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                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1524                 }
1525                 $nottalk = !$this->mTitle->isTalkPage();
1526
1527                 if ( $useLinkPrefixExtension ) {
1528                         $m = array();
1529                         if ( preg_match( $e2, $s, $m ) ) {
1530                                 $first_prefix = $m[2];
1531                         } else {
1532                                 $first_prefix = false;
1533                         }
1534                 } else {
1535                         $prefix = '';
1536                 }
1537
1538                 if($wgContLang->hasVariants()) {
1539                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1540                 } else {
1541                         $selflink = array($this->mTitle->getPrefixedText());
1542                 }
1543                 $useSubpages = $this->areSubpagesAllowed();
1544                 wfProfileOut( $fname.'-setup' );
1545
1546                 # Loop for each link
1547                 for ($k = 0; isset( $a[$k] ); $k++) {
1548                         $line = $a[$k];
1549                         if ( $useLinkPrefixExtension ) {
1550                                 wfProfileIn( $fname.'-prefixhandling' );
1551                                 if ( preg_match( $e2, $s, $m ) ) {
1552                                         $prefix = $m[2];
1553                                         $s = $m[1];
1554                                 } else {
1555                                         $prefix='';
1556                                 }
1557                                 # first link
1558                                 if($first_prefix) {
1559                                         $prefix = $first_prefix;
1560                                         $first_prefix = false;
1561                                 }
1562                                 wfProfileOut( $fname.'-prefixhandling' );
1563                         }
1564
1565                         $might_be_img = false;
1566
1567                         wfProfileIn( "$fname-e1" );
1568                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1569                                 $text = $m[2];
1570                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1571                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1572                                 # the real problem is with the $e1 regex
1573                                 # See bug 1300.
1574                                 #
1575                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1576                                 # and no image is in sight. See bug 2095.
1577                                 #
1578                                 if( $text !== '' &&
1579                                         substr( $m[3], 0, 1 ) === ']' &&
1580                                         strpos($text, '[') !== false
1581                                 )
1582                                 {
1583                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1584                                         $m[3] = substr( $m[3], 1 );
1585                                 }
1586                                 # fix up urlencoded title texts
1587                                 if( strpos( $m[1], '%' ) !== false ) {
1588                                         # Should anchors '#' also be rejected?
1589                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1590                                 }
1591                                 $trail = $m[3];
1592                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1593                                 $might_be_img = true;
1594                                 $text = $m[2];
1595                                 if ( strpos( $m[1], '%' ) !== false ) {
1596                                         $m[1] = urldecode($m[1]);
1597                                 }
1598                                 $trail = "";
1599                         } else { # Invalid form; output directly
1600                                 $s .= $prefix . '[[' . $line ;
1601                                 wfProfileOut( "$fname-e1" );
1602                                 continue;
1603                         }
1604                         wfProfileOut( "$fname-e1" );
1605                         wfProfileIn( "$fname-misc" );
1606
1607                         # Don't allow internal links to pages containing
1608                         # PROTO: where PROTO is a valid URL protocol; these
1609                         # should be external links.
1610                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1611                                 $s .= $prefix . '[[' . $line ;
1612                                 continue;
1613                         }
1614
1615                         # Make subpage if necessary
1616                         if( $useSubpages ) {
1617                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1618                         } else {
1619                                 $link = $m[1];
1620                         }
1621
1622                         $noforce = (substr($m[1], 0, 1) != ':');
1623                         if (!$noforce) {
1624                                 # Strip off leading ':'
1625                                 $link = substr($link, 1);
1626                         }
1627
1628                         wfProfileOut( "$fname-misc" );
1629                         wfProfileIn( "$fname-title" );
1630                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1631                         if( !$nt ) {
1632                                 $s .= $prefix . '[[' . $line;
1633                                 wfProfileOut( "$fname-title" );
1634                                 continue;
1635                         }
1636
1637                         $ns = $nt->getNamespace();
1638                         $iw = $nt->getInterWiki();
1639                         wfProfileOut( "$fname-title" );
1640
1641                         if ($might_be_img) { # if this is actually an invalid link
1642                                 wfProfileIn( "$fname-might_be_img" );
1643                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1644                                         $found = false;
1645                                         while (isset ($a[$k+1]) ) {
1646                                                 #look at the next 'line' to see if we can close it there
1647                                                 $spliced = array_splice( $a, $k + 1, 1 );
1648                                                 $next_line = array_shift( $spliced );
1649                                                 $m = explode( ']]', $next_line, 3 );
1650                                                 if ( count( $m ) == 3 ) {
1651                                                         # the first ]] closes the inner link, the second the image
1652                                                         $found = true;
1653                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1654                                                         $trail = $m[2];
1655                                                         break;
1656                                                 } elseif ( count( $m ) == 2 ) {
1657                                                         #if there's exactly one ]] that's fine, we'll keep looking
1658                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1659                                                 } else {
1660                                                         #if $next_line is invalid too, we need look no further
1661                                                         $text .= '[[' . $next_line;
1662                                                         break;
1663                                                 }
1664                                         }
1665                                         if ( !$found ) {
1666                                                 # we couldn't find the end of this imageLink, so output it raw
1667                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1668                                                 $text = $this->replaceInternalLinks($text);
1669                                                 $s .= "{$prefix}[[$link|$text";
1670                                                 # note: no $trail, because without an end, there *is* no trail
1671                                                 wfProfileOut( "$fname-might_be_img" );
1672                                                 continue;
1673                                         }
1674                                 } else { #it's not an image, so output it raw
1675                                         $s .= "{$prefix}[[$link|$text";
1676                                         # note: no $trail, because without an end, there *is* no trail
1677                                         wfProfileOut( "$fname-might_be_img" );
1678                                         continue;
1679                                 }
1680                                 wfProfileOut( "$fname-might_be_img" );
1681                         }
1682
1683                         $wasblank = ( '' == $text );
1684                         if( $wasblank ) $text = $link;
1685
1686                         # Link not escaped by : , create the various objects
1687                         if( $noforce ) {
1688
1689                                 # Interwikis
1690                                 wfProfileIn( "$fname-interwiki" );
1691                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1692                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1693                                         $s = rtrim($s . $prefix);
1694                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1695                                         wfProfileOut( "$fname-interwiki" );
1696                                         continue;
1697                                 }
1698                                 wfProfileOut( "$fname-interwiki" );
1699
1700                                 if ( $ns == NS_IMAGE ) {
1701                                         wfProfileIn( "$fname-image" );
1702                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1703                                                 # recursively parse links inside the image caption
1704                                                 # actually, this will parse them in any other parameters, too,
1705                                                 # but it might be hard to fix that, and it doesn't matter ATM
1706                                                 $text = $this->replaceExternalLinks($text);
1707                                                 $text = $this->replaceInternalLinks($text);
1708
1709                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1710                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1711                                                 $this->mOutput->addImage( $nt->getDBkey() );
1712
1713                                                 wfProfileOut( "$fname-image" );
1714                                                 continue;
1715                                         } else {
1716                                                 # We still need to record the image's presence on the page
1717                                                 $this->mOutput->addImage( $nt->getDBkey() );
1718                                         }
1719                                         wfProfileOut( "$fname-image" );
1720
1721                                 }
1722
1723                                 if ( $ns == NS_CATEGORY ) {
1724                                         wfProfileIn( "$fname-category" );
1725                                         $s = rtrim($s . "\n"); # bug 87
1726
1727                                         if ( $wasblank ) {
1728                                                 $sortkey = $this->getDefaultSort();
1729                                         } else {
1730                                                 $sortkey = $text;
1731                                         }
1732                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1733                                         $sortkey = str_replace( "\n", '', $sortkey );
1734                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1735                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1736
1737                                         /**
1738                                          * Strip the whitespace Category links produce, see bug 87
1739                                          * @todo We might want to use trim($tmp, "\n") here.
1740                                          */
1741                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1742
1743                                         wfProfileOut( "$fname-category" );
1744                                         continue;
1745                                 }
1746                         }
1747
1748                         # Self-link checking
1749                         if( $nt->getFragment() === '' ) {
1750                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1751                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1752                                         continue;
1753                                 }
1754                         }
1755
1756                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1757                         if( $ns == NS_MEDIA ) {
1758                                 # Give extensions a chance to select the file revision for us
1759                                 $skip = $time = false;
1760                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1761                                 if ( $skip ) {
1762                                         $link = $sk->makeLinkObj( $nt );
1763                                 } else {
1764                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1765                                 }
1766                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1767                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1768                                 $this->mOutput->addImage( $nt->getDBkey() );
1769                                 continue;
1770                         } elseif( $ns == NS_SPECIAL ) {
1771                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1772                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1773                                 } else {
1774                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1775                                 }
1776                                 continue;
1777                         } elseif( $ns == NS_IMAGE ) {
1778                                 $img = wfFindFile( $nt );
1779                                 if( $img ) {
1780                                         // Force a blue link if the file exists; may be a remote
1781                                         // upload on the shared repository, and we want to see its
1782                                         // auto-generated page.
1783                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1784                                         $this->mOutput->addLink( $nt );
1785                                         continue;
1786                                 }
1787                         }
1788                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1789                 }
1790                 wfProfileOut( $fname );
1791                 return $s;
1792         }
1793
1794         /**
1795          * Make a link placeholder. The text returned can be later resolved to a real link with
1796          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1797          * parsing of interwiki links, and secondly to allow all existence checks and
1798          * article length checks (for stub links) to be bundled into a single query.
1799          *
1800          */
1801         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1802                 wfProfileIn( __METHOD__ );
1803                 if ( ! is_object($nt) ) {
1804                         # Fail gracefully
1805                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1806                 } else {
1807                         # Separate the link trail from the rest of the link
1808                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1809
1810                         if ( $nt->isExternal() ) {
1811                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1812                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1813                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1814                         } else {
1815                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1816                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1817                                 $this->mLinkHolders['queries'][] = $query;
1818                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1819                                 $this->mLinkHolders['titles'][] = $nt;
1820
1821                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1822                         }
1823                 }
1824                 wfProfileOut( __METHOD__ );
1825                 return $retVal;
1826         }
1827
1828         /**
1829          * Render a forced-blue link inline; protect against double expansion of
1830          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1831          * Since this little disaster has to split off the trail text to avoid
1832          * breaking URLs in the following text without breaking trails on the
1833          * wiki links, it's been made into a horrible function.
1834          *
1835          * @param Title $nt
1836          * @param string $text
1837          * @param string $query
1838          * @param string $trail
1839          * @param string $prefix
1840          * @return string HTML-wikitext mix oh yuck
1841          */
1842         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1843                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1844                 $sk = $this->mOptions->getSkin();
1845                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1846                 return $this->armorLinks( $link ) . $trail;
1847         }
1848
1849         /**
1850          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1851          * going to go through further parsing steps before inline URL expansion.
1852          *
1853          * In particular this is important when using action=render, which causes
1854          * full URLs to be included.
1855          *
1856          * Oh man I hate our multi-layer parser!
1857          *
1858          * @param string more-or-less HTML
1859          * @return string less-or-more HTML with NOPARSE bits
1860          */
1861         function armorLinks( $text ) {
1862                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1863                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1864         }
1865
1866         /**
1867          * Return true if subpage links should be expanded on this page.
1868          * @return bool
1869          */
1870         function areSubpagesAllowed() {
1871                 # Some namespaces don't allow subpages
1872                 global $wgNamespacesWithSubpages;
1873                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1874         }
1875
1876         /**
1877          * Handle link to subpage if necessary
1878          * @param string $target the source of the link
1879          * @param string &$text the link text, modified as necessary
1880          * @return string the full name of the link
1881          * @private
1882          */
1883         function maybeDoSubpageLink($target, &$text) {
1884                 # Valid link forms:
1885                 # Foobar -- normal
1886                 # :Foobar -- override special treatment of prefix (images, language links)
1887                 # /Foobar -- convert to CurrentPage/Foobar
1888                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1889                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1890                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1891
1892                 $fname = 'Parser::maybeDoSubpageLink';
1893                 wfProfileIn( $fname );
1894                 $ret = $target; # default return value is no change
1895
1896                 # Some namespaces don't allow subpages,
1897                 # so only perform processing if subpages are allowed
1898                 if( $this->areSubpagesAllowed() ) {
1899                         $hash = strpos( $target, '#' );
1900                         if( $hash !== false ) {
1901                                 $suffix = substr( $target, $hash );
1902                                 $target = substr( $target, 0, $hash );
1903                         } else {
1904                                 $suffix = '';
1905                         }
1906                         # bug 7425
1907                         $target = trim( $target );
1908                         # Look at the first character
1909                         if( $target != '' && $target{0} == '/' ) {
1910                                 # / at end means we don't want the slash to be shown
1911                                 $m = array();
1912                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1913                                 if( $trailingSlashes ) {
1914                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1915                                 } else {
1916                                         $noslash = substr( $target, 1 );
1917                                 }
1918
1919                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1920                                 if( '' === $text ) {
1921                                         $text = $target . $suffix;
1922                                 } # this might be changed for ugliness reasons
1923                         } else {
1924                                 # check for .. subpage backlinks
1925                                 $dotdotcount = 0;
1926                                 $nodotdot = $target;
1927                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1928                                         ++$dotdotcount;
1929                                         $nodotdot = substr( $nodotdot, 3 );
1930                                 }
1931                                 if($dotdotcount > 0) {
1932                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1933                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1934                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1935                                                 # / at the end means don't show full path
1936                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1937                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1938                                                         if( '' === $text ) {
1939                                                                 $text = $nodotdot . $suffix;
1940                                                         }
1941                                                 }
1942                                                 $nodotdot = trim( $nodotdot );
1943                                                 if( $nodotdot != '' ) {
1944                                                         $ret .= '/' . $nodotdot;
1945                                                 }
1946                                                 $ret .= $suffix;
1947                                         }
1948                                 }
1949                         }
1950                 }
1951
1952                 wfProfileOut( $fname );
1953                 return $ret;
1954         }
1955
1956         /**#@+
1957          * Used by doBlockLevels()
1958          * @private
1959          */
1960         /* private */ function closeParagraph() {
1961                 $result = '';
1962                 if ( '' != $this->mLastSection ) {
1963                         $result = '</' . $this->mLastSection  . ">\n";
1964                 }
1965                 $this->mInPre = false;
1966                 $this->mLastSection = '';
1967                 return $result;
1968         }
1969         # getCommon() returns the length of the longest common substring
1970         # of both arguments, starting at the beginning of both.
