includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * preprocess()
  24  *   removes HTML comments and expands templates
  25  * cleanSig()
  26  *   Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *   Extracts sections from an article for section editing
  29  *
  30  * Globals used:
  31  *    objects:   $wgLang, $wgContLang
  32  *
  33  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  34  *
  35  * settings:
  36  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  37  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  38  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  39  *  $wgMaxArticleSize*
  40  *
  41  *  * only within ParserOptions
  42  * </pre>
  43  *
  44  * @ingroup Parser
  45  */
  46 class Parser
  47 {
  48         /**
  49          * Update this version number when the ParserOutput format
  50          * changes in an incompatible way, so the parser cache
  51          * can automatically discard old data.
  52          */
  53         const VERSION = '1.6.4';
  54
  55         # Flags for Parser::setFunctionHook
  56         # Also available as global constants from Defines.php
  57         const SFH_NO_HASH = 1;
  58         const SFH_OBJECT_ARGS = 2;
  59
  60         # Constants needed for external link processing
  61         # Everything except bracket, space, or control characters
  62         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  63         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  64                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  65
  66         // State constants for the definition list colon extraction
  67         const COLON_STATE_TEXT = 0;
  68         const COLON_STATE_TAG = 1;
  69         const COLON_STATE_TAGSTART = 2;
  70         const COLON_STATE_CLOSETAG = 3;
  71         const COLON_STATE_TAGSLASH = 4;
  72         const COLON_STATE_COMMENT = 5;
  73         const COLON_STATE_COMMENTDASH = 6;
  74         const COLON_STATE_COMMENTDASHDASH = 7;
  75
  76         // Flags for preprocessToDom
  77         const PTD_FOR_INCLUSION = 1;
  78
  79         // Allowed values for $this->mOutputType
  80         // Parameter to startExternalParse().
  81         const OT_HTML = 1;
  82         const OT_WIKI = 2;
  83         const OT_PREPROCESS = 3;
  84         const OT_MSG = 3;
  85
  86         // Marker Suffix needs to be accessible staticly.
  87         const MARKER_SUFFIX = "-QINU\x7f";
  88
  89         /**#@+
  90          * @private
  91          */
  92         # Persistent:
  93         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  94                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex, $mPreprocessor,
  95                 $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList, $mVarCache, $mConf;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mLinkHolders, $mLinkID;
 102         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 103         var $mTplExpandCache; // empty-frame expansion cache
 104         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 105         var $mExpensiveFunctionCount; // number of expensive parser function calls
 106         var $mFileCache;
 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->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 131                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 132                 $this->mUrlProtocols = wfUrlProtocols();
 133                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 134                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 135                 $this->mVarCache = array();
 136                 if ( isset( $conf['preprocessorClass'] ) ) {
 137                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 138                 } elseif ( extension_loaded( 'domxml' ) ) {
 139                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 140                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 141                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 142                 } elseif ( extension_loaded( 'dom' ) ) {
 143                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 144                 } else {
 145                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 146                 }
 147                 $this->mMarkerIndex = 0;
 148                 $this->mFirstCall = true;
 149         }
 150
 151         /**
 152          * Reduce memory usage to reduce the impact of circular references
 153          */
 154         function __destruct() {
 155                 if ( isset( $this->mLinkHolders ) ) {
 156                         $this->mLinkHolders->__destruct();
 157                 }
 158                 foreach ( $this as $name => $value ) {
 159                         unset( $this->$name );
 160                 }
 161         }
 162
 163         /**
 164          * Do various kinds of initialisation on the first call of the parser
 165          */
 166         function firstCallInit() {
 167                 if ( !$this->mFirstCall ) {
 168                         return;
 169                 }
 170                 $this->mFirstCall = false;
 171
 172                 wfProfileIn( __METHOD__ );
 173
 174                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 175                 CoreParserFunctions::register( $this );
 176                 $this->initialiseVariables();
 177
 178                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 179                 wfProfileOut( __METHOD__ );
 180         }
 181
 182         /**
 183          * Clear Parser state
 184          *
 185          * @private
 186          */
 187         function clearState() {
 188                 wfProfileIn( __METHOD__ );
 189                 if ( $this->mFirstCall ) {
 190                         $this->firstCallInit();
 191                 }
 192                 $this->mOutput = new ParserOutput;
 193                 $this->mAutonumber = 0;
 194                 $this->mLastSection = '';
 195                 $this->mDTopen = false;
 196                 $this->mIncludeCount = array();
 197                 $this->mStripState = new StripState;
 198                 $this->mArgStack = false;
 199                 $this->mInPre = false;
 200                 $this->mLinkHolders = new LinkHolderArray( $this );
 201                 $this->mLinkID = 0;
 202                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 203
 204                 /**
 205                  * Prefix for temporary replacement strings for the multipass parser.
 206                  * \x07 should never appear in input as it's disallowed in XML.
 207                  * Using it at the front also gives us a little extra robustness
 208                  * since it shouldn't match when butted up against identifier-like
 209                  * string constructs.
 210                  *
 211                  * Must not consist of all title characters, or else it will change
 212                  * the behaviour of <nowiki> in a link.
 213                  */
 214                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 215                 # Changed to \x7f to allow XML double-parsing -- TS
 216                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 217
 218
 219                 # Clear these on every parse, bug 4549
 220                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 221
 222                 $this->mShowToc = true;
 223                 $this->mForceTocPosition = false;
 224                 $this->mIncludeSizes = array(
 225                         'post-expand' => 0,
 226                         'arg' => 0,
 227                 );
 228                 $this->mPPNodeCount = 0;
 229                 $this->mDefaultSort = false;
 230                 $this->mHeadings = array();
 231                 $this->mDoubleUnderscores = array();
 232                 $this->mExpensiveFunctionCount = 0;
 233                 $this->mFileCache = array();
 234
 235                 # Fix cloning
 236                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 237                         $this->mPreprocessor = null;
 238                 }
 239
 240                 wfRunHooks( 'ParserClearState', array( &$this ) );
 241                 wfProfileOut( __METHOD__ );
 242         }
 243
 244         function setOutputType( $ot ) {
 245                 $this->mOutputType = $ot;
 246                 // Shortcut alias
 247                 $this->ot = array(
 248                         'html' => $ot == self::OT_HTML,
 249                         'wiki' => $ot == self::OT_WIKI,
 250                         'pre' => $ot == self::OT_PREPROCESS,
 251                 );
 252         }
 253
 254         /**
 255          * Set the context title
 256          */
 257         function setTitle( $t ) {
 258                 if ( !$t || $t instanceof FakeTitle ) {
 259                         $t = Title::newFromText( 'NO TITLE' );
 260                 }
 261                 if ( strval( $t->getFragment() ) !== '' ) {
 262                         # Strip the fragment to avoid various odd effects
 263                         $this->mTitle = clone $t;
 264                         $this->mTitle->setFragment( '' );
 265                 } else {
 266                         $this->mTitle = $t;
 267                 }
 268         }
 269
 270         /**
 271          * Accessor for mUniqPrefix.
 272          *
 273          * @public
 274          */
 275         function uniqPrefix() {
 276                 if( !isset( $this->mUniqPrefix ) ) {
 277                         // @fixme this is probably *horribly wrong*
 278                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 279                         // if this is called for a parser cache hit, the parser may not
 280                         // have ever been initialized in the first place.
 281                         // Not really sure what the heck is supposed to be going on here.
 282                         return '';
 283                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 284                 }
 285                 return $this->mUniqPrefix;
 286         }
 287
 288         /**
 289          * Convert wikitext to HTML
 290          * Do not call this function recursively.
 291          *
 292          * @param $text String: text we want to parse
 293          * @param $title A title object
 294          * @param $options ParserOptions
 295          * @param $linestart boolean
 296          * @param $clearState boolean
 297          * @param $revid Int: number to pass in {{REVISIONID}}
 298          * @return ParserOutput a ParserOutput
 299          */
 300         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 301                 /**
 302                  * First pass--just handle <nowiki> sections, pass the rest off
 303                  * to internalParse() which does all the real work.
 304                  */
 305
 306                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 307                 $fname = __METHOD__.'-' . wfGetCaller();
 308                 wfProfileIn( __METHOD__ );
 309                 wfProfileIn( $fname );
 310
 311                 if ( $clearState ) {
 312                         $this->clearState();
 313                 }
 314
 315                 $this->mOptions = $options;
 316                 $this->setTitle( $title );
 317                 $oldRevisionId = $this->mRevisionId;
 318                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 319                 if( $revid !== null ) {
 320                         $this->mRevisionId = $revid;
 321                         $this->mRevisionTimestamp = null;
 322                 }
 323                 $this->setOutputType( self::OT_HTML );
 324                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 325                 # No more strip!
 326                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 327                 $text = $this->internalParse( $text );
 328                 $text = $this->mStripState->unstripGeneral( $text );
 329
 330                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 331                 $fixtags = array(
 332                         # french spaces, last one Guillemet-left
 333                         # only if there is something before the space
 334                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 335                         # french spaces, Guillemet-right
 336                         '/(\\302\\253) /' => '\\1&nbsp;',
 337                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 338                 );
 339                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 340
 341                 $text = $this->doBlockLevels( $text, $linestart );
 342
 343                 $this->replaceLinkHolders( $text );
 344
 345                 # the position of the parserConvert() call should not be changed. it
 346                 # assumes that the links are all replaced and the only thing left
 347                 # is the <nowiki> mark.
 348                 # Side-effects: this calls $this->mOutput->setTitleText()
 349                 $text = $wgContLang->parserConvert( $text, $this );
 350
 351                 $text = $this->mStripState->unstripNoWiki( $text );
 352
 353                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 354
 355 //!JF Move to its own function
 356
 357                 $uniq_prefix = $this->mUniqPrefix;
 358                 $matches = array();
 359                 $elements = array_keys( $this->mTransparentTagHooks );
 360                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 361
 362                 foreach( $matches as $marker => $data ) {
 363                         list( $element, $content, $params, $tag ) = $data;
 364                         $tagName = strtolower( $element );
 365                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 366                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 367                                         array( $content, $params, $this ) );
 368                         } else {
 369                                 $output = $tag;
 370                         }
 371                         $this->mStripState->general->setPair( $marker, $output );
 372                 }
 373                 $text = $this->mStripState->unstripGeneral( $text );
 374
 375                 $text = Sanitizer::normalizeCharReferences( $text );
 376
 377                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 378                         $text = MWTidy::tidy( $text );
 379                 } else {
 380                         # attempt to sanitize at least some nesting problems
 381                         # (bug #2702 and quite a few others)
 382                         $tidyregs = array(
 383                                 # ''Something [http://www.cool.com cool''] -->
 384                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 385                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 386                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 387                                 # fix up an anchor inside another anchor, only
 388                                 # at least for a single single nested link (bug 3695)
 389                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 390                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 391                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 392                                 # contains a div, so fix it up here; replace
 393                                 # div with escaped text
 394                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 395                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 396                                 # remove empty italic or bold tag pairs, some
 397                                 # introduced by rules above
 398                                 '/<([bi])><\/\\1>/' => '',
 399                         );
 400
 401                         $text = preg_replace(
 402                                 array_keys( $tidyregs ),
 403                                 array_values( $tidyregs ),
 404                                 $text );
 405                 }
 406                 global $wgExpensiveParserFunctionLimit;
 407                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 408                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 409                 }
 410
 411                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 412
 413                 # Information on include size limits, for the benefit of users who try to skirt them
 414                 if ( $this->mOptions->getEnableLimitReport() ) {
 415                         global $wgExpensiveParserFunctionLimit;
 416                         $max = $this->mOptions->getMaxIncludeSize();
 417                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 418                         $limitReport =
 419                                 "NewPP limit report\n" .
 420                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 421                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 422                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 423                                 $PFreport;
 424                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 425                         $text .= "\n<!-- \n$limitReport-->\n";
 426                 }
 427                 $this->mOutput->setText( $text );
 428                 $this->mRevisionId = $oldRevisionId;
 429                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 430                 wfProfileOut( $fname );
 431                 wfProfileOut( __METHOD__ );
 432
 433                 return $this->mOutput;
 434         }
 435
 436         /**
 437          * Recursive parser entry point that can be called from an extension tag
 438          * hook.
