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                 $mFunctionTagHooks;
  97
  98
  99         # Cleared with clearState():
 100         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 101         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 102         var $mLinkHolders, $mLinkID;
 103         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 104         var $mTplExpandCache; // empty-frame expansion cache
 105         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 106         var $mExpensiveFunctionCount; // number of expensive parser function calls
 107
 108         # Temporary
 109         # These are variables reset at least once per parse regardless of $clearState
 110         var $mOptions,      // ParserOptions object
 111                 $mTitle,        // Title context, used for self-link rendering and similar things
 112                 $mOutputType,   // Output type, one of the OT_xxx constants
 113                 $ot,            // Shortcut alias, see setOutputType()
 114                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 115                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 116                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 117
 118         /**#@-*/
 119
 120         /**
 121          * Constructor
 122          *
 123          * @public
 124          */
 125         function __construct( $conf = array() ) {
 126                 $this->mConf = $conf;
 127                 $this->mTagHooks = array();
 128                 $this->mTransparentTagHooks = array();
 129                 $this->mFunctionHooks = array();
 130                 $this->mFunctionTagHooks = array();
 131                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 132                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 133                 $this->mUrlProtocols = wfUrlProtocols();
 134                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 135                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 136                 $this->mVarCache = array();
 137                 if ( isset( $conf['preprocessorClass'] ) ) {
 138                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 139                 } elseif ( extension_loaded( 'domxml' ) ) {
 140                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 141                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 142                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 143                 } elseif ( extension_loaded( 'dom' ) ) {
 144                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 145                 } else {
 146                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 147                 }
 148                 $this->mMarkerIndex = 0;
 149                 $this->mFirstCall = true;
 150         }
 151
 152         /**
 153          * Reduce memory usage to reduce the impact of circular references
 154          */
 155         function __destruct() {
 156                 if ( isset( $this->mLinkHolders ) ) {
 157                         $this->mLinkHolders->__destruct();
 158                 }
 159                 foreach ( $this as $name => $value ) {
 160                         unset( $this->$name );
 161                 }
 162         }
 163
 164         /**
 165          * Do various kinds of initialisation on the first call of the parser
 166          */
 167         function firstCallInit() {
 168                 if ( !$this->mFirstCall ) {
 169                         return;
 170                 }
 171                 $this->mFirstCall = false;
 172
 173                 wfProfileIn( __METHOD__ );
 174
 175                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 176                 CoreParserFunctions::register( $this );
 177                 $this->initialiseVariables();
 178
 179                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 180                 wfProfileOut( __METHOD__ );
 181         }
 182
 183         /**
 184          * Clear Parser state
 185          *
 186          * @private
 187          */
 188         function clearState() {
 189                 wfProfileIn( __METHOD__ );
 190                 if ( $this->mFirstCall ) {
 191                         $this->firstCallInit();
 192                 }
 193                 $this->mOutput = new ParserOutput;
 194                 $this->mAutonumber = 0;
 195                 $this->mLastSection = '';
 196                 $this->mDTopen = false;
 197                 $this->mIncludeCount = array();
 198                 $this->mStripState = new StripState;
 199                 $this->mArgStack = false;
 200                 $this->mInPre = false;
 201                 $this->mLinkHolders = new LinkHolderArray( $this );
 202                 $this->mLinkID = 0;
 203                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 204                 $this->mVarCache = array();
 205
 206                 /**
 207                  * Prefix for temporary replacement strings for the multipass parser.
 208                  * \x07 should never appear in input as it's disallowed in XML.
 209                  * Using it at the front also gives us a little extra robustness
 210                  * since it shouldn't match when butted up against identifier-like
 211                  * string constructs.
 212                  *
 213                  * Must not consist of all title characters, or else it will change
 214                  * the behaviour of <nowiki> in a link.
 215                  */
 216                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 217                 # Changed to \x7f to allow XML double-parsing -- TS
 218                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 219
 220
 221                 # Clear these on every parse, bug 4549
 222                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 223
 224                 $this->mShowToc = true;
 225                 $this->mForceTocPosition = false;
 226                 $this->mIncludeSizes = array(
 227                         'post-expand' => 0,
 228                         'arg' => 0,
 229                 );
 230                 $this->mPPNodeCount = 0;
 231                 $this->mDefaultSort = false;
 232                 $this->mHeadings = array();
 233                 $this->mDoubleUnderscores = array();
 234                 $this->mExpensiveFunctionCount = 0;
 235
 236                 # Fix cloning
 237                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 238                         $this->mPreprocessor = null;
 239                 }
 240
 241                 wfRunHooks( 'ParserClearState', array( &$this ) );
 242                 wfProfileOut( __METHOD__ );
 243         }
 244
 245         function setOutputType( $ot ) {
 246                 $this->mOutputType = $ot;
 247                 // Shortcut alias
 248                 $this->ot = array(
 249                         'html' => $ot == self::OT_HTML,
 250                         'wiki' => $ot == self::OT_WIKI,
 251                         'pre' => $ot == self::OT_PREPROCESS,
 252                 );
 253         }
 254
 255         /**
 256          * Set the context title
 257          */
 258         function setTitle( $t ) {
 259                 if ( !$t || $t instanceof FakeTitle ) {
 260                         $t = Title::newFromText( 'NO TITLE' );
 261                 }
 262                 if ( strval( $t->getFragment() ) !== '' ) {
 263                         # Strip the fragment to avoid various odd effects
 264                         $this->mTitle = clone $t;
 265                         $this->mTitle->setFragment( '' );
 266                 } else {
 267                         $this->mTitle = $t;
 268                 }
 269         }
 270
 271         /**
 272          * Accessor for mUniqPrefix.
 273          *
 274          * @public
 275          */
 276         function uniqPrefix() {
 277                 if( !isset( $this->mUniqPrefix ) ) {
 278                         // @fixme this is probably *horribly wrong*
 279                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 280                         // if this is called for a parser cache hit, the parser may not
 281                         // have ever been initialized in the first place.
 282                         // Not really sure what the heck is supposed to be going on here.
 283                         return '';
 284                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 285                 }
 286                 return $this->mUniqPrefix;
 287         }
 288
 289         /**
 290          * Convert wikitext to HTML
 291          * Do not call this function recursively.
 292          *
 293          * @param $text String: text we want to parse
 294          * @param $title A title object
 295          * @param $options ParserOptions
 296          * @param $linestart boolean
 297          * @param $clearState boolean
 298          * @param $revid Int: number to pass in {{REVISIONID}}
 299          * @return ParserOutput a ParserOutput
 300          */
 301         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 302                 /**
 303                  * First pass--just handle <nowiki> sections, pass the rest off
 304                  * to internalParse() which does all the real work.
 305                  */
 306
 307                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 308                 $fname = __METHOD__.'-' . wfGetCaller();
 309                 wfProfileIn( __METHOD__ );
 310                 wfProfileIn( $fname );
 311
 312                 if ( $clearState ) {
 313                         $this->clearState();
 314                 }
 315
 316                 $this->mOptions = $options;
 317                 $this->setTitle( $title );
 318                 $oldRevisionId = $this->mRevisionId;
 319                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 320                 if( $revid !== null ) {
 321                         $this->mRevisionId = $revid;
 322                         $this->mRevisionTimestamp = null;
 323                 }
 324                 $this->setOutputType( self::OT_HTML );
 325                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 326                 # No more strip!
 327                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 328                 $text = $this->internalParse( $text );
 329                 $text = $this->mStripState->unstripGeneral( $text );
 330
 331                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 332                 $fixtags = array(
 333                         # french spaces, last one Guillemet-left
 334                         # only if there is something before the space
 335                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 336                         # french spaces, Guillemet-right
 337                         '/(\\302\\253) /' => '\\1&nbsp;',
 338                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 339                 );
 340                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 341
 342                 $text = $this->doBlockLevels( $text, $linestart );
 343
 344                 $this->replaceLinkHolders( $text );
 345
 346                 # the position of the parserConvert() call should not be changed. it
 347                 # assumes that the links are all replaced and the only thing left
 348                 # is the <nowiki> mark.
 349                 # Side-effects: this calls $this->mOutput->setTitleText()
 350                 $text = $wgContLang->parserConvert( $text, $this );
 351
 352                 $text = $this->mStripState->unstripNoWiki( $text );
 353
 354                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 355
 356 //!JF Move to its own function
 357
 358                 $uniq_prefix = $this->mUniqPrefix;
 359                 $matches = array();
 360                 $elements = array_keys( $this->mTransparentTagHooks );
 361                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 362
 363                 foreach( $matches as $marker => $data ) {
 364                         list( $element, $content, $params, $tag ) = $data;
 365                         $tagName = strtolower( $element );
 366                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 367                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 368                                         array( $content, $params, $this ) );
 369                         } else {
 370                                 $output = $tag;
 371                         }
 372                         $this->mStripState->general->setPair( $marker, $output );
 373                 }
 374                 $text = $this->mStripState->unstripGeneral( $text );
 375
 376                 $text = Sanitizer::normalizeCharReferences( $text );
 377
 378                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 379                         $text = MWTidy::tidy( $text );
 380                 } else {
 381                         # attempt to sanitize at least some nesting problems
 382                         # (bug #2702 and quite a few others)
 383                         $tidyregs = array(
 384                                 # ''Something [http://www.cool.com cool''] -->
 385                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 386                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 387                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 388                                 # fix up an anchor inside another anchor, only
 389                                 # at least for a single single nested link (bug 3695)
 390                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 391                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 392                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 393                                 # contains a div, so fix it up here; replace
 394                                 # div with escaped text
 395                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 396                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 397                                 # remove empty italic or bold tag pairs, some
 398                                 # introduced by rules above
 399                                 '/<([bi])><\/\\1>/' => '',
 400                         );
 401
 402                         $text = preg_replace(
 403                                 array_keys( $tidyregs ),
 404                                 array_values( $tidyregs ),
 405                                 $text );
 406                 }
 407                 global $wgExpensiveParserFunctionLimit;
 408                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 409                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 410                 }
 411
 412                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 413
 414                 # Information on include size limits, for the benefit of users who try to skirt them
 415                 if ( $this->mOptions->getEnableLimitReport() ) {
 416                         global $wgExpensiveParserFunctionLimit;
 417                         $max = $this->mOptions->getMaxIncludeSize();
 418                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 419                         $limitReport =
 420                                 "NewPP limit report\n" .
 421                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 422                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 423                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 424                                 $PFreport;
 425                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 426                         $text .= "\n<!-- \n$limitReport-->\n";
 427                 }
 428                 $this->mOutput->setText( $text );
 429                 $this->mRevisionId = $oldRevisionId;
 430                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 431                 wfProfileOut( $fname );
 432                 wfProfileOut( __METHOD__ );
 433
 434                 return $this->mOutput;
 435         }
 436
 437         /**
 438          * Recursive parser entry point that can be called from an extension tag
 439          * hook.
 440          *
 441          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 442          *
 443          * @param $text String: text extension wants to have parsed
 444          * @param PPFrame $frame: The frame to use for expanding any template variables
 445          */
 446         function recursiveTagParse( $text, $frame=false ) {
 447                 wfProfileIn( __METHOD__ );
 448                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 449                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 450                 $text = $this->internalParse( $text, false, $frame );
 451                 wfProfileOut( __METHOD__ );
 452                 return $text;
 453         }
 454
 455         /**
 456          * Expand templates and variables in the text, producing valid, static wikitext.
 457          * Also removes comments.
 458          */
 459         function preprocess( $text, $title, $options, $revid = null ) {
 460                 wfProfileIn( __METHOD__ );
 461                 $this->clearState();
 462                 $this->setOutputType( self::OT_PREPROCESS );
 463                 $this->mOptions = $options;
 464                 $this->setTitle( $title );
 465                 if( $revid !== null ) {
 466                         $this->mRevisionId = $revid;
 467                 }
 468                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 469                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 470                 $text = $this->replaceVariables( $text );
 471                 $text = $this->mStripState->unstripBoth( $text );
 472                 wfProfileOut( __METHOD__ );
 473                 return $text;
 474         }
 475
 476         /**
 477          * Get a random string
 478          *
 479          * @private
 480          * @static
 481          */
 482         function getRandomString() {
 483                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 484         }
 485
 486         function &getTitle() { return $this->mTitle; }
 487         function getOptions() { return $this->mOptions; }
 488         function getRevisionId() { return $this->mRevisionId; }
 489         function getOutput() { return $this->mOutput; }
 490         function nextLinkID() { return $this->mLinkID++; }
 491
 492         function getFunctionLang() {
 493                 global $wgLang, $wgContLang;
 494
 495                 $target = $this->mOptions->getTargetLanguage();
 496                 if ( $target !== null ) {
 497                         return $target;
 498                 } else {
 499                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 500                 }
 501         }
 502
 503         /**
 504          * Get a preprocessor object
 505          */
 506         function getPreprocessor() {
 507                 if ( !isset( $this->mPreprocessor ) ) {
 508                         $class = $this->mPreprocessorClass;
 509                         $this->mPreprocessor = new $class( $this );
 510                 }
 511                 return $this->mPreprocessor;
 512         }
 513
 514         /**
 515          * Replaces all occurrences of HTML-style comments and the given tags
 516          * in the text with a random marker and returns the next text. The output
 517          * parameter $matches will be an associative array filled with data in
 518          * the form:
 519          *   'UNIQ-xxxxx' => array(
 520          *     'element',
 521          *     'tag content',
 522          *     array( 'param' => 'x' ),
 523          *     '<element param="x">tag content</element>' ) )
 524          *
 525          * @param $elements list of element names. Comments are always extracted.
 526          * @param $text Source text string.
 527          * @param $uniq_prefix
 528          *
 529          * @public
 530          * @static
 531          */
 532         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 533                 static $n = 1;
 534                 $stripped = '';
 535                 $matches = array();
 536
 537                 $taglist = implode( '|', $elements );
 538                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 539
 540                 while ( '' != $text ) {
 541                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 542                         $stripped .= $p[0];
 543                         if( count( $p ) < 5 ) {
 544                                 break;
 545                         }
 546                         if( count( $p ) > 5 ) {
 547                                 // comment
 548                                 $element    = $p[4];
 549                                 $attributes = '';
 550                                 $close      = '';
 551                                 $inside     = $p[5];
 552                         } else {
 553                                 // tag
 554                                 $element    = $p[1];
 555                                 $attributes = $p[2];
 556                                 $close      = $p[3];
 557                                 $inside     = $p[4];
 558                         }
 559
 560                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 561                         $stripped .= $marker;
 562
 563                         if ( $close === '/>' ) {
 564                                 // Empty element tag, <tag />
 565                                 $content = null;
 566                                 $text = $inside;
 567                                 $tail = null;
 568                         } else {
 569                                 if( $element === '!--' ) {
 570                                         $end = '/(-->)/';
 571                                 } else {
 572                                         $end = "/(<\\/$element\\s*>)/i";
 573                                 }
 574                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 575                                 $content = $q[0];
 576                                 if( count( $q ) < 3 ) {
 577                                         # No end tag -- let it run out to the end of the text.
 578                                         $tail = '';
 579                                         $text = '';
 580                                 } else {
 581                                         $tail = $q[1];
 582                                         $text = $q[2];
 583                                 }
 584                         }
 585
 586                         $matches[$marker] = array( $element,
 587                                 $content,
 588                                 Sanitizer::decodeTagAttributes( $attributes ),
 589                                 "<$element$attributes$close$content$tail" );
 590                 }
 591                 return $stripped;
 592         }
 593
 594         /**
 595          * Get a list of strippable XML-like elements
 596          */
 597         function getStripList() {
 598                 global $wgRawHtml;
 599                 $elements = $this->mStripList;
 600                 if( $wgRawHtml ) {
 601                         $elements[] = 'html';
 602                 }
 603                 if( $this->mOptions->getUseTeX() ) {
 604                         $elements[] = 'math';
 605                 }
 606                 return $elements;
 607         }
 608
 609         /**
 610          * @deprecated use replaceVariables
 611          */
 612         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 613                 return $text;
 614         }
 615
 616         /**
 617          * Restores pre, math, and other extensions removed by strip()
 618          *
 619          * always call unstripNoWiki() after this one
 620          * @private
 621          * @deprecated use $this->mStripState->unstrip()
 622          */
 623         function unstrip( $text, $state ) {
 624                 return $state->unstripGeneral( $text );
 625         }
 626
 627         /**
 628          * Always call this after unstrip() to preserve the order
 629          *
 630          * @private
 631          * @deprecated use $this->mStripState->unstrip()
 632          */
 633         function unstripNoWiki( $text, $state ) {
 634                 return $state->unstripNoWiki( $text );
 635         }
 636
 637         /**
 638          * @deprecated use $this->mStripState->unstripBoth()
 639          */
 640         function unstripForHTML( $text ) {
 641                 return $this->mStripState->unstripBoth( $text );
 642         }
 643
 644         /**
 645          * Add an item to the strip state
 646          * Returns the unique tag which must be inserted into the stripped text
 647          * The tag will be replaced with the original text in unstrip()
 648          *
 649          * @private
 650          */
 651         function insertStripItem( $text ) {
 652                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 653                 $this->mMarkerIndex++;
 654                 $this->mStripState->general->setPair( $rnd, $text );
 655                 return $rnd;
 656         }
 657
 658         /**
 659          * Interface with html tidy
 660          * @deprecated Use MWTidy::tidy()
 661          */
 662         public static function tidy( $text ) {
 663                 wfDeprecated( __METHOD__ );
 664                 return MWTidy::tidy( $text );
 665         }
 666
 667         /**
 668          * parse the wiki syntax used to render tables
 669          *
 670          * @private
 671          */
 672         function doTableStuff ( $text ) {
 673                 wfProfileIn( __METHOD__ );
 674
 675                 $lines = StringUtils::explode( "\n", $text );
 676                 $out = '';
 677                 $td_history = array (); // Is currently a td tag open?
 678                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 679                 $tr_history = array (); // Is currently a tr tag open?
 680                 $tr_attributes = array (); // history of tr attributes
 681                 $has_opened_tr = array(); // Did this table open a <tr> element?
 682                 $indent_level = 0; // indent level of the table
 683
 684                 foreach ( $lines as $outLine ) {
 685                         $line = trim( $outLine );
 686
 687                         if( $line == '' ) { // empty line, go to next line
 688                                 $out .= $outLine."\n";
 689                                 continue;
 690                         }
 691                         $first_character = $line[0];
 692                         $matches = array();
 693
 694                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 695                                 // First check if we are starting a new table
 696                                 $indent_level = strlen( $matches[1] );
 697
 698                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 699                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 700
 701                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 702                                 array_push ( $td_history , false );
 703                                 array_push ( $last_tag_history , '' );
 704                                 array_push ( $tr_history , false );
 705                                 array_push ( $tr_attributes , '' );
 706                                 array_push ( $has_opened_tr , false );
 707                         } else if ( count ( $td_history ) == 0 ) {
 708                                 // Don't do any of the following
 709                                 $out .= $outLine."\n";
 710                                 continue;
 711                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 712                                 // We are ending a table
 713                                 $line = '</table>' . substr ( $line , 2 );
 714                                 $last_tag = array_pop ( $last_tag_history );
 715
 716                                 if ( !array_pop ( $has_opened_tr ) ) {
 717                                         $line = "<tr><td></td></tr>{$line}";
 718                                 }
 719
 720                                 if ( array_pop ( $tr_history ) ) {
 721                                         $line = "</tr>{$line}";
 722                                 }
 723
 724                                 if ( array_pop ( $td_history ) ) {
 725                                         $line = "</{$last_tag}>{$line}";
 726                                 }
 727                                 array_pop ( $tr_attributes );
 728                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 729                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 730                                 // Now we have a table row
 731                                 $line = preg_replace( '#^\|-+#', '', $line );
 732
 733                                 // Whats after the tag is now only attributes
 734                                 $attributes = $this->mStripState->unstripBoth( $line );
 735                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 736                                 array_pop ( $tr_attributes );
 737                                 array_push ( $tr_attributes , $attributes );
 738
 739                                 $line = '';
 740                                 $last_tag = array_pop ( $last_tag_history );
 741                                 array_pop ( $has_opened_tr );
 742                                 array_push ( $has_opened_tr , true );
 743
 744                                 if ( array_pop ( $tr_history ) ) {
 745                                         $line = '</tr>';
 746                                 }
 747
 748                                 if ( array_pop ( $td_history ) ) {
 749                                         $line = "</{$last_tag}>{$line}";
 750                                 }
 751
 752                                 $outLine = $line;
 753                                 array_push ( $tr_history , false );
 754                                 array_push ( $td_history , false );
 755                                 array_push ( $last_tag_history , '' );
 756                         }
 757                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 758                                 // This might be cell elements, td, th or captions
 759                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 760                                         $first_character = '+';
 761                                         $line = substr ( $line , 1 );
 762                                 }
 763
 764                                 $line = substr ( $line , 1 );
 765
 766                                 if ( $first_character === '!' ) {
 767                                         $line = str_replace ( '!!' , '||' , $line );
 768                                 }
 769
 770                                 // Split up multiple cells on the same line.