1971         #
1972         /* private */ function getCommon( $st1, $st2 ) {
1973                 $fl = strlen( $st1 );
1974                 $shorter = strlen( $st2 );
1975                 if ( $fl < $shorter ) { $shorter = $fl; }
1976
1977                 for ( $i = 0; $i < $shorter; ++$i ) {
1978                         if ( $st1{$i} != $st2{$i} ) { break; }
1979                 }
1980                 return $i;
1981         }
1982         # These next three functions open, continue, and close the list
1983         # element appropriate to the prefix character passed into them.
1984         #
1985         /* private */ function openList( $char ) {
1986                 $result = $this->closeParagraph();
1987
1988                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1989                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1990                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1991                 else if ( ';' == $char ) {
1992                         $result .= '<dl><dt>';
1993                         $this->mDTopen = true;
1994                 }
1995                 else { $result = '<!-- ERR 1 -->'; }
1996
1997                 return $result;
1998         }
1999
2000         /* private */ function nextItem( $char ) {
2001                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2002                 else if ( ':' == $char || ';' == $char ) {
2003                         $close = '</dd>';
2004                         if ( $this->mDTopen ) { $close = '</dt>'; }
2005                         if ( ';' == $char ) {
2006                                 $this->mDTopen = true;
2007                                 return $close . '<dt>';
2008                         } else {
2009                                 $this->mDTopen = false;
2010                                 return $close . '<dd>';
2011                         }
2012                 }
2013                 return '<!-- ERR 2 -->';
2014         }
2015
2016         /* private */ function closeList( $char ) {
2017                 if ( '*' == $char ) { $text = '</li></ul>'; }
2018                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2019                 else if ( ':' == $char ) {
2020                         if ( $this->mDTopen ) {
2021                                 $this->mDTopen = false;
2022                                 $text = '</dt></dl>';
2023                         } else {
2024                                 $text = '</dd></dl>';
2025                         }
2026                 }
2027                 else {  return '<!-- ERR 3 -->'; }
2028                 return $text."\n";
2029         }
2030         /**#@-*/
2031
2032         /**
2033          * Make lists from lines starting with ':', '*', '#', etc.
2034          *
2035          * @private
2036          * @return string the lists rendered as HTML
2037          */
2038         function doBlockLevels( $text, $linestart ) {
2039                 $fname = 'Parser::doBlockLevels';
2040                 wfProfileIn( $fname );
2041
2042                 # Parsing through the text line by line.  The main thing
2043                 # happening here is handling of block-level elements p, pre,
2044                 # and making lists from lines starting with * # : etc.
2045                 #
2046                 $textLines = explode( "\n", $text );
2047
2048                 $lastPrefix = $output = '';
2049                 $this->mDTopen = $inBlockElem = false;
2050                 $prefixLength = 0;
2051                 $paragraphStack = false;
2052
2053                 if ( !$linestart ) {
2054                         $output .= array_shift( $textLines );
2055                 }
2056                 foreach ( $textLines as $oLine ) {
2057                         $lastPrefixLength = strlen( $lastPrefix );
2058                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2059                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2060                         if ( !$this->mInPre ) {
2061                                 # Multiple prefixes may abut each other for nested lists.
2062                                 $prefixLength = strspn( $oLine, '*#:;' );
2063                                 $pref = substr( $oLine, 0, $prefixLength );
2064
2065                                 # eh?
2066                                 $pref2 = str_replace( ';', ':', $pref );
2067                                 $t = substr( $oLine, $prefixLength );
2068                                 $this->mInPre = !empty($preOpenMatch);
2069                         } else {
2070                                 # Don't interpret any other prefixes in preformatted text
2071                                 $prefixLength = 0;
2072                                 $pref = $pref2 = '';
2073                                 $t = $oLine;
2074                         }
2075
2076                         # List generation
2077                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2078                                 # Same as the last item, so no need to deal with nesting or opening stuff
2079                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2080                                 $paragraphStack = false;
2081
2082                                 if ( substr( $pref, -1 ) == ';') {
2083                                         # The one nasty exception: definition lists work like this:
2084                                         # ; title : definition text
2085                                         # So we check for : in the remainder text to split up the
2086                                         # title and definition, without b0rking links.
2087                                         $term = $t2 = '';
2088                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2089                                                 $t = $t2;
2090                                                 $output .= $term . $this->nextItem( ':' );
2091                                         }
2092                                 }
2093                         } elseif( $prefixLength || $lastPrefixLength ) {
2094                                 # Either open or close a level...
2095                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2096                                 $paragraphStack = false;
2097
2098                                 while( $commonPrefixLength < $lastPrefixLength ) {
2099                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2100                                         --$lastPrefixLength;
2101                                 }
2102                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2103                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2104                                 }
2105                                 while ( $prefixLength > $commonPrefixLength ) {
2106                                         $char = substr( $pref, $commonPrefixLength, 1 );
2107                                         $output .= $this->openList( $char );
2108
2109                                         if ( ';' == $char ) {
2110                                                 # FIXME: This is dupe of code above
2111                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2112                                                         $t = $t2;
2113                                                         $output .= $term . $this->nextItem( ':' );
2114                                                 }
2115                                         }
2116                                         ++$commonPrefixLength;
2117                                 }
2118                                 $lastPrefix = $pref2;
2119                         }
2120                         if( 0 == $prefixLength ) {
2121                                 wfProfileIn( "$fname-paragraph" );
2122                                 # No prefix (not in list)--go to paragraph mode
2123                                 // XXX: use a stack for nestable elements like span, table and div
2124                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2125                                 $closematch = preg_match(
2126                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2127                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2128                                 if ( $openmatch or $closematch ) {
2129                                         $paragraphStack = false;
2130                                         # TODO bug 5718: paragraph closed
2131                                         $output .= $this->closeParagraph();
2132                                         if ( $preOpenMatch and !$preCloseMatch ) {
2133                                                 $this->mInPre = true;
2134                                         }
2135                                         if ( $closematch ) {
2136                                                 $inBlockElem = false;
2137                                         } else {
2138                                                 $inBlockElem = true;
2139                                         }
2140                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2141                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2142                                                 // pre
2143                                                 if ($this->mLastSection != 'pre') {
2144                                                         $paragraphStack = false;
2145                                                         $output .= $this->closeParagraph().'<pre>';
2146                                                         $this->mLastSection = 'pre';
2147                                                 }
2148                                                 $t = substr( $t, 1 );
2149                                         } else {
2150                                                 // paragraph
2151                                                 if ( '' == trim($t) ) {
2152                                                         if ( $paragraphStack ) {
2153                                                                 $output .= $paragraphStack.'<br />';
2154                                                                 $paragraphStack = false;
2155                                                                 $this->mLastSection = 'p';
2156                                                         } else {
2157                                                                 if ($this->mLastSection != 'p' ) {
2158                                                                         $output .= $this->closeParagraph();
2159                                                                         $this->mLastSection = '';
2160                                                                         $paragraphStack = '<p>';
2161                                                                 } else {
2162                                                                         $paragraphStack = '</p><p>';
2163                                                                 }
2164                                                         }
2165                                                 } else {
2166                                                         if ( $paragraphStack ) {
2167                                                                 $output .= $paragraphStack;
2168                                                                 $paragraphStack = false;
2169                                                                 $this->mLastSection = 'p';
2170                                                         } else if ($this->mLastSection != 'p') {
2171                                                                 $output .= $this->closeParagraph().'<p>';
2172                                                                 $this->mLastSection = 'p';
2173                                                         }
2174                                                 }
2175                                         }
2176                                 }
2177                                 wfProfileOut( "$fname-paragraph" );
2178                         }
2179                         // somewhere above we forget to get out of pre block (bug 785)
2180                         if($preCloseMatch && $this->mInPre) {
2181                                 $this->mInPre = false;
2182                         }
2183                         if ($paragraphStack === false) {
2184                                 $output .= $t."\n";
2185                         }
2186                 }
2187                 while ( $prefixLength ) {
2188                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2189                         --$prefixLength;
2190                 }
2191                 if ( '' != $this->mLastSection ) {
2192                         $output .= '</' . $this->mLastSection . '>';
2193                         $this->mLastSection = '';
2194                 }
2195
2196                 wfProfileOut( $fname );
2197                 return $output;
2198         }
2199
2200         /**
2201          * Split up a string on ':', ignoring any occurences inside tags
2202          * to prevent illegal overlapping.
2203          * @param string $str the string to split
2204          * @param string &$before set to everything before the ':'
2205          * @param string &$after set to everything after the ':'
2206          * return string the position of the ':', or false if none found
2207          */
2208         function findColonNoLinks($str, &$before, &$after) {
2209                 $fname = 'Parser::findColonNoLinks';
2210                 wfProfileIn( $fname );
2211
2212                 $pos = strpos( $str, ':' );
2213                 if( $pos === false ) {
2214                         // Nothing to find!
2215                         wfProfileOut( $fname );
2216                         return false;
2217                 }
2218
2219                 $lt = strpos( $str, '<' );
2220                 if( $lt === false || $lt > $pos ) {
2221                         // Easy; no tag nesting to worry about
2222                         $before = substr( $str, 0, $pos );
2223                         $after = substr( $str, $pos+1 );
2224                         wfProfileOut( $fname );
2225                         return $pos;
2226                 }
2227
2228                 // Ugly state machine to walk through avoiding tags.
2229                 $state = self::COLON_STATE_TEXT;
2230                 $stack = 0;
2231                 $len = strlen( $str );
2232                 for( $i = 0; $i < $len; $i++ ) {
2233                         $c = $str{$i};
2234
2235                         switch( $state ) {
2236                         // (Using the number is a performance hack for common cases)
2237                         case 0: // self::COLON_STATE_TEXT:
2238                                 switch( $c ) {
2239                                 case "<":
2240                                         // Could be either a <start> tag or an </end> tag
2241                                         $state = self::COLON_STATE_TAGSTART;
2242                                         break;
2243                                 case ":":
2244                                         if( $stack == 0 ) {
2245                                                 // We found it!
2246                                                 $before = substr( $str, 0, $i );
2247                                                 $after = substr( $str, $i + 1 );
2248                                                 wfProfileOut( $fname );
2249                                                 return $i;
2250                                         }
2251                                         // Embedded in a tag; don't break it.
2252                                         break;
2253                                 default:
2254                                         // Skip ahead looking for something interesting
2255                                         $colon = strpos( $str, ':', $i );
2256                                         if( $colon === false ) {
2257                                                 // Nothing else interesting
2258                                                 wfProfileOut( $fname );
2259                                                 return false;
2260                                         }
2261                                         $lt = strpos( $str, '<', $i );
2262                                         if( $stack === 0 ) {
2263                                                 if( $lt === false || $colon < $lt ) {
2264                                                         // We found it!
2265                                                         $before = substr( $str, 0, $colon );
2266                                                         $after = substr( $str, $colon + 1 );
2267                                                         wfProfileOut( $fname );
2268                                                         return $i;
2269                                                 }
2270                                         }
2271                                         if( $lt === false ) {
2272                                                 // Nothing else interesting to find; abort!
2273                                                 // We're nested, but there's no close tags left. Abort!
2274                                                 break 2;
2275                                         }
2276                                         // Skip ahead to next tag start
2277                                         $i = $lt;
2278                                         $state = self::COLON_STATE_TAGSTART;
2279                                 }
2280                                 break;
2281                         case 1: // self::COLON_STATE_TAG:
2282                                 // In a <tag>
2283                                 switch( $c ) {
2284                                 case ">":
2285                                         $stack++;
2286                                         $state = self::COLON_STATE_TEXT;
2287                                         break;
2288                                 case "/":
2289                                         // Slash may be followed by >?
2290                                         $state = self::COLON_STATE_TAGSLASH;
2291                                         break;
2292                                 default:
2293                                         // ignore
2294                                 }
2295                                 break;
2296                         case 2: // self::COLON_STATE_TAGSTART:
2297                                 switch( $c ) {
2298                                 case "/":
2299                                         $state = self::COLON_STATE_CLOSETAG;
2300                                         break;
2301                                 case "!":
2302                                         $state = self::COLON_STATE_COMMENT;
2303                                         break;
2304                                 case ">":
2305                                         // Illegal early close? This shouldn't happen D:
2306                                         $state = self::COLON_STATE_TEXT;
2307                                         break;
2308                                 default:
2309                                         $state = self::COLON_STATE_TAG;
2310                                 }
2311                                 break;
2312                         case 3: // self::COLON_STATE_CLOSETAG:
2313                                 // In a </tag>
2314                                 if( $c == ">" ) {
2315                                         $stack--;
2316                                         if( $stack < 0 ) {
2317                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2318                                                 wfProfileOut( $fname );
2319                                                 return false;
2320                                         }
2321                                         $state = self::COLON_STATE_TEXT;
2322                                 }
2323                                 break;
2324                         case self::COLON_STATE_TAGSLASH:
2325                                 if( $c == ">" ) {
2326                                         // Yes, a self-closed tag <blah/>
2327                                         $state = self::COLON_STATE_TEXT;
2328                                 } else {
2329                                         // Probably we're jumping the gun, and this is an attribute
2330                                         $state = self::COLON_STATE_TAG;
2331                                 }
2332                                 break;
2333                         case 5: // self::COLON_STATE_COMMENT:
2334                                 if( $c == "-" ) {
2335                                         $state = self::COLON_STATE_COMMENTDASH;
2336                                 }
2337                                 break;
2338                         case self::COLON_STATE_COMMENTDASH:
2339                                 if( $c == "-" ) {
2340                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2341                                 } else {
2342                                         $state = self::COLON_STATE_COMMENT;
2343                                 }
2344                                 break;
2345                         case self::COLON_STATE_COMMENTDASHDASH:
2346                                 if( $c == ">" ) {
2347                                         $state = self::COLON_STATE_TEXT;
2348                                 } else {
2349                                         $state = self::COLON_STATE_COMMENT;
2350                                 }
2351                                 break;
2352                         default:
2353                                 throw new MWException( "State machine error in $fname" );
2354                         }
2355                 }
2356                 if( $stack > 0 ) {
2357                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2358                         return false;
2359                 }
2360                 wfProfileOut( $fname );
2361                 return false;
2362         }
2363
2364         /**
2365          * Return value of a magic variable (like PAGENAME)
2366          *
2367          * @private
2368          */
2369         function getVariableValue( $index ) {
2370                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2371
2372                 /**
2373                  * Some of these require message or data lookups and can be
2374                  * expensive to check many times.