 439          */
 440         function recursiveTagParse( $text ) {
 441                 wfProfileIn( __METHOD__ );
 442                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 443                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 444                 $text = $this->internalParse( $text );
 445                 wfProfileOut( __METHOD__ );
 446                 return $text;
 447         }
 448
 449         /**
 450          * Expand templates and variables in the text, producing valid, static wikitext.
 451          * Also removes comments.
 452          */
 453         function preprocess( $text, $title, $options, $revid = null ) {
 454                 wfProfileIn( __METHOD__ );
 455                 $this->clearState();
 456                 $this->setOutputType( self::OT_PREPROCESS );
 457                 $this->mOptions = $options;
 458                 $this->setTitle( $title );
 459                 if( $revid !== null ) {
 460                         $this->mRevisionId = $revid;
 461                 }
 462                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 463                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 464                 $text = $this->replaceVariables( $text );
 465                 $text = $this->mStripState->unstripBoth( $text );
 466                 wfProfileOut( __METHOD__ );
 467                 return $text;
 468         }
 469
 470         /**
 471          * Get a random string
 472          *
 473          * @private
 474          * @static
 475          */
 476         function getRandomString() {
 477                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 478         }
 479
 480         function &getTitle() { return $this->mTitle; }
 481         function getOptions() { return $this->mOptions; }
 482         function getRevisionId() { return $this->mRevisionId; }
 483         function getOutput() { return $this->mOutput; }
 484         function nextLinkID() { return $this->mLinkID++; }
 485
 486         function getFunctionLang() {
 487                 global $wgLang, $wgContLang;
 488
 489                 $target = $this->mOptions->getTargetLanguage();
 490                 if ( $target !== null ) {
 491                         return $target;
 492                 } else {
 493                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 494                 }
 495         }
 496
 497         /**
 498          * Get a preprocessor object
 499          */
 500         function getPreprocessor() {
 501                 if ( !isset( $this->mPreprocessor ) ) {
 502                         $class = $this->mPreprocessorClass;
 503                         $this->mPreprocessor = new $class( $this );
 504                 }
 505                 return $this->mPreprocessor;
 506         }
 507
 508         /**
 509          * Replaces all occurrences of HTML-style comments and the given tags
 510          * in the text with a random marker and returns the next text. The output
 511          * parameter $matches will be an associative array filled with data in
 512          * the form:
 513          *   'UNIQ-xxxxx' => array(
 514          *     'element',
 515          *     'tag content',
 516          *     array( 'param' => 'x' ),
 517          *     '<element param="x">tag content</element>' ) )
 518          *
 519          * @param $elements list of element names. Comments are always extracted.
 520          * @param $text Source text string.
 521          * @param $uniq_prefix
 522          *
 523          * @public
 524          * @static
 525          */
 526         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 527                 static $n = 1;
 528                 $stripped = '';
 529                 $matches = array();
 530
 531                 $taglist = implode( '|', $elements );
 532                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 533
 534                 while ( '' != $text ) {
 535                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 536                         $stripped .= $p[0];
 537                         if( count( $p ) < 5 ) {
 538                                 break;
 539                         }
 540                         if( count( $p ) > 5 ) {
 541                                 // comment
 542                                 $element    = $p[4];
 543                                 $attributes = '';
 544                                 $close      = '';
 545                                 $inside     = $p[5];
 546                         } else {
 547                                 // tag
 548                                 $element    = $p[1];
 549                                 $attributes = $p[2];
 550                                 $close      = $p[3];
 551                                 $inside     = $p[4];
 552                         }
 553
 554                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 555                         $stripped .= $marker;
 556
 557                         if ( $close === '/>' ) {
 558                                 // Empty element tag, <tag />
 559                                 $content = null;
 560                                 $text = $inside;
 561                                 $tail = null;
 562                         } else {
 563                                 if( $element === '!--' ) {
 564                                         $end = '/(-->)/';
 565                                 } else {
 566                                         $end = "/(<\\/$element\\s*>)/i";
 567                                 }
 568                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 569                                 $content = $q[0];
 570                                 if( count( $q ) < 3 ) {
 571                                         # No end tag -- let it run out to the end of the text.
 572                                         $tail = '';
 573                                         $text = '';
 574                                 } else {
 575                                         $tail = $q[1];
 576                                         $text = $q[2];
 577                                 }
 578                         }
 579
 580                         $matches[$marker] = array( $element,
 581                                 $content,
 582                                 Sanitizer::decodeTagAttributes( $attributes ),
 583                                 "<$element$attributes$close$content$tail" );
 584                 }
 585                 return $stripped;
 586         }
 587
 588         /**
 589          * Get a list of strippable XML-like elements
 590          */
 591         function getStripList() {
 592                 global $wgRawHtml;
 593                 $elements = $this->mStripList;
 594                 if( $wgRawHtml ) {
 595                         $elements[] = 'html';
 596                 }
 597                 if( $this->mOptions->getUseTeX() ) {
 598                         $elements[] = 'math';
 599                 }
 600                 return $elements;
 601         }
 602
 603         /**
 604          * @deprecated use replaceVariables
 605          */
 606         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 607                 return $text;
 608         }
 609
 610         /**
 611          * Restores pre, math, and other extensions removed by strip()
 612          *
 613          * always call unstripNoWiki() after this one
 614          * @private
 615          * @deprecated use $this->mStripState->unstrip()
 616          */
 617         function unstrip( $text, $state ) {
 618                 return $state->unstripGeneral( $text );
 619         }
 620
 621         /**
 622          * Always call this after unstrip() to preserve the order
 623          *
 624          * @private
 625          * @deprecated use $this->mStripState->unstrip()
 626          */
 627         function unstripNoWiki( $text, $state ) {
 628                 return $state->unstripNoWiki( $text );
 629         }
 630
 631         /**
 632          * @deprecated use $this->mStripState->unstripBoth()
 633          */
 634         function unstripForHTML( $text ) {
 635                 return $this->mStripState->unstripBoth( $text );
 636         }
 637
 638         /**
 639          * Add an item to the strip state
 640          * Returns the unique tag which must be inserted into the stripped text
 641          * The tag will be replaced with the original text in unstrip()
 642          *
 643          * @private
 644          */
 645         function insertStripItem( $text ) {
 646                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 647                 $this->mMarkerIndex++;
 648                 $this->mStripState->general->setPair( $rnd, $text );
 649                 return $rnd;
 650         }
 651
 652         /**
 653          * Interface with html tidy
 654          * @deprecated Use MWTidy::tidy()
 655          */
 656         public static function tidy( $text ) {
 657                 wfDeprecated( __METHOD__ );
 658                 return MWTidy::tidy( $text );
 659         }
 660
 661         /**
 662          * parse the wiki syntax used to render tables
 663          *
 664          * @private
 665          */
 666         function doTableStuff ( $text ) {
 667                 wfProfileIn( __METHOD__ );
 668
 669                 $lines = StringUtils::explode( "\n", $text );
 670                 $out = '';
 671                 $td_history = array (); // Is currently a td tag open?
 672                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 673                 $tr_history = array (); // Is currently a tr tag open?
 674                 $tr_attributes = array (); // history of tr attributes
 675                 $has_opened_tr = array(); // Did this table open a <tr> element?
 676                 $indent_level = 0; // indent level of the table
 677
 678                 foreach ( $lines as $outLine ) {
 679                         $line = trim( $outLine );
 680
 681                         if( $line == '' ) { // empty line, go to next line
 682                                 $out .= $outLine."\n";
 683                                 continue;
 684                         }
 685                         $first_character = $line[0];
 686                         $matches = array();
 687
 688                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 689                                 // First check if we are starting a new table
 690                                 $indent_level = strlen( $matches[1] );
 691
 692                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 693                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 694
 695                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 696                                 array_push ( $td_history , false );
 697                                 array_push ( $last_tag_history , '' );
 698                                 array_push ( $tr_history , false );
 699                                 array_push ( $tr_attributes , '' );
 700                                 array_push ( $has_opened_tr , false );
 701                         } else if ( count ( $td_history ) == 0 ) {
 702                                 // Don't do any of the following
 703                                 $out .= $outLine."\n";
 704                                 continue;
 705                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 706                                 // We are ending a table
 707                                 $line = '</table>' . substr ( $line , 2 );
 708                                 $last_tag = array_pop ( $last_tag_history );
 709
 710                                 if ( !array_pop ( $has_opened_tr ) ) {
 711                                         $line = "<tr><td></td></tr>{$line}";
 712                                 }
 713
 714                                 if ( array_pop ( $tr_history ) ) {
 715                                         $line = "</tr>{$line}";
 716                                 }
 717
 718                                 if ( array_pop ( $td_history ) ) {
 719                                         $line = "</{$last_tag}>{$line}";
 720                                 }
 721                                 array_pop ( $tr_attributes );
 722                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 723                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 724                                 // Now we have a table row
 725                                 $line = preg_replace( '#^\|-+#', '', $line );
 726
 727                                 // Whats after the tag is now only attributes
 728                                 $attributes = $this->mStripState->unstripBoth( $line );
 729                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 730                                 array_pop ( $tr_attributes );
 731                                 array_push ( $tr_attributes , $attributes );
 732
 733                                 $line = '';
 734                                 $last_tag = array_pop ( $last_tag_history );
 735                                 array_pop ( $has_opened_tr );
 736                                 array_push ( $has_opened_tr , true );
 737
 738                                 if ( array_pop ( $tr_history ) ) {
 739                                         $line = '</tr>';
 740                                 }
 741
 742                                 if ( array_pop ( $td_history ) ) {
 743                                         $line = "</{$last_tag}>{$line}";
 744                                 }
 745
 746                                 $outLine = $line;
 747                                 array_push ( $tr_history , false );
 748                                 array_push ( $td_history , false );
 749                                 array_push ( $last_tag_history , '' );
 750                         }
 751                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 752                                 // This might be cell elements, td, th or captions
 753                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 754                                         $first_character = '+';
 755                                         $line = substr ( $line , 1 );
 756                                 }
 757
 758                                 $line = substr ( $line , 1 );
 759
 760                                 if ( $first_character === '!' ) {
 761                                         $line = str_replace ( '!!' , '||' , $line );
 762                                 }
 763
 764                                 // Split up multiple cells on the same line.
 765                                 // FIXME : This can result in improper nesting of tags processed
 766                                 // by earlier parser steps, but should avoid splitting up eg
 767                                 // attribute values containing literal "||".
 768                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 769
 770                                 $outLine = '';
 771
 772                                 // Loop through each table cell
 773                                 foreach ( $cells as $cell )
 774                                 {
 775                                         $previous = '';
 776                                         if ( $first_character !== '+' )
 777                                         {
 778                                                 $tr_after = array_pop ( $tr_attributes );
 779                                                 if ( !array_pop ( $tr_history ) ) {
 780                                                         $previous = "<tr{$tr_after}>\n";
 781                                                 }
 782                                                 array_push ( $tr_history , true );
 783                                                 array_push ( $tr_attributes , '' );
 784                                                 array_pop ( $has_opened_tr );
 785                                                 array_push ( $has_opened_tr , true );
 786                                         }
 787
 788                                         $last_tag = array_pop ( $last_tag_history );
 789
 790                                         if ( array_pop ( $td_history ) ) {
 791                                                 $previous = "</{$last_tag}>{$previous}";
 792                                         }
 793
 794                                         if ( $first_character === '|' ) {
 795                                                 $last_tag = 'td';
 796                                         } else if ( $first_character === '!' ) {
 797                                                 $last_tag = 'th';
 798                                         } else if ( $first_character === '+' ) {
 799                                                 $last_tag = 'caption';
 800                                         } else {
 801                                                 $last_tag = '';
 802                                         }
 803
 804                                         array_push ( $last_tag_history , $last_tag );
 805
 806                                         // A cell could contain both parameters and data
 807                                         $cell_data = explode ( '|' , $cell , 2 );
 808
 809                                         // Bug 553: Note that a '|' inside an invalid link should not
 810                                         // be mistaken as delimiting cell parameters
 811                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 812                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 813                                         } else if ( count ( $cell_data ) == 1 )
 814                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 815                                         else {
 816                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 817                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 818                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 819                                         }
 820
 821                                         $outLine .= $cell;
 822                                         array_push ( $td_history , true );
 823                                 }
 824                         }
 825                         $out .= $outLine . "\n";
 826                 }
 827
 828                 // Closing open td, tr && table
 829                 while ( count ( $td_history ) > 0 )
 830                 {
 831                         if ( array_pop ( $td_history ) ) {
 832                                 $out .= "</td>\n";
 833                         }
 834                         if ( array_pop ( $tr_history ) ) {
 835                                 $out .= "</tr>\n";
 836                         }
 837                         if ( !array_pop ( $has_opened_tr ) ) {
 838                                 $out .= "<tr><td></td></tr>\n" ;
 839                         }
 840
 841                         $out .= "</table>\n";
 842                 }
 843
 844                 // Remove trailing line-ending (b/c)
 845                 if ( substr( $out, -1 ) === "\n" ) {
 846                         $out = substr( $out, 0, -1 );
 847                 }
 848
 849                 // special case: don't return empty table
 850                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 851                         $out = '';
 852                 }
 853
 854                 wfProfileOut( __METHOD__ );
 855
 856                 return $out;
 857         }
 858
 859         /**
 860          * Helper function for parse() that transforms wiki markup into
 861          * HTML. Only called for $mOutputType == self::OT_HTML.
