includes/parser/Parser.php

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