 771                                 // FIXME : This can result in improper nesting of tags processed
 772                                 // by earlier parser steps, but should avoid splitting up eg
 773                                 // attribute values containing literal "||".
 774                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 775
 776                                 $outLine = '';
 777
 778                                 // Loop through each table cell
 779                                 foreach ( $cells as $cell )
 780                                 {
 781                                         $previous = '';
 782                                         if ( $first_character !== '+' )
 783                                         {
 784                                                 $tr_after = array_pop ( $tr_attributes );
 785                                                 if ( !array_pop ( $tr_history ) ) {
 786                                                         $previous = "<tr{$tr_after}>\n";
 787                                                 }
 788                                                 array_push ( $tr_history , true );
 789                                                 array_push ( $tr_attributes , '' );
 790                                                 array_pop ( $has_opened_tr );
 791                                                 array_push ( $has_opened_tr , true );
 792                                         }
 793
 794                                         $last_tag = array_pop ( $last_tag_history );
 795
 796                                         if ( array_pop ( $td_history ) ) {
 797                                                 $previous = "</{$last_tag}>{$previous}";
 798                                         }
 799
 800                                         if ( $first_character === '|' ) {
 801                                                 $last_tag = 'td';
 802                                         } else if ( $first_character === '!' ) {
 803                                                 $last_tag = 'th';
 804                                         } else if ( $first_character === '+' ) {
 805                                                 $last_tag = 'caption';
 806                                         } else {
 807                                                 $last_tag = '';
 808                                         }
 809
 810                                         array_push ( $last_tag_history , $last_tag );
 811
 812                                         // A cell could contain both parameters and data
 813                                         $cell_data = explode ( '|' , $cell , 2 );
 814
 815                                         // Bug 553: Note that a '|' inside an invalid link should not
 816                                         // be mistaken as delimiting cell parameters
 817                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 818                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 819                                         } else if ( count ( $cell_data ) == 1 )
 820                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 821                                         else {
 822                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 823                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 824                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 825                                         }
 826
 827                                         $outLine .= $cell;
 828                                         array_push ( $td_history , true );
 829                                 }
 830                         }
 831                         $out .= $outLine . "\n";
 832                 }
 833
 834                 // Closing open td, tr && table
 835                 while ( count ( $td_history ) > 0 )
 836                 {
 837                         if ( array_pop ( $td_history ) ) {
 838                                 $out .= "</td>\n";
 839                         }
 840                         if ( array_pop ( $tr_history ) ) {
 841                                 $out .= "</tr>\n";
 842                         }
 843                         if ( !array_pop ( $has_opened_tr ) ) {
 844                                 $out .= "<tr><td></td></tr>\n" ;
 845                         }
 846
 847                         $out .= "</table>\n";
 848                 }
 849
 850                 // Remove trailing line-ending (b/c)
 851                 if ( substr( $out, -1 ) === "\n" ) {
 852                         $out = substr( $out, 0, -1 );
 853                 }
 854
 855                 // special case: don't return empty table
 856                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 857                         $out = '';
 858                 }
 859
 860                 wfProfileOut( __METHOD__ );
 861
 862                 return $out;
 863         }
 864
 865         /**
 866          * Helper function for parse() that transforms wiki markup into
 867          * HTML. Only called for $mOutputType == self::OT_HTML.
 868          *
 869          * @private
 870          */
 871         function internalParse( $text, $isMain = true, $frame=false ) {
 872                 wfProfileIn( __METHOD__ );
 873
 874                 $origText = $text;
 875
 876                 # Hook to suspend the parser in this state
 877                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 878                         wfProfileOut( __METHOD__ );
 879                         return $text ;
 880                 }
 881
 882                 // if $frame is provided, then use $frame for replacing any variables
 883                 if ($frame) {
 884                         // use frame depth to infer how include/noinclude tags should be handled
 885                         // depth=0 means this is the top-level document; otherwise it's an included document
 886                         if( !$frame->depth )
 887                                 $flag = 0;
 888                         else
 889                                 $flag = Parser::PTD_FOR_INCLUSION;
 890                         $dom = $this->preprocessToDom( $text, $flag );
 891                         $text = $frame->expand( $dom );
 892                 }
 893                 // if $frame is not provided, then use old-style replaceVariables
 894                 else {
 895                         $text = $this->replaceVariables( $text );
 896                 }
 897
 898                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 899                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 900
 901                 // Tables need to come after variable replacement for things to work
 902                 // properly; putting them before other transformations should keep
 903                 // exciting things like link expansions from showing up in surprising
 904                 // places.
 905                 $text = $this->doTableStuff( $text );
 906
 907                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 908
 909                 $text = $this->doDoubleUnderscore( $text );
 910                 $text = $this->doHeadings( $text );
 911                 if( $this->mOptions->getUseDynamicDates() ) {
 912                         $df = DateFormatter::getInstance();
 913                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 914                 }
 915                 $text = $this->doAllQuotes( $text );
 916                 $text = $this->replaceInternalLinks( $text );
 917                 $text = $this->replaceExternalLinks( $text );
 918
 919                 # replaceInternalLinks may sometimes leave behind
 920                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 921                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 922
 923                 $text = $this->doMagicLinks( $text );
 924                 $text = $this->formatHeadings( $text, $origText, $isMain );
 925
 926                 wfProfileOut( __METHOD__ );
 927                 return $text;
 928         }
 929
 930         /**
 931          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 932          * magic external links.
 933          *
 934          * DML
 935          * @private
 936          */
 937         function doMagicLinks( $text ) {
 938                 wfProfileIn( __METHOD__ );
 939                 $prots = $this->mUrlProtocols;
 940                 $urlChar = self::EXT_LINK_URL_CLASS;
 941                 $text = preg_replace_callback(
 942                         '!(?:                           # Start cases
 943                                 (<a.*?</a>) |               # m[1]: Skip link text
 944                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 945                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 946                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 947                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 948                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 949                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 950                                     [0-9Xx]                 # check digit
 951                                     \b)
 952                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 953                 wfProfileOut( __METHOD__ );
 954                 return $text;
 955         }
 956
 957         function magicLinkCallback( $m ) {
 958                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 959                         # Skip anchor
 960                         return $m[0];
 961                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 962                         # Skip HTML element
 963                         return $m[0];
 964                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 965                         # Free external link
 966                         return $this->makeFreeExternalLink( $m[0] );
 967                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 968                         # RFC or PMID
 969                         $CssClass = '';
 970                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 971                                 $keyword = 'RFC';
 972                                 $urlmsg = 'rfcurl';
 973                                 $CssClass = 'mw-magiclink-rfc';
 974                                 $id = $m[4];
 975                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
 976                                 $keyword = 'PMID';
 977                                 $urlmsg = 'pubmedurl';
 978                                 $CssClass = 'mw-magiclink-pmid';
 979                                 $id = $m[4];
 980                         } else {
 981                                 throw new MWException( __METHOD__.': unrecognised match type "' .
 982                                         substr($m[0], 0, 20 ) . '"' );
 983                         }
 984                         $url = wfMsg( $urlmsg, $id);
 985                         $sk = $this->mOptions->getSkin();
 986                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
 987                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
 988                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
 989                         # ISBN
 990                         $isbn = $m[5];
 991                         $num = strtr( $isbn, array(
 992                                 '-' => '',
 993                                 ' ' => '',
 994                                 'x' => 'X',
 995                         ));
 996                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
 997                         return'<a href="' .
 998                                 $titleObj->escapeLocalUrl() .
 999                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1000                 } else {
1001                         return $m[0];
1002                 }
1003         }
1004
1005         /**
1006          * Make a free external link, given a user-supplied URL
1007          * @return HTML
1008          * @private
1009          */
1010         function makeFreeExternalLink( $url ) {
1011                 global $wgContLang;
1012                 wfProfileIn( __METHOD__ );
1013
1014                 $sk = $this->mOptions->getSkin();
1015                 $trail = '';
1016
1017                 # The characters '<' and '>' (which were escaped by
1018                 # removeHTMLtags()) should not be included in
1019                 # URLs, per RFC 2396.
1020                 $m2 = array();
1021                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1022                         $trail = substr($url, $m2[0][1]) . $trail;
1023                         $url = substr($url, 0, $m2[0][1]);
1024                 }
1025
1026                 # Move trailing punctuation to $trail
1027                 $sep = ',;\.:!?';
1028                 # If there is no left bracket, then consider right brackets fair game too
1029                 if ( strpos( $url, '(' ) === false ) {
1030                         $sep .= ')';
1031                 }
1032
1033                 $numSepChars = strspn( strrev( $url ), $sep );
1034                 if ( $numSepChars ) {
1035                         $trail = substr( $url, -$numSepChars ) . $trail;
1036                         $url = substr( $url, 0, -$numSepChars );
1037                 }
1038
1039                 $url = Sanitizer::cleanUrl( $url );
1040
1041                 # Is this an external image?
1042                 $text = $this->maybeMakeExternalImage( $url );
1043                 if ( $text === false ) {
1044                         # Not an image, make a link
1045                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1046                                 $this->getExternalLinkAttribs( $url ) );
1047                         # Register it in the output object...
1048                         # Replace unnecessary URL escape codes with their equivalent characters
1049                         $pasteurized = self::replaceUnusualEscapes( $url );
1050                         $this->mOutput->addExternalLink( $pasteurized );
1051                 }
1052                 wfProfileOut( __METHOD__ );
1053                 return $text . $trail;
1054         }
1055
1056
1057         /**
1058          * Parse headers and return html
1059          *
1060          * @private
1061          */
1062         function doHeadings( $text ) {
1063                 wfProfileIn( __METHOD__ );
1064                 for ( $i = 6; $i >= 1; --$i ) {
1065                         $h = str_repeat( '=', $i );
1066                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1067                           "<h$i>\\1</h$i>", $text );
1068                 }
1069                 wfProfileOut( __METHOD__ );
1070                 return $text;
1071         }
1072
1073         /**
1074          * Replace single quotes with HTML markup
1075          * @private
1076          * @return string the altered text
1077          */
1078         function doAllQuotes( $text ) {
1079                 wfProfileIn( __METHOD__ );
1080                 $outtext = '';
1081                 $lines = StringUtils::explode( "\n", $text );
1082                 foreach ( $lines as $line ) {
1083                         $outtext .= $this->doQuotes( $line ) . "\n";
1084                 }
1085                 $outtext = substr($outtext, 0,-1);
1086                 wfProfileOut( __METHOD__ );
1087                 return $outtext;
1088         }
1089
1090         /**
1091          * Helper function for doAllQuotes()
1092          */
1093         public function doQuotes( $text ) {
1094                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1095                 if ( count( $arr ) == 1 )
1096                         return $text;
1097                 else
1098                 {
1099                         # First, do some preliminary work. This may shift some apostrophes from
1100                         # being mark-up to being text. It also counts the number of occurrences
1101                         # of bold and italics mark-ups.
1102                         $i = 0;
1103                         $numbold = 0;
1104                         $numitalics = 0;
1105                         foreach ( $arr as $r )
1106                         {
1107                                 if ( ( $i % 2 ) == 1 )
1108                                 {
1109                                         # If there are ever four apostrophes, assume the first is supposed to
1110                                         # be text, and the remaining three constitute mark-up for bold text.
1111                                         if ( strlen( $arr[$i] ) == 4 )
1112                                         {
1113                                                 $arr[$i-1] .= "'";
1114                                                 $arr[$i] = "'''";
1115                                         }
1116                                         # If there are more than 5 apostrophes in a row, assume they're all
1117                                         # text except for the last 5.
1118                                         else if ( strlen( $arr[$i] ) > 5 )
1119                                         {
1120                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1121                                                 $arr[$i] = "'''''";
1122                                         }
1123                                         # Count the number of occurrences of bold and italics mark-ups.
1124                                         # We are not counting sequences of five apostrophes.
1125                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1126                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1127                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1128                                 }
1129                                 $i++;
1130                         }
1131
1132                         # If there is an odd number of both bold and italics, it is likely
1133                         # that one of the bold ones was meant to be an apostrophe followed
1134                         # by italics. Which one we cannot know for certain, but it is more
1135                         # likely to be one that has a single-letter word before it.
1136                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1137                         {
1138                                 $i = 0;
1139                                 $firstsingleletterword = -1;
1140                                 $firstmultiletterword = -1;
1141                                 $firstspace = -1;
1142                                 foreach ( $arr as $r )
1143                                 {
1144                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1145                                         {
1146                                                 $x1 = substr ($arr[$i-1], -1);
1147                                                 $x2 = substr ($arr[$i-1], -2, 1);
1148                                                 if ($x1 === ' ') {
1149                                                         if ($firstspace == -1) $firstspace = $i;
1150                                                 } else if ($x2 === ' ') {
1151                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1152                                                 } else {
1153                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1154                                                 }
1155                                         }
1156                                         $i++;
1157                                 }
1158
1159                                 # If there is a single-letter word, use it!
1160                                 if ($firstsingleletterword > -1)
1161                                 {
1162                                         $arr [ $firstsingleletterword ] = "''";
1163                                         $arr [ $firstsingleletterword-1 ] .= "'";
1164                                 }
1165                                 # If not, but there's a multi-letter word, use that one.
1166                                 else if ($firstmultiletterword > -1)
1167                                 {
1168                                         $arr [ $firstmultiletterword ] = "''";
1169                                         $arr [ $firstmultiletterword-1 ] .= "'";
1170                                 }
1171                                 # ... otherwise use the first one that has neither.
1172                                 # (notice that it is possible for all three to be -1 if, for example,
1173                                 # there is only one pentuple-apostrophe in the line)
1174                                 else if ($firstspace > -1)
1175                                 {
1176                                         $arr [ $firstspace ] = "''";
1177                                         $arr [ $firstspace-1 ] .= "'";
1178                                 }
1179                         }
1180
1181                         # Now let's actually convert our apostrophic mush to HTML!
1182                         $output = '';
1183                         $buffer = '';
1184                         $state = '';
1185                         $i = 0;
1186                         foreach ($arr as $r)
1187                         {
1188                                 if (($i % 2) == 0)
1189                                 {
1190                                         if ($state === 'both')
1191                                                 $buffer .= $r;
1192                                         else
1193                                                 $output .= $r;
1194                                 }
1195                                 else
1196                                 {
1197                                         if (strlen ($r) == 2)
1198                                         {
1199                                                 if ($state === 'i')
1200                                                 { $output .= '</i>'; $state = ''; }
1201                                                 else if ($state === 'bi')
1202                                                 { $output .= '</i>'; $state = 'b'; }
1203                                                 else if ($state === 'ib')
1204                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1205                                                 else if ($state === 'both')
1206                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1207                                                 else # $state can be 'b' or ''
1208                                                 { $output .= '<i>'; $state .= 'i'; }
1209                                         }
1210                                         else if (strlen ($r) == 3)
1211                                         {
1212                                                 if ($state === 'b')
1213                                                 { $output .= '</b>'; $state = ''; }
1214                                                 else if ($state === 'bi')
1215                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1216                                                 else if ($state === 'ib')
1217                                                 { $output .= '</b>'; $state = 'i'; }
1218                                                 else if ($state === 'both')
1219                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1220                                                 else # $state can be 'i' or ''
1221                                                 { $output .= '<b>'; $state .= 'b'; }
1222                                         }
1223                                         else if (strlen ($r) == 5)
1224                                         {
1225                                                 if ($state === 'b')
1226                                                 { $output .= '</b><i>'; $state = 'i'; }
1227                                                 else if ($state === 'i')
1228                                                 { $output .= '</i><b>'; $state = 'b'; }
1229                                                 else if ($state === 'bi')
1230                                                 { $output .= '</i></b>'; $state = ''; }
1231                                                 else if ($state === 'ib')
1232                                                 { $output .= '</b></i>'; $state = ''; }
1233                                                 else if ($state === 'both')
1234                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1235                                                 else # ($state == '')
1236                                                 { $buffer = ''; $state = 'both'; }
1237                                         }
1238                                 }
1239                                 $i++;
1240                         }
1241                         # Now close all remaining tags.  Notice that the order is important.
1242                         if ($state === 'b' || $state === 'ib')
1243                                 $output .= '</b>';
1244                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1245                                 $output .= '</i>';
1246                         if ($state === 'bi')
1247                                 $output .= '</b>';
1248                         # There might be lonely ''''', so make sure we have a buffer
1249                         if ($state === 'both' && $buffer)
1250                                 $output .= '<b><i>'.$buffer.'</i></b>';
1251                         return $output;
1252                 }
1253         }
1254
1255         /**
1256          * Replace external links (REL)
1257          *
1258          * Note: this is all very hackish and the order of execution matters a lot.
1259          * Make sure to run maintenance/parserTests.php if you change this code.
1260          *
1261          * @private
1262          */
1263         function replaceExternalLinks( $text ) {
1264                 global $wgContLang;
1265                 wfProfileIn( __METHOD__ );
1266
1267                 $sk = $this->mOptions->getSkin();
1268
1269                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1270                 $s = array_shift( $bits );
1271
1272                 $i = 0;
1273                 while ( $i<count( $bits ) ) {
1274                         $url = $bits[$i++];
1275                         $protocol = $bits[$i++];
1276                         $text = $bits[$i++];
1277                         $trail = $bits[$i++];
1278
1279                         # The characters '<' and '>' (which were escaped by
1280                         # removeHTMLtags()) should not be included in
1281                         # URLs, per RFC 2396.
1282                         $m2 = array();
1283                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1284                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1285                                 $url = substr($url, 0, $m2[0][1]);
1286                         }
1287
1288                         # If the link text is an image URL, replace it with an <img> tag
1289                         # This happened by accident in the original parser, but some people used it extensively
1290                         $img = $this->maybeMakeExternalImage( $text );
1291                         if ( $img !== false ) {
1292                                 $text = $img;
1293                         }
1294
1295                         $dtrail = '';
1296
1297                         # Set linktype for CSS - if URL==text, link is essentially free
1298                         $linktype = ($text === $url) ? 'free' : 'text';
1299
1300                         # No link text, e.g. [http://domain.tld/some.link]
1301                         if ( $text == '' ) {
1302                                 # Autonumber if allowed. See bug #5918
1303                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1304                                         $langObj = $this->getFunctionLang();
1305                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1306                                         $linktype = 'autonumber';
1307                                 } else {
1308                                         # Otherwise just use the URL
1309                                         $text = htmlspecialchars( $url );
1310                                         $linktype = 'free';
1311                                 }
1312                         } else {
1313                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1314                                 # Check for trail
1315                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1316                         }
1317
1318                         $text = $wgContLang->markNoConversion($text);
1319
1320                         $url = Sanitizer::cleanUrl( $url );
1321
1322                         # Use the encoded URL
1323                         # This means that users can paste URLs directly into the text
1324                         # Funny characters like &ouml; aren't valid in URLs anyway
1325                         # This was changed in August 2004
1326                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1327                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1328
1329                         # Register link in the output object.
1330                         # Replace unnecessary URL escape codes with the referenced character
1331                         # This prevents spammers from hiding links from the filters
1332                         $pasteurized = self::replaceUnusualEscapes( $url );
1333                         $this->mOutput->addExternalLink( $pasteurized );
1334                 }
1335
1336                 wfProfileOut( __METHOD__ );
1337                 return $s;
1338         }
1339
1340         /**
1341          * Get an associative array of additional HTML attributes appropriate for a
1342          * particular external link.  This currently may include rel => nofollow
1343          * (depending on configuration, namespace, and the URL's domain) and/or a
1344          * target attribute (depending on configuration).
1345          *
1346          * @param string $url Optional URL, to extract the domain from for rel =>
1347          *   nofollow if appropriate
1348          * @return array Associative array of HTML attributes
1349          */
1350         function getExternalLinkAttribs( $url = false ) {
1351                 $attribs = array();
1352                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1353                 $ns = $this->mTitle->getNamespace();
1354                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1355                         $attribs['rel'] = 'nofollow';
1356
1357                         global $wgNoFollowDomainExceptions;
1358                         if ( $wgNoFollowDomainExceptions ) {
1359                                 $bits = wfParseUrl( $url );
1360                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1361                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1362                                                 if( substr( $bits['host'], -strlen( $domain ) )
1363                                                 == $domain ) {
1364                                                         unset( $attribs['rel'] );
1365                                                         break;
1366                                                 }
1367                                         }
1368                                 }
1369                         }
1370                 }
1371                 if ( $this->mOptions->getExternalLinkTarget() ) {
1372                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1373                 }
1374                 return $attribs;
1375         }
1376
1377
1378         /**
1379          * Replace unusual URL escape codes with their equivalent characters
1380          * @param string
1381          * @return string
1382          * @static
1383          * @todo  This can merge genuinely required bits in the path or query string,
1384          *        breaking legit URLs. A proper fix would treat the various parts of
1385          *        the URL differently; as a workaround, just use the output for
1386          *        statistical records, not for actual linking/output.