2375                  */
2376                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2377                         if ( isset( $this->mVarCache[$index] ) ) {
2378                                 return $this->mVarCache[$index];
2379                         }
2380                 }
2381
2382                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2383                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2384
2385                 # Use the time zone
2386                 global $wgLocaltimezone;
2387                 if ( isset( $wgLocaltimezone ) ) {
2388                         $oldtz = getenv( 'TZ' );
2389                         putenv( 'TZ='.$wgLocaltimezone );
2390                 }
2391
2392                 wfSuppressWarnings(); // E_STRICT system time bitching
2393                 $localTimestamp = date( 'YmdHis', $ts );
2394                 $localMonth = date( 'm', $ts );
2395                 $localMonthName = date( 'n', $ts );
2396                 $localDay = date( 'j', $ts );
2397                 $localDay2 = date( 'd', $ts );
2398                 $localDayOfWeek = date( 'w', $ts );
2399                 $localWeek = date( 'W', $ts );
2400                 $localYear = date( 'Y', $ts );
2401                 $localHour = date( 'H', $ts );
2402                 if ( isset( $wgLocaltimezone ) ) {
2403                         putenv( 'TZ='.$oldtz );
2404                 }
2405                 wfRestoreWarnings();
2406
2407                 switch ( $index ) {
2408                         case 'currentmonth':
2409                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2410                         case 'currentmonthname':
2411                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2412                         case 'currentmonthnamegen':
2413                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2414                         case 'currentmonthabbrev':
2415                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2416                         case 'currentday':
2417                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2418                         case 'currentday2':
2419                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2420                         case 'localmonth':
2421                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2422                         case 'localmonthname':
2423                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2424                         case 'localmonthnamegen':
2425                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2426                         case 'localmonthabbrev':
2427                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2428                         case 'localday':
2429                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2430                         case 'localday2':
2431                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2432                         case 'pagename':
2433                                 return wfEscapeWikiText( $this->mTitle->getText() );
2434                         case 'pagenamee':
2435                                 return $this->mTitle->getPartialURL();
2436                         case 'fullpagename':
2437                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2438                         case 'fullpagenamee':
2439                                 return $this->mTitle->getPrefixedURL();
2440                         case 'subpagename':
2441                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2442                         case 'subpagenamee':
2443                                 return $this->mTitle->getSubpageUrlForm();
2444                         case 'basepagename':
2445                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2446                         case 'basepagenamee':
2447                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2448                         case 'talkpagename':
2449                                 if( $this->mTitle->canTalk() ) {
2450                                         $talkPage = $this->mTitle->getTalkPage();
2451                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2452                                 } else {
2453                                         return '';
2454                                 }
2455                         case 'talkpagenamee':
2456                                 if( $this->mTitle->canTalk() ) {
2457                                         $talkPage = $this->mTitle->getTalkPage();
2458                                         return $talkPage->getPrefixedUrl();
2459                                 } else {
2460                                         return '';
2461                                 }
2462                         case 'subjectpagename':
2463                                 $subjPage = $this->mTitle->getSubjectPage();
2464                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2465                         case 'subjectpagenamee':
2466                                 $subjPage = $this->mTitle->getSubjectPage();
2467                                 return $subjPage->getPrefixedUrl();
2468                         case 'revisionid':
2469                                 // Let the edit saving system know we should parse the page
2470                                 // *after* a revision ID has been assigned.
2471                                 $this->mOutput->setFlag( 'vary-revision' );
2472                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2473                                 return $this->mRevisionId;
2474                         case 'revisionday':
2475                                 // Let the edit saving system know we should parse the page
2476                                 // *after* a revision ID has been assigned. This is for null edits.
2477                                 $this->mOutput->setFlag( 'vary-revision' );
2478                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2479                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2480                         case 'revisionday2':
2481                                 // Let the edit saving system know we should parse the page
2482                                 // *after* a revision ID has been assigned. This is for null edits.
2483                                 $this->mOutput->setFlag( 'vary-revision' );
2484                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2485                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2486                         case 'revisionmonth':
2487                                 // Let the edit saving system know we should parse the page
2488                                 // *after* a revision ID has been assigned. This is for null edits.
2489                                 $this->mOutput->setFlag( 'vary-revision' );
2490                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2491                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2492                         case 'revisionyear':
2493                                 // Let the edit saving system know we should parse the page
2494                                 // *after* a revision ID has been assigned. This is for null edits.
2495                                 $this->mOutput->setFlag( 'vary-revision' );
2496                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2497                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2498                         case 'revisiontimestamp':
2499                                 // Let the edit saving system know we should parse the page
2500                                 // *after* a revision ID has been assigned. This is for null edits.
2501                                 $this->mOutput->setFlag( 'vary-revision' );
2502                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2503                                 return $this->getRevisionTimestamp();
2504                         case 'namespace':
2505                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2506                         case 'namespacee':
2507                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2508                         case 'talkspace':
2509                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2510                         case 'talkspacee':
2511                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2512                         case 'subjectspace':
2513                                 return $this->mTitle->getSubjectNsText();
2514                         case 'subjectspacee':
2515                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2516                         case 'currentdayname':
2517                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2518                         case 'currentyear':
2519                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2520                         case 'currenttime':
2521                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2522                         case 'currenthour':
2523                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2524                         case 'currentweek':
2525                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2526                                 // int to remove the padding
2527                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2528                         case 'currentdow':
2529                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2530                         case 'localdayname':
2531                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2532                         case 'localyear':
2533                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2534                         case 'localtime':
2535                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2536                         case 'localhour':
2537                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2538                         case 'localweek':
2539                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2540                                 // int to remove the padding
2541                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2542                         case 'localdow':
2543                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2544                         case 'numberofarticles':
2545                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2546                         case 'numberoffiles':
2547                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2548                         case 'numberofusers':
2549                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2550                         case 'numberofpages':
2551                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2552                         case 'numberofadmins':
2553                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2554                         case 'numberofedits':
2555                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2556                         case 'currenttimestamp':
2557                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2558                         case 'localtimestamp':
2559                                 return $this->mVarCache[$index] = $localTimestamp;
2560                         case 'currentversion':
2561                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2562                         case 'sitename':
2563                                 return $wgSitename;
2564                         case 'server':
2565                                 return $wgServer;
2566                         case 'servername':
2567                                 return $wgServerName;
2568                         case 'scriptpath':
2569                                 return $wgScriptPath;
2570                         case 'directionmark':
2571                                 return $wgContLang->getDirMark();
2572                         case 'contentlanguage':
2573                                 global $wgContLanguageCode;
2574                                 return $wgContLanguageCode;
2575                         default:
2576                                 $ret = null;
2577                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2578                                         return $ret;
2579                                 else
2580                                         return null;
2581                 }
2582         }
2583
2584         /**
2585          * initialise the magic variables (like CURRENTMONTHNAME)
2586          *
2587          * @private
2588          */
2589         function initialiseVariables() {
2590                 $fname = 'Parser::initialiseVariables';
2591                 wfProfileIn( $fname );
2592                 $variableIDs = MagicWord::getVariableIDs();
2593
2594                 $this->mVariables = new MagicWordArray( $variableIDs );
2595                 wfProfileOut( $fname );
2596         }
2597
2598         /**
2599          * Preprocess some wikitext and return the document tree.
2600          * This is the ghost of replace_variables().
2601          *
2602          * @param string $text The text to parse
2603          * @param integer flags Bitwise combination of:
2604          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2605          *                                     included. Default is to assume a direct page view.
2606          *
2607          * The generated DOM tree must depend only on the input text and the flags.
2608          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2609          *
2610          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2611          * change in the DOM tree for a given text, must be passed through the section identifier
2612          * in the section edit link and thus back to extractSections().
2613          *
2614          * The output of this function is currently only cached in process memory, but a persistent
2615          * cache may be implemented at a later date which takes further advantage of these strict
2616          * dependency requirements.
2617          *
2618          * @private
2619          */
2620         function preprocessToDom ( $text, $flags = 0 ) {
2621                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2622                 return $dom;
2623         }
2624
2625         /*
2626          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2627          */
2628         public static function splitWhitespace( $s ) {
2629                 $ltrimmed = ltrim( $s );
2630                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2631                 $trimmed = rtrim( $ltrimmed );
2632                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2633                 if ( $diff > 0 ) {
2634                         $w2 = substr( $ltrimmed, -$diff );
2635                 } else {
2636                         $w2 = '';
2637                 }
2638                 return array( $w1, $trimmed, $w2 );
2639         }
2640
2641         /**
2642          * Replace magic variables, templates, and template arguments
2643          * with the appropriate text. Templates are substituted recursively,
2644          * taking care to avoid infinite loops.
2645          *
2646          * Note that the substitution depends on value of $mOutputType:
2647          *  self::OT_WIKI: only {{subst:}} templates
2648          *  self::OT_PREPROCESS: templates but not extension tags
2649          *  self::OT_HTML: all templates and extension tags
2650          *
2651          * @param string $tex The text to transform
2652          * @param PPFrame $frame Object describing the arguments passed to the template
2653          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2654          * @private
2655          */
2656         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2657                 # Prevent too big inclusions
2658                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2659                         return $text;
2660                 }
2661
2662                 $fname = __METHOD__;
2663                 wfProfileIn( $fname );
2664
2665                 if ( $frame === false ) {
2666                         $frame = $this->getPreprocessor()->newFrame();
2667                 } elseif ( !( $frame instanceof PPFrame ) ) {
2668                         throw new MWException( __METHOD__ . ' called using the old argument format' );
2669                 }
2670
2671                 $dom = $this->preprocessToDom( $text );
2672                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2673                 $text = $frame->expand( $dom, $flags );
2674
2675                 wfProfileOut( $fname );
2676                 return $text;
2677         }
2678
2679         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2680         static function createAssocArgs( $args ) {
2681                 $assocArgs = array();
2682                 $index = 1;
2683                 foreach( $args as $arg ) {
2684                         $eqpos = strpos( $arg, '=' );
2685                         if ( $eqpos === false ) {
2686                                 $assocArgs[$index++] = $arg;
2687                         } else {
2688                                 $name = trim( substr( $arg, 0, $eqpos ) );
2689                                 $value = trim( substr( $arg, $eqpos+1 ) );
2690                                 if ( $value === false ) {
2691                                         $value = '';
2692                                 }
2693                                 if ( $name !== false ) {
2694                                         $assocArgs[$name] = $value;
2695                                 }
2696                         }
2697                 }
2698
2699                 return $assocArgs;
2700         }
2701
2702         /**
2703          * Return the text of a template, after recursively
2704          * replacing any variables or templates within the template.
2705          *
2706          * @param array $piece The parts of the template
2707          *  $piece['title']: the title, i.e. the part before the |
2708          *  $piece['parts']: the parameter array
2709          *  $piece['lineStart']: whether the brace was at the start of a line
2710          * @param PPFrame The current frame, contains template arguments
2711          * @return string the text of the template
2712          * @private
2713          */
2714         function braceSubstitution( $piece, $frame ) {
2715                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2716                 $fname = __METHOD__;
2717                 wfProfileIn( $fname );
2718                 wfProfileIn( __METHOD__.'-setup' );
2719
2720                 # Flags
2721                 $found = false;             # $text has been filled
2722                 $nowiki = false;            # wiki markup in $text should be escaped
2723                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2724                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2725                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2726                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2727
2728                 # Title object, where $text came from
2729                 $title = NULL;
2730
2731                 # $part1 is the bit before the first |, and must contain only title characters.
2732                 # Various prefixes will be stripped from it later.