 862          *
 863          * @private
 864          */
 865         function internalParse( $text ) {
 866                 $isMain = true;
 867                 wfProfileIn( __METHOD__ );
 868
 869                 # Hook to suspend the parser in this state
 870                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 871                         wfProfileOut( __METHOD__ );
 872                         return $text ;
 873                 }
 874
 875                 $text = $this->replaceVariables( $text );
 876                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 877                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 878
 879                 // Tables need to come after variable replacement for things to work
 880                 // properly; putting them before other transformations should keep
 881                 // exciting things like link expansions from showing up in surprising
 882                 // places.
 883                 $text = $this->doTableStuff( $text );
 884
 885                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 886
 887                 $text = $this->doDoubleUnderscore( $text );
 888                 $text = $this->doHeadings( $text );
 889                 if( $this->mOptions->getUseDynamicDates() ) {
 890                         $df = DateFormatter::getInstance();
 891                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 892                 }
 893                 $text = $this->doAllQuotes( $text );
 894                 $text = $this->replaceInternalLinks( $text );
 895                 $text = $this->replaceExternalLinks( $text );
 896
 897                 # replaceInternalLinks may sometimes leave behind
 898                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 899                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 900
 901                 $text = $this->doMagicLinks( $text );
 902                 $text = $this->formatHeadings( $text, $isMain );
 903
 904                 wfProfileOut( __METHOD__ );
 905                 return $text;
 906         }
 907
 908         /**
 909          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 910          * magic external links.
 911          *
 912          * DML
 913          * @private
 914          */
 915         function doMagicLinks( $text ) {
 916                 wfProfileIn( __METHOD__ );
 917                 $prots = $this->mUrlProtocols;
 918                 $urlChar = self::EXT_LINK_URL_CLASS;
 919                 $text = preg_replace_callback(
 920                         '!(?:                           # Start cases
 921                                 (<a.*?</a>) |               # m[1]: Skip link text
 922                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 923                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 924                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 925                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 926                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 927                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 928                                     [0-9Xx]                 # check digit
 929                                     \b)
 930                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 931                 wfProfileOut( __METHOD__ );
 932                 return $text;
 933         }
 934
 935         function magicLinkCallback( $m ) {
 936                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 937                         # Skip anchor
 938                         return $m[0];
 939                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 940                         # Skip HTML element
 941                         return $m[0];
 942                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 943                         # Free external link
 944                         return $this->makeFreeExternalLink( $m[0] );
 945                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 946                         # RFC or PMID
 947                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 948                                 $keyword = 'RFC';
 949                                 $urlmsg = 'rfcurl';
 950                                 $id = $m[4];
 951                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
 952                                 $keyword = 'PMID';
 953                                 $urlmsg = 'pubmedurl';
 954                                 $id = $m[4];
 955                         } else {
 956                                 throw new MWException( __METHOD__.': unrecognised match type "' .
 957                                         substr($m[0], 0, 20 ) . '"' );
 958                         }
 959                         $url = wfMsg( $urlmsg, $id);
 960                         $sk = $this->mOptions->getSkin();
 961                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
 962                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
 963                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
 964                         # ISBN
 965                         $isbn = $m[5];
 966                         $num = strtr( $isbn, array(
 967                                 '-' => '',
 968                                 ' ' => '',
 969                                 'x' => 'X',
 970                         ));
 971                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
 972                         return'<a href="' .
 973                                 $titleObj->escapeLocalUrl() .
 974                                 "\" class=\"internal\">ISBN $isbn</a>";
 975                 } else {
 976                         return $m[0];
 977                 }
 978         }
 979
 980         /**
 981          * Make a free external link, given a user-supplied URL
 982          * @return HTML
 983          * @private
 984          */
 985         function makeFreeExternalLink( $url ) {
 986                 global $wgContLang;
 987                 wfProfileIn( __METHOD__ );
 988
 989                 $sk = $this->mOptions->getSkin();
 990                 $trail = '';
 991
 992                 # The characters '<' and '>' (which were escaped by
 993                 # removeHTMLtags()) should not be included in
 994                 # URLs, per RFC 2396.
 995                 $m2 = array();
 996                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
 997                         $trail = substr($url, $m2[0][1]) . $trail;
 998                         $url = substr($url, 0, $m2[0][1]);
 999                 }
1000
1001                 # Move trailing punctuation to $trail
1002                 $sep = ',;\.:!?';
1003                 # If there is no left bracket, then consider right brackets fair game too
1004                 if ( strpos( $url, '(' ) === false ) {
1005                         $sep .= ')';
1006                 }
1007
1008                 $numSepChars = strspn( strrev( $url ), $sep );
1009                 if ( $numSepChars ) {
1010                         $trail = substr( $url, -$numSepChars ) . $trail;
1011                         $url = substr( $url, 0, -$numSepChars );
1012                 }
1013
1014                 $url = Sanitizer::cleanUrl( $url );
1015
1016                 # Is this an external image?
1017                 $text = $this->maybeMakeExternalImage( $url );
1018                 if ( $text === false ) {
1019                         # Not an image, make a link
1020                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1021                                 $this->getExternalLinkAttribs( $url ) );
1022                         # Register it in the output object...
1023                         # Replace unnecessary URL escape codes with their equivalent characters
1024                         $pasteurized = self::replaceUnusualEscapes( $url );
1025                         $this->mOutput->addExternalLink( $pasteurized );
1026                 }
1027                 wfProfileOut( __METHOD__ );
1028                 return $text . $trail;
1029         }
1030
1031
1032         /**
1033          * Parse headers and return html
1034          *
1035          * @private
1036          */
1037         function doHeadings( $text ) {
1038                 wfProfileIn( __METHOD__ );
1039                 for ( $i = 6; $i >= 1; --$i ) {
1040                         $h = str_repeat( '=', $i );
1041                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1042                           "<h$i>\\1</h$i>", $text );
1043                 }
1044                 wfProfileOut( __METHOD__ );
1045                 return $text;
1046         }
1047
1048         /**
1049          * Replace single quotes with HTML markup
1050          * @private
1051          * @return string the altered text
1052          */
1053         function doAllQuotes( $text ) {
1054                 wfProfileIn( __METHOD__ );
1055                 $outtext = '';
1056                 $lines = StringUtils::explode( "\n", $text );
1057                 foreach ( $lines as $line ) {
1058                         $outtext .= $this->doQuotes( $line ) . "\n";
1059                 }
1060                 $outtext = substr($outtext, 0,-1);
1061                 wfProfileOut( __METHOD__ );
1062                 return $outtext;
1063         }
1064
1065         /**
1066          * Helper function for doAllQuotes()
1067          */
1068         public function doQuotes( $text ) {
1069                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1070                 if ( count( $arr ) == 1 )
1071                         return $text;
1072                 else
1073                 {
1074                         # First, do some preliminary work. This may shift some apostrophes from
1075                         # being mark-up to being text. It also counts the number of occurrences
1076                         # of bold and italics mark-ups.
1077                         $i = 0;
1078                         $numbold = 0;
1079                         $numitalics = 0;
1080                         foreach ( $arr as $r )
1081                         {
1082                                 if ( ( $i % 2 ) == 1 )
1083                                 {
1084                                         # If there are ever four apostrophes, assume the first is supposed to
1085                                         # be text, and the remaining three constitute mark-up for bold text.
1086                                         if ( strlen( $arr[$i] ) == 4 )
1087                                         {
1088                                                 $arr[$i-1] .= "'";
1089                                                 $arr[$i] = "'''";
1090                                         }
1091                                         # If there are more than 5 apostrophes in a row, assume they're all
1092                                         # text except for the last 5.
1093                                         else if ( strlen( $arr[$i] ) > 5 )
1094                                         {
1095                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1096                                                 $arr[$i] = "'''''";
1097                                         }
1098                                         # Count the number of occurrences of bold and italics mark-ups.
1099                                         # We are not counting sequences of five apostrophes.
1100                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1101                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1102                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1103                                 }
1104                                 $i++;
1105                         }
1106
1107                         # If there is an odd number of both bold and italics, it is likely
1108                         # that one of the bold ones was meant to be an apostrophe followed
1109                         # by italics. Which one we cannot know for certain, but it is more
1110                         # likely to be one that has a single-letter word before it.
1111                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1112                         {
1113                                 $i = 0;
1114                                 $firstsingleletterword = -1;
1115                                 $firstmultiletterword = -1;
1116                                 $firstspace = -1;
1117                                 foreach ( $arr as $r )
1118                                 {
1119                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1120                                         {
1121                                                 $x1 = substr ($arr[$i-1], -1);
1122                                                 $x2 = substr ($arr[$i-1], -2, 1);
1123                                                 if ($x1 === ' ') {
1124                                                         if ($firstspace == -1) $firstspace = $i;
1125                                                 } else if ($x2 === ' ') {
1126                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1127                                                 } else {
1128                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1129                                                 }
1130                                         }
1131                                         $i++;
1132                                 }
1133
1134                                 # If there is a single-letter word, use it!
1135                                 if ($firstsingleletterword > -1)
1136                                 {
1137                                         $arr [ $firstsingleletterword ] = "''";
1138                                         $arr [ $firstsingleletterword-1 ] .= "'";
1139                                 }
1140                                 # If not, but there's a multi-letter word, use that one.
1141                                 else if ($firstmultiletterword > -1)
1142                                 {
1143                                         $arr [ $firstmultiletterword ] = "''";
1144                                         $arr [ $firstmultiletterword-1 ] .= "'";
1145                                 }
1146                                 # ... otherwise use the first one that has neither.
1147                                 # (notice that it is possible for all three to be -1 if, for example,
1148                                 # there is only one pentuple-apostrophe in the line)
1149                                 else if ($firstspace > -1)
1150                                 {
1151                                         $arr [ $firstspace ] = "''";
1152                                         $arr [ $firstspace-1 ] .= "'";
1153                                 }
1154                         }
1155
1156                         # Now let's actually convert our apostrophic mush to HTML!