1387          */
1388         static function replaceUnusualEscapes( $url ) {
1389                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1390                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1391         }
1392
1393         /**
1394          * Callback function used in replaceUnusualEscapes().
1395          * Replaces unusual URL escape codes with their equivalent character
1396          * @static
1397          * @private
1398          */
1399         private static function replaceUnusualEscapesCallback( $matches ) {
1400                 $char = urldecode( $matches[0] );
1401                 $ord = ord( $char );
1402                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1403                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1404                         // No, shouldn't be escaped
1405                         return $char;
1406                 } else {
1407                         // Yes, leave it escaped
1408                         return $matches[0];
1409                 }
1410         }
1411
1412         /**
1413          * make an image if it's allowed, either through the global
1414          * option, through the exception, or through the on-wiki whitelist
1415          * @private
1416          */
1417         function maybeMakeExternalImage( $url ) {
1418                 $sk = $this->mOptions->getSkin();
1419                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1420                 $imagesexception = !empty($imagesfrom);
1421                 $text = false;
1422                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1423                 if( $imagesexception && is_array( $imagesfrom ) ) {
1424                         $imagematch = false;
1425                         foreach( $imagesfrom as $match ) {
1426                                 if( strpos( $url, $match ) === 0 ) {
1427                                         $imagematch = true;
1428                                         break;
1429                                 }
1430                         }
1431                 } elseif( $imagesexception ) {
1432                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1433                 } else {
1434                         $imagematch = false;
1435                 }
1436                 if ( $this->mOptions->getAllowExternalImages()
1437                      || ( $imagesexception && $imagematch ) ) {
1438                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1439                                 # Image found
1440                                 $text = $sk->makeExternalImage( $url );
1441                         }
1442                 }
1443                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1444                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1445                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1446                         foreach( $whitelist as $entry ) {
1447                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1448                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1449                                         continue;
1450                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1451                                         # Image matches a whitelist entry
1452                                         $text = $sk->makeExternalImage( $url );
1453                                         break;
1454                                 }
1455                         }
1456                 }
1457                 return $text;
1458         }
1459
1460         /**
1461          * Process [[ ]] wikilinks
1462          * @return processed text
1463          *
1464          * @private
1465          */
1466         function replaceInternalLinks( $s ) {
1467                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1468                 return $s;
1469         }
1470
1471         /**
1472          * Process [[ ]] wikilinks (RIL)
1473          * @return LinkHolderArray
1474          *
1475          * @private
1476          */
1477         function replaceInternalLinks2( &$s ) {
1478                 global $wgContLang;
1479
1480                 wfProfileIn( __METHOD__ );
1481
1482                 wfProfileIn( __METHOD__.'-setup' );
1483                 static $tc = FALSE, $e1, $e1_img;
1484                 # the % is needed to support urlencoded titles as well
1485                 if ( !$tc ) {
1486                         $tc = Title::legalChars() . '#%';
1487                         # Match a link having the form [[namespace:link|alternate]]trail
1488                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1489                         # Match cases where there is no "]]", which might still be images
1490                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1491                 }
1492
1493                 $sk = $this->mOptions->getSkin();
1494                 $holders = new LinkHolderArray( $this );
1495
1496                 #split the entire text string on occurences of [[
1497                 $a = StringUtils::explode( '[[', ' ' . $s );
1498                 #get the first element (all text up to first [[), and remove the space we added
1499                 $s = $a->current();
1500                 $a->next();
1501                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1502                 $s = substr( $s, 1 );
1503
1504                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1505                 $e2 = null;
1506                 if ( $useLinkPrefixExtension ) {
1507                         # Match the end of a line for a word that's not followed by whitespace,
1508                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1509                         $e2 = wfMsgForContent( 'linkprefix' );
1510                 }
1511
1512                 if( is_null( $this->mTitle ) ) {
1513                         wfProfileOut( __METHOD__.'-setup' );
1514                         wfProfileOut( __METHOD__ );
1515                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1516                 }
1517                 $nottalk = !$this->mTitle->isTalkPage();
1518
1519                 if ( $useLinkPrefixExtension ) {
1520                         $m = array();
1521                         if ( preg_match( $e2, $s, $m ) ) {
1522                                 $first_prefix = $m[2];
1523                         } else {
1524                                 $first_prefix = false;
1525                         }
1526                 } else {
1527                         $prefix = '';
1528                 }
1529
1530                 if($wgContLang->hasVariants()) {
1531                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1532                 } else {
1533                         $selflink = array($this->mTitle->getPrefixedText());
1534                 }
1535                 $useSubpages = $this->areSubpagesAllowed();
1536                 wfProfileOut( __METHOD__.'-setup' );
1537
1538                 # Loop for each link
1539                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1540                         # Check for excessive memory usage
1541                         if ( $holders->isBig() ) {
1542                                 # Too big
1543                                 # Do the existence check, replace the link holders and clear the array
1544                                 $holders->replace( $s );
1545                                 $holders->clear();
1546                         }
1547
1548                         if ( $useLinkPrefixExtension ) {
1549                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1550                                 if ( preg_match( $e2, $s, $m ) ) {
1551                                         $prefix = $m[2];
1552                                         $s = $m[1];
1553                                 } else {
1554                                         $prefix='';
1555                                 }
1556                                 # first link
1557                                 if($first_prefix) {
1558                                         $prefix = $first_prefix;
1559                                         $first_prefix = false;
1560                                 }
1561                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1562                         }
1563
1564                         $might_be_img = false;
1565
1566                         wfProfileIn( __METHOD__."-e1" );
1567                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1568                                 $text = $m[2];
1569                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1570                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1571                                 # the real problem is with the $e1 regex
1572                                 # See bug 1300.
1573                                 #
1574                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1575                                 # and no image is in sight. See bug 2095.
1576                                 #
1577                                 if( $text !== '' &&
1578                                         substr( $m[3], 0, 1 ) === ']' &&
1579                                         strpos($text, '[') !== false
1580                                 )
1581                                 {
1582                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1583                                         $m[3] = substr( $m[3], 1 );
1584                                 }
1585                                 # fix up urlencoded title texts
1586                                 if( strpos( $m[1], '%' ) !== false ) {
1587                                         # Should anchors '#' also be rejected?
1588                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1589                                 }
1590                                 $trail = $m[3];
1591                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1592                                 $might_be_img = true;
1593                                 $text = $m[2];
1594                                 if ( strpos( $m[1], '%' ) !== false ) {
1595                                         $m[1] = urldecode($m[1]);
1596                                 }
1597                                 $trail = "";
1598                         } else { # Invalid form; output directly
1599                                 $s .= $prefix . '[[' . $line ;
1600                                 wfProfileOut( __METHOD__."-e1" );
1601                                 continue;
1602                         }
1603                         wfProfileOut( __METHOD__."-e1" );
1604                         wfProfileIn( __METHOD__."-misc" );
1605
1606                         # Don't allow internal links to pages containing
1607                         # PROTO: where PROTO is a valid URL protocol; these
1608                         # should be external links.
1609                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1610                                 $s .= $prefix . '[[' . $line ;
1611                                 wfProfileOut( __METHOD__."-misc" );
1612                                 continue;
1613                         }
1614
1615                         # Make subpage if necessary
1616                         if ( $useSubpages ) {
1617                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1618                         } else {
1619                                 $link = $m[1];
1620                         }
1621
1622                         $noforce = (substr( $m[1], 0, 1 ) !== ':');
1623                         if (!$noforce) {
1624                                 # Strip off leading ':'
1625                                 $link = substr( $link, 1 );
1626                         }
1627
1628                         wfProfileOut( __METHOD__."-misc" );
1629                         wfProfileIn( __METHOD__."-title" );
1630                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1631                         if ( $nt === NULL ) {
1632                                 $s .= $prefix . '[[' . $line;
1633                                 wfProfileOut( __METHOD__."-title" );
1634                                 continue;
1635                         }
1636
1637                         $ns = $nt->getNamespace();
1638                         $iw = $nt->getInterWiki();
1639                         wfProfileOut( __METHOD__."-title" );
1640
1641                         if ( $might_be_img ) { # if this is actually an invalid link
1642                                 wfProfileIn( __METHOD__."-might_be_img" );
1643                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1644                                         $found = false;
1645                                         while ( true ) {
1646                                                 #look at the next 'line' to see if we can close it there
1647                                                 $a->next();
1648                                                 $next_line = $a->current();
1649                                                 if ( $next_line === false || $next_line === null ) {
1650                                                         break;
1651                                                 }
1652                                                 $m = explode( ']]', $next_line, 3 );
1653                                                 if ( count( $m ) == 3 ) {
1654                                                         # the first ]] closes the inner link, the second the image
1655                                                         $found = true;
1656                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1657                                                         $trail = $m[2];
1658                                                         break;
1659                                                 } elseif ( count( $m ) == 2 ) {
1660                                                         #if there's exactly one ]] that's fine, we'll keep looking
1661                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1662                                                 } else {
1663                                                         #if $next_line is invalid too, we need look no further
1664                                                         $text .= '[[' . $next_line;
1665                                                         break;
1666                                                 }
1667                                         }
1668                                         if ( !$found ) {
1669                                                 # we couldn't find the end of this imageLink, so output it raw
1670                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1671                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1672                                                 $s .= "{$prefix}[[$link|$text";
1673                                                 # note: no $trail, because without an end, there *is* no trail
1674                                                 wfProfileOut( __METHOD__."-might_be_img" );
1675                                                 continue;
1676                                         }
1677                                 } else { #it's not an image, so output it raw
1678                                         $s .= "{$prefix}[[$link|$text";
1679                                         # note: no $trail, because without an end, there *is* no trail
1680                                         wfProfileOut( __METHOD__."-might_be_img" );
1681                                         continue;
1682                                 }
1683                                 wfProfileOut( __METHOD__."-might_be_img" );
1684                         }
1685
1686                         $wasblank = ( '' == $text );
1687                         if ( $wasblank ) $text = $link;
1688
1689                         # Link not escaped by : , create the various objects
1690                         if ( $noforce ) {
1691
1692                                 # Interwikis
1693                                 wfProfileIn( __METHOD__."-interwiki" );
1694                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1695                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1696                                         $s = rtrim($s . $prefix);
1697                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1698                                         wfProfileOut( __METHOD__."-interwiki" );
1699                                         continue;
1700                                 }
1701                                 wfProfileOut( __METHOD__."-interwiki" );
1702
1703                                 if ( $ns == NS_FILE ) {
1704                                         wfProfileIn( __METHOD__."-image" );
1705                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1706                                                 if ( $wasblank ) {
1707                                                         # if no parameters were passed, $text
1708                                                         # becomes something like "File:Foo.png",
1709                                                         # which we don't want to pass on to the
1710                                                         # image generator
1711                                                         $text = '';
1712                                                 } else {
1713                                                         # recursively parse links inside the image caption
1714                                                         # actually, this will parse them in any other parameters, too,
1715                                                         # but it might be hard to fix that, and it doesn't matter ATM
1716                                                         $text = $this->replaceExternalLinks($text);
1717                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1718                                                 }
1719                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1720                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1721                                         }
1722                                         $this->mOutput->addImage( $nt->getDBkey() );
1723                                         wfProfileOut( __METHOD__."-image" );
1724                                         continue;
1725
1726                                 }
1727
1728                                 if ( $ns == NS_CATEGORY ) {
1729                                         wfProfileIn( __METHOD__."-category" );
1730                                         $s = rtrim($s . "\n"); # bug 87
1731
1732                                         if ( $wasblank ) {
1733                                                 $sortkey = $this->getDefaultSort();
1734                                         } else {
1735                                                 $sortkey = $text;
1736                                         }
1737                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1738                                         $sortkey = str_replace( "\n", '', $sortkey );
1739                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1740                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1741
1742                                         /**
1743                                          * Strip the whitespace Category links produce, see bug 87
1744                                          * @todo We might want to use trim($tmp, "\n") here.
1745                                          */
1746                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1747
1748                                         wfProfileOut( __METHOD__."-category" );
1749                                         continue;
1750                                 }
1751                         }
1752
1753                         # Self-link checking
1754                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1755                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1756                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1757                                         continue;
1758                                 }
1759                         }
1760
1761                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1762                         # FIXME: Should do batch file existence checks, see comment below
1763                         if( $ns == NS_MEDIA ) {
1764                                 wfProfileIn( __METHOD__."-media" );
1765                                 # Give extensions a chance to select the file revision for us
1766                                 $skip = $time = false;
1767                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1768                                 if ( $skip ) {
1769                                         $link = $sk->link( $nt );
1770                                 } else {
1771                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1772                                 }
1773                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1774                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1775                                 $this->mOutput->addImage( $nt->getDBkey() );
1776                                 wfProfileOut( __METHOD__."-media" );
1777                                 continue;
1778                         }
1779
1780                         wfProfileIn( __METHOD__."-always_known" );
1781                         # Some titles, such as valid special pages or files in foreign repos, should
1782                         # be shown as bluelinks even though they're not included in the page table
1783                         #
1784                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1785                         # batch file existence checks for NS_FILE and NS_MEDIA
1786                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1787                                 $this->mOutput->addLink( $nt );
1788                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1789                         } else {
1790                                 # Links will be added to the output link list after checking
1791                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1792                         }
1793                         wfProfileOut( __METHOD__."-always_known" );
1794                 }
1795                 wfProfileOut( __METHOD__ );
1796                 return $holders;
1797         }
1798
1799         /**
1800          * Make a link placeholder. The text returned can be later resolved to a real link with
1801          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1802          * parsing of interwiki links, and secondly to allow all existence checks and
1803          * article length checks (for stub links) to be bundled into a single query.
1804          *
1805          * @deprecated
1806          */
1807         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1808                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1809         }
1810
1811         /**
1812          * Render a forced-blue link inline; protect against double expansion of
1813          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1814          * Since this little disaster has to split off the trail text to avoid
1815          * breaking URLs in the following text without breaking trails on the
1816          * wiki links, it's been made into a horrible function.
1817          *
1818          * @param Title $nt
1819          * @param string $text
1820          * @param string $query
1821          * @param string $trail
1822          * @param string $prefix
1823          * @return string HTML-wikitext mix oh yuck
1824          */
1825         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1826                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1827                 $sk = $this->mOptions->getSkin();
1828                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1829                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1830                 return $this->armorLinks( $link ) . $trail;
1831         }
1832
1833         /**
1834          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1835          * going to go through further parsing steps before inline URL expansion.
1836          *
1837          * Not needed quite as much as it used to be since free links are a bit
1838          * more sensible these days. But bracketed links are still an issue.
1839          *
1840          * @param string more-or-less HTML
1841          * @return string less-or-more HTML with NOPARSE bits
1842          */
1843         function armorLinks( $text ) {
1844                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1845                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1846         }
1847
1848         /**
1849          * Return true if subpage links should be expanded on this page.
1850          * @return bool
1851          */
1852         function areSubpagesAllowed() {
1853                 # Some namespaces don't allow subpages
1854                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1855         }
1856
1857         /**
1858          * Handle link to subpage if necessary
1859          * @param string $target the source of the link
1860          * @param string &$text the link text, modified as necessary
1861          * @return string the full name of the link
1862          * @private
1863          */
1864         function maybeDoSubpageLink($target, &$text) {
1865                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1866         }
1867
1868         /**#@+
1869          * Used by doBlockLevels()
1870          * @private
1871          */
1872         /* private */ function closeParagraph() {
1873                 $result = '';
1874                 if ( '' != $this->mLastSection ) {
1875                         $result = '</' . $this->mLastSection  . ">\n";
1876                 }
1877                 $this->mInPre = false;
1878                 $this->mLastSection = '';
1879                 return $result;
1880         }
1881         # getCommon() returns the length of the longest common substring
1882         # of both arguments, starting at the beginning of both.
1883         #
1884         /* private */ function getCommon( $st1, $st2 ) {
1885                 $fl = strlen( $st1 );
1886                 $shorter = strlen( $st2 );
1887                 if ( $fl < $shorter ) { $shorter = $fl; }
1888
1889                 for ( $i = 0; $i < $shorter; ++$i ) {
1890                         if ( $st1{$i} != $st2{$i} ) { break; }
1891                 }
1892                 return $i;
1893         }
1894         # These next three functions open, continue, and close the list
1895         # element appropriate to the prefix character passed into them.
1896         #
1897         /* private */ function openList( $char ) {
1898                 $result = $this->closeParagraph();
1899
1900                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1901                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1902                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1903                 elseif ( ';' === $char ) {
1904                         $result .= '<dl><dt>';
1905                         $this->mDTopen = true;
1906                 }
1907                 else { $result = '<!-- ERR 1 -->'; }
1908
1909                 return $result;
1910         }
1911
1912         /* private */ function nextItem( $char ) {
1913                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1914                 elseif ( ':' === $char || ';' === $char ) {
1915                         $close = '</dd>';
1916                         if ( $this->mDTopen ) { $close = '</dt>'; }
1917                         if ( ';' === $char ) {
1918                                 $this->mDTopen = true;
1919                                 return $close . '<dt>';
1920                         } else {
1921                                 $this->mDTopen = false;
1922                                 return $close . '<dd>';
1923                         }
1924                 }
1925                 return '<!-- ERR 2 -->';
1926         }
1927
1928         /* private */ function closeList( $char ) {
1929                 if ( '*' === $char ) { $text = '</li></ul>'; }
1930                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
1931                 elseif ( ':' === $char ) {
1932                         if ( $this->mDTopen ) {
1933                                 $this->mDTopen = false;
1934                                 $text = '</dt></dl>';
1935                         } else {
1936                                 $text = '</dd></dl>';
1937                         }
1938                 }
1939                 else {  return '<!-- ERR 3 -->'; }
1940                 return $text."\n";
1941         }
1942         /**#@-*/
1943
1944         /**
1945          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1946          *
1947          * @param $linestart bool whether or not this is at the start of a line.
1948          * @private
1949          * @return string the lists rendered as HTML
1950          */
1951         function doBlockLevels( $text, $linestart ) {
1952                 wfProfileIn( __METHOD__ );
1953
1954                 # Parsing through the text line by line.  The main thing
1955                 # happening here is handling of block-level elements p, pre,
1956                 # and making lists from lines starting with * # : etc.
1957                 #
1958                 $textLines = StringUtils::explode( "\n", $text );
1959
1960                 $lastPrefix = $output = '';
1961                 $this->mDTopen = $inBlockElem = false;
1962                 $prefixLength = 0;
1963                 $paragraphStack = false;
1964
1965                 foreach ( $textLines as $oLine ) {
1966                         # Fix up $linestart
1967                         if ( !$linestart ) {
1968                                 $output .= $oLine;
1969                                 $linestart = true;
1970                                 continue;
1971                         }
1972                         // * = ul
1973                         // # = ol
1974                         // ; = dt
1975                         // : = dd
1976
1977                         $lastPrefixLength = strlen( $lastPrefix );
1978                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1979                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1980                         // If not in a <pre> element, scan for and figure out what prefixes are there.
1981                         if ( !$this->mInPre ) {
1982                                 # Multiple prefixes may abut each other for nested lists.
1983                                 $prefixLength = strspn( $oLine, '*#:;' );
1984                                 $prefix = substr( $oLine, 0, $prefixLength );
1985
1986                                 # eh?
1987                                 // ; and : are both from definition-lists, so they're equivalent
1988                                 //  for the purposes of determining whether or not we need to open/close
1989                                 //  elements.
1990                                 $prefix2 = str_replace( ';', ':', $prefix );
1991                                 $t = substr( $oLine, $prefixLength );
1992                                 $this->mInPre = (bool)$preOpenMatch;
1993                         } else {
1994                                 # Don't interpret any other prefixes in preformatted text
1995                                 $prefixLength = 0;
1996                                 $prefix = $prefix2 = '';
1997                                 $t = $oLine;
1998                         }
1999
2000                         # List generation
2001                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2002                                 # Same as the last item, so no need to deal with nesting or opening stuff
2003                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2004                                 $paragraphStack = false;
2005
2006                                 if ( substr( $prefix, -1 ) === ';') {
2007                                         # The one nasty exception: definition lists work like this:
2008                                         # ; title : definition text
2009                                         # So we check for : in the remainder text to split up the
2010                                         # title and definition, without b0rking links.
2011                                         $term = $t2 = '';
2012                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2013                                                 $t = $t2;
2014                                                 $output .= $term . $this->nextItem( ':' );
2015                                         }
2016                                 }
2017                         } elseif( $prefixLength || $lastPrefixLength ) {
2018                                 // We need to open or close prefixes, or both.