2733                 $titleWithSpaces = $frame->expand( $piece['title'] );
2734                 $part1 = trim( $titleWithSpaces );
2735                 $titleText = false;
2736
2737                 # Original title text preserved for various purposes
2738                 $originalTitle = $part1;
2739
2740                 # $args is a list of argument nodes, starting from index 0, not including $part1
2741                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2742                 wfProfileOut( __METHOD__.'-setup' );
2743
2744                 # SUBST
2745                 wfProfileIn( __METHOD__.'-modifiers' );
2746                 if ( !$found ) {
2747                         $mwSubst = MagicWord::get( 'subst' );
2748                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2749                                 # One of two possibilities is true:
2750                                 # 1) Found SUBST but not in the PST phase
2751                                 # 2) Didn't find SUBST and in the PST phase
2752                                 # In either case, return without further processing
2753                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2754                                 $isLocalObj = true;
2755                                 $found = true;
2756                         }
2757                 }
2758
2759                 # Variables
2760                 if ( !$found && $args->getLength() == 0 ) {
2761                         $id = $this->mVariables->matchStartToEnd( $part1 );
2762                         if ( $id !== false ) {
2763                                 $text = $this->getVariableValue( $id );
2764                                 if (MagicWord::getCacheTTL($id)>-1)
2765                                         $this->mOutput->mContainsOldMagic = true;
2766                                 $found = true;
2767                         }
2768                 }
2769
2770                 # MSG, MSGNW and RAW
2771                 if ( !$found ) {
2772                         # Check for MSGNW:
2773                         $mwMsgnw = MagicWord::get( 'msgnw' );
2774                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2775                                 $nowiki = true;
2776                         } else {
2777                                 # Remove obsolete MSG:
2778                                 $mwMsg = MagicWord::get( 'msg' );
2779                                 $mwMsg->matchStartAndRemove( $part1 );
2780                         }
2781
2782                         # Check for RAW:
2783                         $mwRaw = MagicWord::get( 'raw' );
2784                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2785                                 $forceRawInterwiki = true;
2786                         }
2787                 }
2788                 wfProfileOut( __METHOD__.'-modifiers' );
2789
2790                 # Parser functions
2791                 if ( !$found ) {
2792                         wfProfileIn( __METHOD__ . '-pfunc' );
2793
2794                         $colonPos = strpos( $part1, ':' );
2795                         if ( $colonPos !== false ) {
2796                                 # Case sensitive functions
2797                                 $function = substr( $part1, 0, $colonPos );
2798                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2799                                         $function = $this->mFunctionSynonyms[1][$function];
2800                                 } else {
2801                                         # Case insensitive functions
2802                                         $function = strtolower( $function );
2803                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2804                                                 $function = $this->mFunctionSynonyms[0][$function];
2805                                         } else {
2806                                                 $function = false;
2807                                         }
2808                                 }
2809                                 if ( $function ) {
2810                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2811                                         $initialArgs = array( &$this );
2812                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2813                                         if ( $flags & SFH_OBJECT_ARGS ) {
2814                                                 # Add a frame parameter, and pass the arguments as an array
2815                                                 $allArgs = $initialArgs;
2816                                                 $allArgs[] = $frame;
2817                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2818                                                         $funcArgs[] = $args->item( $i );
2819                                                 }
2820                                                 $allArgs[] = $funcArgs;
2821                                         } else {
2822                                                 # Convert arguments to plain text
2823                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2824                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2825                                                 }
2826                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2827                                         }
2828
2829                                         # Workaround for PHP bug 35229 and similar
2830                                         if ( !is_callable( $callback ) ) {
2831                                                 throw new MWException( "Tag hook for $name is not callable\n" );
2832                                         }
2833                                         $result = call_user_func_array( $callback, $allArgs );
2834                                         $found = true;
2835
2836                                         if ( is_array( $result ) ) {
2837                                                 if ( isset( $result[0] ) ) {
2838                                                         $text = $result[0];
2839                                                         unset( $result[0] );
2840                                                 }
2841
2842                                                 // Extract flags into the local scope
2843                                                 // This allows callers to set flags such as nowiki, found, etc.
2844                                                 extract( $result );
2845                                         } else {
2846                                                 $text = $result;
2847                                         }
2848                                 }
2849                         }
2850                         wfProfileOut( __METHOD__ . '-pfunc' );
2851                 }
2852
2853                 # Finish mangling title and then check for loops.
2854                 # Set $title to a Title object and $titleText to the PDBK
2855                 if ( !$found ) {
2856                         $ns = NS_TEMPLATE;
2857                         # Split the title into page and subpage
2858                         $subpage = '';
2859                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2860                         if ($subpage !== '') {
2861                                 $ns = $this->mTitle->getNamespace();
2862                         }
2863                         $title = Title::newFromText( $part1, $ns );
2864                         if ( $title ) {
2865                                 $titleText = $title->getPrefixedText();
2866                                 # Check for language variants if the template is not found
2867                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2868                                         $wgContLang->findVariantLink($part1, $title);
2869                                 }
2870                                 # Do infinite loop check
2871                                 if ( !$frame->loopCheck( $title ) ) {
2872                                         $found = true;
2873                                         $text = "<span class=\"error\">Template loop detected: [[$titleText]]</span>";
2874                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2875                                 }
2876                                 # Do recursion depth check
2877                                 $limit = $this->mOptions->getMaxTemplateDepth();
2878                                 if ( $frame->depth >= $limit ) {
2879                                         $found = true;
2880                                         $text = "<span class=\"error\">Template recursion depth limit exceeded ($limit)</span>";
2881                                 }
2882                         }
2883                 }
2884
2885                 # Load from database
2886                 if ( !$found && $title ) {
2887                         wfProfileIn( __METHOD__ . '-loadtpl' );
2888                         if ( !$title->isExternal() ) {
2889                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2890                                         $text = SpecialPage::capturePath( $title );
2891                                         if ( is_string( $text ) ) {
2892                                                 $found = true;
2893                                                 $isHTML = true;
2894                                                 $this->disableCache();
2895                                         }
2896                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2897                                         $found = false; //access denied
2898                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
2899                                 } else {
2900                                         list( $text, $title ) = $this->getTemplateDom( $title );
2901                                         if ( $text !== false ) {
2902                                                 $found = true;
2903                                                 $isChildObj = true;
2904                                         }
2905                                 }
2906
2907                                 # If the title is valid but undisplayable, make a link to it
2908                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2909                                         $text = "[[:$titleText]]";
2910                                         $found = true;
2911                                 }
2912                         } elseif ( $title->isTrans() ) {
2913                                 // Interwiki transclusion
2914                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2915                                         $text = $this->interwikiTransclude( $title, 'render' );
2916                                         $isHTML = true;
2917                                 } else {
2918                                         $text = $this->interwikiTransclude( $title, 'raw' );
2919                                         // Preprocess it like a template
2920                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2921                                         $isChildObj = true;
2922                                 }
2923                                 $found = true;
2924                         }
2925                         wfProfileOut( __METHOD__ . '-loadtpl' );
2926                 }
2927
2928                 # If we haven't found text to substitute by now, we're done
2929                 # Recover the source wikitext and return it
2930                 if ( !$found ) {
2931                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2932                         wfProfileOut( $fname );
2933                         return array( 'object' => $text );
2934                 }
2935
2936                 # Expand DOM-style return values in a child frame
2937                 if ( $isChildObj ) {
2938                         # Clean up argument array
2939                         $newFrame = $frame->newChild( $args, $title );
2940
2941                         if ( $nowiki ) {
2942                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2943                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2944                                 # Expansion is eligible for the empty-frame cache
2945                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2946                                         $text = $this->mTplExpandCache[$titleText];
2947                                 } else {
2948                                         $text = $newFrame->expand( $text );
2949                                         $this->mTplExpandCache[$titleText] = $text;
2950                                 }
2951                         } else {
2952                                 # Uncached expansion
2953                                 $text = $newFrame->expand( $text );
2954                         }
2955                 }
2956                 if ( $isLocalObj && $nowiki ) {
2957                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2958                         $isLocalObj = false;
2959                 }
2960
2961                 # Replace raw HTML by a placeholder
2962                 # Add a blank line preceding, to prevent it from mucking up
2963                 # immediately preceding headings
2964                 if ( $isHTML ) {
2965                         $text = "\n\n" . $this->insertStripItem( $text );
2966                 }
2967                 # Escape nowiki-style return values
2968                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2969                         $text = wfEscapeWikiText( $text );
2970                 }
2971                 # Bug 529: if the template begins with a table or block-level
2972                 # element, it should be treated as beginning a new line.
2973                 # This behaviour is somewhat controversial.
2974                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
2975                         $text = "\n" . $text;
2976                 }
2977
2978                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
2979                         # Error, oversize inclusion
2980                         $text = "[[$originalTitle]]" .
2981                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
2982                 }
2983
2984                 if ( $isLocalObj ) {
2985                         $ret = array( 'object' => $text );
2986                 } else {
2987                         $ret = array( 'text' => $text );
2988                 }
2989
2990                 wfProfileOut( $fname );
2991                 return $ret;
2992         }
2993
2994         /**
2995          * Get the semi-parsed DOM representation of a template with a given title,
2996          * and its redirect destination title. Cached.
2997          */
2998         function getTemplateDom( $title ) {
2999                 $cacheTitle = $title;
3000                 $titleText = $title->getPrefixedDBkey();
3001
3002                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3003                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3004                         $title = Title::makeTitle( $ns, $dbk );
3005                         $titleText = $title->getPrefixedDBkey();
3006                 }
3007                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3008                         return array( $this->mTplDomCache[$titleText], $title );
3009                 }
3010
3011                 // Cache miss, go to the database
3012                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3013
3014                 if ( $text === false ) {
3015                         $this->mTplDomCache[$titleText] = false;
3016                         return array( false, $title );
3017                 }
3018
3019                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3020                 $this->mTplDomCache[ $titleText ] = $dom;
3021
3022                 if (! $title->equals($cacheTitle)) {
3023                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3024                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3025                 }
3026
3027                 return array( $dom, $title );
3028         }
3029
3030         /**
3031          * Fetch the unparsed text of a template and register a reference to it.
3032          */
3033         function fetchTemplateAndTitle( $title ) {
3034                 $templateCb = $this->mOptions->getTemplateCallback();
3035                 $stuff = call_user_func( $templateCb, $title );
3036                 $text = $stuff['text'];
3037                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3038                 if ( isset( $stuff['deps'] ) ) {
3039                         foreach ( $stuff['deps'] as $dep ) {
3040                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3041                         }
3042                 }
3043                 return array($text,$finalTitle);
3044         }
3045
3046         function fetchTemplate( $title ) {
3047                 $rv = $this->fetchTemplateAndTitle($title);
3048                 return $rv[0];
3049         }
3050
3051         /**
3052          * Static function to get a template
3053          * Can be overridden via ParserOptions::setTemplateCallback().
3054          */
3055         static function statelessFetchTemplate( $title ) {
3056                 $text = $skip = false;
3057                 $finalTitle = $title;
3058                 $deps = array();
3059
3060                 // Loop to fetch the article, with up to 1 redirect
3061                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3062                         # Give extensions a chance to select the revision instead
3063                         $id = false; // Assume current
3064                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3065
3066                         if( $skip ) {
3067                                 $text = false;
3068                                 $deps[] = array(
3069                                         'title' => $title,
3070                                         'page_id' => $title->getArticleID(),
3071                                         'rev_id' => null );
3072                                 break;
3073                         }
3074                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3075                         $rev_id = $rev ? $rev->getId() : 0;
3076
3077                         $deps[] = array(
3078                                 'title' => $title,
3079                                 'page_id' => $title->getArticleID(),
3080                                 'rev_id' => $rev_id );
3081
3082                         if( $rev ) {
3083                                 $text = $rev->getText();
3084                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3085                                 global $wgLang;
3086                                 $message = $wgLang->lcfirst( $title->getText() );
3087                                 $text = wfMsgForContentNoTrans( $message );
3088                                 if( wfEmptyMsg( $message, $text ) ) {
3089                                         $text = false;
3090                                         break;
3091                                 }
3092                         } else {
3093                                 break;
3094                         }
3095                         if ( $text === false ) {
3096                                 break;
3097                         }
3098                         // Redirect?
3099                         $finalTitle = $title;
3100                         $title = Title::newFromRedirect( $text );
3101                 }
3102                 return array(
3103                         'text' => $text,
3104                         'finalTitle' => $finalTitle,
3105                         'deps' => $deps );
3106         }
3107
3108         /**
3109          * Transclude an interwiki link.
3110          */
3111         function interwikiTransclude( $title, $action ) {
3112                 global $wgEnableScaryTranscluding;
3113
3114                 if (!$wgEnableScaryTranscluding)
3115                         return wfMsg('scarytranscludedisabled');
3116
3117                 $url = $title->getFullUrl( "action=$action" );
3118
3119                 if (strlen($url) > 255)
3120                         return wfMsg('scarytranscludetoolong');
3121                 return $this->fetchScaryTemplateMaybeFromCache($url);
3122         }
3123
3124         function fetchScaryTemplateMaybeFromCache($url) {
3125                 global $wgTranscludeCacheExpiry;
3126                 $dbr = wfGetDB(DB_SLAVE);
3127                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3128                                 array('tc_url' => $url));
3129                 if ($obj) {
3130                         $time = $obj->tc_time;
3131                         $text = $obj->tc_contents;
3132                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3133                                 return $text;
3134                         }
3135                 }
3136
3137                 $text = Http::get($url);
3138                 if (!$text)
3139                         return wfMsg('scarytranscludefailed', $url);
3140
3141                 $dbw = wfGetDB(DB_MASTER);
3142                 $dbw->replace('transcache', array('tc_url'), array(
3143                         'tc_url' => $url,
3144                         'tc_time' => time(),
3145                         'tc_contents' => $text));
3146                 return $text;
3147         }
3148
3149
3150         /**
3151          * Triple brace replacement -- used for template arguments
3152          * @private
3153          */
3154         function argSubstitution( $piece, $frame ) {
3155                 wfProfileIn( __METHOD__ );
3156
3157                 $error = false;
3158                 $parts = $piece['parts'];
3159                 $nameWithSpaces = $frame->expand( $piece['title'] );
3160                 $argName = trim( $nameWithSpaces );
3161                 $object = false;
3162                 $text = $frame->getArgument( $argName );
3163                 if (  $text === false && $parts->getLength() > 0
3164                   && (
3165                     $this->ot['html']
3166                     || $this->ot['pre']
3167                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3168                   )
3169                 ) {
3170                         # No match in frame, use the supplied default
3171                         $object = $parts->item( 0 )->getChildren();
3172                 }
3173                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3174                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3175                 }
3176
3177                 if ( $text === false && $object === false ) {
3178                         # No match anywhere
3179                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3180                 }
3181                 if ( $error !== false ) {
3182                         $text .= $error;
3183                 }
3184                 if ( $object !== false ) {
3185                         $ret = array( 'object' => $object );
3186                 } else {
3187                         $ret = array( 'text' => $text );
3188                 }
3189
3190                 wfProfileOut( __METHOD__ );
3191                 return $ret;
3192         }
3193
3194         /**
3195          * Return the text to be used for a given extension tag.