1157                         $output = '';
1158                         $buffer = '';
1159                         $state = '';
1160                         $i = 0;
1161                         foreach ($arr as $r)
1162                         {
1163                                 if (($i % 2) == 0)
1164                                 {
1165                                         if ($state === 'both')
1166                                                 $buffer .= $r;
1167                                         else
1168                                                 $output .= $r;
1169                                 }
1170                                 else
1171                                 {
1172                                         if (strlen ($r) == 2)
1173                                         {
1174                                                 if ($state === 'i')
1175                                                 { $output .= '</i>'; $state = ''; }
1176                                                 else if ($state === 'bi')
1177                                                 { $output .= '</i>'; $state = 'b'; }
1178                                                 else if ($state === 'ib')
1179                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1180                                                 else if ($state === 'both')
1181                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1182                                                 else # $state can be 'b' or ''
1183                                                 { $output .= '<i>'; $state .= 'i'; }
1184                                         }
1185                                         else if (strlen ($r) == 3)
1186                                         {
1187                                                 if ($state === 'b')
1188                                                 { $output .= '</b>'; $state = ''; }
1189                                                 else if ($state === 'bi')
1190                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1191                                                 else if ($state === 'ib')
1192                                                 { $output .= '</b>'; $state = 'i'; }
1193                                                 else if ($state === 'both')
1194                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1195                                                 else # $state can be 'i' or ''
1196                                                 { $output .= '<b>'; $state .= 'b'; }
1197                                         }
1198                                         else if (strlen ($r) == 5)
1199                                         {
1200                                                 if ($state === 'b')
1201                                                 { $output .= '</b><i>'; $state = 'i'; }
1202                                                 else if ($state === 'i')
1203                                                 { $output .= '</i><b>'; $state = 'b'; }
1204                                                 else if ($state === 'bi')
1205                                                 { $output .= '</i></b>'; $state = ''; }
1206                                                 else if ($state === 'ib')
1207                                                 { $output .= '</b></i>'; $state = ''; }
1208                                                 else if ($state === 'both')
1209                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1210                                                 else # ($state == '')
1211                                                 { $buffer = ''; $state = 'both'; }
1212                                         }
1213                                 }
1214                                 $i++;
1215                         }
1216                         # Now close all remaining tags.  Notice that the order is important.
1217                         if ($state === 'b' || $state === 'ib')
1218                                 $output .= '</b>';
1219                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1220                                 $output .= '</i>';
1221                         if ($state === 'bi')
1222                                 $output .= '</b>';
1223                         # There might be lonely ''''', so make sure we have a buffer
1224                         if ($state === 'both' && $buffer)
1225                                 $output .= '<b><i>'.$buffer.'</i></b>';
1226                         return $output;
1227                 }
1228         }
1229
1230         /**
1231          * Replace external links (REL)
1232          *
1233          * Note: this is all very hackish and the order of execution matters a lot.
1234          * Make sure to run maintenance/parserTests.php if you change this code.
1235          *
1236          * @private
1237          */
1238         function replaceExternalLinks( $text ) {
1239                 global $wgContLang;
1240                 wfProfileIn( __METHOD__ );
1241
1242                 $sk = $this->mOptions->getSkin();
1243
1244                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1245                 $s = array_shift( $bits );
1246
1247                 $i = 0;
1248                 while ( $i<count( $bits ) ) {
1249                         $url = $bits[$i++];
1250                         $protocol = $bits[$i++];
1251                         $text = $bits[$i++];
1252                         $trail = $bits[$i++];
1253
1254                         # The characters '<' and '>' (which were escaped by
1255                         # removeHTMLtags()) should not be included in
1256                         # URLs, per RFC 2396.
1257                         $m2 = array();
1258                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1259                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1260                                 $url = substr($url, 0, $m2[0][1]);
1261                         }
1262
1263                         # If the link text is an image URL, replace it with an <img> tag
1264                         # This happened by accident in the original parser, but some people used it extensively
1265                         $img = $this->maybeMakeExternalImage( $text );
1266                         if ( $img !== false ) {
1267                                 $text = $img;
1268                         }
1269
1270                         $dtrail = '';
1271
1272                         # Set linktype for CSS - if URL==text, link is essentially free
1273                         $linktype = ($text === $url) ? 'free' : 'text';
1274
1275                         # No link text, e.g. [http://domain.tld/some.link]
1276                         if ( $text == '' ) {
1277                                 # Autonumber if allowed. See bug #5918
1278                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1279                                         $langObj = $this->getFunctionLang();
1280                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1281                                         $linktype = 'autonumber';
1282                                 } else {
1283                                         # Otherwise just use the URL
1284                                         $text = htmlspecialchars( $url );
1285                                         $linktype = 'free';
1286                                 }
1287                         } else {
1288                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1289                                 # Check for trail
1290                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1291                         }
1292
1293                         $text = $wgContLang->markNoConversion($text);
1294
1295                         $url = Sanitizer::cleanUrl( $url );
1296
1297                         # Use the encoded URL
1298                         # This means that users can paste URLs directly into the text
1299                         # Funny characters like &ouml; aren't valid in URLs anyway
1300                         # This was changed in August 2004
1301                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1302                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1303
1304                         # Register link in the output object.
1305                         # Replace unnecessary URL escape codes with the referenced character
1306                         # This prevents spammers from hiding links from the filters
1307                         $pasteurized = self::replaceUnusualEscapes( $url );
1308                         $this->mOutput->addExternalLink( $pasteurized );
1309                 }
1310
1311                 wfProfileOut( __METHOD__ );
1312                 return $s;
1313         }
1314
1315         /**
1316          * Get an associative array of additional HTML attributes appropriate for a
1317          * particular external link.  This currently may include rel => nofollow
1318          * (depending on configuration, namespace, and the URL's domain) and/or a
1319          * target attribute (depending on configuration).
1320          *
1321          * @param string $url Optional URL, to extract the domain from for rel =>
1322          *   nofollow if appropriate
1323          * @return array Associative array of HTML attributes
1324          */
1325         function getExternalLinkAttribs( $url = false ) {
1326                 $attribs = array();
1327                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1328                 $ns = $this->mTitle->getNamespace();
1329                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1330                         $attribs['rel'] = 'nofollow';
1331
1332                         global $wgNoFollowDomainExceptions;
1333                         if ( $wgNoFollowDomainExceptions ) {
1334                                 $bits = wfParseUrl( $url );
1335                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1336                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1337                                                 if( substr( $bits['host'], -strlen( $domain ) )
1338                                                 == $domain ) {
1339                                                         unset( $attribs['rel'] );
1340                                                         break;
1341                                                 }
1342                                         }
1343                                 }
1344                         }
1345                 }
1346                 if ( $this->mOptions->getExternalLinkTarget() ) {
1347                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1348                 }
1349                 return $attribs;
1350         }
1351
1352
1353         /**
1354          * Replace unusual URL escape codes with their equivalent characters
1355          * @param string
1356          * @return string
1357          * @static
1358          * @todo  This can merge genuinely required bits in the path or query string,
1359          *        breaking legit URLs. A proper fix would treat the various parts of
1360          *        the URL differently; as a workaround, just use the output for
1361          *        statistical records, not for actual linking/output.
1362          */
1363         static function replaceUnusualEscapes( $url ) {
1364                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1365                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1366         }
1367
1368         /**
1369          * Callback function used in replaceUnusualEscapes().
1370          * Replaces unusual URL escape codes with their equivalent character
1371          * @static
1372          * @private
1373          */
1374         private static function replaceUnusualEscapesCallback( $matches ) {
1375                 $char = urldecode( $matches[0] );
1376                 $ord = ord( $char );
1377                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1378                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1379                         // No, shouldn't be escaped
1380                         return $char;
1381                 } else {
1382                         // Yes, leave it escaped
1383                         return $matches[0];
1384                 }
1385         }
1386
1387         /**
1388          * make an image if it's allowed, either through the global
1389          * option, through the exception, or through the on-wiki whitelist
1390          * @private
1391          */
1392         function maybeMakeExternalImage( $url ) {
1393                 $sk = $this->mOptions->getSkin();
1394                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1395                 $imagesexception = !empty($imagesfrom);
1396                 $text = false;
1397                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1398                 if( $imagesexception && is_array( $imagesfrom ) ) {
1399                         $imagematch = false;
1400                         foreach( $imagesfrom as $match ) {
1401                                 if( strpos( $url, $match ) === 0 ) {
1402                                         $imagematch = true;
1403                                         break;
1404                                 }
1405                         }
1406                 } elseif( $imagesexception ) {
1407                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1408                 } else {
1409                         $imagematch = false;
1410                 }
1411                 if ( $this->mOptions->getAllowExternalImages()
1412                      || ( $imagesexception && $imagematch ) ) {
1413                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1414                                 # Image found
1415                                 $text = $sk->makeExternalImage( $url );
1416                         }
1417                 }
1418                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1419                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1420                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1421                         foreach( $whitelist as $entry ) {
1422                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1423                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1424                                         continue;
1425                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1426                                         # Image matches a whitelist entry
1427                                         $text = $sk->makeExternalImage( $url );
1428                                         break;
1429                                 }
1430                         }
1431                 }
1432                 return $text;
1433         }
1434
1435         /**
1436          * Process [[ ]] wikilinks
1437          * @return processed text
1438          *
1439          * @private
1440          */
1441         function replaceInternalLinks( $s ) {
1442                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1443                 return $s;
1444         }
1445
1446         /**
1447          * Process [[ ]] wikilinks (RIL)
1448          * @return LinkHolderArray
1449          *
1450          * @private
1451          */
1452         function replaceInternalLinks2( &$s ) {
1453                 global $wgContLang;
1454
1455                 wfProfileIn( __METHOD__ );
1456
1457                 wfProfileIn( __METHOD__.'-setup' );
1458                 static $tc = FALSE, $e1, $e1_img;
1459                 # the % is needed to support urlencoded titles as well
1460                 if ( !$tc ) {
1461                         $tc = Title::legalChars() . '#%';
1462                         # Match a link having the form [[namespace:link|alternate]]trail
1463                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1464                         # Match cases where there is no "]]", which might still be images
1465                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1466                 }
1467
1468                 $sk = $this->mOptions->getSkin();
1469                 $holders = new LinkHolderArray( $this );
1470
1471                 #split the entire text string on occurences of [[
1472                 $a = StringUtils::explode( '[[', ' ' . $s );
1473                 #get the first element (all text up to first [[), and remove the space we added
1474                 $s = $a->current();
1475                 $a->next();
1476                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1477                 $s = substr( $s, 1 );
1478
1479                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1480                 $e2 = null;
1481                 if ( $useLinkPrefixExtension ) {
1482                         # Match the end of a line for a word that's not followed by whitespace,
1483                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1484                         $e2 = wfMsgForContent( 'linkprefix' );
1485                 }
1486
1487                 if( is_null( $this->mTitle ) ) {
1488                         wfProfileOut( __METHOD__.'-setup' );
1489                         wfProfileOut( __METHOD__ );
1490                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1491                 }
1492                 $nottalk = !$this->mTitle->isTalkPage();
1493
1494                 if ( $useLinkPrefixExtension ) {
1495                         $m = array();
1496                         if ( preg_match( $e2, $s, $m ) ) {
1497                                 $first_prefix = $m[2];
1498                         } else {
1499                                 $first_prefix = false;
1500                         }
1501                 } else {
1502                         $prefix = '';
1503                 }
1504
1505                 if($wgContLang->hasVariants()) {
1506                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1507                 } else {
1508                         $selflink = array($this->mTitle->getPrefixedText());
1509                 }
1510                 $useSubpages = $this->areSubpagesAllowed();
1511                 wfProfileOut( __METHOD__.'-setup' );
1512
1513                 # Loop for each link
1514                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1515                         # Check for excessive memory usage
1516                         if ( $holders->isBig() ) {
1517                                 # Too big
1518                                 # Do the existence check, replace the link holders and clear the array
1519                                 $holders->replace( $s );
1520                                 $holders->clear();
1521                         }
1522
1523                         if ( $useLinkPrefixExtension ) {
1524                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1525                                 if ( preg_match( $e2, $s, $m ) ) {
1526                                         $prefix = $m[2];
1527                                         $s = $m[1];
1528                                 } else {
1529                                         $prefix='';
1530                                 }
1531                                 # first link
1532                                 if($first_prefix) {
1533                                         $prefix = $first_prefix;
1534                                         $first_prefix = false;
1535                                 }
1536                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1537                         }
1538
1539                         $might_be_img = false;
1540
1541                         wfProfileIn( __METHOD__."-e1" );
1542                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1543                                 $text = $m[2];
1544                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1545                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1546                                 # the real problem is with the $e1 regex
1547                                 # See bug 1300.
1548                                 #
1549                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1550                                 # and no image is in sight. See bug 2095.
1551                                 #
1552                                 if( $text !== '' &&
1553                                         substr( $m[3], 0, 1 ) === ']' &&
1554                                         strpos($text, '[') !== false
1555                                 )
1556                                 {
1557                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1558                                         $m[3] = substr( $m[3], 1 );
1559                                 }
1560                                 # fix up urlencoded title texts
1561                                 if( strpos( $m[1], '%' ) !== false ) {
1562                                         # Should anchors '#' also be rejected?