2019
2020                                 # Either open or close a level...
2021                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2022                                 $paragraphStack = false;
2023
2024                                 // Close all the prefixes which aren't shared.
2025                                 while( $commonPrefixLength < $lastPrefixLength ) {
2026                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2027                                         --$lastPrefixLength;
2028                                 }
2029
2030                                 // Continue the current prefix if appropriate.
2031                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2032                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2033                                 }
2034
2035                                 // Open prefixes where appropriate.
2036                                 while ( $prefixLength > $commonPrefixLength ) {
2037                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2038                                         $output .= $this->openList( $char );
2039
2040                                         if ( ';' === $char ) {
2041                                                 # FIXME: This is dupe of code above
2042                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2043                                                         $t = $t2;
2044                                                         $output .= $term . $this->nextItem( ':' );
2045                                                 }
2046                                         }
2047                                         ++$commonPrefixLength;
2048                                 }
2049                                 $lastPrefix = $prefix2;
2050                         }
2051
2052                         // If we have no prefixes, go to paragraph mode.
2053                         if( 0 == $prefixLength ) {
2054                                 wfProfileIn( __METHOD__."-paragraph" );
2055                                 # No prefix (not in list)--go to paragraph mode
2056                                 // XXX: use a stack for nestable elements like span, table and div
2057                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2058                                 $closematch = preg_match(
2059                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2060                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2061                                 if ( $openmatch or $closematch ) {
2062                                         $paragraphStack = false;
2063                                         # TODO bug 5718: paragraph closed
2064                                         $output .= $this->closeParagraph();
2065                                         if ( $preOpenMatch and !$preCloseMatch ) {
2066                                                 $this->mInPre = true;
2067                                         }
2068                                         if ( $closematch ) {
2069                                                 $inBlockElem = false;
2070                                         } else {
2071                                                 $inBlockElem = true;
2072                                         }
2073                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2074                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2075                                                 // pre
2076                                                 if ($this->mLastSection !== 'pre') {
2077                                                         $paragraphStack = false;
2078                                                         $output .= $this->closeParagraph().'<pre>';
2079                                                         $this->mLastSection = 'pre';
2080                                                 }
2081                                                 $t = substr( $t, 1 );
2082                                         } else {
2083                                                 // paragraph
2084                                                 if ( '' == trim($t) ) {
2085                                                         if ( $paragraphStack ) {
2086                                                                 $output .= $paragraphStack.'<br />';
2087                                                                 $paragraphStack = false;
2088                                                                 $this->mLastSection = 'p';
2089                                                         } else {
2090                                                                 if ($this->mLastSection !== 'p' ) {
2091                                                                         $output .= $this->closeParagraph();
2092                                                                         $this->mLastSection = '';
2093                                                                         $paragraphStack = '<p>';
2094                                                                 } else {
2095                                                                         $paragraphStack = '</p><p>';
2096                                                                 }
2097                                                         }
2098                                                 } else {
2099                                                         if ( $paragraphStack ) {
2100                                                                 $output .= $paragraphStack;
2101                                                                 $paragraphStack = false;
2102                                                                 $this->mLastSection = 'p';
2103                                                         } else if ($this->mLastSection !== 'p') {
2104                                                                 $output .= $this->closeParagraph().'<p>';
2105                                                                 $this->mLastSection = 'p';
2106                                                         }
2107                                                 }
2108                                         }
2109                                 }
2110                                 wfProfileOut( __METHOD__."-paragraph" );
2111                         }
2112                         // somewhere above we forget to get out of pre block (bug 785)
2113                         if($preCloseMatch && $this->mInPre) {
2114                                 $this->mInPre = false;
2115                         }
2116                         if ($paragraphStack === false) {
2117                                 $output .= $t."\n";
2118                         }
2119                 }
2120                 while ( $prefixLength ) {
2121                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2122                         --$prefixLength;
2123                 }
2124                 if ( '' != $this->mLastSection ) {
2125                         $output .= '</' . $this->mLastSection . '>';
2126                         $this->mLastSection = '';
2127                 }
2128
2129                 wfProfileOut( __METHOD__ );
2130                 return $output;
2131         }
2132
2133         /**
2134          * Split up a string on ':', ignoring any occurences inside tags
2135          * to prevent illegal overlapping.
2136          * @param string $str the string to split
2137          * @param string &$before set to everything before the ':'
2138          * @param string &$after set to everything after the ':'
2139          * return string the position of the ':', or false if none found
2140          */
2141         function findColonNoLinks($str, &$before, &$after) {
2142                 wfProfileIn( __METHOD__ );
2143
2144                 $pos = strpos( $str, ':' );
2145                 if( $pos === false ) {
2146                         // Nothing to find!
2147                         wfProfileOut( __METHOD__ );
2148                         return false;
2149                 }
2150
2151                 $lt = strpos( $str, '<' );
2152                 if( $lt === false || $lt > $pos ) {
2153                         // Easy; no tag nesting to worry about
2154                         $before = substr( $str, 0, $pos );
2155                         $after = substr( $str, $pos+1 );
2156                         wfProfileOut( __METHOD__ );
2157                         return $pos;
2158                 }
2159
2160                 // Ugly state machine to walk through avoiding tags.
2161                 $state = self::COLON_STATE_TEXT;
2162                 $stack = 0;
2163                 $len = strlen( $str );
2164                 for( $i = 0; $i < $len; $i++ ) {
2165                         $c = $str{$i};
2166
2167                         switch( $state ) {
2168                         // (Using the number is a performance hack for common cases)
2169                         case 0: // self::COLON_STATE_TEXT:
2170                                 switch( $c ) {
2171                                 case "<":
2172                                         // Could be either a <start> tag or an </end> tag
2173                                         $state = self::COLON_STATE_TAGSTART;
2174                                         break;
2175                                 case ":":
2176                                         if( $stack == 0 ) {
2177                                                 // We found it!
2178                                                 $before = substr( $str, 0, $i );
2179                                                 $after = substr( $str, $i + 1 );
2180                                                 wfProfileOut( __METHOD__ );
2181                                                 return $i;
2182                                         }
2183                                         // Embedded in a tag; don't break it.
2184                                         break;
2185                                 default:
2186                                         // Skip ahead looking for something interesting
2187                                         $colon = strpos( $str, ':', $i );
2188                                         if( $colon === false ) {
2189                                                 // Nothing else interesting
2190                                                 wfProfileOut( __METHOD__ );
2191                                                 return false;
2192                                         }
2193                                         $lt = strpos( $str, '<', $i );
2194                                         if( $stack === 0 ) {
2195                                                 if( $lt === false || $colon < $lt ) {
2196                                                         // We found it!
2197                                                         $before = substr( $str, 0, $colon );
2198                                                         $after = substr( $str, $colon + 1 );
2199                                                         wfProfileOut( __METHOD__ );
2200                                                         return $i;
2201                                                 }
2202                                         }
2203                                         if( $lt === false ) {
2204                                                 // Nothing else interesting to find; abort!
2205                                                 // We're nested, but there's no close tags left. Abort!
2206                                                 break 2;
2207                                         }
2208                                         // Skip ahead to next tag start
2209                                         $i = $lt;
2210                                         $state = self::COLON_STATE_TAGSTART;
2211                                 }
2212                                 break;
2213                         case 1: // self::COLON_STATE_TAG:
2214                                 // In a <tag>
2215                                 switch( $c ) {
2216                                 case ">":
2217                                         $stack++;
2218                                         $state = self::COLON_STATE_TEXT;
2219                                         break;
2220                                 case "/":
2221                                         // Slash may be followed by >?
2222                                         $state = self::COLON_STATE_TAGSLASH;
2223                                         break;
2224                                 default:
2225                                         // ignore
2226                                 }
2227                                 break;
2228                         case 2: // self::COLON_STATE_TAGSTART:
2229                                 switch( $c ) {
2230                                 case "/":
2231                                         $state = self::COLON_STATE_CLOSETAG;
2232                                         break;
2233                                 case "!":
2234                                         $state = self::COLON_STATE_COMMENT;
2235                                         break;
2236                                 case ">":
2237                                         // Illegal early close? This shouldn't happen D:
2238                                         $state = self::COLON_STATE_TEXT;
2239                                         break;
2240                                 default:
2241                                         $state = self::COLON_STATE_TAG;
2242                                 }
2243                                 break;
2244                         case 3: // self::COLON_STATE_CLOSETAG:
2245                                 // In a </tag>
2246                                 if( $c === ">" ) {
2247                                         $stack--;
2248                                         if( $stack < 0 ) {
2249                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2250                                                 wfProfileOut( __METHOD__ );
2251                                                 return false;
2252                                         }
2253                                         $state = self::COLON_STATE_TEXT;
2254                                 }
2255                                 break;
2256                         case self::COLON_STATE_TAGSLASH:
2257                                 if( $c === ">" ) {
2258                                         // Yes, a self-closed tag <blah/>
2259                                         $state = self::COLON_STATE_TEXT;
2260                                 } else {
2261                                         // Probably we're jumping the gun, and this is an attribute
2262                                         $state = self::COLON_STATE_TAG;
2263                                 }
2264                                 break;
2265                         case 5: // self::COLON_STATE_COMMENT:
2266                                 if( $c === "-" ) {
2267                                         $state = self::COLON_STATE_COMMENTDASH;
2268                                 }
2269                                 break;
2270                         case self::COLON_STATE_COMMENTDASH:
2271                                 if( $c === "-" ) {
2272                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2273                                 } else {
2274                                         $state = self::COLON_STATE_COMMENT;
2275                                 }
2276                                 break;
2277                         case self::COLON_STATE_COMMENTDASHDASH:
2278                                 if( $c === ">" ) {
2279                                         $state = self::COLON_STATE_TEXT;
2280                                 } else {
2281                                         $state = self::COLON_STATE_COMMENT;
2282                                 }
2283                                 break;
2284                         default:
2285                                 throw new MWException( "State machine error in " . __METHOD__ );
2286                         }
2287                 }
2288                 if( $stack > 0 ) {
2289                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2290                         return false;
2291                 }
2292                 wfProfileOut( __METHOD__ );
2293                 return false;
2294         }
2295
2296         /**
2297          * Return value of a magic variable (like PAGENAME)
2298          *
2299          * @private
2300          */
2301         function getVariableValue( $index, $frame=false ) {
2302                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2303                 global $wgScriptPath, $wgStylePath;
2304
2305                 /**
2306                  * Some of these require message or data lookups and can be
2307                  * expensive to check many times.
2308                  */
2309                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2310                         if ( isset( $this->mVarCache[$index] ) ) {
2311                                 return $this->mVarCache[$index];
2312                         }
2313                 }
2314
2315                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2316                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2317
2318                 # Use the time zone
2319                 global $wgLocaltimezone;
2320                 if ( isset( $wgLocaltimezone ) ) {
2321                         $oldtz = getenv( 'TZ' );
2322                         putenv( 'TZ='.$wgLocaltimezone );
2323                 }
2324
2325                 wfSuppressWarnings(); // E_STRICT system time bitching
2326                 $localTimestamp = date( 'YmdHis', $ts );
2327                 $localMonth = date( 'm', $ts );
2328                 $localMonth1 = date( 'n', $ts );
2329                 $localMonthName = date( 'n', $ts );
2330                 $localDay = date( 'j', $ts );
2331                 $localDay2 = date( 'd', $ts );
2332                 $localDayOfWeek = date( 'w', $ts );
2333                 $localWeek = date( 'W', $ts );
2334                 $localYear = date( 'Y', $ts );
2335                 $localHour = date( 'H', $ts );
2336                 if ( isset( $wgLocaltimezone ) ) {
2337                         putenv( 'TZ='.$oldtz );
2338                 }
2339                 wfRestoreWarnings();
2340
2341                 switch ( $index ) {
2342                         case 'currentmonth':
2343                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2344                                 break;
2345                         case 'currentmonth1':
2346                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2347                                 break;
2348                         case 'currentmonthname':
2349                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2350                                 break;
2351                         case 'currentmonthnamegen':
2352                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2353                                 break;
2354                         case 'currentmonthabbrev':
2355                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2356                                 break;
2357                         case 'currentday':
2358                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2359                                 break;
2360                         case 'currentday2':
2361                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2362                                 break;
2363                         case 'localmonth':
2364                                 $value = $wgContLang->formatNum( $localMonth );
2365                                 break;
2366                         case 'localmonth1':
2367                                 $value = $wgContLang->formatNum( $localMonth1 );
2368                                 break;
2369                         case 'localmonthname':
2370                                 $value = $wgContLang->getMonthName( $localMonthName );
2371                                 break;
2372                         case 'localmonthnamegen':
2373                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2374                                 break;
2375                         case 'localmonthabbrev':
2376                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2377                                 break;
2378                         case 'localday':
2379                                 $value = $wgContLang->formatNum( $localDay );
2380                                 break;
2381                         case 'localday2':
2382                                 $value = $wgContLang->formatNum( $localDay2 );
2383                                 break;
2384                         case 'pagename':
2385                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2386                                 break;
2387                         case 'pagenamee':
2388                                 $value = $this->mTitle->getPartialURL();
2389                                 break;
2390                         case 'fullpagename':
2391                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2392                                 break;
2393                         case 'fullpagenamee':
2394                                 $value = $this->mTitle->getPrefixedURL();
2395                                 break;
2396                         case 'subpagename':
2397                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2398                                 break;
2399                         case 'subpagenamee':
2400                                 $value = $this->mTitle->getSubpageUrlForm();
2401                                 break;
2402                         case 'basepagename':
2403                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2404                                 break;
2405                         case 'basepagenamee':
2406                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2407                                 break;
2408                         case 'talkpagename':
2409                                 if( $this->mTitle->canTalk() ) {
2410                                         $talkPage = $this->mTitle->getTalkPage();
2411                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2412                                 } else {
2413                                         $value = '';
2414                                 }
2415                                 break;
2416                         case 'talkpagenamee':
2417                                 if( $this->mTitle->canTalk() ) {
2418                                         $talkPage = $this->mTitle->getTalkPage();
2419                                         $value = $talkPage->getPrefixedUrl();
2420                                 } else {
2421                                         $value = '';
2422                                 }
2423                                 break;
2424                         case 'subjectpagename':
2425                                 $subjPage = $this->mTitle->getSubjectPage();
2426                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2427                                 break;
2428                         case 'subjectpagenamee':
2429                                 $subjPage = $this->mTitle->getSubjectPage();
2430                                 $value = $subjPage->getPrefixedUrl();
2431                                 break;
2432                         case 'revisionid':
2433                                 // Let the edit saving system know we should parse the page
2434                                 // *after* a revision ID has been assigned.
2435                                 $this->mOutput->setFlag( 'vary-revision' );
2436                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2437                                 $value = $this->mRevisionId;
2438                                 break;
2439                         case 'revisionday':
2440                                 // Let the edit saving system know we should parse the page
2441                                 // *after* a revision ID has been assigned. This is for null edits.
2442                                 $this->mOutput->setFlag( 'vary-revision' );
2443                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2444                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2445                                 break;
2446                         case 'revisionday2':
2447                                 // Let the edit saving system know we should parse the page
2448                                 // *after* a revision ID has been assigned. This is for null edits.
2449                                 $this->mOutput->setFlag( 'vary-revision' );
2450                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2451                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2452                                 break;
2453                         case 'revisionmonth':
2454                                 // Let the edit saving system know we should parse the page
2455                                 // *after* a revision ID has been assigned. This is for null edits.
2456                                 $this->mOutput->setFlag( 'vary-revision' );
2457                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2458                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2459                                 break;
2460                         case 'revisionyear':
2461                                 // Let the edit saving system know we should parse the page
2462                                 // *after* a revision ID has been assigned. This is for null edits.
2463                                 $this->mOutput->setFlag( 'vary-revision' );
2464                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2465                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2466                                 break;
2467                         case 'revisiontimestamp':
2468                                 // Let the edit saving system know we should parse the page
2469                                 // *after* a revision ID has been assigned. This is for null edits.
2470                                 $this->mOutput->setFlag( 'vary-revision' );
2471                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2472                                 $value = $this->getRevisionTimestamp();
2473                                 break;
2474                         case 'revisionuser':
2475                                 // Let the edit saving system know we should parse the page
2476                                 // *after* a revision ID has been assigned. This is for null edits.
2477                                 $this->mOutput->setFlag( 'vary-revision' );
2478                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2479                                 $value = $this->getRevisionUser();
2480                                 break;
2481                         case 'namespace':
2482                                 $value = str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2483                                 break;
2484                         case 'namespacee':
2485                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2486                                 break;
2487                         case 'talkspace':
2488                                 $value = $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2489                                 break;
2490                         case 'talkspacee':
2491                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2492                                 break;
2493                         case 'subjectspace':
2494                                 $value = $this->mTitle->getSubjectNsText();
2495                                 break;
2496                         case 'subjectspacee':
2497                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2498                                 break;
2499                         case 'currentdayname':
2500                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2501                                 break;
2502                         case 'currentyear':
2503                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2504                                 break;
2505                         case 'currenttime':
2506                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2507                                 break;
2508                         case 'currenthour':
2509                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2510                                 break;
2511                         case 'currentweek':
2512                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2513                                 // int to remove the padding
2514                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2515                                 break;
2516                         case 'currentdow':
2517                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2518                                 break;
2519                         case 'localdayname':
2520                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2521                                 break;
2522                         case 'localyear':
2523                                 $value = $wgContLang->formatNum( $localYear, true );
2524                                 break;
2525                         case 'localtime':
2526                                 $value = $wgContLang->time( $localTimestamp, false, false );
2527                                 break;
2528                         case 'localhour':
2529                                 $value = $wgContLang->formatNum( $localHour, true );
2530                                 break;
2531                         case 'localweek':
2532                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2533                                 // int to remove the padding
2534                                 $value = $wgContLang->formatNum( (int)$localWeek );
2535                                 break;
2536                         case 'localdow':
2537                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2538                                 break;
2539                         case 'numberofarticles':
2540                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2541                                 break;
2542                         case 'numberoffiles':
2543                                 $value = $wgContLang->formatNum( SiteStats::images() );
2544                                 break;
2545                         case 'numberofusers':
2546                                 $value = $wgContLang->formatNum( SiteStats::users() );
2547                                 break;
2548                         case 'numberofactiveusers':
2549                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2550                                 break;
2551                         case 'numberofpages':
2552                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2553                                 break;
2554                         case 'numberofadmins':
2555                                 $value = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2556                                 break;
2557                         case 'numberofedits':
2558                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2559                                 break;
2560                         case 'numberofviews':
2561                                 $value = $wgContLang->formatNum( SiteStats::views() );
2562                                 break;
2563                         case 'currenttimestamp':
2564                                 $value = wfTimestamp( TS_MW, $ts );
2565                                 break;
2566                         case 'localtimestamp':
2567                                 $value = $localTimestamp;
2568                                 break;
2569                         case 'currentversion':
2570                                 $value = SpecialVersion::getVersion();
2571                                 break;
2572                         case 'sitename':
2573                                 return $wgSitename;
2574                         case 'server':
2575                                 return $wgServer;
2576                         case 'servername':
2577                                 return $wgServerName;
2578                         case 'scriptpath':
2579                                 return $wgScriptPath;
2580                         case 'directionmark':
2581                                 return $wgContLang->getDirMark();
2582                         case 'contentlanguage':
2583                                 global $wgContLanguageCode;
2584                                 return $wgContLanguageCode;
2585                         default:
2586                                 $ret = null;
2587                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) )
2588                                         return $ret;
2589                                 else
2590                                         return null;
2591                 }
2592
2593                 if ( $index )
2594                         $this->mVarCache[$index] = $value;
2595
2596                 return $value;
2597         }
2598
2599         /**
2600          * initialise the magic variables (like CURRENTMONTHNAME)
2601          *
2602          * @private
2603          */
2604         function initialiseVariables() {
2605                 wfProfileIn( __METHOD__ );
2606                 $variableIDs = MagicWord::getVariableIDs();
2607
2608                 $this->mVariables = new MagicWordArray( $variableIDs );
2609                 wfProfileOut( __METHOD__ );
2610         }
2611
2612         /**
2613          * Preprocess some wikitext and return the document tree.
2614          * This is the ghost of replace_variables().
2615          *
2616          * @param string $text The text to parse
2617          * @param integer flags Bitwise combination of:
2618          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2619          *                                     included. Default is to assume a direct page view.
2620          *
2621          * The generated DOM tree must depend only on the input text and the flags.
2622          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2623          *
2624          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2625          * change in the DOM tree for a given text, must be passed through the section identifier
2626          * in the section edit link and thus back to extractSections().
2627          *
2628          * The output of this function is currently only cached in process memory, but a persistent
2629          * cache may be implemented at a later date which takes further advantage of these strict
2630          * dependency requirements.