3196          * This is the ghost of strip().
3197          *
3198          * @param array $params Associative array of parameters:
3199          *     name       PPNode for the tag name
3200          *     attr       PPNode for unparsed text where tag attributes are thought to be
3201          *     attributes Optional associative array of parsed attributes
3202          *     inner      Contents of extension element
3203          *     noClose    Original text did not have a close tag
3204          * @param PPFrame $frame
3205          */
3206         function extensionSubstitution( $params, $frame ) {
3207                 global $wgRawHtml, $wgContLang;
3208
3209                 $name = $frame->expand( $params['name'] );
3210                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3211                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3212
3213                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3214
3215                 if ( $this->ot['html'] ) {
3216                         $name = strtolower( $name );
3217
3218                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3219                         if ( isset( $params['attributes'] ) ) {
3220                                 $attributes = $attributes + $params['attributes'];
3221                         }
3222                         switch ( $name ) {
3223                                 case 'html':
3224                                         if( $wgRawHtml ) {
3225                                                 $output = $content;
3226                                                 break;
3227                                         } else {
3228                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3229                                         }
3230                                 case 'nowiki':
3231                                         $output = Xml::escapeTagsOnly( $content );
3232                                         break;
3233                                 case 'math':
3234                                         $output = $wgContLang->armourMath(
3235                                                 MathRenderer::renderMath( $content, $attributes ) );
3236                                         break;
3237                                 case 'gallery':
3238                                         $output = $this->renderImageGallery( $content, $attributes );
3239                                         break;
3240                                 default:
3241                                         if( isset( $this->mTagHooks[$name] ) ) {
3242                                                 # Workaround for PHP bug 35229 and similar
3243                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3244                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3245                                                 }
3246                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3247                                                         array( $content, $attributes, $this ) );
3248                                         } else {
3249                                                 throw new MWException( "Invalid call hook $name" );
3250                                         }
3251                         }
3252                 } else {
3253                         if ( is_null( $attrText ) ) {
3254                                 $attrText = '';
3255                         }
3256                         if ( isset( $params['attributes'] ) ) {
3257                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3258                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3259                                                 htmlspecialchars( $attrValue ) . '"';
3260                                 }
3261                         }
3262                         if ( $content === null ) {
3263                                 $output = "<$name$attrText/>";
3264                         } else {
3265                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3266                                 $output = "<$name$attrText>$content$close";
3267                         }
3268                 }
3269
3270                 if ( $name == 'html' || $name == 'nowiki' ) {
3271                         $this->mStripState->nowiki->setPair( $marker, $output );
3272                 } else {
3273                         $this->mStripState->general->setPair( $marker, $output );
3274                 }
3275                 return $marker;
3276         }
3277
3278         /**
3279          * Increment an include size counter
3280          *
3281          * @param string $type The type of expansion
3282          * @param integer $size The size of the text
3283          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3284          */
3285         function incrementIncludeSize( $type, $size ) {
3286                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3287                         return false;
3288                 } else {
3289                         $this->mIncludeSizes[$type] += $size;
3290                         return true;
3291                 }
3292         }
3293
3294         /**
3295          * Increment the expensive function count
3296          *
3297          * @return boolean False if the limit has been exceeded
3298          */
3299         function incrementExpensiveFunctionCount() {
3300                 global $wgExpensiveParserFunctionLimit;
3301                 $this->mExpensiveFunctionCount++;
3302                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3303                         return true;
3304                 }
3305                 return false;
3306         }
3307
3308         /**
3309          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3310          * Fills $this->mDoubleUnderscores, returns the modified text
3311          */
3312         function doDoubleUnderscore( $text ) {
3313                 // The position of __TOC__ needs to be recorded
3314                 $mw = MagicWord::get( 'toc' );
3315                 if( $mw->match( $text ) ) {
3316                         $this->mShowToc = true;
3317                         $this->mForceTocPosition = true;
3318
3319                         // Set a placeholder. At the end we'll fill it in with the TOC.
3320                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3321
3322                         // Only keep the first one.
3323                         $text = $mw->replace( '', $text );
3324                 }
3325
3326                 // Now match and remove the rest of them
3327                 $mwa = MagicWord::getDoubleUnderscoreArray();
3328                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3329
3330                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3331                         $this->mOutput->mNoGallery = true;
3332                 }
3333                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3334                         $this->mShowToc = false;
3335                 }
3336                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3337                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3338
3339                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3340                         if ( $containerCategory ) {
3341                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3342                         } else {
3343                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3344                         }
3345                 }
3346                 return $text;
3347         }
3348
3349         /**
3350          * This function accomplishes several tasks:
3351          * 1) Auto-number headings if that option is enabled
3352          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3353          * 3) Add a Table of contents on the top for users who have enabled the option
3354          * 4) Auto-anchor headings
3355          *
3356          * It loops through all headlines, collects the necessary data, then splits up the
3357          * string and re-inserts the newly formatted headlines.
3358          *
3359          * @param string $text
3360          * @param boolean $isMain
3361          * @private
3362          */
3363         function formatHeadings( $text, $isMain=true ) {
3364                 global $wgMaxTocLevel, $wgContLang;
3365
3366                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3367                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3368                         $showEditLink = 0;
3369                 } else {
3370                         $showEditLink = $this->mOptions->getEditSection();
3371                 }
3372
3373                 # Inhibit editsection links if requested in the page
3374                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3375                         $showEditLink = 0;
3376                 }
3377
3378                 # Get all headlines for numbering them and adding funky stuff like [edit]
3379                 # links - this is for later, but we need the number of headlines right now
3380                 $matches = array();
3381                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3382
3383                 # if there are fewer than 4 headlines in the article, do not show TOC
3384                 # unless it's been explicitly enabled.
3385                 $enoughToc = $this->mShowToc &&
3386                         (($numMatches >= 4) || $this->mForceTocPosition);
3387
3388                 # Allow user to stipulate that a page should have a "new section"
3389                 # link added via __NEWSECTIONLINK__
3390                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3391                         $this->mOutput->setNewSection( true );
3392                 }
3393
3394                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3395                 # override above conditions and always show TOC above first header
3396                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3397                         $this->mShowToc = true;
3398                         $enoughToc = true;
3399                 }
3400
3401                 # We need this to perform operations on the HTML
3402                 $sk = $this->mOptions->getSkin();
3403
3404                 # headline counter
3405                 $headlineCount = 0;
3406                 $numVisible = 0;
3407
3408                 # Ugh .. the TOC should have neat indentation levels which can be
3409                 # passed to the skin functions. These are determined here
3410                 $toc = '';
3411                 $full = '';
3412                 $head = array();
3413                 $sublevelCount = array();
3414                 $levelCount = array();
3415                 $toclevel = 0;
3416                 $level = 0;
3417                 $prevlevel = 0;
3418                 $toclevel = 0;
3419                 $prevtoclevel = 0;
3420                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3421                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3422                 $tocraw = array();
3423
3424                 foreach( $matches[3] as $headline ) {
3425                         $isTemplate = false;
3426                         $titleText = false;
3427                         $sectionIndex = false;
3428                         $numbering = '';
3429                         $markerMatches = array();
3430                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3431                                 $serial = $markerMatches[1];
3432                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3433                                 $isTemplate = ($titleText != $baseTitleText);
3434                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3435                         }
3436
3437                         if( $toclevel ) {
3438                                 $prevlevel = $level;
3439                                 $prevtoclevel = $toclevel;
3440                         }
3441                         $level = $matches[1][$headlineCount];
3442
3443                         if( $doNumberHeadings || $enoughToc ) {
3444
3445                                 if ( $level > $prevlevel ) {
3446                                         # Increase TOC level
3447                                         $toclevel++;
3448                                         $sublevelCount[$toclevel] = 0;
3449                                         if( $toclevel<$wgMaxTocLevel ) {
3450                                                 $prevtoclevel = $toclevel;
3451                                                 $toc .= $sk->tocIndent();
3452                                                 $numVisible++;
3453                                         }
3454                                 }
3455                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3456                                         # Decrease TOC level, find level to jump to
3457
3458                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3459                                                 # Can only go down to level 1
3460                                                 $toclevel = 1;
3461                                         } else {
3462                                                 for ($i = $toclevel; $i > 0; $i--) {
3463                                                         if ( $levelCount[$i] == $level ) {
3464                                                                 # Found last matching level
3465                                                                 $toclevel = $i;
3466                                                                 break;
3467                                                         }
3468                                                         elseif ( $levelCount[$i] < $level ) {
3469                                                                 # Found first matching level below current level
3470                                                                 $toclevel = $i + 1;
3471                                                                 break;
3472                                                         }
3473                                                 }
3474                                         }
3475                                         if( $toclevel<$wgMaxTocLevel ) {
3476                                                 if($prevtoclevel < $wgMaxTocLevel) {
3477                                                         # Unindent only if the previous toc level was shown :p
3478                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3479                                                         $prevtoclevel = $toclevel;
3480                                                 } else {
3481                                                         $toc .= $sk->tocLineEnd();
3482                                                 }
3483                                         }
3484                                 }
3485                                 else {
3486                                         # No change in level, end TOC line
3487                                         if( $toclevel<$wgMaxTocLevel ) {
3488                                                 $toc .= $sk->tocLineEnd();
3489                                         }
3490                                 }
3491
3492                                 $levelCount[$toclevel] = $level;
3493
3494                                 # count number of headlines for each level
3495                                 @$sublevelCount[$toclevel]++;
3496                                 $dot = 0;
3497                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3498                                         if( !empty( $sublevelCount[$i] ) ) {
3499                                                 if( $dot ) {
3500                                                         $numbering .= '.';
3501                                                 }
3502                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3503                                                 $dot = 1;
3504                                         }
3505                                 }
3506                         }
3507
3508                         # The safe header is a version of the header text safe to use for links
3509                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3510                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3511
3512                         # Remove link placeholders by the link text.
3513                         #     <!--LINK number-->
3514                         # turns into
3515                         #     link text with suffix
3516                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3517                                                             "\$this->mLinkHolders['texts'][\$1]",
3518                                                             $safeHeadline );
3519                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3520                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3521                                                             $safeHeadline );
3522
3523                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3524                         $tocline = preg_replace(
3525                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3526                                 array( '',                          '<$1>'),
3527                                 $safeHeadline
3528                         );
3529                         $tocline = trim( $tocline );
3530
3531                         # For the anchor, strip out HTML-y stuff period
3532                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3533                         $safeHeadline = trim( $safeHeadline );
3534
3535                         # Save headline for section edit hint before it's escaped
3536                         $headlineHint = $safeHeadline;
3537                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3538                         # HTML names must be case-insensitively unique (bug 10721)
3539                         $arrayKey = strtolower( $safeHeadline );
3540
3541                         # XXX : Is $refers[$headlineCount] ever accessed, actually ?
3542                         $refers[$headlineCount] = $safeHeadline;
3543
3544                         # count how many in assoc. array so we can track dupes in anchors
3545                         isset( $refers[$arrayKey] ) ? $refers[$arrayKey]++ : $refers[$arrayKey] = 1;
3546                         $refcount[$headlineCount] = $refers[$arrayKey];
3547
3548                         # Don't number the heading if it is the only one (looks silly)
3549                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3550                                 # the two are different if the line contains a link
3551                                 $headline=$numbering . ' ' . $headline;
3552                         }
3553
3554                         # Create the anchor for linking from the TOC to the section
3555                         $anchor = $safeHeadline;
3556                         if($refcount[$headlineCount] > 1 ) {
3557                                 $anchor .= '_' . $refcount[$headlineCount];
3558                         }
3559                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3560                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3561                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3562                         }
3563                         # give headline the correct <h#> tag
3564                         if( $showEditLink && $sectionIndex !== false ) {
3565                                 if( $isTemplate ) {
3566                                         # Put a T flag in the section identifier, to indicate to extractSections()
3567                                         # that sections inside <includeonly> should be counted.
3568                                         $editlink = $sk->editSectionLinkForOther($titleText, "T-$sectionIndex");
3569                                 } else {
3570                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3571                                 }
3572                         } else {
3573                                 $editlink = '';
3574                         }
3575                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3576
3577                         $headlineCount++;
3578                 }
3579
3580                 $this->mOutput->setSections( $tocraw );
3581
3582                 # Never ever show TOC if no headers
3583                 if( $numVisible < 1 ) {
3584                         $enoughToc = false;
3585                 }
3586
3587                 if( $enoughToc ) {
3588                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3589                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3590                         }
3591                         $toc = $sk->tocList( $toc );
3592                 }
3593
3594                 # split up and insert constructed headlines
3595
3596                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3597                 $i = 0;
3598
3599                 foreach( $blocks as $block ) {
3600                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3601                                 # This is the [edit] link that appears for the top block of text when
3602                                 # section editing is enabled
3603
3604                                 # Disabled because it broke block formatting
3605                                 # For example, a bullet point in the top line
3606                                 # $full .= $sk->editSectionLink(0);
3607                         }
3608                         $full .= $block;
3609                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3610                                 # Top anchor now in skin
3611                                 $full = $full.$toc;
3612                         }
3613
3614                         if( !empty( $head[$i] ) ) {
3615                                 $full .= $head[$i];
3616                         }
3617                         $i++;
3618                 }
3619                 if( $this->mForceTocPosition ) {
3620                         return str_replace( '<!--MWTOC-->', $toc, $full );
3621                 } else {
3622                         return $full;
3623                 }
3624         }
3625
3626         /**
3627          * Transform wiki markup when saving a page by doing \r\n -> \n
3628          * conversion, substitting signatures, {{subst:}} templates, etc.