1563                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1564                                 }
1565                                 $trail = $m[3];
1566                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1567                                 $might_be_img = true;
1568                                 $text = $m[2];
1569                                 if ( strpos( $m[1], '%' ) !== false ) {
1570                                         $m[1] = urldecode($m[1]);
1571                                 }
1572                                 $trail = "";
1573                         } else { # Invalid form; output directly
1574                                 $s .= $prefix . '[[' . $line ;
1575                                 wfProfileOut( __METHOD__."-e1" );
1576                                 continue;
1577                         }
1578                         wfProfileOut( __METHOD__."-e1" );
1579                         wfProfileIn( __METHOD__."-misc" );
1580
1581                         # Don't allow internal links to pages containing
1582                         # PROTO: where PROTO is a valid URL protocol; these
1583                         # should be external links.
1584                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1585                                 $s .= $prefix . '[[' . $line ;
1586                                 wfProfileOut( __METHOD__."-misc" );
1587                                 continue;
1588                         }
1589
1590                         # Make subpage if necessary
1591                         if( $useSubpages ) {
1592                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1593                         } else {
1594                                 $link = $m[1];
1595                         }
1596
1597                         $noforce = (substr($m[1], 0, 1) !== ':');
1598                         if (!$noforce) {
1599                                 # Strip off leading ':'
1600                                 $link = substr($link, 1);
1601                         }
1602
1603                         wfProfileOut( __METHOD__."-misc" );
1604                         wfProfileIn( __METHOD__."-title" );
1605                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1606                         if( $nt === NULL ) {
1607                                 $s .= $prefix . '[[' . $line;
1608                                 wfProfileOut( __METHOD__."-title" );
1609                                 continue;
1610                         }
1611
1612                         $ns = $nt->getNamespace();
1613                         $iw = $nt->getInterWiki();
1614                         wfProfileOut( __METHOD__."-title" );
1615
1616                         if ($might_be_img) { # if this is actually an invalid link
1617                                 wfProfileIn( __METHOD__."-might_be_img" );
1618                                 if ($ns == NS_FILE && $noforce) { #but might be an image
1619                                         $found = false;
1620                                         while ( true ) {
1621                                                 #look at the next 'line' to see if we can close it there
1622                                                 $a->next();
1623                                                 $next_line = $a->current();
1624                                                 if ( $next_line === false || $next_line === null ) {
1625                                                         break;
1626                                                 }
1627                                                 $m = explode( ']]', $next_line, 3 );
1628                                                 if ( count( $m ) == 3 ) {
1629                                                         # the first ]] closes the inner link, the second the image
1630                                                         $found = true;
1631                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1632                                                         $trail = $m[2];
1633                                                         break;
1634                                                 } elseif ( count( $m ) == 2 ) {
1635                                                         #if there's exactly one ]] that's fine, we'll keep looking
1636                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1637                                                 } else {
1638                                                         #if $next_line is invalid too, we need look no further
1639                                                         $text .= '[[' . $next_line;
1640                                                         break;
1641                                                 }
1642                                         }
1643                                         if ( !$found ) {
1644                                                 # we couldn't find the end of this imageLink, so output it raw
1645                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1646                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1647                                                 $s .= "{$prefix}[[$link|$text";
1648                                                 # note: no $trail, because without an end, there *is* no trail
1649                                                 wfProfileOut( __METHOD__."-might_be_img" );
1650                                                 continue;
1651                                         }
1652                                 } else { #it's not an image, so output it raw
1653                                         $s .= "{$prefix}[[$link|$text";
1654                                         # note: no $trail, because without an end, there *is* no trail
1655                                         wfProfileOut( __METHOD__."-might_be_img" );
1656                                         continue;
1657                                 }
1658                                 wfProfileOut( __METHOD__."-might_be_img" );
1659                         }
1660
1661                         $wasblank = ( '' == $text );
1662                         if( $wasblank ) $text = $link;
1663
1664                         # Link not escaped by : , create the various objects
1665                         if( $noforce ) {
1666
1667                                 # Interwikis
1668                                 wfProfileIn( __METHOD__."-interwiki" );
1669                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1670                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1671                                         $s = rtrim($s . $prefix);
1672                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1673                                         wfProfileOut( __METHOD__."-interwiki" );
1674                                         continue;
1675                                 }
1676                                 wfProfileOut( __METHOD__."-interwiki" );
1677
1678                                 if ( $ns == NS_FILE ) {
1679                                         wfProfileIn( __METHOD__."-image" );
1680                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1681                                                 # recursively parse links inside the image caption
1682                                                 # actually, this will parse them in any other parameters, too,
1683                                                 # but it might be hard to fix that, and it doesn't matter ATM
1684                                                 $text = $this->replaceExternalLinks($text);
1685                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1686
1687                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1688                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1689                                         }
1690                                         $this->mOutput->addImage( $nt->getDBkey() );
1691                                         wfProfileOut( __METHOD__."-image" );
1692                                         continue;
1693
1694                                 }
1695
1696                                 if ( $ns == NS_CATEGORY ) {
1697                                         wfProfileIn( __METHOD__."-category" );
1698                                         $s = rtrim($s . "\n"); # bug 87
1699
1700                                         if ( $wasblank ) {
1701                                                 $sortkey = $this->getDefaultSort();
1702                                         } else {
1703                                                 $sortkey = $text;
1704                                         }
1705                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1706                                         $sortkey = str_replace( "\n", '', $sortkey );
1707                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1708                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1709
1710                                         /**
1711                                          * Strip the whitespace Category links produce, see bug 87
1712                                          * @todo We might want to use trim($tmp, "\n") here.
1713                                          */
1714                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1715
1716                                         wfProfileOut( __METHOD__."-category" );
1717                                         continue;
1718                                 }
1719                         }
1720
1721                         # Self-link checking
1722                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1723                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1724                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1725                                         continue;
1726                                 }
1727                         }
1728
1729                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1730                         # FIXME: Should do batch file existence checks, see comment below
1731                         if( $ns == NS_MEDIA ) {
1732                                 wfProfileIn( __METHOD__."-media" );
1733                                 # Give extensions a chance to select the file revision for us
1734                                 $skip = $time = false;
1735                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1736                                 if ( $skip ) {
1737                                         $link = $sk->link( $nt );
1738                                 } else {
1739                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1740                                 }
1741                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1742                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1743                                 $this->mOutput->addImage( $nt->getDBkey() );
1744                                 wfProfileOut( __METHOD__."-media" );
1745                                 continue;
1746                         }
1747
1748                         wfProfileIn( __METHOD__."-always_known" );
1749                         # Some titles, such as valid special pages or files in foreign repos, should
1750                         # be shown as bluelinks even though they're not included in the page table
1751                         #
1752                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1753                         # batch file existence checks for NS_FILE and NS_MEDIA
1754                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1755                                 $this->mOutput->addLink( $nt );
1756                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1757                         } else {
1758                                 # Links will be added to the output link list after checking
1759                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1760                         }
1761                         wfProfileOut( __METHOD__."-always_known" );
1762                 }
1763                 wfProfileOut( __METHOD__ );
1764                 return $holders;
1765         }
1766
1767         /**
1768          * Make a link placeholder. The text returned can be later resolved to a real link with
1769          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1770          * parsing of interwiki links, and secondly to allow all existence checks and
1771          * article length checks (for stub links) to be bundled into a single query.
1772          *
1773          * @deprecated
1774          */
1775         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1776                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1777         }
1778
1779         /**
1780          * Render a forced-blue link inline; protect against double expansion of
1781          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1782          * Since this little disaster has to split off the trail text to avoid
1783          * breaking URLs in the following text without breaking trails on the
1784          * wiki links, it's been made into a horrible function.
1785          *
1786          * @param Title $nt
1787          * @param string $text
1788          * @param string $query
1789          * @param string $trail
1790          * @param string $prefix
1791          * @return string HTML-wikitext mix oh yuck
1792          */
1793         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1794                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1795                 $sk = $this->mOptions->getSkin();
1796                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1797                 return $this->armorLinks( $link ) . $trail;
1798         }
1799
1800         /**
1801          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1802          * going to go through further parsing steps before inline URL expansion.
1803          *
1804          * Not needed quite as much as it used to be since free links are a bit
1805          * more sensible these days. But bracketed links are still an issue.
1806          *
1807          * @param string more-or-less HTML
1808          * @return string less-or-more HTML with NOPARSE bits
1809          */
1810         function armorLinks( $text ) {
1811                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1812                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1813         }
1814
1815         /**
1816          * Return true if subpage links should be expanded on this page.
1817          * @return bool
1818          */
1819         function areSubpagesAllowed() {
1820                 # Some namespaces don't allow subpages
1821                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1822         }
1823
1824         /**
1825          * Handle link to subpage if necessary
1826          * @param string $target the source of the link
1827          * @param string &$text the link text, modified as necessary
1828          * @return string the full name of the link
1829          * @private
1830          */
1831         function maybeDoSubpageLink($target, &$text) {
1832                 # Valid link forms:
1833                 # Foobar -- normal
1834                 # :Foobar -- override special treatment of prefix (images, language links)
1835                 # /Foobar -- convert to CurrentPage/Foobar
1836                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1837                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1838                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1839
1840                 wfProfileIn( __METHOD__ );
1841                 $ret = $target; # default return value is no change
1842
1843                 # Some namespaces don't allow subpages,
1844                 # so only perform processing if subpages are allowed
1845                 if( $this->areSubpagesAllowed() ) {
1846                         $hash = strpos( $target, '#' );
1847                         if( $hash !== false ) {
1848                                 $suffix = substr( $target, $hash );
1849                                 $target = substr( $target, 0, $hash );
1850                         } else {
1851                                 $suffix = '';
1852                         }
1853                         # bug 7425
1854                         $target = trim( $target );
1855                         # Look at the first character
1856                         if( $target != '' && $target{0} === '/' ) {
1857                                 # / at end means we don't want the slash to be shown
1858                                 $m = array();
1859                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1860                                 if( $trailingSlashes ) {
1861                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1862                                 } else {
1863                                         $noslash = substr( $target, 1 );
1864                                 }
1865
1866                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1867                                 if( '' === $text ) {
1868                                         $text = $target . $suffix;
1869                                 } # this might be changed for ugliness reasons
1870                         } else {
1871                                 # check for .. subpage backlinks
1872                                 $dotdotcount = 0;
1873                                 $nodotdot = $target;
1874                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1875                                         ++$dotdotcount;
1876                                         $nodotdot = substr( $nodotdot, 3 );
1877                                 }
1878                                 if($dotdotcount > 0) {
1879                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1880                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1881                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1882                                                 # / at the end means don't show full path
1883                                                 if( substr( $nodotdot, -1, 1 ) === '/' ) {
1884                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1885                                                         if( '' === $text ) {
1886                                                                 $text = $nodotdot . $suffix;
1887                                                         }
1888                                                 }
1889                                                 $nodotdot = trim( $nodotdot );
1890                                                 if( $nodotdot != '' ) {
1891                                                         $ret .= '/' . $nodotdot;
1892                                                 }
1893                                                 $ret .= $suffix;
1894                                         }
1895                                 }
1896                         }
1897                 }
1898
1899                 wfProfileOut( __METHOD__ );
1900                 return $ret;
1901         }
1902
1903         /**#@+
1904          * Used by doBlockLevels()
1905          * @private
1906          */
1907         /* private */ function closeParagraph() {
1908                 $result = '';
1909                 if ( '' != $this->mLastSection ) {
1910                         $result = '</' . $this->mLastSection  . ">\n";
1911                 }
1912                 $this->mInPre = false;
1913                 $this->mLastSection = '';
1914                 return $result;
1915         }
1916         # getCommon() returns the length of the longest common substring
1917         # of both arguments, starting at the beginning of both.
1918         #
1919         /* private */ function getCommon( $st1, $st2 ) {
1920                 $fl = strlen( $st1 );
1921                 $shorter = strlen( $st2 );
1922                 if ( $fl < $shorter ) { $shorter = $fl; }
1923
1924                 for ( $i = 0; $i < $shorter; ++$i ) {
1925                         if ( $st1{$i} != $st2{$i} ) { break; }
1926                 }
1927                 return $i;
1928         }
1929         # These next three functions open, continue, and close the list
1930         # element appropriate to the prefix character passed into them.
