDefaultSort
4757          * Will use the title/prefixed title if none is set
4758          *
4759          * @return string
4760          */
4761         public function getDefaultSort() {
4762                 if( $this->mDefaultSort !== false ) {
4763                         return $this->mDefaultSort;
4764                 } else {
4765                         return $this->mTitle->getNamespace() == NS_CATEGORY
4766                                         ? $this->mTitle->getText()
4767                                         : $this->mTitle->getPrefixedText();
4768                 }
4769         }
4770
4771         /**
4772          * Try to guess the section anchor name based on a wikitext fragment
4773          * presumably extracted from a heading, for example "Header" from
4774          * "== Header ==".
4775          */
4776         public function guessSectionNameFromWikiText( $text ) {
4777                 # Strip out wikitext links(they break the anchor)
4778                 $text = $this->stripSectionName( $text );
4779                 $headline = Sanitizer::decodeCharReferences( $text );
4780                 # strip out HTML
4781                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4782                 $headline = trim( $headline );
4783                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4784                 $replacearray = array(
4785                         '%3A' => ':',
4786                         '%' => '.'
4787                 );
4788                 return str_replace(
4789                         array_keys( $replacearray ),
4790                         array_values( $replacearray ),
4791                         $sectionanchor );
4792         }
4793
4794         /**
4795          * Strips a text string of wikitext for use in a section anchor
4796          *
4797          * Accepts a text string and then removes all wikitext from the
4798          * string and leaves only the resultant text (i.e. the result of
4799          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4800          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4801          * to create valid section anchors by mimicing the output of the
4802          * parser when headings are parsed.
4803          *
4804          * @param $text string Text string to be stripped of wikitext
4805          * for use in a Section anchor
4806          * @return Filtered text string
4807          */
4808         public function stripSectionName( $text ) {
4809                 # Strip internal link markup
4810                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4811                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4812
4813                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4814                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4815                 # on how many empty links there are on the page - need to figure that out.
4816                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4817
4818                 # Parse wikitext quotes (italics & bold)
4819                 $text = $this->doQuotes($text);
4820
4821                 # Strip HTML tags
4822                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4823                 return $text;
4824         }
4825
4826         function srvus( $text ) {
4827                 return $this->testSrvus( $text, $this->mOutputType );
4828         }
4829
4830         /**
4831          * strip/replaceVariables/unstrip for preprocessor regression testing
4832          */
4833         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4834                 $this->clearState();
4835                 if ( ! ( $title instanceof Title ) ) {
4836                         $title = Title::newFromText( $title );
4837                 }
4838                 $this->mTitle = $title;
4839                 $this->mOptions = $options;
4840                 $this->setOutputType( $outputType );
4841                 $text = $this->replaceVariables( $text );
4842                 $text = $this->mStripState->unstripBoth( $text );
4843                 $text = Sanitizer::removeHTMLtags( $text );
4844                 return $text;
4845         }
4846
4847         function testPst( $text, $title, $options ) {
4848                 global $wgUser;
4849                 if ( ! ( $title instanceof Title ) ) {
4850                         $title = Title::newFromText( $title );
4851                 }
4852                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4853         }
4854
4855         function testPreprocess( $text, $title, $options ) {
4856                 if ( ! ( $title instanceof Title ) ) {
4857                         $title = Title::newFromText( $title );
4858                 }
4859                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4860         }
4861
4862         function markerSkipCallback( $s, $callback ) {
4863                 $i = 0;
4864                 $out = '';
4865                 while ( $i < strlen( $s ) ) {
4866                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4867                         if ( $markerStart === false ) {
4868                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4869                                 break;
4870                         } else {
4871                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4872                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4873                                 if ( $markerEnd === false ) {
4874                                         $out .= substr( $s, $markerStart );
4875                                         break;
4876                                 } else {
4877                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4878                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4879                                         $i = $markerEnd;
4880                                 }
4881                         }
4882                 }
4883                 return $out;
4884         }
4885 }
4886
4887 /**
4888  * @todo document, briefly.
4889  * @addtogroup Parser
4890  */
4891 class StripState {
4892         var $general, $nowiki;
4893
4894         function __construct() {
4895                 $this->general = new ReplacementArray;
4896                 $this->nowiki = new ReplacementArray;
4897         }
4898
4899         function unstripGeneral( $text ) {
4900                 wfProfileIn( __METHOD__ );
4901                 do {
4902                         $oldText = $text;
4903                         $text = $this->general->replace( $text );
4904                 } while ( $text != $oldText );
4905                 wfProfileOut( __METHOD__ );
4906                 return $text;
4907         }
4908
4909         function unstripNoWiki( $text ) {
4910                 wfProfileIn( __METHOD__ );
4911                 do {
4912                         $oldText = $text;
4913                         $text = $this->nowiki->replace( $text );
4914                 } while ( $text != $oldText );
4915                 wfProfileOut( __METHOD__ );
4916                 return $text;
4917         }
4918
4919         function unstripBoth( $text ) {
4920                 wfProfileIn( __METHOD__ );
4921                 do {
4922                         $oldText = $text;
4923                         $text = $this->general->replace( $text );
4924                         $text = $this->nowiki->replace( $text );
4925                 } while ( $text != $oldText );
4926                 wfProfileOut( __METHOD__ );
4927                 return $text;
4928         }
4929 }
4930
4931 /**
4932  * @todo document, briefly.
4933  * @addtogroup Parser
4934  */
4935 class OnlyIncludeReplacer {
4936         var $output = '';
4937
4938         function replace( $matches ) {
4939                 if ( substr( $matches[1], -1 ) == "\n" ) {
4940                         $this->output .= substr( $matches[1], 0, -1 );
4941                 } else {
4942                         $this->output .= $matches[1];
4943                 }
4944         }
4945 }