3629          *
3630          * @param string $text the text to transform
3631          * @param Title &$title the Title object for the current article
3632          * @param User &$user the User object describing the current user
3633          * @param ParserOptions $options parsing options
3634          * @param bool $clearState whether to clear the parser state first
3635          * @return string the altered wiki markup
3636          * @public
3637          */
3638         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3639                 $this->mOptions = $options;
3640                 $this->setTitle( $title );
3641                 $this->setOutputType( self::OT_WIKI );
3642
3643                 if ( $clearState ) {
3644                         $this->clearState();
3645                 }
3646
3647                 $pairs = array(
3648                         "\r\n" => "\n",
3649                 );
3650                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3651                 $text = $this->pstPass2( $text, $user );
3652                 $text = $this->mStripState->unstripBoth( $text );
3653                 return $text;
3654         }
3655
3656         /**
3657          * Pre-save transform helper function
3658          * @private
3659          */
3660         function pstPass2( $text, $user ) {
3661                 global $wgContLang, $wgLocaltimezone;
3662
3663                 /* Note: This is the timestamp saved as hardcoded wikitext to
3664                  * the database, we use $wgContLang here in order to give
3665                  * everyone the same signature and use the default one rather
3666                  * than the one selected in each user's preferences.
3667                  *
3668                  * (see also bug 12815)
3669                  */
3670                 $ts = $this->mOptions->getTimestamp();
3671                 $tz = 'UTC';
3672                 if ( isset( $wgLocaltimezone ) ) {
3673                         $unixts = wfTimestamp( TS_UNIX, $ts );
3674                         $oldtz = getenv( 'TZ' );
3675                         putenv( 'TZ='.$wgLocaltimezone );
3676                         $ts = date( 'YmdHis', $unixts );
3677                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3678                         putenv( 'TZ='.$oldtz );
3679                 }
3680                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3681
3682                 # Variable replacement
3683                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3684                 $text = $this->replaceVariables( $text );
3685
3686                 # Signatures
3687                 $sigText = $this->getUserSig( $user );
3688                 $text = strtr( $text, array(
3689                         '~~~~~' => $d,
3690                         '~~~~' => "$sigText $d",
3691                         '~~~' => $sigText
3692                 ) );
3693
3694                 # Context links: [[|name]] and [[name (context)|]]
3695                 #
3696                 global $wgLegalTitleChars;
3697                 $tc = "[$wgLegalTitleChars]";
3698                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3699
3700                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3701                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3702                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3703
3704                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3705                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3706                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3707
3708                 $t = $this->mTitle->getText();
3709                 $m = array();
3710                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3711                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3712                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3713                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3714                 } else {
3715                         # if there's no context, don't bother duplicating the title
3716                         $text = preg_replace( $p2, '[[\\1]]', $text );
3717                 }
3718
3719                 # Trim trailing whitespace
3720                 $text = rtrim( $text );
3721
3722                 return $text;
3723         }
3724
3725         /**
3726          * Fetch the user's signature text, if any, and normalize to
3727          * validated, ready-to-insert wikitext.
3728          *
3729          * @param User $user
3730          * @return string
3731          * @private
3732          */
3733         function getUserSig( &$user ) {
3734                 global $wgMaxSigChars;
3735
3736                 $username = $user->getName();
3737                 $nickname = $user->getOption( 'nickname' );
3738                 $nickname = $nickname === '' ? $username : $nickname;
3739
3740                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3741                         $nickname = $username;
3742                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3743                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3744                         # Sig. might contain markup; validate this
3745                         if( $this->validateSig( $nickname ) !== false ) {
3746                                 # Validated; clean up (if needed) and return it
3747                                 return $this->cleanSig( $nickname, true );
3748                         } else {
3749                                 # Failed to validate; fall back to the default
3750                                 $nickname = $username;
3751                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3752                         }
3753                 }
3754
3755                 // Make sure nickname doesnt get a sig in a sig
3756                 $nickname = $this->cleanSigInSig( $nickname );
3757
3758                 # If we're still here, make it a link to the user page
3759                 $userText = wfEscapeWikiText( $username );
3760                 $nickText = wfEscapeWikiText( $nickname );
3761                 if ( $user->isAnon() )  {
3762                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3763                 } else {
3764                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3765                 }
3766         }
3767
3768         /**
3769          * Check that the user's signature contains no bad XML
3770          *
3771          * @param string $text
3772          * @return mixed An expanded string, or false if invalid.
3773          */
3774         function validateSig( $text ) {
3775                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3776         }
3777
3778         /**
3779          * Clean up signature text
3780          *
3781          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3782          * 2) Substitute all transclusions
3783          *
3784          * @param string $text
3785          * @param $parsing Whether we're cleaning (preferences save) or parsing
3786          * @return string Signature text
3787          */
3788         function cleanSig( $text, $parsing = false ) {
3789                 if ( !$parsing ) {
3790                         global $wgTitle;
3791                         $this->clearState();
3792                         $this->setTitle( $wgTitle );
3793                         $this->mOptions = new ParserOptions;
3794                         $this->setOutputType = self::OT_PREPROCESS;
3795                 }
3796
3797                 # FIXME: regex doesn't respect extension tags or nowiki
3798                 #  => Move this logic to braceSubstitution()
3799                 $substWord = MagicWord::get( 'subst' );
3800                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3801                 $substText = '{{' . $substWord->getSynonym( 0 );
3802
3803                 $text = preg_replace( $substRegex, $substText, $text );
3804                 $text = $this->cleanSigInSig( $text );
3805                 $dom = $this->preprocessToDom( $text );
3806                 $frame = $this->getPreprocessor()->newFrame();
3807                 $text = $frame->expand( $dom );
3808
3809                 if ( !$parsing ) {
3810                         $text = $this->mStripState->unstripBoth( $text );
3811                 }
3812
3813                 return $text;
3814         }
3815
3816         /**
3817          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3818          * @param string $text
3819          * @return string Signature text with /~{3,5}/ removed
3820          */
3821         function cleanSigInSig( $text ) {
3822                 $text = preg_replace( '/~{3,5}/', '', $text );
3823                 return $text;
3824         }
3825
3826         /**
3827          * Set up some variables which are usually set up in parse()
3828          * so that an external function can call some class members with confidence
3829          * @public
3830          */
3831         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3832                 $this->setTitle( $title );
3833                 $this->mOptions = $options;
3834                 $this->setOutputType( $outputType );
3835                 if ( $clearState ) {
3836                         $this->clearState();
3837                 }
3838         }
3839
3840         /**
3841          * Wrapper for preprocess()
3842          *
3843          * @param string $text the text to preprocess
3844          * @param ParserOptions $options  options
3845          * @return string
3846          * @public
3847          */
3848         function transformMsg( $text, $options ) {
3849                 global $wgTitle;
3850                 static $executing = false;
3851
3852                 $fname = "Parser::transformMsg";
3853
3854                 # Guard against infinite recursion
3855                 if ( $executing ) {
3856                         return $text;
3857                 }
3858                 $executing = true;
3859
3860                 wfProfileIn($fname);
3861                 $text = $this->preprocess( $text, $wgTitle, $options );
3862
3863                 $executing = false;
3864                 wfProfileOut($fname);
3865                 return $text;
3866         }
3867
3868         /**
3869          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3870          * The callback should have the following form:
3871          *    function myParserHook( $text, $params, &$parser ) { ... }
3872          *
3873          * Transform and return $text. Use $parser for any required context, e.g. use
3874          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3875          *
3876          * @public
3877          *
3878          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3879          * @param mixed $callback The callback function (and object) to use for the tag
3880          *
3881          * @return The old value of the mTagHooks array associated with the hook
3882          */
3883         function setHook( $tag, $callback ) {
3884                 $tag = strtolower( $tag );
3885                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3886                 $this->mTagHooks[$tag] = $callback;
3887                 if( !in_array( $tag, $this->mStripList ) ) {
3888                         $this->mStripList[] = $tag;
3889                 }
3890
3891                 return $oldVal;
3892         }
3893
3894         function setTransparentTagHook( $tag, $callback ) {
3895                 $tag = strtolower( $tag );
3896                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3897                 $this->mTransparentTagHooks[$tag] = $callback;
3898
3899                 return $oldVal;
3900         }
3901
3902         /**
3903          * Remove all tag hooks
3904          */
3905         function clearTagHooks() {
3906                 $this->mTagHooks = array();
3907                 $this->mStripList = $this->mDefaultStripList;
3908         }
3909
3910         /**
3911          * Create a function, e.g. {{sum:1|2|3}}
3912          * The callback function should have the form:
3913          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3914          *
3915          * The callback may either return the text result of the function, or an array with the text
3916          * in element 0, and a number of flags in the other elements. The names of the flags are
3917          * specified in the keys. Valid flags are:
3918          *   found                     The text returned is valid, stop processing the template. This
3919          *                             is on by default.
3920          *   nowiki                    Wiki markup in the return value should be escaped
3921          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3922          *
3923          * @public
3924          *
3925          * @param string $id The magic word ID
3926          * @param mixed $callback The callback function (and object) to use
3927          * @param integer $flags a combination of the following flags:
3928          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3929          *
3930          * @return The old callback function for this name, if any
3931          */
3932         function setFunctionHook( $id, $callback, $flags = 0 ) {
3933                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
3934                 $this->mFunctionHooks[$id] = array( $callback, $flags );
3935
3936                 # Add to function cache
3937                 $mw = MagicWord::get( $id );
3938                 if( !$mw )
3939                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3940
3941                 $synonyms = $mw->getSynonyms();
3942                 $sensitive = intval( $mw->isCaseSensitive() );
3943
3944                 foreach ( $synonyms as $syn ) {
3945                         # Case
3946                         if ( !$sensitive ) {
3947                                 $syn = strtolower( $syn );
3948                         }
3949                         # Add leading hash
3950                         if ( !( $flags & SFH_NO_HASH ) ) {
3951                                 $syn = '#' . $syn;
3952                         }
3953                         # Remove trailing colon
3954                         if ( substr( $syn, -1, 1 ) == ':' ) {
3955                                 $syn = substr( $syn, 0, -1 );
3956                         }
3957                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3958                 }
3959                 return $oldVal;
3960         }
3961
3962         /**
3963          * Get all registered function hook identifiers
3964          *
3965          * @return array
3966          */
3967         function getFunctionHooks() {
3968                 return array_keys( $this->mFunctionHooks );
3969         }
3970
3971         /**
3972          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3973          * Placeholders created in Skin::makeLinkObj()
3974          * Returns an array of link CSS classes, indexed by PDBK.
3975          * $options is a bit field, RLH_FOR_UPDATE to select for update
3976          */
3977         function replaceLinkHolders( &$text, $options = 0 ) {
3978                 global $wgUser;
3979                 global $wgContLang;
3980
3981                 $fname = 'Parser::replaceLinkHolders';
3982                 wfProfileIn( $fname );
3983
3984                 $pdbks = array();
3985                 $colours = array();
3986                 $linkcolour_ids = array();
3987                 $sk = $this->mOptions->getSkin();
3988                 $linkCache = LinkCache::singleton();
3989
3990                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3991                         wfProfileIn( $fname.'-check' );
3992                         $dbr = wfGetDB( DB_SLAVE );
3993                         $page = $dbr->tableName( 'page' );
3994                         $threshold = $wgUser->getOption('stubthreshold');
3995
3996                         # Sort by namespace
3997                         asort( $this->mLinkHolders['namespaces'] );
3998
3999                         # Generate query
4000                         $query = false;
4001                         $current = null;
4002                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4003                                 # Make title object
4004                                 $title = $this->mLinkHolders['titles'][$key];
4005
4006                                 # Skip invalid entries.
4007                                 # Result will be ugly, but prevents crash.