1931         #
1932         /* private */ function openList( $char ) {
1933                 $result = $this->closeParagraph();
1934
1935                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1936                 else if ( '#' === $char ) { $result .= '<ol><li>'; }
1937                 else if ( ':' === $char ) { $result .= '<dl><dd>'; }
1938                 else if ( ';' === $char ) {
1939                         $result .= '<dl><dt>';
1940                         $this->mDTopen = true;
1941                 }
1942                 else { $result = '<!-- ERR 1 -->'; }
1943
1944                 return $result;
1945         }
1946
1947         /* private */ function nextItem( $char ) {
1948                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1949                 else if ( ':' === $char || ';' === $char ) {
1950                         $close = '</dd>';
1951                         if ( $this->mDTopen ) { $close = '</dt>'; }
1952                         if ( ';' === $char ) {
1953                                 $this->mDTopen = true;
1954                                 return $close . '<dt>';
1955                         } else {
1956                                 $this->mDTopen = false;
1957                                 return $close . '<dd>';
1958                         }
1959                 }
1960                 return '<!-- ERR 2 -->';
1961         }
1962
1963         /* private */ function closeList( $char ) {
1964                 if ( '*' === $char ) { $text = '</li></ul>'; }
1965                 else if ( '#' === $char ) { $text = '</li></ol>'; }
1966                 else if ( ':' === $char ) {
1967                         if ( $this->mDTopen ) {
1968                                 $this->mDTopen = false;
1969                                 $text = '</dt></dl>';
1970                         } else {
1971                                 $text = '</dd></dl>';
1972                         }
1973                 }
1974                 else {  return '<!-- ERR 3 -->'; }
1975                 return $text."\n";
1976         }
1977         /**#@-*/
1978
1979         /**
1980          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1981          *
1982          * @private
1983          * @return string the lists rendered as HTML
1984          */
1985         function doBlockLevels( $text, $linestart ) {
1986                 wfProfileIn( __METHOD__ );
1987
1988                 # Parsing through the text line by line.  The main thing
1989                 # happening here is handling of block-level elements p, pre,
1990                 # and making lists from lines starting with * # : etc.
1991                 #
1992                 $textLines = StringUtils::explode( "\n", $text );
1993
1994                 $lastPrefix = $output = '';
1995                 $this->mDTopen = $inBlockElem = false;
1996                 $prefixLength = 0;
1997                 $paragraphStack = false;
1998
1999                 foreach ( $textLines as $oLine ) {
2000                         # Fix up $linestart
2001                         if ( !$linestart ) {
2002                                 $output .= $oLine;
2003                                 $linestart = true;
2004                                 continue;
2005                         }
2006
2007                         $lastPrefixLength = strlen( $lastPrefix );
2008                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2009                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2010                         if ( !$this->mInPre ) {
2011                                 # Multiple prefixes may abut each other for nested lists.
2012                                 $prefixLength = strspn( $oLine, '*#:;' );
2013                                 $prefix = substr( $oLine, 0, $prefixLength );
2014
2015                                 # eh?
2016                                 $prefix2 = str_replace( ';', ':', $prefix );
2017                                 $t = substr( $oLine, $prefixLength );
2018                                 $this->mInPre = (bool)$preOpenMatch;
2019                         } else {
2020                                 # Don't interpret any other prefixes in preformatted text
2021                                 $prefixLength = 0;
2022                                 $prefix = $prefix2 = '';
2023                                 $t = $oLine;
2024                         }
2025
2026                         # List generation
2027                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2028                                 # Same as the last item, so no need to deal with nesting or opening stuff
2029                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2030                                 $paragraphStack = false;
2031
2032                                 if ( substr( $prefix, -1 ) === ';') {
2033                                         # The one nasty exception: definition lists work like this:
2034                                         # ; title : definition text
2035                                         # So we check for : in the remainder text to split up the
2036                                         # title and definition, without b0rking links.
2037                                         $term = $t2 = '';
2038                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2039                                                 $t = $t2;
2040                                                 $output .= $term . $this->nextItem( ':' );
2041                                         }
2042                                 }
2043                         } elseif( $prefixLength || $lastPrefixLength ) {
2044                                 # Either open or close a level...
2045                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2046                                 $paragraphStack = false;
2047
2048                                 while( $commonPrefixLength < $lastPrefixLength ) {
2049                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2050                                         --$lastPrefixLength;
2051                                 }
2052                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2053                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2054                                 }
2055                                 while ( $prefixLength > $commonPrefixLength ) {
2056                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2057                                         $output .= $this->openList( $char );
2058
2059                                         if ( ';' === $char ) {
2060                                                 # FIXME: This is dupe of code above
2061                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2062                                                         $t = $t2;
2063                                                         $output .= $term . $this->nextItem( ':' );
2064                                                 }
2065                                         }
2066                                         ++$commonPrefixLength;
2067                                 }
2068                                 $lastPrefix = $prefix2;
2069                         }
2070                         if( 0 == $prefixLength ) {
2071                                 wfProfileIn( __METHOD__."-paragraph" );
2072                                 # No prefix (not in list)--go to paragraph mode
2073                                 // XXX: use a stack for nestable elements like span, table and div
2074                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2075                                 $closematch = preg_match(
2076                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2077                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2078                                 if ( $openmatch or $closematch ) {
2079                                         $paragraphStack = false;
2080                                         # TODO bug 5718: paragraph closed
2081                                         $output .= $this->closeParagraph();
2082                                         if ( $preOpenMatch and !$preCloseMatch ) {
2083                                                 $this->mInPre = true;
2084                                         }
2085                                         if ( $closematch ) {
2086                                                 $inBlockElem = false;
2087                                         } else {
2088                                                 $inBlockElem = true;
2089                                         }
2090                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2091                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2092                                                 // pre
2093                                                 if ($this->mLastSection !== 'pre') {
2094                                                         $paragraphStack = false;
2095                                                         $output .= $this->closeParagraph().'<pre>';
2096                                                         $this->mLastSection = 'pre';
2097                                                 }
2098                                                 $t = substr( $t, 1 );
2099                                         } else {
2100                                                 // paragraph
2101                                                 if ( '' == trim($t) ) {
2102                                                         if ( $paragraphStack ) {
2103                                                                 $output .= $paragraphStack.'<br />';
2104                                                                 $paragraphStack = false;
2105                                                                 $this->mLastSection = 'p';
2106                                                         } else {
2107                                                                 if ($this->mLastSection !== 'p' ) {
2108                                                                         $output .= $this->closeParagraph();
2109                                                                         $this->mLastSection = '';
2110                                                                         $paragraphStack = '<p>';
2111                                                                 } else {
2112                                                                         $paragraphStack = '</p><p>';
2113                                                                 }
2114                                                         }
2115                                                 } else {
2116                                                         if ( $paragraphStack ) {
2117                                                                 $output .= $paragraphStack;
2118                                                                 $paragraphStack = false;
2119                                                                 $this->mLastSection = 'p';
2120                                                         } else if ($this->mLastSection !== 'p') {
2121                                                                 $output .= $this->closeParagraph().'<p>';
2122                                                                 $this->mLastSection = 'p';
2123                                                         }
2124                                                 }
2125                                         }
2126                                 }
2127                                 wfProfileOut( __METHOD__."-paragraph" );
2128                         }
2129                         // somewhere above we forget to get out of pre block (bug 785)
2130                         if($preCloseMatch && $this->mInPre) {
2131                                 $this->mInPre = false;
2132                         }
2133                         if ($paragraphStack === false) {
2134                                 $output .= $t."\n";
2135                         }
2136                 }
2137                 while ( $prefixLength ) {
2138                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2139                         --$prefixLength;
2140                 }
2141                 if ( '' != $this->mLastSection ) {
2142                         $output .= '</' . $this->mLastSection . '>';
2143                         $this->mLastSection = '';
2144                 }
2145
2146                 wfProfileOut( __METHOD__ );
2147                 return $output;
2148         }
2149
2150         /**
2151          * Split up a string on ':', ignoring any occurences inside tags
2152          * to prevent illegal overlapping.
2153          * @param string $str the string to split
2154          * @param string &$before set to everything before the ':'
2155          * @param string &$after set to everything after the ':'
2156          * return string the position of the ':', or false if none found
2157          */
2158         function findColonNoLinks($str, &$before, &$after) {
2159                 wfProfileIn( __METHOD__ );
2160
2161                 $pos = strpos( $str, ':' );
2162                 if( $pos === false ) {
2163                         // Nothing to find!
2164                         wfProfileOut( __METHOD__ );
2165                         return false;
2166                 }
2167
2168                 $lt = strpos( $str, '<' );
2169                 if( $lt === false || $lt > $pos ) {
2170                         // Easy; no tag nesting to worry about
2171                         $before = substr( $str, 0, $pos );
2172                         $after = substr( $str, $pos+1 );
2173                         wfProfileOut( __METHOD__ );
2174                         return $pos;
2175                 }
2176
2177                 // Ugly state machine to walk through avoiding tags.
2178                 $state = self::COLON_STATE_TEXT;
2179                 $stack = 0;
2180                 $len = strlen( $str );
2181                 for( $i = 0; $i < $len; $i++ ) {
2182                         $c = $str{$i};
2183
2184                         switch( $state ) {
2185                         // (Using the number is a performance hack for common cases)
2186                         case 0: // self::COLON_STATE_TEXT:
2187                                 switch( $c ) {
2188                                 case "<":
2189                                         // Could be either a <start> tag or an </end> tag
2190                                         $state = self::COLON_STATE_TAGSTART;
2191                                         break;
2192                                 case ":":
2193                                         if( $stack == 0 ) {
2194                                                 // We found it!
2195                                                 $before = substr( $str, 0, $i );
2196                                                 $after = substr( $str, $i + 1 );
2197                                                 wfProfileOut( __METHOD__ );
2198                                                 return $i;
2199                                         }
2200                                         // Embedded in a tag; don't break it.
2201                                         break;
2202                                 default:
2203                                         // Skip ahead looking for something interesting
2204                                         $colon = strpos( $str, ':', $i );
2205                                         if( $colon === false ) {
2206                                                 // Nothing else interesting
2207                                                 wfProfileOut( __METHOD__ );
2208                                                 return false;
2209                                         }
2210                                         $lt = strpos( $str, '<', $i );
2211                                         if( $stack === 0 ) {
2212                                                 if( $lt === false || $colon < $lt ) {
2213                                                         // We found it!
2214                                                         $before = substr( $str, 0, $colon );
2215                                                         $after = substr( $str, $colon + 1 );
2216                                                         wfProfileOut( __METHOD__ );
2217                                                         return $i;
2218                                                 }
2219                                         }
2220                                         if( $lt === false ) {
2221                                                 // Nothing else interesting to find; abort!
2222                                                 // We're nested, but there's no close tags left. Abort!
2223                                                 break 2;
2224                                         }
2225                                         // Skip ahead to next tag start
2226                                         $i = $lt;
2227                                         $state = self::COLON_STATE_TAGSTART;
2228                                 }
2229                                 break;
2230                         case 1: // self::COLON_STATE_TAG:
2231                                 // In a <tag>
2232                                 switch( $c ) {
2233                                 case ">":
2234                                         $stack++;
2235                                         $state = self::COLON_STATE_TEXT;
2236                                         break;
2237                                 case "/":
2238                                         // Slash may be followed by >?