2631          *
2632          * @private
2633          */
2634         function preprocessToDom ( $text, $flags = 0 ) {
2635                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2636                 return $dom;
2637         }
2638
2639         /*
2640          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2641          */
2642         public static function splitWhitespace( $s ) {
2643                 $ltrimmed = ltrim( $s );
2644                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2645                 $trimmed = rtrim( $ltrimmed );
2646                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2647                 if ( $diff > 0 ) {
2648                         $w2 = substr( $ltrimmed, -$diff );
2649                 } else {
2650                         $w2 = '';
2651                 }
2652                 return array( $w1, $trimmed, $w2 );
2653         }
2654
2655         /**
2656          * Replace magic variables, templates, and template arguments
2657          * with the appropriate text. Templates are substituted recursively,
2658          * taking care to avoid infinite loops.
2659          *
2660          * Note that the substitution depends on value of $mOutputType:
2661          *  self::OT_WIKI: only {{subst:}} templates
2662          *  self::OT_PREPROCESS: templates but not extension tags
2663          *  self::OT_HTML: all templates and extension tags
2664          *
2665          * @param string $tex The text to transform
2666          * @param PPFrame $frame Object describing the arguments passed to the template.
2667          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2668          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2669          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2670          * @private
2671          */
2672         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2673                 # Is there any text? Also, Prevent too big inclusions!
2674                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2675                         return $text;
2676                 }
2677                 wfProfileIn( __METHOD__ );
2678
2679                 if ( $frame === false ) {
2680                         $frame = $this->getPreprocessor()->newFrame();
2681                 } elseif ( !( $frame instanceof PPFrame ) ) {
2682                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2683                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2684                 }
2685
2686                 $dom = $this->preprocessToDom( $text );
2687                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2688                 $text = $frame->expand( $dom, $flags );
2689
2690                 wfProfileOut( __METHOD__ );
2691                 return $text;
2692         }
2693
2694         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2695         static function createAssocArgs( $args ) {
2696                 $assocArgs = array();
2697                 $index = 1;
2698                 foreach( $args as $arg ) {
2699                         $eqpos = strpos( $arg, '=' );
2700                         if ( $eqpos === false ) {
2701                                 $assocArgs[$index++] = $arg;
2702                         } else {
2703                                 $name = trim( substr( $arg, 0, $eqpos ) );
2704                                 $value = trim( substr( $arg, $eqpos+1 ) );
2705                                 if ( $value === false ) {
2706                                         $value = '';
2707                                 }
2708                                 if ( $name !== false ) {
2709                                         $assocArgs[$name] = $value;
2710                                 }
2711                         }
2712                 }
2713
2714                 return $assocArgs;
2715         }
2716
2717         /**
2718          * Warn the user when a parser limitation is reached
2719          * Will warn at most once the user per limitation type
2720          *
2721          * @param string $limitationType, should be one of:
2722          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2723          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2724          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2725          * @params int $current, $max When an explicit limit has been
2726          *       exceeded, provide the values (optional)
2727          */
2728         function limitationWarn( $limitationType, $current=null, $max=null) {
2729                 //does no harm if $current and $max are present but are unnecessary for the message
2730                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2731                 $this->mOutput->addWarning( $warning );
2732                 $this->addTrackingCategory( "$limitationType-category" );
2733         }
2734
2735         /**
2736          * Return the text of a template, after recursively
2737          * replacing any variables or templates within the template.
2738          *
2739          * @param array $piece The parts of the template
2740          *  $piece['title']: the title, i.e. the part before the |
2741          *  $piece['parts']: the parameter array
2742          *  $piece['lineStart']: whether the brace was at the start of a line
2743          * @param PPFrame The current frame, contains template arguments
2744          * @return string the text of the template
2745          * @private
2746          */
2747         function braceSubstitution( $piece, $frame ) {
2748                 global $wgContLang, $wgNonincludableNamespaces;
2749                 wfProfileIn( __METHOD__ );
2750                 wfProfileIn( __METHOD__.'-setup' );
2751
2752                 # Flags
2753                 $found = false;             # $text has been filled
2754                 $nowiki = false;            # wiki markup in $text should be escaped
2755                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2756                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2757                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2758                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2759
2760                 # Title object, where $text came from
2761                 $title = NULL;
2762
2763                 # $part1 is the bit before the first |, and must contain only title characters.
2764                 # Various prefixes will be stripped from it later.
2765                 $titleWithSpaces = $frame->expand( $piece['title'] );
2766                 $part1 = trim( $titleWithSpaces );
2767                 $titleText = false;
2768
2769                 # Original title text preserved for various purposes
2770                 $originalTitle = $part1;
2771
2772                 # $args is a list of argument nodes, starting from index 0, not including $part1
2773                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2774                 wfProfileOut( __METHOD__.'-setup' );
2775
2776                 # SUBST
2777                 wfProfileIn( __METHOD__.'-modifiers' );
2778                 if ( !$found ) {
2779                         $mwSubst = MagicWord::get( 'subst' );
2780                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2781                                 # One of two possibilities is true:
2782                                 # 1) Found SUBST but not in the PST phase
2783                                 # 2) Didn't find SUBST and in the PST phase
2784                                 # In either case, return without further processing
2785                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2786                                 $isLocalObj = true;
2787                                 $found = true;
2788                         }
2789                 }
2790
2791                 # Variables
2792                 if ( !$found && $args->getLength() == 0 ) {
2793                         $id = $this->mVariables->matchStartToEnd( $part1 );
2794                         if ( $id !== false ) {
2795                                 $text = $this->getVariableValue( $id, $frame );
2796                                 if (MagicWord::getCacheTTL($id)>-1)
2797                                         $this->mOutput->mContainsOldMagic = true;
2798                                 $found = true;
2799                         }
2800                 }
2801
2802                 # MSG, MSGNW and RAW
2803                 if ( !$found ) {
2804                         # Check for MSGNW:
2805                         $mwMsgnw = MagicWord::get( 'msgnw' );
2806                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2807                                 $nowiki = true;
2808                         } else {
2809                                 # Remove obsolete MSG:
2810                                 $mwMsg = MagicWord::get( 'msg' );
2811                                 $mwMsg->matchStartAndRemove( $part1 );
2812                         }
2813
2814                         # Check for RAW:
2815                         $mwRaw = MagicWord::get( 'raw' );
2816                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2817                                 $forceRawInterwiki = true;
2818                         }
2819                 }
2820                 wfProfileOut( __METHOD__.'-modifiers' );
2821
2822                 # Parser functions
2823                 if ( !$found ) {
2824                         wfProfileIn( __METHOD__ . '-pfunc' );
2825
2826                         $colonPos = strpos( $part1, ':' );
2827                         if ( $colonPos !== false ) {
2828                                 # Case sensitive functions
2829                                 $function = substr( $part1, 0, $colonPos );
2830                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2831                                         $function = $this->mFunctionSynonyms[1][$function];
2832                                 } else {
2833                                         # Case insensitive functions
2834                                         $function = $wgContLang->lc( $function );
2835                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2836                                                 $function = $this->mFunctionSynonyms[0][$function];
2837                                         } else {
2838                                                 $function = false;
2839                                         }
2840                                 }
2841                                 if ( $function ) {
2842                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2843                                         $initialArgs = array( &$this );
2844                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2845                                         if ( $flags & SFH_OBJECT_ARGS ) {
2846                                                 # Add a frame parameter, and pass the arguments as an array
2847                                                 $allArgs = $initialArgs;
2848                                                 $allArgs[] = $frame;
2849                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2850                                                         $funcArgs[] = $args->item( $i );
2851                                                 }
2852                                                 $allArgs[] = $funcArgs;
2853                                         } else {
2854                                                 # Convert arguments to plain text
2855                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2856                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2857                                                 }
2858                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2859                                         }
2860
2861                                         # Workaround for PHP bug 35229 and similar
2862                                         if ( !is_callable( $callback ) ) {
2863                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2864                                                 wfProfileOut( __METHOD__ );
2865                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2866                                         }
2867                                         $result = call_user_func_array( $callback, $allArgs );
2868                                         $found = true;
2869                                         $noparse = true;
2870                                         $preprocessFlags = 0;
2871
2872                                         if ( is_array( $result ) ) {
2873                                                 if ( isset( $result[0] ) ) {
2874                                                         $text = $result[0];
2875                                                         unset( $result[0] );
2876                                                 }
2877
2878                                                 // Extract flags into the local scope
2879                                                 // This allows callers to set flags such as nowiki, found, etc.
2880                                                 extract( $result );
2881                                         } else {
2882                                                 $text = $result;
2883                                         }
2884                                         if ( !$noparse ) {
2885                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2886                                                 $isChildObj = true;
2887                                         }
2888                                 }
2889                         }
2890                         wfProfileOut( __METHOD__ . '-pfunc' );
2891                 }
2892
2893                 # Finish mangling title and then check for loops.
2894                 # Set $title to a Title object and $titleText to the PDBK
2895                 if ( !$found ) {
2896                         $ns = NS_TEMPLATE;
2897                         # Split the title into page and subpage
2898                         $subpage = '';
2899                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2900                         if ($subpage !== '') {
2901                                 $ns = $this->mTitle->getNamespace();
2902                         }
2903                         $title = Title::newFromText( $part1, $ns );
2904                         if ( $title ) {
2905                                 $titleText = $title->getPrefixedText();
2906                                 # Check for language variants if the template is not found
2907                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2908                                         $wgContLang->findVariantLink( $part1, $title, true );
2909                                 }
2910                                 # Do recursion depth check
2911                                 $limit = $this->mOptions->getMaxTemplateDepth();
2912                                 if ( $frame->depth >= $limit ) {
2913                                         $found = true;
2914                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2915                                 }
2916                         }
2917                 }
2918
2919                 # Load from database
2920                 if ( !$found && $title ) {
2921                         wfProfileIn( __METHOD__ . '-loadtpl' );
2922                         if ( !$title->isExternal() ) {
2923                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2924                                         $text = SpecialPage::capturePath( $title );
2925                                         if ( is_string( $text ) ) {
2926                                                 $found = true;
2927                                                 $isHTML = true;
2928                                                 $this->disableCache();
2929                                         }
2930                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2931                                         $found = false; //access denied
2932                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2933                                 } else {
2934                                         list( $text, $title ) = $this->getTemplateDom( $title );
2935                                         if ( $text !== false ) {
2936                                                 $found = true;
2937                                                 $isChildObj = true;
2938                                         }
2939                                 }
2940
2941                                 # If the title is valid but undisplayable, make a link to it
2942                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2943                                         $text = "[[:$titleText]]";
2944                                         $found = true;
2945                                 }
2946                         } elseif ( $title->isTrans() ) {
2947                                 // Interwiki transclusion
2948                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2949                                         $text = $this->interwikiTransclude( $title, 'render' );
2950                                         $isHTML = true;
2951                                 } else {
2952                                         $text = $this->interwikiTransclude( $title, 'raw' );
2953                                         // Preprocess it like a template
2954                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2955                                         $isChildObj = true;
2956                                 }
2957                                 $found = true;
2958                         }
2959
2960                         # Do infinite loop check
2961                         # This has to be done after redirect resolution to avoid infinite loops via redirects
2962                         if ( !$frame->loopCheck( $title ) ) {
2963                                 $found = true;
2964                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2965                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2966                         }
2967                         wfProfileOut( __METHOD__ . '-loadtpl' );
2968                 }
2969
2970                 # If we haven't found text to substitute by now, we're done
2971                 # Recover the source wikitext and return it
2972                 if ( !$found ) {
2973                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2974                         wfProfileOut( __METHOD__ );
2975                         return array( 'object' => $text );
2976                 }
2977
2978                 # Expand DOM-style return values in a child frame
2979                 if ( $isChildObj ) {
2980                         # Clean up argument array
2981                         $newFrame = $frame->newChild( $args, $title );
2982
2983                         if ( $nowiki ) {
2984                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2985                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2986                                 # Expansion is eligible for the empty-frame cache
2987                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2988                                         $text = $this->mTplExpandCache[$titleText];
2989                                 } else {
2990                                         $text = $newFrame->expand( $text );
2991                                         $this->mTplExpandCache[$titleText] = $text;
2992                                 }
2993                         } else {
2994                                 # Uncached expansion
2995                                 $text = $newFrame->expand( $text );
2996                         }
2997                 }
2998                 if ( $isLocalObj && $nowiki ) {
2999                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3000                         $isLocalObj = false;
3001                 }
3002
3003                 # Replace raw HTML by a placeholder
3004                 # Add a blank line preceding, to prevent it from mucking up
3005                 # immediately preceding headings
3006                 if ( $isHTML ) {
3007                         $text = "\n\n" . $this->insertStripItem( $text );
3008                 }
3009                 # Escape nowiki-style return values
3010                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3011                         $text = wfEscapeWikiText( $text );
3012                 }
3013                 # Bug 529: if the template begins with a table or block-level
3014                 # element, it should be treated as beginning a new line.
3015                 # This behaviour is somewhat controversial.
3016                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3017                         $text = "\n" . $text;
3018                 }
3019
3020                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3021                         # Error, oversize inclusion
3022                         $text = "[[$originalTitle]]" .
3023                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3024                         $this->limitationWarn( 'post-expand-template-inclusion' );
3025                 }
3026
3027                 if ( $isLocalObj ) {
3028                         $ret = array( 'object' => $text );
3029                 } else {
3030                         $ret = array( 'text' => $text );
3031                 }
3032
3033                 wfProfileOut( __METHOD__ );
3034                 return $ret;
3035         }
3036
3037         /**
3038          * Get the semi-parsed DOM representation of a template with a given title,
3039          * and its redirect destination title. Cached.
3040          */
3041         function getTemplateDom( $title ) {
3042                 $cacheTitle = $title;
3043                 $titleText = $title->getPrefixedDBkey();
3044
3045                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3046                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3047                         $title = Title::makeTitle( $ns, $dbk );
3048                         $titleText = $title->getPrefixedDBkey();
3049                 }
3050                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3051                         return array( $this->mTplDomCache[$titleText], $title );
3052                 }
3053
3054                 // Cache miss, go to the database
3055                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3056
3057                 if ( $text === false ) {
3058                         $this->mTplDomCache[$titleText] = false;
3059                         return array( false, $title );
3060                 }
3061
3062                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3063                 $this->mTplDomCache[ $titleText ] = $dom;
3064
3065                 if (! $title->equals($cacheTitle)) {
3066                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3067                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3068                 }
3069
3070                 return array( $dom, $title );
3071         }
3072
3073         /**
3074          * Fetch the unparsed text of a template and register a reference to it.
3075          */
3076         function fetchTemplateAndTitle( $title ) {
3077                 $templateCb = $this->mOptions->getTemplateCallback();
3078                 $stuff = call_user_func( $templateCb, $title, $this );
3079                 $text = $stuff['text'];
3080                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3081                 if ( isset( $stuff['deps'] ) ) {
3082                         foreach ( $stuff['deps'] as $dep ) {
3083                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3084                         }
3085                 }
3086                 return array($text,$finalTitle);
3087         }
3088
3089         function fetchTemplate( $title ) {
3090                 $rv = $this->fetchTemplateAndTitle($title);
3091                 return $rv[0];
3092         }
3093
3094         /**
3095          * Static function to get a template
3096          * Can be overridden via ParserOptions::setTemplateCallback().
3097          */
3098         static function statelessFetchTemplate( $title, $parser=false ) {
3099                 $text = $skip = false;
3100                 $finalTitle = $title;
3101                 $deps = array();
3102
3103                 // Loop to fetch the article, with up to 1 redirect
3104                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3105                         # Give extensions a chance to select the revision instead
3106                         $id = false; // Assume current
3107                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3108
3109                         if( $skip ) {
3110                                 $text = false;
3111                                 $deps[] = array(
3112                                         'title' => $title,
3113                                         'page_id' => $title->getArticleID(),
3114                                         'rev_id' => null );
3115                                 break;
3116                         }
3117                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3118                         $rev_id = $rev ? $rev->getId() : 0;
3119                         // If there is no current revision, there is no page
3120                         if( $id === false && !$rev ) {
3121                                 $linkCache = LinkCache::singleton();
3122                                 $linkCache->addBadLinkObj( $title );
3123                         }
3124
3125                         $deps[] = array(
3126                                 'title' => $title,
3127                                 'page_id' => $title->getArticleID(),
3128                                 'rev_id' => $rev_id );
3129
3130                         if( $rev ) {
3131                                 $text = $rev->getText();
3132                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3133                                 global $wgContLang;
3134                                 $message = $wgContLang->lcfirst( $title->getText() );
3135                                 $text = wfMsgForContentNoTrans( $message );
3136                                 if( wfEmptyMsg( $message, $text ) ) {
3137                                         $text = false;
3138                                         break;
3139                                 }
3140                         } else {
3141                                 break;
3142                         }
3143                         if ( $text === false ) {
3144                                 break;
3145                         }
3146                         // Redirect?
3147                         $finalTitle = $title;
3148                         $title = Title::newFromRedirect( $text );
3149                 }
3150                 return array(
3151                         'text' => $text,
3152                         'finalTitle' => $finalTitle,
3153                         'deps' => $deps );
3154         }
3155
3156         /**
3157          * Transclude an interwiki link.
3158          */
3159         function interwikiTransclude( $title, $action ) {
3160                 global $wgEnableScaryTranscluding;
3161
3162                 if (!$wgEnableScaryTranscluding)
3163                         return wfMsg('scarytranscludedisabled');
3164
3165                 $url = $title->getFullUrl( "action=$action" );
3166
3167                 if (strlen($url) > 255)
3168                         return wfMsg('scarytranscludetoolong');
3169                 return $this->fetchScaryTemplateMaybeFromCache($url);
3170         }
3171
3172         function fetchScaryTemplateMaybeFromCache($url) {
3173                 global $wgTranscludeCacheExpiry;
3174                 $dbr = wfGetDB(DB_SLAVE);
3175                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3176                                 array('tc_url' => $url));
3177                 if ($obj) {
3178                         $time = $obj->tc_time;
3179                         $text = $obj->tc_contents;
3180                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3181                                 return $text;
3182                         }
3183                 }
3184
3185                 $text = Http::get($url);
3186                 if (!$text)
3187                         return wfMsg('scarytranscludefailed', $url);
3188
3189                 $dbw = wfGetDB(DB_MASTER);
3190                 $dbw->replace('transcache', array('tc_url'), array(
3191                         'tc_url' => $url,
3192                         'tc_time' => time(),
3193                         'tc_contents' => $text));
3194                 return $text;
3195         }
3196
3197
3198         /**
3199          * Triple brace replacement -- used for template arguments
3200          * @private
3201          */
3202         function argSubstitution( $piece, $frame ) {
3203                 wfProfileIn( __METHOD__ );
3204
3205                 $error = false;
3206                 $parts = $piece['parts'];
3207                 $nameWithSpaces = $frame->expand( $piece['title'] );
3208                 $argName = trim( $nameWithSpaces );
3209                 $object = false;
3210                 $text = $frame->getArgument( $argName );
3211                 if (  $text === false && $parts->getLength() > 0
3212                   && (
3213                     $this->ot['html']
3214                     || $this->ot['pre']
3215                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3216                   )
3217                 ) {
3218                         # No match in frame, use the supplied default
3219                         $object = $parts->item( 0 )->getChildren();
3220                 }
3221                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3222                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3223                         $this->limitationWarn( 'post-expand-template-argument' );
3224                 }
3225
3226                 if ( $text === false && $object === false ) {
3227                         # No match anywhere
3228                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3229                 }
3230                 if ( $error !== false ) {
3231                         $text .= $error;
3232                 }
3233                 if ( $object !== false ) {
3234                         $ret = array( 'object' => $object );
3235                 } else {
3236                         $ret = array( 'text' => $text );
3237                 }
3238
3239                 wfProfileOut( __METHOD__ );
3240                 return $ret;
3241         }
3242
3243         /**
3244          * Return the text to be used for a given extension tag.
3245          * This is the ghost of strip().