4008                                 if ( is_null( $title ) ) {
4009                                         continue;
4010                                 }
4011                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4012
4013                                 # Check if it's a static known link, e.g. interwiki
4014                                 if ( $title->isAlwaysKnown() ) {
4015                                         $colours[$pdbk] = '';
4016                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4017                                         $colours[$pdbk] = '';
4018                                         $this->mOutput->addLink( $title, $id );
4019                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4020                                         $colours[$pdbk] = 'new';
4021                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4022                                         $colours[$pdbk] = 'new';
4023                                 } else {
4024                                         # Not in the link cache, add it to the query
4025                                         if ( !isset( $current ) ) {
4026                                                 $current = $ns;
4027                                                 $query =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
4028                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4029                                         } elseif ( $current != $ns ) {
4030                                                 $current = $ns;
4031                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4032                                         } else {
4033                                                 $query .= ', ';
4034                                         }
4035
4036                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4037                                 }
4038                         }
4039                         if ( $query ) {
4040                                 $query .= '))';
4041                                 if ( $options & RLH_FOR_UPDATE ) {
4042                                         $query .= ' FOR UPDATE';
4043                                 }
4044
4045                                 $res = $dbr->query( $query, $fname );
4046
4047                                 # Fetch data and form into an associative array
4048                                 # non-existent = broken
4049                                 while ( $s = $dbr->fetchObject($res) ) {
4050                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4051                                         $pdbk = $title->getPrefixedDBkey();
4052                                         $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect );
4053                                         $this->mOutput->addLink( $title, $s->page_id );
4054                                         $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
4055                                         //add id to the extension todolist
4056                                         $linkcolour_ids[$s->page_id] = $pdbk;
4057                                 }
4058                                 //pass an array of page_ids to an extension
4059                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4060                         }
4061                         wfProfileOut( $fname.'-check' );
4062
4063                         # Do a second query for different language variants of links and categories
4064                         if($wgContLang->hasVariants()){
4065                                 $linkBatch = new LinkBatch();
4066                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4067                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4068                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4069
4070                                 $categories = $this->mOutput->getCategoryLinks();
4071
4072                                 // Add variants of links to link batch
4073                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4074                                         $title = $this->mLinkHolders['titles'][$key];
4075                                         if ( is_null( $title ) )
4076                                                 continue;
4077
4078                                         $pdbk = $title->getPrefixedDBkey();
4079                                         $titleText = $title->getText();
4080
4081                                         // generate all variants of the link title text
4082                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4083
4084                                         // if link was not found (in first query), add all variants to query
4085                                         if ( !isset($colours[$pdbk]) ){
4086                                                 foreach($allTextVariants as $textVariant){
4087                                                         if($textVariant != $titleText){
4088                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4089                                                                 if(is_null($variantTitle)) continue;
4090                                                                 $linkBatch->addObj( $variantTitle );
4091                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4092                                                         }
4093                                                 }
4094                                         }
4095                                 }
4096
4097                                 // process categories, check if a category exists in some variant
4098                                 foreach( $categories as $category ){
4099                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4100                                         foreach($variants as $variant){
4101                                                 if($variant != $category){
4102                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4103                                                         if(is_null($variantTitle)) continue;
4104                                                         $linkBatch->addObj( $variantTitle );
4105                                                         $categoryMap[$variant] = $category;
4106                                                 }
4107                                         }
4108                                 }
4109
4110
4111                                 if(!$linkBatch->isEmpty()){
4112                                         // construct query
4113                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4114
4115                                         $variantQuery =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
4116
4117                                         $variantQuery .= " FROM $page WHERE $titleClause";
4118                                         if ( $options & RLH_FOR_UPDATE ) {
4119                                                 $variantQuery .= ' FOR UPDATE';
4120                                         }
4121
4122                                         $varRes = $dbr->query( $variantQuery, $fname );
4123
4124                                         // for each found variants, figure out link holders and replace
4125                                         while ( $s = $dbr->fetchObject($varRes) ) {
4126
4127                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4128                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4129                                                 $vardbk = $variantTitle->getDBkey();
4130
4131                                                 $holderKeys = array();
4132                                                 if(isset($variantMap[$varPdbk])){
4133                                                         $holderKeys = $variantMap[$varPdbk];
4134                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
4135                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4136                                                 }
4137
4138                                                 // loop over link holders
4139                                                 foreach($holderKeys as $key){
4140                                                         $title = $this->mLinkHolders['titles'][$key];
4141                                                         if ( is_null( $title ) ) continue;
4142
4143                                                         $pdbk = $title->getPrefixedDBkey();
4144
4145                                                         if(!isset($colours[$pdbk])){
4146                                                                 // found link in some of the variants, replace the link holder data
4147                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4148                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4149
4150                                                                 // set pdbk and colour
4151                                                                 $pdbks[$key] = $varPdbk;
4152                                                                 $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );
4153                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4154                                                         }
4155                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4156                                                 }
4157
4158                                                 // check if the object is a variant of a category
4159                                                 if(isset($categoryMap[$vardbk])){
4160                                                         $oldkey = $categoryMap[$vardbk];
4161                                                         if($oldkey != $vardbk)
4162                                                                 $varCategories[$oldkey]=$vardbk;
4163                                                 }
4164                                         }
4165
4166                                         // rebuild the categories in original order (if there are replacements)
4167                                         if(count($varCategories)>0){
4168                                                 $newCats = array();
4169                                                 $originalCats = $this->mOutput->getCategories();
4170                                                 foreach($originalCats as $cat => $sortkey){
4171                                                         // make the replacement
4172                                                         if( array_key_exists($cat,$varCategories) )
4173                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4174                                                         else $newCats[$cat] = $sortkey;
4175                                                 }
4176                                                 $this->mOutput->setCategoryLinks($newCats);
4177                                         }
4178                                 }
4179                         }
4180
4181                         # Construct search and replace arrays
4182                         wfProfileIn( $fname.'-construct' );
4183                         $replacePairs = array();
4184                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4185                                 $pdbk = $pdbks[$key];
4186                                 $searchkey = "<!--LINK $key-->";
4187                                 $title = $this->mLinkHolders['titles'][$key];
4188                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4189                                         $linkCache->addBadLinkObj( $title );
4190                                         $colours[$pdbk] = 'new';
4191                                         $this->mOutput->addLink( $title, 0 );
4192                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4193                                                                         $this->mLinkHolders['texts'][$key],
4194                                                                         $this->mLinkHolders['queries'][$key] );
4195                                 } else {
4196                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4197                                                                         $this->mLinkHolders['texts'][$key],
4198                                                                         $this->mLinkHolders['queries'][$key] );
4199                                 }
4200                         }
4201                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4202                         wfProfileOut( $fname.'-construct' );
4203
4204                         # Do the thing
4205                         wfProfileIn( $fname.'-replace' );
4206                         $text = preg_replace_callback(
4207                                 '/(<!--LINK .*?-->)/',
4208                                 $replacer->cb(),
4209                                 $text);
4210
4211                         wfProfileOut( $fname.'-replace' );
4212                 }
4213
4214                 # Now process interwiki link holders
4215                 # This is quite a bit simpler than internal links
4216                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4217                         wfProfileIn( $fname.'-interwiki' );
4218                         # Make interwiki link HTML
4219                         $replacePairs = array();
4220                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4221                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4222                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4223                         }
4224                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4225
4226                         $text = preg_replace_callback(
4227                                 '/<!--IWLINK (.*?)-->/',
4228                                 $replacer->cb(),
4229                                 $text );
4230                         wfProfileOut( $fname.'-interwiki' );
4231                 }
4232
4233                 wfProfileOut( $fname );
4234                 return $colours;
4235         }
4236
4237         /**
4238          * Replace <!--LINK--> link placeholders with plain text of links
4239          * (not HTML-formatted).
4240          * @param string $text
4241          * @return string
4242          */
4243         function replaceLinkHoldersText( $text ) {
4244                 $fname = 'Parser::replaceLinkHoldersText';
4245                 wfProfileIn( $fname );
4246
4247                 $text = preg_replace_callback(
4248                         '/<!--(LINK|IWLINK) (.*?)-->/',
4249                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4250                         $text );
4251
4252                 wfProfileOut( $fname );
4253                 return $text;
4254         }
4255
4256         /**
4257          * @param array $matches
4258          * @return string
4259          * @private
4260          */
4261         function replaceLinkHoldersTextCallback( $matches ) {
4262                 $type = $matches[1];
4263                 $key  = $matches[2];
4264                 if( $type == 'LINK' ) {
4265                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4266                                 return $this->mLinkHolders['texts'][$key];
4267                         }
4268                 } elseif( $type == 'IWLINK' ) {
4269                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4270                                 return $this->mInterwikiLinkHolders['texts'][$key];
4271                         }
4272                 }
4273                 return $matches[0];
4274         }
4275
4276         /**
4277          * Tag hook handler for 'pre'.
4278          */
4279         function renderPreTag( $text, $attribs ) {
4280                 // Backwards-compatibility hack
4281                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4282
4283                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4284                 return wfOpenElement( 'pre', $attribs ) .
4285                         Xml::escapeTagsOnly( $content ) .
4286                         '</pre>';
4287         }
4288
4289         /**
4290          * Renders an image gallery from a text with one line per image.
4291          * text labels may be given by using |-style alternative text. E.g.
4292          *   Image:one.jpg|The number "1"
4293          *   Image:tree.jpg|A tree
4294          * given as text will return the HTML of a gallery with two images,
4295          * labeled 'The number "1"' and
4296          * 'A tree'.
4297          */
4298         function renderImageGallery( $text, $params ) {
4299                 $ig = new ImageGallery();
4300                 $ig->setContextTitle( $this->mTitle );
4301                 $ig->setShowBytes( false );
4302                 $ig->setShowFilename( false );
4303                 $ig->setParser( $this );
4304                 $ig->setHideBadImages();
4305                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4306                 $ig->useSkin( $this->mOptions->getSkin() );
4307                 $ig->mRevisionId = $this->mRevisionId;
4308
4309                 if( isset( $params['caption'] ) ) {
4310                         $caption = $params['caption'];
4311                         $caption = htmlspecialchars( $caption );
4312                         $caption = $this->replaceInternalLinks( $caption );
4313                         $ig->setCaptionHtml( $caption );
4314                 }
4315                 if( isset( $params['perrow'] ) ) {
4316                         $ig->setPerRow( $params['perrow'] );
4317                 }
4318                 if( isset( $params['widths'] ) ) {
4319                         $ig->setWidths( $params['widths'] );
4320                 }
4321                 if( isset( $params['heights'] ) ) {
4322                         $ig->setHeights( $params['heights'] );
4323                 }
4324
4325                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4326
4327                 $lines = explode( "\n", $text );
4328                 foreach ( $lines as $line ) {
4329                         # match lines like these:
4330                         # Image:someimage.jpg|This is some image
4331                         $matches = array();
4332                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4333                         # Skip empty lines
4334                         if ( count( $matches ) == 0 ) {
4335                                 continue;
4336                         }
4337                         $tp = Title::newFromText( $matches[1] );
4338                         $nt =& $tp;
4339                         if( is_null( $nt ) ) {
4340                                 # Bogus title. Ignore these so we don't bomb out later.
4341                                 continue;
4342                         }
4343                         if ( isset( $matches[3] ) ) {
4344                                 $label = $matches[3];
4345                         } else {
4346                                 $label = '';
4347                         }
4348
4349                         $html = $this->recursiveTagParse( trim( $label ) );
4350
4351                         $ig->add( $nt, $html );
4352
4353                         # Only add real images (bug #5586)
4354                         if ( $nt->getNamespace() == NS_IMAGE ) {
4355                                 $this->mOutput->addImage( $nt->getDBkey() );
4356                         }
4357                 }
4358                 return $ig->toHTML();
4359         }
4360
4361         function getImageParams( $handler ) {
4362                 if ( $handler ) {
4363                         $handlerClass = get_class( $handler );
4364                 } else {
4365                         $handlerClass = '';
4366                 }
4367                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4368                         // Initialise static lists
4369                         static $internalParamNames = array(
4370                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4371                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4372                                         'bottom', 'text-bottom' ),
4373                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4374                                         'upright', 'border' ),
4375                         );
4376                         static $internalParamMap;
4377                         if ( !$internalParamMap ) {
4378                                 $internalParamMap = array();
4379                                 foreach ( $internalParamNames as $type => $names ) {
4380                                         foreach ( $names as $name ) {
4381                                                 $magicName = str_replace( '-', '_', "img_$name" );
4382                                                 $internalParamMap[$magicName] = array( $type, $name );
4383                                         }
4384                                 }
4385                         }
4386
4387                         // Add handler params
4388                         $paramMap = $internalParamMap;
4389                         if ( $handler ) {
4390                                 $handlerParamMap = $handler->getParamMap();
4391                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4392                                         $paramMap[$magic] = array( 'handler', $paramName );
4393                                 }
4394                         }
4395                         $this->mImageParams[$handlerClass] = $paramMap;
4396                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4397                 }
4398                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4399         }
4400
4401         /**
4402          * Parse image options text and use it to make an image
4403          */
4404         function makeImage( $title, $options ) {
4405                 # Check if the options text is of the form "options|alt text"
4406                 # Options are:
4407                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4408                 #  * left               no resizing, just left align. label is used for alt= only
4409                 #  * right              same, but right aligned
4410                 #  * none               same, but not aligned
4411                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4412                 #  * center             center the image
4413                 #  * framed             Keep original image size, no magnify-button.
4414                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4415                 #  * upright            reduce width for upright images, rounded to full __0 px
4416                 #  * border             draw a 1px border around the image
4417                 # vertical-align values (no % or length right now):
4418                 #  * baseline
4419                 #  * sub
4420                 #  * super
4421                 #  * top
4422                 #  * text-top
4423                 #  * middle
4424                 #  * bottom
4425                 #  * text-bottom
4426
4427                 $parts = array_map( 'trim', explode( '|', $options) );
4428                 $sk = $this->mOptions->getSkin();
4429
4430                 # Give extensions a chance to select the file revision for us
4431                 $skip = $time = false;
4432                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4433
4434                 if ( $skip ) {
4435                         return $sk->makeLinkObj( $title );
4436                 }
4437
4438                 # Get parameter map
4439                 $file = wfFindFile( $title, $time );
4440                 $handler = $file ? $file->getHandler() : false;
4441
4442                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4443
4444                 # Process the input parameters
4445                 $caption = '';
4446                 $params = array( 'frame' => array(), 'handler' => array(),
4447                         'horizAlign' => array(), 'vertAlign' => array() );
4448                 foreach( $parts as $part ) {
4449                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4450                         $validated = false;
4451                         if( isset( $paramMap[$magicName] ) ) {
4452                                 list( $type, $paramName ) = $paramMap[$magicName];
4453
4454                                 // Special case; width and height come in one variable together
4455                                 if( $type == 'handler' && $paramName == 'width' ) {
4456                                         $m = array();
4457                                         # (bug 13500) In both cases (width/height and width only),
4458                                         # permit trailing "px" for backward compatibility.
4459                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4460                                                 $width = intval( $m[1] );
4461                                                 $height = intval( $m[2] );
4462                                                 if ( $handler->validateParam( 'width', $width ) ) {
4463                                                         $params[$type]['width'] = $width;
4464                                                         $validated = true;
4465                                                 }
4466                                                 if ( $handler->validateParam( 'height', $height ) ) {
4467                                                         $params[$type]['height'] = $height;
4468                                                         $validated = true;
4469                                                 }
4470                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4471                                                 $width = intval( $value );
4472                                                 if ( $handler->validateParam( 'width', $width ) ) {
4473                                                         $params[$type]['width'] = $width;
4474                                                         $validated = true;
4475                                                 }
4476                                         } // else no validation -- bug 13436
4477                                 } else {
4478                                         if ( $type == 'handler' ) {
4479                                                 # Validate handler parameter
4480                                                 $validated = $handler->validateParam( $paramName, $value );
4481                                         } else {
4482                                                 # Validate internal parameters
4483                                                 switch( $paramName ) {
4484                                                 case "manualthumb":
4485                                                         /// @fixme - possibly check validity here?
4486                                                         /// downstream behavior seems odd with missing manual thumbs.
4487                                                         $validated = true;
4488                                                         break;
4489                                                 default:
4490                                                         // Most other things appear to be empty or numeric...