2239                                         $state = self::COLON_STATE_TAGSLASH;
2240                                         break;
2241                                 default:
2242                                         // ignore
2243                                 }
2244                                 break;
2245                         case 2: // self::COLON_STATE_TAGSTART:
2246                                 switch( $c ) {
2247                                 case "/":
2248                                         $state = self::COLON_STATE_CLOSETAG;
2249                                         break;
2250                                 case "!":
2251                                         $state = self::COLON_STATE_COMMENT;
2252                                         break;
2253                                 case ">":
2254                                         // Illegal early close? This shouldn't happen D:
2255                                         $state = self::COLON_STATE_TEXT;
2256                                         break;
2257                                 default:
2258                                         $state = self::COLON_STATE_TAG;
2259                                 }
2260                                 break;
2261                         case 3: // self::COLON_STATE_CLOSETAG:
2262                                 // In a </tag>
2263                                 if( $c === ">" ) {
2264                                         $stack--;
2265                                         if( $stack < 0 ) {
2266                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2267                                                 wfProfileOut( __METHOD__ );
2268                                                 return false;
2269                                         }
2270                                         $state = self::COLON_STATE_TEXT;
2271                                 }
2272                                 break;
2273                         case self::COLON_STATE_TAGSLASH:
2274                                 if( $c === ">" ) {
2275                                         // Yes, a self-closed tag <blah/>
2276                                         $state = self::COLON_STATE_TEXT;
2277                                 } else {
2278                                         // Probably we're jumping the gun, and this is an attribute
2279                                         $state = self::COLON_STATE_TAG;
2280                                 }
2281                                 break;
2282                         case 5: // self::COLON_STATE_COMMENT:
2283                                 if( $c === "-" ) {
2284                                         $state = self::COLON_STATE_COMMENTDASH;
2285                                 }
2286                                 break;
2287                         case self::COLON_STATE_COMMENTDASH:
2288                                 if( $c === "-" ) {
2289                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2290                                 } else {
2291                                         $state = self::COLON_STATE_COMMENT;
2292                                 }
2293                                 break;
2294                         case self::COLON_STATE_COMMENTDASHDASH:
2295                                 if( $c === ">" ) {
2296                                         $state = self::COLON_STATE_TEXT;
2297                                 } else {
2298                                         $state = self::COLON_STATE_COMMENT;
2299                                 }
2300                                 break;
2301                         default:
2302                                 throw new MWException( "State machine error in " . __METHOD__ );
2303                         }
2304                 }
2305                 if( $stack > 0 ) {
2306                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2307                         return false;
2308                 }
2309                 wfProfileOut( __METHOD__ );
2310                 return false;
2311         }
2312
2313         /**
2314          * Return value of a magic variable (like PAGENAME)
2315          *
2316          * @private
2317          */
2318         function getVariableValue( $index ) {
2319                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2320
2321                 /**
2322                  * Some of these require message or data lookups and can be
2323                  * expensive to check many times.
2324                  */
2325                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2326                         if ( isset( $this->mVarCache[$index] ) ) {
2327                                 return $this->mVarCache[$index];
2328                         }
2329                 }
2330
2331                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2332                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2333
2334                 # Use the time zone
2335                 global $wgLocaltimezone;
2336                 if ( isset( $wgLocaltimezone ) ) {
2337                         $oldtz = getenv( 'TZ' );
2338                         putenv( 'TZ='.$wgLocaltimezone );
2339                 }
2340
2341                 wfSuppressWarnings(); // E_STRICT system time bitching
2342                 $localTimestamp = date( 'YmdHis', $ts );
2343                 $localMonth = date( 'm', $ts );
2344                 $localMonthName = date( 'n', $ts );
2345                 $localDay = date( 'j', $ts );
2346                 $localDay2 = date( 'd', $ts );
2347                 $localDayOfWeek = date( 'w', $ts );
2348                 $localWeek = date( 'W', $ts );
2349                 $localYear = date( 'Y', $ts );
2350                 $localHour = date( 'H', $ts );
2351                 if ( isset( $wgLocaltimezone ) ) {
2352                         putenv( 'TZ='.$oldtz );
2353                 }
2354                 wfRestoreWarnings();
2355
2356                 switch ( $index ) {
2357                         case 'currentmonth':
2358                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2359                         case 'currentmonthname':
2360                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2361                         case 'currentmonthnamegen':
2362                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2363                         case 'currentmonthabbrev':
2364                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2365                         case 'currentday':
2366                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2367                         case 'currentday2':
2368                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2369                         case 'localmonth':
2370                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2371                         case 'localmonthname':
2372                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2373                         case 'localmonthnamegen':
2374                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2375                         case 'localmonthabbrev':
2376                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2377                         case 'localday':
2378                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2379                         case 'localday2':
2380                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2381                         case 'pagename':
2382                                 return wfEscapeWikiText( $this->mTitle->getText() );
2383                         case 'pagenamee':
2384                                 return $this->mTitle->getPartialURL();
2385                         case 'fullpagename':
2386                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2387                         case 'fullpagenamee':
2388                                 return $this->mTitle->getPrefixedURL();
2389                         case 'subpagename':
2390                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2391                         case 'subpagenamee':
2392                                 return $this->mTitle->getSubpageUrlForm();
2393                         case 'basepagename':
2394                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2395                         case 'basepagenamee':
2396                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2397                         case 'talkpagename':
2398                                 if( $this->mTitle->canTalk() ) {
2399                                         $talkPage = $this->mTitle->getTalkPage();
2400                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2401                                 } else {
2402                                         return '';
2403                                 }
2404                         case 'talkpagenamee':
2405                                 if( $this->mTitle->canTalk() ) {
2406                                         $talkPage = $this->mTitle->getTalkPage();
2407                                         return $talkPage->getPrefixedUrl();
2408                                 } else {
2409                                         return '';
2410                                 }
2411                         case 'subjectpagename':
2412                                 $subjPage = $this->mTitle->getSubjectPage();
2413                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2414                         case 'subjectpagenamee':
2415                                 $subjPage = $this->mTitle->getSubjectPage();
2416                                 return $subjPage->getPrefixedUrl();
2417                         case 'revisionid':
2418                                 // Let the edit saving system know we should parse the page
2419                                 // *after* a revision ID has been assigned.
2420                                 $this->mOutput->setFlag( 'vary-revision' );
2421                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2422                                 return $this->mRevisionId;
2423                         case 'revisionday':
2424                                 // Let the edit saving system know we should parse the page
2425                                 // *after* a revision ID has been assigned. This is for null edits.
2426                                 $this->mOutput->setFlag( 'vary-revision' );
2427                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2428                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2429                         case 'revisionday2':
2430                                 // Let the edit saving system know we should parse the page
2431                                 // *after* a revision ID has been assigned. This is for null edits.
2432                                 $this->mOutput->setFlag( 'vary-revision' );
2433                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2434                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2435                         case 'revisionmonth':
2436                                 // Let the edit saving system know we should parse the page
2437                                 // *after* a revision ID has been assigned. This is for null edits.
2438                                 $this->mOutput->setFlag( 'vary-revision' );
2439                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2440                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2441                         case 'revisionyear':
2442                                 // Let the edit saving system know we should parse the page
2443                                 // *after* a revision ID has been assigned. This is for null edits.
2444                                 $this->mOutput->setFlag( 'vary-revision' );
2445                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2446                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2447                         case 'revisiontimestamp':
2448                                 // Let the edit saving system know we should parse the page
2449                                 // *after* a revision ID has been assigned. This is for null edits.
2450                                 $this->mOutput->setFlag( 'vary-revision' );
2451                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2452                                 return $this->getRevisionTimestamp();
2453                         case 'revisionuser':
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__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2458                                 return $this->getRevisionUser();
2459                         case 'namespace':
2460                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2461                         case 'namespacee':
2462                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2463                         case 'talkspace':
2464                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2465                         case 'talkspacee':
2466                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2467                         case 'subjectspace':
2468                                 return $this->mTitle->getSubjectNsText();
2469                         case 'subjectspacee':
2470                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2471                         case 'currentdayname':
2472                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2473                         case 'currentyear':
2474                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2475                         case 'currenttime':
2476                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2477                         case 'currenthour':
2478                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2479                         case 'currentweek':
2480                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2481                                 // int to remove the padding
2482                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2483                         case 'currentdow':
2484                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2485                         case 'localdayname':
2486                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2487                         case 'localyear':
2488                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2489                         case 'localtime':
2490                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2491                         case 'localhour':
2492                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2493                         case 'localweek':
2494                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2495                                 // int to remove the padding
2496                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2497                         case 'localdow':
2498                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2499                         case 'numberofarticles':
2500                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2501                         case 'numberoffiles':
2502                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2503                         case 'numberofusers':
2504                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2505                         case 'numberofactiveusers':
2506                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::activeUsers() );
2507                         case 'numberofpages':
2508                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2509                         case 'numberofadmins':
2510                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2511                         case 'numberofedits':
2512                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2513                         case 'numberofviews':
2514                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::views() );
2515                         case 'currenttimestamp':
2516                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2517                         case 'localtimestamp':
2518                                 return $this->mVarCache[$index] = $localTimestamp;
2519                         case 'currentversion':
2520                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2521                         case 'sitename':
2522                                 return $wgSitename;
2523                         case 'server':
2524                                 return $wgServer;
2525                         case 'servername':
2526                                 return $wgServerName;
2527                         case 'scriptpath':
2528                                 return $wgScriptPath;
2529                         case 'directionmark':
2530                                 return $wgContLang->getDirMark();
2531                         case 'contentlanguage':
2532                                 global $wgContLanguageCode;
2533                                 return $wgContLanguageCode;
2534                         default:
2535                                 $ret = null;
2536                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2537                                         return $ret;
2538                                 else
2539                                         return null;
2540                 }
2541         }
2542
2543         /**
2544          * initialise the magic variables (like CURRENTMONTHNAME)
2545          *
2546          * @private
2547          */
2548         function initialiseVariables() {
2549                 wfProfileIn( __METHOD__ );
2550                 $variableIDs = MagicWord::getVariableIDs();
2551
2552                 $this->mVariables = new MagicWordArray( $variableIDs );
2553                 wfProfileOut( __METHOD__ );
2554         }
2555
2556         /**
2557          * Preprocess some wikitext and return the document tree.
2558          * This is the ghost of replace_variables().
2559          *
2560          * @param string $text The text to parse
2561          * @param integer flags Bitwise combination of:
2562          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2563          *                                     included. Default is to assume a direct page view.
2564          *
2565          * The generated DOM tree must depend only on the input text and the flags.
2566          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2567          *
2568          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2569          * change in the DOM tree for a given text, must be passed through the section identifier
2570          * in the section edit link and thus back to extractSections().
2571          *
2572          * The output of this function is currently only cached in process memory, but a persistent
2573          * cache may be implemented at a later date which takes further advantage of these strict
2574          * dependency requirements.
2575          *
2576          * @private
2577          */
2578         function preprocessToDom ( $text, $flags = 0 ) {
2579                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2580                 return $dom;
2581         }
2582
2583         /*
2584          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2585          */
2586         public static function splitWhitespace( $s ) {
2587                 $ltrimmed = ltrim( $s );
2588                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2589                 $trimmed = rtrim( $ltrimmed );
2590                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2591                 if ( $diff > 0 ) {
2592                         $w2 = substr( $ltrimmed, -$diff );
2593                 } else {
2594                         $w2 = '';
2595                 }
2596                 return array( $w1, $trimmed, $w2 );
2597         }
2598
2599         /**
2600          * Replace magic variables, templates, and template arguments
2601          * with the appropriate text. Templates are substituted recursively,
2602          * taking care to avoid infinite loops.
2603          *
2604          * Note that the substitution depends on value of $mOutputType:
2605          *  self::OT_WIKI: only {{subst:}} templates
2606          *  self::OT_PREPROCESS: templates but not extension tags
2607          *  self::OT_HTML: all templates and extension tags
2608          *
2609          * @param string $tex The text to transform
2610          * @param PPFrame $frame Object describing the arguments passed to the template.
2611          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2612          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2613          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2614          * @private
2615          */
2616         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2617                 # Is there any text? Also, Prevent too big inclusions!