3246          *
3247          * @param array $params Associative array of parameters:
3248          *     name       PPNode for the tag name
3249          *     attr       PPNode for unparsed text where tag attributes are thought to be
3250          *     attributes Optional associative array of parsed attributes
3251          *     inner      Contents of extension element
3252          *     noClose    Original text did not have a close tag
3253          * @param PPFrame $frame
3254          */
3255         function extensionSubstitution( $params, $frame ) {
3256                 global $wgRawHtml, $wgContLang;
3257
3258                 $name = $frame->expand( $params['name'] );
3259                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3260                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3261
3262                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3263
3264                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3265                         ( $this->ot['html'] || $this->ot['pre'] );
3266                 if ( $this->ot['html'] || $isFunctionTag ) {
3267                         $name = strtolower( $name );
3268                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3269                         if ( isset( $params['attributes'] ) ) {
3270                                 $attributes = $attributes + $params['attributes'];
3271                         }
3272                         switch ( $name ) {
3273                                 case 'html':
3274                                         if( $wgRawHtml ) {
3275                                                 $output = $content;
3276                                                 break;
3277                                         } else {
3278                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3279                                         }
3280                                 case 'nowiki':
3281                                         $content = strtr($content, array('-{' => '-&#123;', '}-' => '&#125;-'));
3282                                         $output = Xml::escapeTagsOnly( $content );
3283                                         break;
3284                                 case 'math':
3285                                         if ( $this->mOptions->getUseTeX() ) {
3286                                                 $output = $wgContLang->armourMath(
3287                                                         MathRenderer::renderMath( $content, $attributes ) );
3288                                         }
3289                                         /* else let a tag hook handle it (bug 21222) */
3290                                         break;
3291                                 case 'gallery':
3292                                         $output = $this->renderImageGallery( $content, $attributes );
3293                                         break;
3294                                 default:
3295                                         if( isset( $this->mTagHooks[$name] ) ) {
3296                                                 # Workaround for PHP bug 35229 and similar
3297                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3298                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3299                                                 }
3300                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3301                                                         array( $content, $attributes, $this, $frame ) );
3302                                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3303                                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3304                                                 if( !is_callable( $callback ) )
3305                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3306
3307                                                 $output = call_user_func_array( $callback,
3308                                                         array( &$this, $frame, $content, $attributes ) );
3309                                         } else {
3310                                                 $output = '<span class="error">Invalid tag extension name: ' .
3311                                                         htmlspecialchars( $name ) . '</span>';
3312                                         }
3313                         }
3314                 } else {
3315                         if ( is_null( $attrText ) ) {
3316                                 $attrText = '';
3317                         }
3318                         if ( isset( $params['attributes'] ) ) {
3319                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3320                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3321                                                 htmlspecialchars( $attrValue ) . '"';
3322                                 }
3323                         }
3324                         if ( $content === null ) {
3325                                 $output = "<$name$attrText/>";
3326                         } else {
3327                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3328                                 $output = "<$name$attrText>$content$close";
3329                         }
3330                 }
3331
3332                 if( $isFunctionTag ) {
3333                         return $output;
3334                 } elseif ( $name === 'html' || $name === 'nowiki' ) {
3335                         $this->mStripState->nowiki->setPair( $marker, $output );
3336                 } else {
3337                         $this->mStripState->general->setPair( $marker, $output );
3338                 }
3339                 return $marker;
3340         }
3341
3342         /**
3343          * Increment an include size counter
3344          *
3345          * @param string $type The type of expansion
3346          * @param integer $size The size of the text
3347          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3348          */
3349         function incrementIncludeSize( $type, $size ) {
3350                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3351                         return false;
3352                 } else {
3353                         $this->mIncludeSizes[$type] += $size;
3354                         return true;
3355                 }
3356         }
3357
3358         /**
3359          * Increment the expensive function count
3360          *
3361          * @return boolean False if the limit has been exceeded
3362          */
3363         function incrementExpensiveFunctionCount() {
3364                 global $wgExpensiveParserFunctionLimit;
3365                 $this->mExpensiveFunctionCount++;
3366                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3367                         return true;
3368                 }
3369                 return false;
3370         }
3371
3372         /**
3373          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3374          * Fills $this->mDoubleUnderscores, returns the modified text
3375          */
3376         function doDoubleUnderscore( $text ) {
3377                 wfProfileIn( __METHOD__ );
3378                 // The position of __TOC__ needs to be recorded
3379                 $mw = MagicWord::get( 'toc' );
3380                 if( $mw->match( $text ) ) {
3381                         $this->mShowToc = true;
3382                         $this->mForceTocPosition = true;
3383
3384                         // Set a placeholder. At the end we'll fill it in with the TOC.
3385                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3386
3387                         // Only keep the first one.
3388                         $text = $mw->replace( '', $text );
3389                 }
3390
3391                 // Now match and remove the rest of them
3392                 $mwa = MagicWord::getDoubleUnderscoreArray();
3393                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3394
3395                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3396                         $this->mOutput->mNoGallery = true;
3397                 }
3398                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3399                         $this->mShowToc = false;
3400                 }
3401                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3402                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3403                         $this->addTrackingCategory( 'hidden-category-category' );
3404                 }
3405                 # (bug 8068) Allow control over whether robots index a page.
3406                 #
3407                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3408                 # is not desirable, the last one on the page should win.
3409                 if( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3410                         $this->mOutput->setIndexPolicy( 'noindex' );
3411                         $this->addTrackingCategory( 'noindex-category' );
3412                 }
3413                 if( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ){
3414                         $this->mOutput->setIndexPolicy( 'index' );
3415                         $this->addTrackingCategory( 'index-category' );
3416                 }
3417                 wfProfileOut( __METHOD__ );
3418                 return $text;
3419         }
3420
3421         /**
3422          * Add a tracking category, getting the title from a system message,
3423          * or print a debug message if the title is invalid.
3424          * @param $msg String message key
3425          * @return Bool whether the addition was successful
3426          */
3427         protected function addTrackingCategory( $msg ){
3428                 $cat = wfMsgForContent( $msg );
3429
3430                 # Allow tracking categories to be disabled by setting them to "-"
3431                 if( $cat === '-' ) return false;
3432
3433                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3434                 if ( $containerCategory ) {
3435                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3436                         return true;
3437                 } else {
3438                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3439                         return false;
3440                 }
3441         }
3442
3443         /**
3444          * This function accomplishes several tasks:
3445          * 1) Auto-number headings if that option is enabled
3446          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3447          * 3) Add a Table of contents on the top for users who have enabled the option
3448          * 4) Auto-anchor headings
3449          *
3450          * It loops through all headlines, collects the necessary data, then splits up the
3451          * string and re-inserts the newly formatted headlines.
3452          *
3453          * @param string $text
3454          * @param string $origText Original, untouched wikitext
3455          * @param boolean $isMain
3456          * @private
3457          */
3458         function formatHeadings( $text, $origText, $isMain=true ) {
3459                 global $wgMaxTocLevel, $wgContLang, $wgEnforceHtmlIds;
3460
3461                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3462                 $showEditLink = $this->mOptions->getEditSection();
3463
3464                 // Do not call quickUserCan unless necessary
3465                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3466                         $showEditLink = 0;
3467                 }
3468
3469                 # Inhibit editsection links if requested in the page
3470                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3471                         $showEditLink = 0;
3472                 }
3473
3474                 # Get all headlines for numbering them and adding funky stuff like [edit]
3475                 # links - this is for later, but we need the number of headlines right now
3476                 $matches = array();
3477                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3478
3479                 # if there are fewer than 4 headlines in the article, do not show TOC
3480                 # unless it's been explicitly enabled.
3481                 $enoughToc = $this->mShowToc &&
3482                         (($numMatches >= 4) || $this->mForceTocPosition);
3483
3484                 # Allow user to stipulate that a page should have a "new section"
3485                 # link added via __NEWSECTIONLINK__
3486                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3487                         $this->mOutput->setNewSection( true );
3488                 }
3489
3490                 # Allow user to remove the "new section"
3491                 # link via __NONEWSECTIONLINK__
3492                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3493                         $this->mOutput->hideNewSection( true );
3494                 }
3495
3496                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3497                 # override above conditions and always show TOC above first header
3498                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3499                         $this->mShowToc = true;
3500                         $enoughToc = true;
3501                 }
3502
3503                 # We need this to perform operations on the HTML
3504                 $sk = $this->mOptions->getSkin();
3505
3506                 # headline counter
3507                 $headlineCount = 0;
3508                 $numVisible = 0;
3509
3510                 # Ugh .. the TOC should have neat indentation levels which can be
3511                 # passed to the skin functions. These are determined here
3512                 $toc = '';
3513                 $full = '';
3514                 $head = array();
3515                 $sublevelCount = array();
3516                 $levelCount = array();
3517                 $toclevel = 0;
3518                 $level = 0;
3519                 $prevlevel = 0;
3520                 $toclevel = 0;
3521                 $prevtoclevel = 0;
3522                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3523                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3524                 $oldType = $this->mOutputType;
3525                 $this->setOutputType( self::OT_WIKI );
3526                 $frame = $this->getPreprocessor()->newFrame();
3527                 $root = $this->preprocessToDom( $origText );
3528                 $node = $root->getFirstChild();
3529                 $byteOffset = 0;
3530                 $tocraw = array();
3531
3532                 foreach( $matches[3] as $headline ) {
3533                         $isTemplate = false;
3534                         $titleText = false;
3535                         $sectionIndex = false;
3536                         $numbering = '';
3537                         $markerMatches = array();
3538                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3539                                 $serial = $markerMatches[1];
3540                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3541                                 $isTemplate = ($titleText != $baseTitleText);
3542                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3543                         }
3544
3545                         if( $toclevel ) {
3546                                 $prevlevel = $level;
3547                                 $prevtoclevel = $toclevel;
3548                         }
3549                         $level = $matches[1][$headlineCount];
3550
3551                         if ( $level > $prevlevel ) {
3552                                 # Increase TOC level
3553                                 $toclevel++;
3554                                 $sublevelCount[$toclevel] = 0;
3555                                 if( $toclevel<$wgMaxTocLevel ) {
3556                                         $prevtoclevel = $toclevel;
3557                                         $toc .= $sk->tocIndent();
3558                                         $numVisible++;
3559                                 }
3560                         }
3561                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3562                                 # Decrease TOC level, find level to jump to
3563
3564                                 for ($i = $toclevel; $i > 0; $i--) {
3565                                         if ( $levelCount[$i] == $level ) {
3566                                                 # Found last matching level
3567                                                 $toclevel = $i;
3568                                                 break;
3569                                         }
3570                                         elseif ( $levelCount[$i] < $level ) {
3571                                                 # Found first matching level below current level
3572                                                 $toclevel = $i + 1;
3573                                                 break;
3574                                         }
3575                                 }
3576                                 if( $i == 0 ) $toclevel = 1;
3577                                 if( $toclevel<$wgMaxTocLevel ) {
3578                                         if($prevtoclevel < $wgMaxTocLevel) {
3579                                                 # Unindent only if the previous toc level was shown :p
3580                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3581                                                 $prevtoclevel = $toclevel;
3582                                         } else {
3583                                                 $toc .= $sk->tocLineEnd();
3584                                         }
3585                                 }
3586                         }
3587                         else {
3588                                 # No change in level, end TOC line
3589                                 if( $toclevel<$wgMaxTocLevel ) {
3590                                         $toc .= $sk->tocLineEnd();
3591                                 }
3592                         }
3593
3594                         $levelCount[$toclevel] = $level;
3595
3596                         # count number of headlines for each level
3597                         @$sublevelCount[$toclevel]++;
3598                         $dot = 0;
3599                         for( $i = 1; $i <= $toclevel; $i++ ) {
3600                                 if( !empty( $sublevelCount[$i] ) ) {
3601                                         if( $dot ) {
3602                                                 $numbering .= '.';
3603                                         }
3604                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3605                                         $dot = 1;
3606                                 }
3607                         }
3608
3609                         # The safe header is a version of the header text safe to use for links
3610                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3611                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3612
3613                         # Remove link placeholders by the link text.
3614                         #     <!--LINK number-->
3615                         # turns into
3616                         #     link text with suffix
3617                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3618
3619                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3620                         $tocline = preg_replace(
3621                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3622                                 array( '',                          '<$1>'),
3623                                 $safeHeadline
3624                         );
3625                         $tocline = trim( $tocline );
3626
3627                         # For the anchor, strip out HTML-y stuff period
3628                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3629                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3630                         $safeHeadline = trim( $safeHeadline );
3631
3632                         # Save headline for section edit hint before it's escaped
3633                         $headlineHint = $safeHeadline;
3634
3635                         if ( $wgEnforceHtmlIds ) {
3636                                 $legacyHeadline = false;
3637                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3638                                         'noninitial' );
3639                         } else {
3640                                 # For reverse compatibility, provide an id that's
3641                                 # HTML4-compatible, like we used to.
3642                                 #
3643                                 # It may be worth noting, academically, that it's possible for
3644                                 # the legacy anchor to conflict with a non-legacy headline
3645                                 # anchor on the page.  In this case likely the "correct" thing
3646                                 # would be to either drop the legacy anchors or make sure
3647                                 # they're numbered first.  However, this would require people
3648                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3649                                 # manually, so let's not bother worrying about it.
3650                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3651                                         'noninitial' );
3652                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline, 'xml' );
3653
3654                                 if ( $legacyHeadline == $safeHeadline ) {
3655                                         # No reason to have both (in fact, we can't)
3656                                         $legacyHeadline = false;
3657                                 } elseif ( $legacyHeadline != Sanitizer::escapeId(
3658                                 $legacyHeadline, 'xml' ) ) {
3659                                         # The legacy id is invalid XML.  We used to allow this, but
3660                                         # there's no reason to do so anymore.  Backward
3661                                         # compatibility will fail slightly in this case, but it's
3662                                         # no big deal.
3663                                         $legacyHeadline = false;
3664                                 }
3665                         }
3666
3667                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3668                         # Does this apply to Unicode characters?  Because we aren't
3669                         # handling those here.
3670                         $arrayKey = strtolower( $safeHeadline );
3671                         if ( $legacyHeadline === false ) {
3672                                 $legacyArrayKey = false;
3673                         } else {
3674                                 $legacyArrayKey = strtolower( $legacyHeadline );
3675                         }
3676
3677                         # count how many in assoc. array so we can track dupes in anchors
3678                         if ( isset( $refers[$arrayKey] ) ) {
3679                                 $refers[$arrayKey]++;
3680                         } else {
3681                                 $refers[$arrayKey] = 1;
3682                         }
3683                         if ( isset( $refers[$legacyArrayKey] ) ) {
3684                                 $refers[$legacyArrayKey]++;
3685                         } else {
3686                                 $refers[$legacyArrayKey] = 1;
3687                         }
3688
3689                         # Don't number the heading if it is the only one (looks silly)
3690                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3691                                 # the two are different if the line contains a link
3692                                 $headline=$numbering . ' ' . $headline;
3693                         }
3694
3695                         # Create the anchor for linking from the TOC to the section
3696                         $anchor = $safeHeadline;
3697                         $legacyAnchor = $legacyHeadline;
3698                         if ( $refers[$arrayKey] > 1 ) {
3699                                 $anchor .= '_' . $refers[$arrayKey];
3700                         }
3701                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3702                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3703                         }
3704                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3705                                 $toc .= $sk->tocLine($anchor, $tocline,
3706                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3707                         }
3708
3709                         # Add the section to the section tree
3710                         # Find the DOM node for this header
3711                         while ( $node && !$isTemplate ) {
3712                                 if ( $node->getName() === 'h' ) {
3713                                         $bits = $node->splitHeading();
3714                                         if ( $bits['i'] == $sectionIndex )
3715                                                 break;
3716                                 }
3717                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3718                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3719                                 $node = $node->getNextSibling();
3720                         }
3721                         $tocraw[] = array(
3722                                 'toclevel' => $toclevel,
3723                                 'level' => $level,
3724                                 'line' => $tocline,
3725                                 'number' => $numbering,
3726                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3727                                 'fromtitle' => $titleText,
3728                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3729                                 'anchor' => $anchor,
3730                         );
3731
3732                         # give headline the correct <h#> tag
3733                         if( $showEditLink && $sectionIndex !== false ) {
3734                                 if( $isTemplate ) {
3735                                         # Put a T flag in the section identifier, to indicate to extractSections()
3736                                         # that sections inside <includeonly> should be counted.
3737                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3738                                 } else {
3739                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3740                                 }
3741                         } else {
3742                                 $editlink = '';
3743                         }
3744                         $head[$headlineCount] = $sk->makeHeadline( $level,
3745                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3746                                 $editlink, $legacyAnchor );
3747
3748                         $headlineCount++;
3749                 }
3750
3751                 $this->setOutputType( $oldType );
3752
3753                 # Never ever show TOC if no headers
3754                 if( $numVisible < 1 ) {
3755                         $enoughToc = false;
3756                 }
3757
3758                 if( $enoughToc ) {
3759                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3760                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3761                         }
3762                         $toc = $sk->tocList( $toc );
3763                         $this->mOutput->setTOCHTML( $toc );
3764                 }
3765
3766                 if ( $isMain ) {
3767                         $this->mOutput->setSections( $tocraw );
3768                 }
3769
3770                 # split up and insert constructed headlines
3771
3772                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3773                 $i = 0;
3774
3775                 foreach( $blocks as $block ) {
3776                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3777                                 # This is the [edit] link that appears for the top block of text when
3778                                 # section editing is enabled
3779
3780                                 # Disabled because it broke block formatting
3781                                 # For example, a bullet point in the top line
3782                                 # $full .= $sk->editSectionLink(0);
3783                         }
3784                         $full .= $block;
3785                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3786                                 # Top anchor now in skin
3787                                 $full = $full.$toc;
3788                         }
3789
3790                         if( !empty( $head[$i] ) ) {
3791                                 $full .= $head[$i];
3792                         }
3793                         $i++;
3794                 }
3795                 if( $this->mForceTocPosition ) {
3796                         return str_replace( '<!--MWTOC-->', $toc, $full );
3797                 } else {
3798                         return $full;
3799                 }
3800         }
3801
3802         /**
3803          * Merge $tree2 into $tree1 by replacing the section with index
3804          * $section in $tree1 and its descendants with the sections in $tree2.
3805          * Note that in the returned section tree, only the 'index' and
3806          * 'byteoffset' fields are guaranteed to be correct.
3807          * @param $tree1 array Section tree from ParserOutput::getSectons()
3808         *  @param $tree2 array Section tree
3809          * @param $section int Section index
3810          * @param $title Title Title both section trees come from
3811          * @param $len2 int Length of the original wikitext for $tree2
3812          * @return array Merged section tree
3813          */
3814         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3815                 global $wgContLang;
3816                 $newTree = array();
3817                 $targetLevel = false;
3818                 $merged = false;
3819                 $lastLevel = 1;
3820                 $nextIndex = 1;
3821                 $numbering = array( 0 );
3822                 $titletext = $title->getPrefixedDBkey();
3823                 foreach ( $tree1 as $s ) {
3824                         if ( $targetLevel !== false ) {
3825                                 if ( $s['level'] <= $targetLevel )
3826                                         // We've skipped enough
3827                                         $targetLevel = false;
3828                                 else
3829                                         continue;
3830                         }
3831                         if ( $s['index'] != $section ||
3832                                         $s['fromtitle'] != $titletext ) {
3833                                 self::incrementNumbering( $numbering,
3834                                         $s['toclevel'], $lastLevel );
3835
3836                                 // Rewrite index, byteoffset and number
3837                                 if ( $s['fromtitle'] == $titletext ) {
3838                                         $s['index'] = $nextIndex++;
3839                                         if ( $merged )
3840                                                 $s['byteoffset'] += $len2;
3841                                 }
3842                                 $s['number']  = implode( '.', array_map(
3843                                         array( $wgContLang, 'formatnum' ),
3844                                         $numbering ) );
3845                                 $lastLevel = $s['toclevel'];
3846                                 $newTree[] = $s;
3847                         } else {
3848                                 // We're at $section
3849                                 // Insert sections from $tree2 here
3850                                 foreach ( $tree2 as $s2 ) {
3851                                         // Rewrite the fields in $s2
3852                                         // before inserting it
3853                                         $s2['toclevel'] += $s['toclevel'] - 1;
3854                                         $s2['level'] += $s['level'] - 1;
3855                                         $s2['index'] = $nextIndex++;
3856                                         $s2['byteoffset'] += $s['byteoffset'];
3857
3858                                         self::incrementNumbering( $numbering,
3859                                                 $s2['toclevel'], $lastLevel );
3860                                         $s2['number']  = implode( '.', array_map(
3861                                                 array( $wgContLang, 'formatnum' ),
3862                                                 $numbering ) );
3863                                         $lastLevel = $s2['toclevel'];
3864                                         $newTree[] = $s2;
3865                                 }
3866                                 // Skip all descendants of $section in $tree1
3867                                 $targetLevel = $s['level'];
3868                                 $merged = true;
3869                         }
3870                 }
3871                 return $newTree;
3872         }
3873
3874         /**
3875          * Increment a section number. Helper function for mergeSectionTrees()
3876          * @param $number array Array representing a section number
3877          * @param $level int Current TOC level (depth)
3878          * @param $lastLevel int Level of previous TOC entry
3879          */
3880         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3881                 if ( $level > $lastLevel )
3882                         $number[$level - 1] = 1;
3883                 else if ( $level < $lastLevel ) {
3884                         foreach ( $number as $key => $unused )
3885                                 if ( $key >= $level )
3886                                         unset( $number[$key] );
3887                         $number[$level - 1]++;
3888                 } else
3889                         $number[$level - 1]++;
3890         }
3891
3892         /**
3893          * Transform wiki markup when saving a page by doing \r\n -> \n
3894          * conversion, substitting signatures, {{subst:}} templates, etc.