4491                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4492                                                 }
4493                                         }
4494
4495                                         if ( $validated ) {
4496                                                 $params[$type][$paramName] = $value;
4497                                         }
4498                                 }
4499                         }
4500                         if ( !$validated ) {
4501                                 $caption = $part;
4502                         }
4503                 }
4504
4505                 # Process alignment parameters
4506                 if ( $params['horizAlign'] ) {
4507                         $params['frame']['align'] = key( $params['horizAlign'] );
4508                 }
4509                 if ( $params['vertAlign'] ) {
4510                         $params['frame']['valign'] = key( $params['vertAlign'] );
4511                 }
4512
4513                 # Strip bad stuff out of the alt text
4514                 $alt = $this->replaceLinkHoldersText( $caption );
4515
4516                 # make sure there are no placeholders in thumbnail attributes
4517                 # that are later expanded to html- so expand them now and
4518                 # remove the tags
4519                 $alt = $this->mStripState->unstripBoth( $alt );
4520                 $alt = Sanitizer::stripAllTags( $alt );
4521
4522                 $params['frame']['alt'] = $alt;
4523                 $params['frame']['caption'] = $caption;
4524
4525                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4526
4527                 # Linker does the rest
4528                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time );
4529
4530                 # Give the handler a chance to modify the parser object
4531                 if ( $handler ) {
4532                         $handler->parserTransformHook( $this, $file );
4533                 }
4534
4535                 return $ret;
4536         }
4537
4538         /**
4539          * Set a flag in the output object indicating that the content is dynamic and
4540          * shouldn't be cached.
4541          */
4542         function disableCache() {
4543                 wfDebug( "Parser output marked as uncacheable.\n" );
4544                 $this->mOutput->mCacheTime = -1;
4545         }
4546
4547         /**#@+
4548          * Callback from the Sanitizer for expanding items found in HTML attribute
4549          * values, so they can be safely tested and escaped.
4550          * @param string $text
4551          * @param PPFrame $frame
4552          * @return string
4553          * @private
4554          */
4555         function attributeStripCallback( &$text, $frame = false ) {
4556                 $text = $this->replaceVariables( $text, $frame );
4557                 $text = $this->mStripState->unstripBoth( $text );
4558                 return $text;
4559         }
4560
4561         /**#@-*/
4562
4563         /**#@+
4564          * Accessor/mutator
4565          */
4566         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4567         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4568         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4569         /**#@-*/
4570
4571         /**#@+
4572          * Accessor
4573          */
4574         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4575         /**#@-*/
4576
4577
4578         /**
4579          * Break wikitext input into sections, and either pull or replace
4580          * some particular section's text.
4581          *
4582          * External callers should use the getSection and replaceSection methods.
4583          *
4584          * @param string $text Page wikitext
4585          * @param string $section A section identifier string of the form:
4586          *   <flag1> - <flag2> - ... - <section number>
4587          *
4588          * Currently the only recognised flag is "T", which means the target section number
4589          * was derived during a template inclusion parse, in other words this is a template
4590          * section edit link. If no flags are given, it was an ordinary section edit link.
4591          * This flag is required to avoid a section numbering mismatch when a section is
4592          * enclosed by <includeonly> (bug 6563).
4593          *
4594          * The section number 0 pulls the text before the first heading; other numbers will
4595          * pull the given section along with its lower-level subsections. If the section is
4596          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4597          *
4598          * @param string $mode One of "get" or "replace"
4599          * @param string $newText Replacement text for section data.
4600          * @return string for "get", the extracted section text.
4601          *                for "replace", the whole page with the section replaced.
4602          */
4603         private function extractSections( $text, $section, $mode, $newText='' ) {
4604                 global $wgTitle;
4605                 $this->clearState();
4606                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4607                 $this->mOptions = new ParserOptions;
4608                 $this->setOutputType( self::OT_WIKI );
4609                 $outText = '';
4610                 $frame = $this->getPreprocessor()->newFrame();
4611
4612                 // Process section extraction flags
4613                 $flags = 0;
4614                 $sectionParts = explode( '-', $section );
4615                 $sectionIndex = array_pop( $sectionParts );
4616                 foreach ( $sectionParts as $part ) {
4617                         if ( $part == 'T' ) {
4618                                 $flags |= self::PTD_FOR_INCLUSION;
4619                         }
4620                 }
4621                 // Preprocess the text
4622                 $root = $this->preprocessToDom( $text, $flags );
4623
4624                 // <h> nodes indicate section breaks
4625                 // They can only occur at the top level, so we can find them by iterating the root's children
4626                 $node = $root->getFirstChild();
4627
4628                 // Find the target section
4629                 if ( $sectionIndex == 0 ) {
4630                         // Section zero doesn't nest, level=big
4631                         $targetLevel = 1000;
4632                 } else {
4633             while ( $node ) {
4634                 if ( $node->getName() == 'h' ) {
4635                     $bits = $node->splitHeading();
4636                                         if ( $bits['i'] == $sectionIndex ) {
4637                                         $targetLevel = $bits['level'];
4638                                                 break;
4639                                         }
4640                                 }
4641                                 if ( $mode == 'replace' ) {
4642                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4643                                 }
4644                                 $node = $node->getNextSibling();
4645                         }
4646                 }
4647
4648                 if ( !$node ) {
4649                         // Not found
4650                         if ( $mode == 'get' ) {
4651                                 return $newText;
4652                         } else {
4653                                 return $text;
4654                         }
4655                 }
4656
4657                 // Find the end of the section, including nested sections
4658                 do {
4659                         if ( $node->getName() == 'h' ) {
4660                                 $bits = $node->splitHeading();
4661                                 $curLevel = $bits['level'];
4662                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4663                                         break;
4664                                 }
4665                         }
4666                         if ( $mode == 'get' ) {
4667                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4668                         }
4669                         $node = $node->getNextSibling();
4670                 } while ( $node );
4671
4672                 // Write out the remainder (in replace mode only)
4673                 if ( $mode == 'replace' ) {
4674                         // Output the replacement text
4675                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4676                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4677                         $outText .= $newText . "\n\n";
4678                         while ( $node ) {
4679                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4680                                 $node = $node->getNextSibling();
4681                         }
4682                 }
4683
4684                 if ( is_string( $outText ) ) {
4685                         // Re-insert stripped tags
4686                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4687                 }
4688
4689                 return $outText;
4690         }
4691
4692         /**
4693          * This function returns the text of a section, specified by a number ($section).
4694          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4695          * the first section before any such heading (section 0).
4696          *
4697          * If a section contains subsections, these are also returned.
4698          *
4699          * @param string $text text to look in
4700          * @param string $section section identifier
4701          * @param string $deftext default to return if section is not found
4702          * @return string text of the requested section
4703          */
4704         public function getSection( $text, $section, $deftext='' ) {
4705                 return $this->extractSections( $text, $section, "get", $deftext );
4706         }
4707
4708         public function replaceSection( $oldtext, $section, $text ) {
4709                 return $this->extractSections( $oldtext, $section, "replace", $text );
4710         }
4711
4712         /**
4713          * Get the timestamp associated with the current revision, adjusted for
4714          * the default server-local timestamp
4715          */
4716         function getRevisionTimestamp() {
4717                 if ( is_null( $this->mRevisionTimestamp ) ) {
4718                         wfProfileIn( __METHOD__ );
4719                         global $wgContLang;
4720                         $dbr = wfGetDB( DB_SLAVE );
4721                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4722                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4723
4724                         // Normalize timestamp to internal MW format for timezone processing.
4725                         // This has the added side-effect of replacing a null value with
4726                         // the current time, which gives us more sensible behavior for
4727                         // previews.
4728                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4729
4730                         // The cryptic '' timezone parameter tells to use the site-default
4731                         // timezone offset instead of the user settings.
4732                         //
4733                         // Since this value will be saved into the parser cache, served
4734                         // to other users, and potentially even used inside links and such,
4735                         // it needs to be consistent for all visitors.
4736                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4737
4738                         wfProfileOut( __METHOD__ );
4739                 }
4740                 return $this->mRevisionTimestamp;
4741         }
4742
4743         /**
4744          * Mutator for $mDefaultSort
4745          *
4746          * @param $sort New value
4747          */
4748         public function setDefaultSort( $sort ) {
4749                 $this->mDefaultSort = $sort;
4750         }
4751
4752         /**
4753          * Accessor for $mDefaultSort
4754          * Will use the title/prefixed title if none is set
4755          *
4756          * @return string
4757          */
4758         public function getDefaultSort() {
4759                 if( $this->mDefaultSort !== false ) {
4760                         return $this->mDefaultSort;
4761                 } else {
4762                         return $this->mTitle->getNamespace() == NS_CATEGORY
4763                                         ? $this->mTitle->getText()
4764                                         : $this->mTitle->getPrefixedText();
4765                 }
4766         }
4767
4768         /**
4769          * Try to guess the section anchor name based on a wikitext fragment
4770          * presumably extracted from a heading, for example "Header" from
4771          * "== Header ==".
4772          */
4773         public function guessSectionNameFromWikiText( $text ) {
4774                 # Strip out wikitext links(they break the anchor)
4775                 $text = $this->stripSectionName( $text );
4776                 $headline = Sanitizer::decodeCharReferences( $text );
4777                 # strip out HTML
4778                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4779                 $headline = trim( $headline );
4780                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4781                 $replacearray = array(
4782                         '%3A' => ':',
4783                         '%' => '.'
4784                 );
4785                 return str_replace(
4786                         array_keys( $replacearray ),
4787                         array_values( $replacearray ),
4788                         $sectionanchor );
4789         }
4790
4791         /**
4792          * Strips a text string of wikitext for use in a section anchor
4793          *
4794          * Accepts a text string and then removes all wikitext from the
4795          * string and leaves only the resultant text (i.e. the result of
4796          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4797          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4798          * to create valid section anchors by mimicing the output of the
4799          * parser when headings are parsed.
4800          *
4801          * @param $text string Text string to be stripped of wikitext
4802          * for use in a Section anchor
4803          * @return Filtered text string
4804          */
4805         public function stripSectionName( $text ) {
4806                 # Strip internal link markup
4807                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4808                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4809
4810                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4811                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4812                 # on how many empty links there are on the page - need to figure that out.
4813                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4814
4815                 # Parse wikitext quotes (italics & bold)
4816                 $text = $this->doQuotes($text);
4817
4818                 # Strip HTML tags
4819                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4820                 return $text;
4821         }
4822
4823         function srvus( $text ) {
4824                 return $this->testSrvus( $text, $this->mOutputType );
4825         }
4826
4827         /**
4828          * strip/replaceVariables/unstrip for preprocessor regression testing
4829          */
4830         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4831                 $this->clearState();
4832                 if ( ! ( $title instanceof Title ) ) {
4833                         $title = Title::newFromText( $title );
4834                 }
4835                 $this->mTitle = $title;
4836                 $this->mOptions = $options;
4837                 $this->setOutputType( $outputType );
4838                 $text = $this->replaceVariables( $text );
4839                 $text = $this->mStripState->unstripBoth( $text );
4840                 $text = Sanitizer::removeHTMLtags( $text );
4841                 return $text;
4842         }
4843
4844         function testPst( $text, $title, $options ) {
4845                 global $wgUser;
4846                 if ( ! ( $title instanceof Title ) ) {
4847                         $title = Title::newFromText( $title );
4848                 }
4849                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4850         }
4851
4852         function testPreprocess( $text, $title, $options ) {
4853                 if ( ! ( $title instanceof Title ) ) {
4854                         $title = Title::newFromText( $title );
4855                 }
4856                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4857         }
4858
4859         function markerSkipCallback( $s, $callback ) {
4860                 $i = 0;
4861                 $out = '';
4862                 while ( $i < strlen( $s ) ) {
4863                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4864                         if ( $markerStart === false ) {
4865                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4866                                 break;
4867                         } else {
4868                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4869                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4870                                 if ( $markerEnd === false ) {
4871                                         $out .= substr( $s, $markerStart );
4872                                         break;
4873                                 } else {
4874                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4875                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4876                                         $i = $markerEnd;
4877                                 }
4878                         }
4879                 }
4880                 return $out;
4881         }
4882 }
4883
4884 /**
4885  * @todo document, briefly.
4886  * @addtogroup Parser
4887  */
4888 class StripState {
4889         var $general, $nowiki;
4890
4891         function __construct() {
4892                 $this->general = new ReplacementArray;
4893                 $this->nowiki = new ReplacementArray;
4894         }
4895
4896         function unstripGeneral( $text ) {
4897                 wfProfileIn( __METHOD__ );
4898                 do {
4899                         $oldText = $text;
4900                         $text = $this->general->replace( $text );
4901                 } while ( $text != $oldText );
4902                 wfProfileOut( __METHOD__ );
4903                 return $text;
4904         }
4905
4906         function unstripNoWiki( $text ) {
4907                 wfProfileIn( __METHOD__ );
4908                 do {
4909                         $oldText = $text;
4910                         $text = $this->nowiki->replace( $text );
4911                 } while ( $text != $oldText );
4912                 wfProfileOut( __METHOD__ );
4913                 return $text;
4914         }
4915
4916         function unstripBoth( $text ) {
4917                 wfProfileIn( __METHOD__ );
4918                 do {
4919                         $oldText = $text;
4920                         $text = $this->general->replace( $text );
4921                         $text = $this->nowiki->replace( $text );
4922                 } while ( $text != $oldText );
4923                 wfProfileOut( __METHOD__ );
4924                 return $text;
4925         }
4926 }
4927
4928 /**
4929  * @todo document, briefly.
4930  * @addtogroup Parser
4931  */
4932 class OnlyIncludeReplacer {
4933         var $output = '';
4934
4935         function replace( $matches ) {
4936                 if ( substr( $matches[1], -1 ) == "\n" ) {
4937                         $this->output .= substr( $matches[1], 0, -1 );
4938                 } else {
4939                         $this->output .= $matches[1];
4940                 }
4941         }
4942 }