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