3895          *
3896          * @param string $text the text to transform
3897          * @param Title &$title the Title object for the current article
3898          * @param User $user the User object describing the current user
3899          * @param ParserOptions $options parsing options
3900          * @param bool $clearState whether to clear the parser state first
3901          * @return string the altered wiki markup
3902          * @public
3903          */
3904         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3905                 $this->mOptions = $options;
3906                 $this->setTitle( $title );
3907                 $this->setOutputType( self::OT_WIKI );
3908
3909                 if ( $clearState ) {
3910                         $this->clearState();
3911                 }
3912
3913                 $pairs = array(
3914                         "\r\n" => "\n",
3915                 );
3916                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3917                 $text = $this->pstPass2( $text, $user );
3918                 $text = $this->mStripState->unstripBoth( $text );
3919                 return $text;
3920         }
3921
3922         /**
3923          * Pre-save transform helper function
3924          * @private
3925          */
3926         function pstPass2( $text, $user ) {
3927                 global $wgContLang, $wgLocaltimezone;
3928
3929                 /* Note: This is the timestamp saved as hardcoded wikitext to
3930                  * the database, we use $wgContLang here in order to give
3931                  * everyone the same signature and use the default one rather
3932                  * than the one selected in each user's preferences.
3933                  *
3934                  * (see also bug 12815)
3935                  */
3936                 $ts = $this->mOptions->getTimestamp();
3937                 $tz = wfMsgForContent( 'timezone-utc' );
3938                 if ( isset( $wgLocaltimezone ) ) {
3939                         $unixts = wfTimestamp( TS_UNIX, $ts );
3940                         $oldtz = getenv( 'TZ' );
3941                         putenv( 'TZ='.$wgLocaltimezone );
3942                         $ts = date( 'YmdHis', $unixts );
3943                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3944
3945                         /* Allow translation of timezones trough wiki. date() can return
3946                          * whatever crap the system uses, localised or not, so we cannot
3947                          * ship premade translations.
3948                          */
3949                         $key = 'timezone-' . strtolower( trim( $tz ) );
3950                         $value = wfMsgForContent( $key );
3951                         if ( !wfEmptyMsg( $key, $value ) ) $tz = $value;
3952
3953                         putenv( 'TZ='.$oldtz );
3954                 }
3955
3956                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3957
3958                 # Variable replacement
3959                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3960                 $text = $this->replaceVariables( $text );
3961
3962                 # Signatures
3963                 $sigText = $this->getUserSig( $user );
3964                 $text = strtr( $text, array(
3965                         '~~~~~' => $d,
3966                         '~~~~' => "$sigText $d",
3967                         '~~~' => $sigText
3968                 ) );
3969
3970                 # Context links: [[|name]] and [[name (context)|]]
3971                 #
3972                 global $wgLegalTitleChars;
3973                 $tc = "[$wgLegalTitleChars]";
3974                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3975
3976                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3977                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3978                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3979                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3980
3981                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3982                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3983                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3984                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3985
3986                 $t = $this->mTitle->getText();
3987                 $m = array();
3988                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3989                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3990                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3991                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3992                 } else {
3993                         # if there's no context, don't bother duplicating the title
3994                         $text = preg_replace( $p2, '[[\\1]]', $text );
3995                 }
3996
3997                 # Trim trailing whitespace
3998                 $text = rtrim( $text );
3999
4000                 return $text;
4001         }
4002
4003         /**
4004          * Fetch the user's signature text, if any, and normalize to
4005          * validated, ready-to-insert wikitext.
4006          * If you have pre-fetched the nickname or the fancySig option, you can
4007          * specify them here to save a database query.
4008          *
4009          * @param User $user
4010          * @return string
4011          */
4012         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4013                 global $wgMaxSigChars;
4014
4015                 $username = $user->getName();
4016
4017                 // If not given, retrieve from the user object.
4018                 if ( $nickname === false )
4019                         $nickname = $user->getOption( 'nickname' );
4020
4021                 if ( is_null( $fancySig ) )
4022                         $fancySig = $user->getBoolOption( 'fancysig' );
4023
4024                 $nickname = $nickname == null ? $username : $nickname;
4025
4026                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4027                         $nickname = $username;
4028                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4029                 } elseif( $fancySig !== false ) {
4030                         # Sig. might contain markup; validate this
4031                         if( $this->validateSig( $nickname ) !== false ) {
4032                                 # Validated; clean up (if needed) and return it
4033                                 return $this->cleanSig( $nickname, true );
4034                         } else {
4035                                 # Failed to validate; fall back to the default
4036                                 $nickname = $username;
4037                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4038                         }
4039                 }
4040
4041                 // Make sure nickname doesnt get a sig in a sig
4042                 $nickname = $this->cleanSigInSig( $nickname );
4043
4044                 # If we're still here, make it a link to the user page
4045                 $userText = wfEscapeWikiText( $username );
4046                 $nickText = wfEscapeWikiText( $nickname );
4047                 if ( $user->isAnon() )  {
4048                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4049                 } else {
4050                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4051                 }
4052         }
4053
4054         /**
4055          * Check that the user's signature contains no bad XML
4056          *
4057          * @param string $text
4058          * @return mixed An expanded string, or false if invalid.
4059          */
4060         function validateSig( $text ) {
4061                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4062         }
4063
4064         /**
4065          * Clean up signature text
4066          *
4067          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4068          * 2) Substitute all transclusions
4069          *
4070          * @param string $text
4071          * @param $parsing Whether we're cleaning (preferences save) or parsing
4072          * @return string Signature text
4073          */
4074         function cleanSig( $text, $parsing = false ) {
4075                 if ( !$parsing ) {
4076                         global $wgTitle;
4077                         $this->clearState();
4078                         $this->setTitle( $wgTitle );
4079                         $this->mOptions = new ParserOptions;
4080                         $this->setOutputType = self::OT_PREPROCESS;
4081                 }
4082
4083                 # Option to disable this feature
4084                 if ( !$this->mOptions->getCleanSignatures() ) {
4085                         return $text;
4086                 }
4087
4088                 # FIXME: regex doesn't respect extension tags or nowiki
4089                 #  => Move this logic to braceSubstitution()
4090                 $substWord = MagicWord::get( 'subst' );
4091                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4092                 $substText = '{{' . $substWord->getSynonym( 0 );
4093
4094                 $text = preg_replace( $substRegex, $substText, $text );
4095                 $text = $this->cleanSigInSig( $text );
4096                 $dom = $this->preprocessToDom( $text );
4097                 $frame = $this->getPreprocessor()->newFrame();
4098                 $text = $frame->expand( $dom );
4099
4100                 if ( !$parsing ) {
4101                         $text = $this->mStripState->unstripBoth( $text );
4102                 }
4103
4104                 return $text;
4105         }
4106
4107         /**
4108          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4109          * @param string $text
4110          * @return string Signature text with /~{3,5}/ removed
4111          */
4112         function cleanSigInSig( $text ) {
4113                 $text = preg_replace( '/~{3,5}/', '', $text );
4114                 return $text;
4115         }
4116
4117         /**
4118          * Set up some variables which are usually set up in parse()
4119          * so that an external function can call some class members with confidence
4120          * @public
4121          */
4122         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4123                 $this->setTitle( $title );
4124                 $this->mOptions = $options;
4125                 $this->setOutputType( $outputType );
4126                 if ( $clearState ) {
4127                         $this->clearState();
4128                 }
4129         }
4130
4131         /**
4132          * Wrapper for preprocess()
4133          *
4134          * @param string $text the text to preprocess
4135          * @param ParserOptions $options  options
4136          * @return string
4137          * @public
4138          */
4139         function transformMsg( $text, $options ) {
4140                 global $wgTitle;
4141                 static $executing = false;
4142
4143                 # Guard against infinite recursion
4144                 if ( $executing ) {
4145                         return $text;
4146                 }
4147                 $executing = true;
4148
4149                 wfProfileIn(__METHOD__);
4150                 $text = $this->preprocess( $text, $wgTitle, $options );
4151
4152                 $executing = false;
4153                 wfProfileOut(__METHOD__);
4154                 return $text;
4155         }
4156
4157         /**
4158          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4159          * The callback should have the following form:
4160          *    function myParserHook( $text, $params, &$parser ) { ... }
4161          *
4162          * Transform and return $text. Use $parser for any required context, e.g. use
4163          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4164          *
4165          * @public
4166          *
4167          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4168          * @param mixed $callback The callback function (and object) to use for the tag
4169          *
4170          * @return The old value of the mTagHooks array associated with the hook
4171          */
4172         function setHook( $tag, $callback ) {
4173                 $tag = strtolower( $tag );
4174                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4175                 $this->mTagHooks[$tag] = $callback;
4176                 if( !in_array( $tag, $this->mStripList ) ) {
4177                         $this->mStripList[] = $tag;
4178                 }
4179
4180                 return $oldVal;
4181         }
4182
4183         function setTransparentTagHook( $tag, $callback ) {
4184                 $tag = strtolower( $tag );
4185                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4186                 $this->mTransparentTagHooks[$tag] = $callback;
4187
4188                 return $oldVal;
4189         }
4190
4191         /**
4192          * Remove all tag hooks
4193          */
4194         function clearTagHooks() {
4195                 $this->mTagHooks = array();
4196                 $this->mStripList = $this->mDefaultStripList;
4197         }
4198
4199         /**
4200          * Create a function, e.g. {{sum:1|2|3}}
4201          * The callback function should have the form:
4202          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4203          *
4204          * Or with SFH_OBJECT_ARGS:
4205          *    function myParserFunction( $parser, $frame, $args ) { ... }
4206          *
4207          * The callback may either return the text result of the function, or an array with the text
4208          * in element 0, and a number of flags in the other elements. The names of the flags are
4209          * specified in the keys. Valid flags are:
4210          *   found                     The text returned is valid, stop processing the template. This
4211          *                             is on by default.
4212          *   nowiki                    Wiki markup in the return value should be escaped
4213          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4214          *
4215          * @public
4216          *
4217          * @param string $id The magic word ID
4218          * @param mixed $callback The callback function (and object) to use
4219          * @param integer $flags a combination of the following flags:
4220          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4221          *
4222          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4223          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4224          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4225          *     the arguments, and to control the way they are expanded.
4226          *
4227          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4228          *     arguments, for instance:
4229          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4230          *
4231          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4232          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4233          *     working if/when this is changed.
4234          *
4235          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4236          *     expansion.
4237          *
4238          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4239          *     about the methods available in PPFrame and PPNode.
4240          *
4241          * @return The old callback function for this name, if any
4242          */
4243         function setFunctionHook( $id, $callback, $flags = 0 ) {
4244                 global $wgContLang;
4245
4246                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4247                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4248
4249                 # Add to function cache
4250                 $mw = MagicWord::get( $id );
4251                 if( !$mw )
4252                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4253
4254                 $synonyms = $mw->getSynonyms();
4255                 $sensitive = intval( $mw->isCaseSensitive() );
4256
4257                 foreach ( $synonyms as $syn ) {
4258                         # Case
4259                         if ( !$sensitive ) {
4260                                 $syn = $wgContLang->lc( $syn );
4261                         }
4262                         # Add leading hash
4263                         if ( !( $flags & SFH_NO_HASH ) ) {
4264                                 $syn = '#' . $syn;
4265                         }
4266                         # Remove trailing colon
4267                         if ( substr( $syn, -1, 1 ) === ':' ) {
4268                                 $syn = substr( $syn, 0, -1 );
4269                         }
4270                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4271                 }
4272                 return $oldVal;
4273         }
4274
4275         /**
4276          * Get all registered function hook identifiers
4277          *
4278          * @return array
4279          */
4280         function getFunctionHooks() {
4281                 return array_keys( $this->mFunctionHooks );
4282         }
4283
4284         /**
4285          * Create a tag function, e.g. <test>some stuff</test>.
4286          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4287          * Unlike parser functions, their content is not preprocessed.
4288          */
4289         function setFunctionTagHook( $tag, $callback, $flags ) {
4290                 $tag = strtolower( $tag );
4291                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4292                         $this->mFunctionTagHooks[$tag] : null;
4293                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4294
4295                 if( !in_array( $tag, $this->mStripList ) ) {
4296                         $this->mStripList[] = $tag;
4297                 }
4298
4299                 return $old;
4300         }
4301
4302         /**
4303          * FIXME: update documentation. makeLinkObj() is deprecated.
4304          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4305          * Placeholders created in Skin::makeLinkObj()
4306          * Returns an array of link CSS classes, indexed by PDBK.
4307          */
4308         function replaceLinkHolders( &$text, $options = 0 ) {
4309                 return $this->mLinkHolders->replace( $text );
4310         }
4311
4312         /**
4313          * Replace <!--LINK--> link placeholders with plain text of links
4314          * (not HTML-formatted).
4315          * @param string $text
4316          * @return string
4317          */
4318         function replaceLinkHoldersText( $text ) {
4319                 return $this->mLinkHolders->replaceText( $text );
4320         }
4321
4322         /**
4323          * Tag hook handler for 'pre'.
4324          */
4325         function renderPreTag( $text, $attribs ) {
4326                 // Backwards-compatibility hack
4327                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4328
4329                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4330                 return Xml::openElement( 'pre', $attribs ) .
4331                         Xml::escapeTagsOnly( $content ) .
4332                         '</pre>';
4333         }
4334
4335         /**
4336          * Renders an image gallery from a text with one line per image.
4337          * text labels may be given by using |-style alternative text. E.g.
4338          *   Image:one.jpg|The number "1"
4339          *   Image:tree.jpg|A tree
4340          * given as text will return the HTML of a gallery with two images,
4341          * labeled 'The number "1"' and
4342          * 'A tree'.
4343          */
4344         function renderImageGallery( $text, $params ) {
4345                 $ig = new ImageGallery();
4346                 $ig->setContextTitle( $this->mTitle );
4347                 $ig->setShowBytes( false );
4348                 $ig->setShowFilename( false );
4349                 $ig->setParser( $this );
4350                 $ig->setHideBadImages();
4351                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4352                 $ig->useSkin( $this->mOptions->getSkin() );
4353                 $ig->mRevisionId = $this->mRevisionId;
4354
4355                 if( isset( $params['caption'] ) ) {
4356                         $caption = $params['caption'];
4357                         $caption = htmlspecialchars( $caption );
4358                         $caption = $this->replaceInternalLinks( $caption );
4359                         $ig->setCaptionHtml( $caption );
4360                 }
4361                 if( isset( $params['perrow'] ) ) {
4362                         $ig->setPerRow( $params['perrow'] );
4363                 }
4364                 if( isset( $params['widths'] ) ) {
4365                         $ig->setWidths( $params['widths'] );
4366                 }
4367                 if( isset( $params['heights'] ) ) {
4368                         $ig->setHeights( $params['heights'] );
4369                 }
4370
4371                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4372
4373                 $lines = StringUtils::explode( "\n", $text );
4374                 foreach ( $lines as $line ) {
4375                         # match lines like these:
4376                         # Image:someimage.jpg|This is some image
4377                         $matches = array();
4378                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4379                         # Skip empty lines
4380                         if ( count( $matches ) == 0 ) {
4381                                 continue;
4382                         }
4383
4384                         if ( strpos( $matches[0], '%' ) !== false )
4385                                 $matches[1] = urldecode( $matches[1] );
4386                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4387                         $nt =& $tp;
4388                         if( is_null( $nt ) ) {
4389                                 # Bogus title. Ignore these so we don't bomb out later.
4390                                 continue;
4391                         }
4392                         if ( isset( $matches[3] ) ) {
4393                                 $label = $matches[3];
4394                         } else {
4395                                 $label = '';
4396                         }
4397
4398                         $html = $this->recursiveTagParse( trim( $label ) );
4399
4400                         $ig->add( $nt, $html );
4401
4402                         # Only add real images (bug #5586)
4403                         if ( $nt->getNamespace() == NS_FILE ) {
4404                                 $this->mOutput->addImage( $nt->getDBkey() );
4405                         }
4406                 }
4407                 return $ig->toHTML();
4408         }
4409
4410         function getImageParams( $handler ) {
4411                 if ( $handler ) {
4412                         $handlerClass = get_class( $handler );
4413                 } else {
4414                         $handlerClass = '';
4415                 }
4416                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4417                         // Initialise static lists
4418                         static $internalParamNames = array(
4419                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4420                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4421                                         'bottom', 'text-bottom' ),
4422                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4423                                         'upright', 'border', 'link', 'alt' ),
4424                         );
4425                         static $internalParamMap;
4426                         if ( !$internalParamMap ) {
4427                                 $internalParamMap = array();
4428                                 foreach ( $internalParamNames as $type => $names ) {
4429                                         foreach ( $names as $name ) {
4430                                                 $magicName = str_replace( '-', '_', "img_$name" );
4431                                                 $internalParamMap[$magicName] = array( $type, $name );
4432                                         }
4433                                 }
4434                         }
4435
4436                         // Add handler params
4437                         $paramMap = $internalParamMap;
4438                         if ( $handler ) {
4439                                 $handlerParamMap = $handler->getParamMap();
4440                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4441                                         $paramMap[$magic] = array( 'handler', $paramName );
4442                                 }
4443                         }
4444                         $this->mImageParams[$handlerClass] = $paramMap;
4445                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4446                 }
4447                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4448         }
4449
4450         /**
4451          * Parse image options text and use it to make an image
4452          * @param Title $title
4453          * @param string $options
4454          * @param LinkHolderArray $holders
4455          */
4456         function makeImage( $title, $options, $holders = false ) {
4457                 # Check if the options text is of the form "options|alt text"
4458                 # Options are:
4459                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4460                 #  * left       no resizing, just left align. label is used for alt= only
4461                 #  * right      same, but right aligned
4462                 #  * none       same, but not aligned
4463                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4464                 #  * center     center the image
4465                 #  * frame      Keep original image size, no magnify-button.
4466                 #  * framed     Same as "frame"
4467                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4468                 #  * upright    reduce width for upright images, rounded to full __0 px
4469                 #  * border     draw a 1px border around the image
4470                 #  * alt        Text for HTML alt attribute (defaults to empty)
4471                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4472                 # vertical-align values (no % or length right now):
4473                 #  * baseline
4474                 #  * sub
4475                 #  * super
4476                 #  * top
4477                 #  * text-top
4478                 #  * middle
4479                 #  * bottom
4480                 #  * text-bottom
4481
4482                 $parts = StringUtils::explode( "|", $options );
4483                 $sk = $this->mOptions->getSkin();
4484
4485                 # Give extensions a chance to select the file revision for us
4486                 $skip = $time = $descQuery = false;
4487                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4488
4489                 if ( $skip ) {
4490                         return $sk->link( $title );
4491                 }
4492
4493                 # Get the file
4494                 $imagename = $title->getDBkey();
4495                 $file = wfFindFile( $title, array( 'time' => $time ) );
4496                 # Get parameter map
4497                 $handler = $file ? $file->getHandler() : false;
4498
4499                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4500
4501                 # Process the input parameters
4502                 $caption = '';
4503                 $params = array( 'frame' => array(), 'handler' => array(),
4504                         'horizAlign' => array(), 'vertAlign' => array() );
4505                 foreach( $parts as $part ) {
4506                         $part = trim( $part );
4507                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4508                         $validated = false;
4509                         if( isset( $paramMap[$magicName] ) ) {
4510                                 list( $type, $paramName ) = $paramMap[$magicName];
4511
4512                                 // Special case; width and height come in one variable together
4513                                 if( $type === 'handler' && $paramName === 'width' ) {
4514                                         $m = array();
4515                                         # (bug 13500) In both cases (width/height and width only),
4516                                         # permit trailing "px" for backward compatibility.
4517                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4518                                                 $width = intval( $m[1] );
4519                                                 $height = intval( $m[2] );
4520                                                 if ( $handler->validateParam( 'width', $width ) ) {
4521                                                         $params[$type]['width'] = $width;
4522                                                         $validated = true;
4523                                                 }
4524                                                 if ( $handler->validateParam( 'height', $height ) ) {
4525                                                         $params[$type]['height'] = $height;
4526                                                         $validated = true;
4527                                                 }
4528                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4529                                                 $width = intval( $value );
4530                                                 if ( $handler->validateParam( 'width', $width ) ) {
4531                                                         $params[$type]['width'] = $width;
4532                                                         $validated = true;
4533                                                 }
4534                                         } // else no validation -- bug 13436
4535                                 } else {
4536                                         if ( $type === 'handler' ) {
4537                                                 # Validate handler parameter
4538                                                 $validated = $handler->validateParam( $paramName, $value );
4539                                         } else {
4540                                                 # Validate internal parameters
4541                                                 switch( $paramName ) {
4542                                                 case 'manualthumb':
4543                                                 case 'alt':
4544                                                         // @fixme - possibly check validity here for
4545                                                         // manualthumb? downstream behavior seems odd with
4546                                                         // missing manual thumbs.
4547                                                         $validated = true;
4548                                                         $value = $this->stripAltText( $value, $holders );
4549                                                         break;
4550                                                 case 'link':
4551                                                         $chars = self::EXT_LINK_URL_CLASS;
4552                                                         $prots = $this->mUrlProtocols;
4553                                                         if ( $value === '' ) {
4554                                                                 $paramName = 'no-link';
4555                                                                 $value = true;
4556                                                                 $validated = true;
4557                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4558                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4559                                                                         $paramName = 'link-url';
4560                                                                         $this->mOutput->addExternalLink( $value );
4561                                                                         $validated = true;
4562                                                                 }
4563                                                         } else {
4564                                                                 $linkTitle = Title::newFromText( $value );
4565                                                                 if ( $linkTitle ) {
4566                                                                         $paramName = 'link-title';
4567                                                                         $value = $linkTitle;
4568                                                                         $this->mOutput->addLink( $linkTitle );
4569                                                                         $validated = true;
4570                                                                 }
4571                                                         }
4572                                                         break;
4573                                                 default:
4574                                                         // Most other things appear to be empty or numeric...
4575                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4576                                                 }
4577                                         }
4578
4579                                         if ( $validated ) {
4580                                                 $params[$type][$paramName] = $value;
4581                                         }
4582                                 }
4583                         }
4584                         if ( !$validated ) {
4585                                 $caption = $part;
4586                         }
4587                 }
4588
4589                 # Process alignment parameters
4590                 if ( $params['horizAlign'] ) {
4591                         $params['frame']['align'] = key( $params['horizAlign'] );
4592                 }
4593                 if ( $params['vertAlign'] ) {
4594                         $params['frame']['valign'] = key( $params['vertAlign'] );
4595                 }
4596
4597                 $params['frame']['caption'] = $caption;
4598
4599                 # Will the image be presented in a frame, with the caption below?
4600                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4601                                  isset( $params['frame']['framed'] ) ||
4602                                  isset( $params['frame']['thumbnail'] ) ||
4603                                  isset( $params['frame']['manualthumb'] );
4604
4605                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4606                 # came to also set the caption, ordinary text after the image -- which
4607                 # makes no sense, because that just repeats the text multiple times in
4608                 # screen readers.  It *also* came to set the title attribute.
4609                 #
4610                 # Now that we have an alt attribute, we should not set the alt text to
4611                 # equal the caption: that's worse than useless, it just repeats the
4612                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4613                 # use the unnamed parameter for alt text as well, just for the time be-
4614                 # ing, if the unnamed param is set and the alt param is not.
4615                 #
4616                 # For the future, we need to figure out if we want to tweak this more,
4617                 # e.g., introducing a title= parameter for the title; ignoring the un-
4618                 # named parameter entirely for images without a caption; adding an ex-
4619                 # plicit caption= parameter and preserving the old magic unnamed para-
4620                 # meter for BC; ...
4621                 if ( $imageIsFramed ) { # Framed image
4622                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4623                                 # No caption or alt text, add the filename as the alt text so
4624                                 # that screen readers at least get some description of the image
4625                                 $params['frame']['alt'] = $title->getText();
4626                         }
4627                         # Do not set $params['frame']['title'] because tooltips don't make sense
4628                         # for framed images
4629                 } else { # Inline image
4630                         if ( !isset( $params['frame']['alt'] ) ) {
4631                                 # No alt text, use the "caption" for the alt text
4632                                 if ( $caption !== '') {
4633                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4634                                 } else {
4635                                         # No caption, fall back to using the filename for the
4636                                         # alt text
4637                                         $params['frame']['alt'] = $title->getText();
4638                                 }
4639                         }
4640                         # Use the "caption" for the tooltip text
4641                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4642                 }
4643
4644                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4645
4646                 # Linker does the rest
4647                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4648
4649                 # Give the handler a chance to modify the parser object
4650                 if ( $handler ) {
4651                         $handler->parserTransformHook( $this, $file );
4652                 }
4653
4654                 return $ret;
4655         }
4656
4657         protected function stripAltText( $caption, $holders ) {
4658                 # Strip bad stuff out of the title (tooltip).  We can't just use
4659                 # replaceLinkHoldersText() here, because if this function is called
4660                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4661                 if ( $holders ) {
4662                         $tooltip = $holders->replaceText( $caption );
4663                 } else {
4664                         $tooltip = $this->replaceLinkHoldersText( $caption );
4665                 }
4666
4667                 # make sure there are no placeholders in thumbnail attributes
4668                 # that are later expanded to html- so expand them now and
4669                 # remove the tags
4670                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4671                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4672
4673                 return $tooltip;
4674         }
4675
4676         /**
4677          * Set a flag in the output object indicating that the content is dynamic and
4678          * shouldn't be cached.
4679          */
4680         function disableCache() {
4681                 wfDebug( "Parser output marked as uncacheable.\n" );
4682                 $this->mOutput->mCacheTime = -1;
4683         }
4684
4685         /**#@+
4686          * Callback from the Sanitizer for expanding items found in HTML attribute
4687          * values, so they can be safely tested and escaped.
4688          * @param string $text
4689          * @param PPFrame $frame
4690          * @return string
4691          * @private
4692          */
4693         function attributeStripCallback( &$text, $frame = false ) {
4694                 $text = $this->replaceVariables( $text, $frame );
4695                 $text = $this->mStripState->unstripBoth( $text );
4696                 return $text;
4697         }
4698
4699         /**#@-*/
4700
4701         /**#@+
4702          * Accessor/mutator
4703          */
4704         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4705         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4706         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4707         /**#@-*/
4708
4709         /**#@+
4710          * Accessor
4711          */
4712         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4713         /**#@-*/
4714
4715
4716         /**
4717          * Break wikitext input into sections, and either pull or replace
4718          * some particular section's text.
4719          *
4720          * External callers should use the getSection and replaceSection methods.
4721          *
4722          * @param string $text Page wikitext
4723          * @param string $section A section identifier string of the form:
4724          *   <flag1> - <flag2> - ... - <section number>
4725          *
4726          * Currently the only recognised flag is "T", which means the target section number
4727          * was derived during a template inclusion parse, in other words this is a template
4728          * section edit link. If no flags are given, it was an ordinary section edit link.
4729          * This flag is required to avoid a section numbering mismatch when a section is
4730          * enclosed by <includeonly> (bug 6563).
4731          *
4732          * The section number 0 pulls the text before the first heading; other numbers will
4733          * pull the given section along with its lower-level subsections. If the section is
4734          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4735          *
4736          * @param string $mode One of "get" or "replace"
4737          * @param string $newText Replacement text for section data.
4738          * @return string for "get", the extracted section text.
4739          *                for "replace", the whole page with the section replaced.
4740          */
4741         private function extractSections( $text, $section, $mode, $newText='' ) {
4742                 global $wgTitle;
4743                 $this->clearState();
4744                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4745                 $this->mOptions = new ParserOptions;
4746                 $this->setOutputType( self::OT_WIKI );
4747                 $outText = '';
4748                 $frame = $this->getPreprocessor()->newFrame();
4749
4750                 // Process section extraction flags
4751                 $flags = 0;
4752                 $sectionParts = explode( '-', $section );
4753                 $sectionIndex = array_pop( $sectionParts );
4754                 foreach ( $sectionParts as $part ) {
4755                         if ( $part === 'T' ) {
4756                                 $flags |= self::PTD_FOR_INCLUSION;
4757                         }
4758                 }
4759                 // Preprocess the text
4760                 $root = $this->preprocessToDom( $text, $flags );
4761
4762                 // <h> nodes indicate section breaks
4763                 // They can only occur at the top level, so we can find them by iterating the root's children
4764                 $node = $root->getFirstChild();
4765
4766                 // Find the target section
4767                 if ( $sectionIndex == 0 ) {
4768                         // Section zero doesn't nest, level=big
4769                         $targetLevel = 1000;
4770                 } else {
4771             while ( $node ) {
4772                 if ( $node->getName() === 'h' ) {
4773                     $bits = $node->splitHeading();
4774                                         if ( $bits['i'] == $sectionIndex ) {
4775                                         $targetLevel = $bits['level'];
4776                                                 break;
4777                                         }
4778                                 }
4779                                 if ( $mode === 'replace' ) {
4780                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4781                                 }
4782                                 $node = $node->getNextSibling();
4783                         }
4784                 }
4785
4786                 if ( !$node ) {
4787                         // Not found
4788                         if ( $mode === 'get' ) {
4789                                 return $newText;
4790                         } else {
4791                                 return $text;
4792                         }
4793                 }
4794
4795                 // Find the end of the section, including nested sections
4796                 do {
4797                         if ( $node->getName() === 'h' ) {
4798                                 $bits = $node->splitHeading();
4799                                 $curLevel = $bits['level'];
4800                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4801                                         break;
4802                                 }
4803                         }
4804                         if ( $mode === 'get' ) {
4805                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4806                         }
4807                         $node = $node->getNextSibling();
4808                 } while ( $node );
4809
4810                 // Write out the remainder (in replace mode only)
4811                 if ( $mode === 'replace' ) {
4812                         // Output the replacement text
4813                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4814                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4815                         // Only add trailing whitespace if there is newText
4816                         if($newText != "") {
4817                                 $outText .= $newText . "\n\n";
4818                         }
4819
4820                         while ( $node ) {
4821                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4822                                 $node = $node->getNextSibling();
4823                         }
4824                 }
4825
4826                 if ( is_string( $outText ) ) {
4827                         // Re-insert stripped tags
4828                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4829                 }
4830
4831                 return $outText;
4832         }
4833
4834         /**
4835          * This function returns the text of a section, specified by a number ($section).
4836          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4837          * the first section before any such heading (section 0).
4838          *
4839          * If a section contains subsections, these are also returned.
4840          *
4841          * @param string $text text to look in
4842          * @param string $section section identifier
4843          * @param string $deftext default to return if section is not found
4844          * @return string text of the requested section
4845          */
4846         public function getSection( $text, $section, $deftext='' ) {
4847                 return $this->extractSections( $text, $section, "get", $deftext );
4848         }
4849
4850         public function replaceSection( $oldtext, $section, $text ) {
4851                 return $this->extractSections( $oldtext, $section, "replace", $text );
4852         }
4853
4854         /**
4855          * Get the timestamp associated with the current revision, adjusted for
4856          * the default server-local timestamp
4857          */
4858         function getRevisionTimestamp() {
4859                 if ( is_null( $this->mRevisionTimestamp ) ) {
4860                         wfProfileIn( __METHOD__ );
4861                         global $wgContLang;
4862                         $dbr = wfGetDB( DB_SLAVE );
4863                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4864                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4865
4866                         // Normalize timestamp to internal MW format for timezone processing.
4867                         // This has the added side-effect of replacing a null value with
4868                         // the current time, which gives us more sensible behavior for
4869                         // previews.
4870                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4871
4872                         // The cryptic '' timezone parameter tells to use the site-default
4873                         // timezone offset instead of the user settings.
4874                         //
4875                         // Since this value will be saved into the parser cache, served
4876                         // to other users, and potentially even used inside links and such,
4877                         // it needs to be consistent for all visitors.
4878                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4879
4880                         wfProfileOut( __METHOD__ );
4881                 }
4882                 return $this->mRevisionTimestamp;
4883         }
4884
4885         /**
4886          * Get the name of the user that edited the last revision
4887          */
4888         function getRevisionUser() {
4889                 // if this template is subst: the revision id will be blank,
4890                 // so just use the current user's name
4891                 if( $this->mRevisionId ) {
4892                         $revision = Revision::newFromId( $this->mRevisionId );
4893                         $revuser = $revision->getUserText();
4894                 } else {
4895                         global $wgUser;
4896                         $revuser = $wgUser->getName();
4897                 }
4898                 return $revuser;
4899         }
4900
4901         /**
4902          * Mutator for $mDefaultSort
4903          *
4904          * @param $sort New value
4905          */
4906         public function setDefaultSort( $sort ) {
4907                 $this->mDefaultSort = $sort;
4908         }
4909
4910         /**
4911          * Accessor for $mDefaultSort
4912          * Will use the title/prefixed title if none is set
4913          *
4914          * @return string
4915          */
4916         public function getDefaultSort() {
4917                 global $wgCategoryPrefixedDefaultSortkey;
4918                 if( $this->mDefaultSort !== false ) {
4919                         return $this->mDefaultSort;
4920                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4921                         !$wgCategoryPrefixedDefaultSortkey) {
4922                         return $this->mTitle->getText();
4923                 } else {
4924                         return $this->mTitle->getPrefixedText();
4925                 }
4926         }
4927
4928         /**
4929          * Accessor for $mDefaultSort
4930          * Unlike getDefaultSort(), will return false if none is set
4931          *
4932          * @return string or false
4933          */
4934         public function getCustomDefaultSort() {
4935                 return $this->mDefaultSort;
4936         }
4937
4938         /**
4939          * Try to guess the section anchor name based on a wikitext fragment
4940          * presumably extracted from a heading, for example "Header" from
4941          * "== Header ==".
4942          */
4943         public function guessSectionNameFromWikiText( $text ) {
4944                 # Strip out wikitext links(they break the anchor)
4945                 $text = $this->stripSectionName( $text );
4946                 $headline = Sanitizer::decodeCharReferences( $text );
4947                 # strip out HTML
4948                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4949                 $headline = trim( $headline );
4950                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4951                 $replacearray = array(
4952                         '%3A' => ':',
4953                         '%' => '.'
4954                 );
4955                 return str_replace(
4956                         array_keys( $replacearray ),
4957                         array_values( $replacearray ),
4958                         $sectionanchor );
4959         }
4960
4961         /**
4962          * Strips a text string of wikitext for use in a section anchor
4963          *
4964          * Accepts a text string and then removes all wikitext from the
4965          * string and leaves only the resultant text (i.e. the result of
4966          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4967          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4968          * to create valid section anchors by mimicing the output of the
4969          * parser when headings are parsed.
4970          *
4971          * @param $text string Text string to be stripped of wikitext
4972          * for use in a Section anchor
4973          * @return Filtered text string
4974          */
4975         public function stripSectionName( $text ) {
4976                 # Strip internal link markup
4977                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4978                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4979
4980                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4981                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4982                 # on how many empty links there are on the page - need to figure that out.
4983                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4984
4985                 # Parse wikitext quotes (italics & bold)
4986                 $text = $this->doQuotes($text);
4987
4988                 # Strip HTML tags
4989                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4990                 return $text;
4991         }
4992
4993         function srvus( $text ) {
4994                 return $this->testSrvus( $text, $this->mOutputType );
4995         }
4996
4997         /**
4998          * strip/replaceVariables/unstrip for preprocessor regression testing
4999          */
5000         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5001                 $this->clearState();
5002                 if ( ! ( $title instanceof Title ) ) {
5003                         $title = Title::newFromText( $title );
5004                 }
5005                 $this->mTitle = $title;
5006                 $this->mOptions = $options;
5007                 $this->setOutputType( $outputType );
5008                 $text = $this->replaceVariables( $text );
5009                 $text = $this->mStripState->unstripBoth( $text );
5010                 $text = Sanitizer::removeHTMLtags( $text );
5011                 return $text;
5012         }
5013
5014         function testPst( $text, $title, $options ) {
5015                 global $wgUser;
5016                 if ( ! ( $title instanceof Title ) ) {
5017                         $title = Title::newFromText( $title );
5018                 }
5019                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5020         }
5021
5022         function testPreprocess( $text, $title, $options ) {
5023                 if ( ! ( $title instanceof Title ) ) {
5024                         $title = Title::newFromText( $title );
5025                 }
5026                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5027         }
5028
5029         function markerSkipCallback( $s, $callback ) {
5030                 $i = 0;
5031                 $out = '';
5032                 while ( $i < strlen( $s ) ) {
5033                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5034                         if ( $markerStart === false ) {
5035                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5036                                 break;
5037                         } else {
5038                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5039                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5040                                 if ( $markerEnd === false ) {
5041                                         $out .= substr( $s, $markerStart );
5042                                         break;
5043                                 } else {
5044                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5045                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5046                                         $i = $markerEnd;
5047                                 }
5048                         }
5049                 }
5050                 return $out;
5051         }
5052
5053         function serialiseHalfParsedText( $text ) {
5054                 $data = array();
5055                 $data['text'] = $text;
5056
5057                 // First, find all strip markers, and store their
5058                 //  data in an array.
5059                 $stripState = new StripState;
5060                 $pos = 0;
5061                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
5062                         $end_pos += strlen( self::MARKER_SUFFIX );
5063                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5064
5065                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5066                                 $replaceArray = $stripState->general;
5067                                 $stripText = $this->mStripState->general->data[$marker];
5068                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5069                                 $replaceArray = $stripState->nowiki;
5070                                 $stripText = $this->mStripState->nowiki->data[$marker];
5071                         } else {
5072                                 throw new MWException( "Hanging strip marker: '$marker'." );
5073                         }
5074
5075                         $replaceArray->setPair( $marker, $stripText );
5076                         $pos = $end_pos;
5077                 }
5078                 $data['stripstate'] = $stripState;
5079
5080                 // Now, find all of our links, and store THEIR
5081                 //  data in an array! :)
5082                 $links = array( 'internal' => array(), 'interwiki' => array() );
5083                 $pos = 0;
5084
5085                 // Internal links
5086                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5087                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5088
5089                         $ns = trim($ns);
5090                         if (empty( $links['internal'][$ns] )) {
5091                                 $links['internal'][$ns] = array();
5092                         }
5093
5094                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5095                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5096                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5097                 }
5098
5099                 $pos = 0;
5100
5101                 // Interwiki links
5102                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5103                         $data = substr( $text, $start_pos );
5104                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5105                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5106                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5107                 }
5108
5109                 $data['linkholder'] = $links;
5110
5111                 return $data;
5112         }
5113
5114         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5115                 if (!$intPrefix)
5116                         $intPrefix = $this->getRandomString();
5117
5118                 // First, extract the strip state.
5119                 $stripState = $data['stripstate'];
5120                 $this->mStripState->general->merge( $stripState->general );
5121                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5122
5123                 // Now, extract the text, and renumber links
5124                 $text = $data['text'];
5125                 $links = $data['linkholder'];
5126
5127                 // Internal...
5128                 foreach( $links['internal'] as $ns => $nsLinks ) {
5129                         foreach( $nsLinks as $key => $entry ) {
5130                                 $newKey = $intPrefix . '-' . $key;
5131                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5132
5133                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5134                         }
5135                 }
5136
5137                 // Interwiki...
5138                 foreach( $links['interwiki'] as $key => $entry ) {
5139                         $newKey = "$intPrefix-$key";
5140                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5141
5142                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5143                 }
5144
5145                 // Should be good to go.
5146                 return $text;
5147         }
5148 }
5149
5150 /**
5151  * @todo document, briefly.
5152  * @ingroup Parser
5153  */
5154 class StripState {
5155         var $general, $nowiki;
5156
5157         function __construct() {
5158                 $this->general = new ReplacementArray;
5159                 $this->nowiki = new ReplacementArray;
5160         }
5161
5162         function unstripGeneral( $text ) {
5163                 wfProfileIn( __METHOD__ );
5164                 do {
5165                         $oldText = $text;
5166                         $text = $this->general->replace( $text );
5167                 } while ( $text !== $oldText );
5168                 wfProfileOut( __METHOD__ );
5169                 return $text;
5170         }
5171
5172         function unstripNoWiki( $text ) {
5173                 wfProfileIn( __METHOD__ );
5174                 do {
5175                         $oldText = $text;
5176                         $text = $this->nowiki->replace( $text );
5177                 } while ( $text !== $oldText );
5178                 wfProfileOut( __METHOD__ );
5179                 return $text;
5180         }
5181
5182         function unstripBoth( $text ) {
5183                 wfProfileIn( __METHOD__ );
5184                 do {
5185                         $oldText = $text;
5186                         $text = $this->general->replace( $text );
5187                         $text = $this->nowiki->replace( $text );
5188                 } while ( $text !== $oldText );
5189                 wfProfileOut( __METHOD__ );
5190                 return $text;
5191         }
5192 }
5193
5194 /**
5195  * @todo document, briefly.
5196  * @ingroup Parser
5197  */
5198 class OnlyIncludeReplacer {
5199         var $output = '';
5200
5201         function replace( $matches ) {
5202                 if ( substr( $matches[1], -1 ) === "\n" ) {
5203                         $this->output .= substr( $matches[1], 0, -1 );
5204                 } else {
5205                         $this->output .= $matches[1];
5206                 }
5207         }
5208 }