2618                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2619                         return $text;
2620                 }
2621                 wfProfileIn( __METHOD__ );
2622
2623                 if ( $frame === false ) {
2624                         $frame = $this->getPreprocessor()->newFrame();
2625                 } elseif ( !( $frame instanceof PPFrame ) ) {
2626                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2627                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2628                 }
2629
2630                 $dom = $this->preprocessToDom( $text );
2631                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2632                 $text = $frame->expand( $dom, $flags );
2633
2634                 wfProfileOut( __METHOD__ );
2635                 return $text;
2636         }
2637
2638         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2639         static function createAssocArgs( $args ) {
2640                 $assocArgs = array();
2641                 $index = 1;
2642                 foreach( $args as $arg ) {
2643                         $eqpos = strpos( $arg, '=' );
2644                         if ( $eqpos === false ) {
2645                                 $assocArgs[$index++] = $arg;
2646                         } else {
2647                                 $name = trim( substr( $arg, 0, $eqpos ) );
2648                                 $value = trim( substr( $arg, $eqpos+1 ) );
2649                                 if ( $value === false ) {
2650                                         $value = '';
2651                                 }
2652                                 if ( $name !== false ) {
2653                                         $assocArgs[$name] = $value;
2654                                 }
2655                         }
2656                 }
2657
2658                 return $assocArgs;
2659         }
2660
2661         /**
2662          * Warn the user when a parser limitation is reached
2663          * Will warn at most once the user per limitation type
2664          *
2665          * @param string $limitationType, should be one of:
2666          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2667          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2668          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2669          * @params int $current, $max When an explicit limit has been
2670          *       exceeded, provide the values (optional)
2671          */
2672         function limitationWarn( $limitationType, $current=null, $max=null) {
2673                 $msgName = $limitationType . '-warning';
2674                 //does no harm if $current and $max are present but are unnecessary for the message
2675                 $warning = wfMsgExt( $msgName, array( 'parsemag', 'escape' ), $current, $max );
2676                 $this->mOutput->addWarning( $warning );
2677                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2678                 if ( $cat ) {
2679                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2680                 }
2681         }
2682
2683         /**
2684          * Return the text of a template, after recursively
2685          * replacing any variables or templates within the template.
2686          *
2687          * @param array $piece The parts of the template
2688          *  $piece['title']: the title, i.e. the part before the |
2689          *  $piece['parts']: the parameter array
2690          *  $piece['lineStart']: whether the brace was at the start of a line
2691          * @param PPFrame The current frame, contains template arguments
2692          * @return string the text of the template
2693          * @private
2694          */
2695         function braceSubstitution( $piece, $frame ) {
2696                 global $wgContLang, $wgNonincludableNamespaces;
2697                 wfProfileIn( __METHOD__ );
2698                 wfProfileIn( __METHOD__.'-setup' );
2699
2700                 # Flags
2701                 $found = false;             # $text has been filled
2702                 $nowiki = false;            # wiki markup in $text should be escaped
2703                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2704                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2705                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2706                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2707
2708                 # Title object, where $text came from
2709                 $title = NULL;
2710
2711                 # $part1 is the bit before the first |, and must contain only title characters.
2712                 # Various prefixes will be stripped from it later.
2713                 $titleWithSpaces = $frame->expand( $piece['title'] );
2714                 $part1 = trim( $titleWithSpaces );
2715                 $titleText = false;
2716
2717                 # Original title text preserved for various purposes
2718                 $originalTitle = $part1;
2719
2720                 # $args is a list of argument nodes, starting from index 0, not including $part1
2721                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2722                 wfProfileOut( __METHOD__.'-setup' );
2723
2724                 # SUBST
2725                 wfProfileIn( __METHOD__.'-modifiers' );
2726                 if ( !$found ) {
2727                         $mwSubst = MagicWord::get( 'subst' );
2728                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2729                                 # One of two possibilities is true:
2730                                 # 1) Found SUBST but not in the PST phase
2731                                 # 2) Didn't find SUBST and in the PST phase
2732                                 # In either case, return without further processing
2733                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2734                                 $isLocalObj = true;
2735                                 $found = true;
2736                         }
2737                 }
2738
2739                 # Variables
2740                 if ( !$found && $args->getLength() == 0 ) {
2741                         $id = $this->mVariables->matchStartToEnd( $part1 );
2742                         if ( $id !== false ) {
2743                                 $text = $this->getVariableValue( $id );
2744                                 if (MagicWord::getCacheTTL($id)>-1)
2745                                         $this->mOutput->mContainsOldMagic = true;
2746                                 $found = true;
2747                         }
2748                 }
2749
2750                 # MSG, MSGNW and RAW
2751                 if ( !$found ) {
2752                         # Check for MSGNW:
2753                         $mwMsgnw = MagicWord::get( 'msgnw' );
2754                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2755                                 $nowiki = true;
2756                         } else {
2757                                 # Remove obsolete MSG:
2758                                 $mwMsg = MagicWord::get( 'msg' );
2759                                 $mwMsg->matchStartAndRemove( $part1 );
2760                         }
2761
2762                         # Check for RAW:
2763                         $mwRaw = MagicWord::get( 'raw' );
2764                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2765                                 $forceRawInterwiki = true;
2766                         }
2767                 }
2768                 wfProfileOut( __METHOD__.'-modifiers' );
2769
2770                 # Parser functions
2771                 if ( !$found ) {
2772                         wfProfileIn( __METHOD__ . '-pfunc' );
2773
2774                         $colonPos = strpos( $part1, ':' );
2775                         if ( $colonPos !== false ) {
2776                                 # Case sensitive functions
2777                                 $function = substr( $part1, 0, $colonPos );
2778                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2779                                         $function = $this->mFunctionSynonyms[1][$function];
2780                                 } else {
2781                                         # Case insensitive functions
2782                                         $function = strtolower( $function );
2783                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2784                                                 $function = $this->mFunctionSynonyms[0][$function];
2785                                         } else {
2786                                                 $function = false;
2787                                         }
2788                                 }
2789                                 if ( $function ) {
2790                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2791                                         $initialArgs = array( &$this );
2792                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2793                                         if ( $flags & SFH_OBJECT_ARGS ) {
2794                                                 # Add a frame parameter, and pass the arguments as an array
2795                                                 $allArgs = $initialArgs;
2796                                                 $allArgs[] = $frame;
2797                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2798                                                         $funcArgs[] = $args->item( $i );
2799                                                 }
2800                                                 $allArgs[] = $funcArgs;
2801                                         } else {
2802                                                 # Convert arguments to plain text
2803                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2804                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2805                                                 }
2806                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2807                                         }
2808
2809                                         # Workaround for PHP bug 35229 and similar
2810                                         if ( !is_callable( $callback ) ) {
2811                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2812                                         }
2813                                         $result = call_user_func_array( $callback, $allArgs );
2814                                         $found = true;
2815                                         $noparse = true;
2816                                         $preprocessFlags = 0;
2817
2818                                         if ( is_array( $result ) ) {
2819                                                 if ( isset( $result[0] ) ) {
2820                                                         $text = $result[0];
2821                                                         unset( $result[0] );
2822                                                 }
2823
2824                                                 // Extract flags into the local scope
2825                                                 // This allows callers to set flags such as nowiki, found, etc.
2826                                                 extract( $result );
2827                                         } else {
2828                                                 $text = $result;
2829                                         }
2830                                         if ( !$noparse ) {
2831                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2832                                                 $isChildObj = true;
2833                                         }
2834                                 }
2835                         }
2836                         wfProfileOut( __METHOD__ . '-pfunc' );
2837                 }
2838
2839                 # Finish mangling title and then check for loops.
2840                 # Set $title to a Title object and $titleText to the PDBK
2841                 if ( !$found ) {
2842                         $ns = NS_TEMPLATE;
2843                         # Split the title into page and subpage
2844                         $subpage = '';
2845                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2846                         if ($subpage !== '') {
2847                                 $ns = $this->mTitle->getNamespace();
2848                         }
2849                         $title = Title::newFromText( $part1, $ns );
2850                         if ( $title ) {
2851                                 $titleText = $title->getPrefixedText();
2852                                 # Check for language variants if the template is not found
2853                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2854                                         $wgContLang->findVariantLink( $part1, $title, true );
2855                                 }
2856                                 # Do recursion depth check
2857                                 $limit = $this->mOptions->getMaxTemplateDepth();
2858                                 if ( $frame->depth >= $limit ) {
2859                                         $found = true;
2860                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2861                                 }
2862                         }
2863                 }
2864
2865                 # Load from database
2866                 if ( !$found && $title ) {
2867                         wfProfileIn( __METHOD__ . '-loadtpl' );
2868                         if ( !$title->isExternal() ) {
2869                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2870                                         $text = SpecialPage::capturePath( $title );
2871                                         if ( is_string( $text ) ) {
2872                                                 $found = true;
2873                                                 $isHTML = true;
2874                                                 $this->disableCache();
2875                                         }
2876                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2877                                         $found = false; //access denied
2878                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2879                                 } else {
2880                                         list( $text, $title ) = $this->getTemplateDom( $title );
2881                                         if ( $text !== false ) {
2882                                                 $found = true;
2883                                                 $isChildObj = true;
2884                                         }
2885                                 }
2886
2887                                 # If the title is valid but undisplayable, make a link to it
2888                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2889                                         $text = "[[:$titleText]]";
2890                                         $found = true;
2891                                 }
2892                         } elseif ( $title->isTrans() ) {
2893                                 // Interwiki transclusion
2894                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2895                                         $text = $this->interwikiTransclude( $title, 'render' );
2896                                         $isHTML = true;
2897                                 } else {
2898                                         $text = $this->interwikiTransclude( $title, 'raw' );
2899                                         // Preprocess it like a template
2900                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2901                                         $isChildObj = true;
2902                                 }
2903                                 $found = true;
2904                         }
2905
2906                         # Do infinite loop check
2907                         # This has to be done after redirect resolution to avoid infinite loops via redirects
2908                         if ( !$frame->loopCheck( $title ) ) {
2909                                 $found = true;
2910                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2911                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2912                         }
2913                         wfProfileOut( __METHOD__ . '-loadtpl' );
2914                 }
2915
2916                 # If we haven't found text to substitute by now, we're done
2917                 # Recover the source wikitext and return it
2918                 if ( !$found ) {
2919                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2920                         wfProfileOut( __METHOD__ );
2921                         return array( 'object' => $text );
2922                 }
2923
2924                 # Expand DOM-style return values in a child frame
2925                 if ( $isChildObj ) {
2926                         # Clean up argument array
2927                         $newFrame = $frame->newChild( $args, $title );
2928
2929                         if ( $nowiki ) {
2930                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2931                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2932                                 # Expansion is eligible for the empty-frame cache
2933                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2934                                         $text = $this->mTplExpandCache[$titleText];
2935                                 } else {
2936                                         $text = $newFrame->expand( $text );
2937                                         $this->mTplExpandCache[$titleText] = $text;
2938                                 }
2939                         } else {
2940                                 # Uncached expansion
2941                                 $text = $newFrame->expand( $text );
2942                         }
2943                 }
2944                 if ( $isLocalObj && $nowiki ) {
2945                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2946                         $isLocalObj = false;
2947                 }
2948
2949                 # Replace raw HTML by a placeholder
2950                 # Add a blank line preceding, to prevent it from mucking up
2951                 # immediately preceding headings
2952                 if ( $isHTML ) {
2953                         $text = "\n\n" . $this->insertStripItem( $text );
2954                 }
2955                 # Escape nowiki-style return values
2956                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2957                         $text = wfEscapeWikiText( $text );
2958                 }
2959                 # Bug 529: if the template begins with a table or block-level
2960                 # element, it should be treated as beginning a new line.
2961                 # This behaviour is somewhat controversial.
2962                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
2963                         $text = "\n" . $text;
2964                 }
2965
2966                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
2967                         # Error, oversize inclusion
2968                         $text = "[[$originalTitle]]" .
2969                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
2970                         $this->limitationWarn( 'post-expand-template-inclusion' );
2971                 }
2972
2973                 if ( $isLocalObj ) {
2974                         $ret = array( 'object' => $text );
2975                 } else {
2976                         $ret = array( 'text' => $text );
2977                 }
2978
2979                 wfProfileOut( __METHOD__ );
2980                 return $ret;
2981         }
2982
2983         /**
2984          * Get the semi-parsed DOM representation of a template with a given title,
2985          * and its redirect destination title. Cached.
2986          */
2987         function getTemplateDom( $title ) {
2988                 $cacheTitle = $title;
2989                 $titleText = $title->getPrefixedDBkey();
2990
2991                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
2992                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
2993                         $title = Title::makeTitle( $ns, $dbk );
2994                         $titleText = $title->getPrefixedDBkey();
2995                 }
2996                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
2997                         return array( $this->mTplDomCache[$titleText], $title );
2998                 }
2999
3000                 // Cache miss, go to the database
3001                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3002
3003                 if ( $text === false ) {
3004                         $this->mTplDomCache[$titleText] = false;
3005                         return array( false, $title );
3006                 }
3007
3008                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3009                 $this->mTplDomCache[ $titleText ] = $dom;
3010
3011                 if (! $title->equals($cacheTitle)) {
3012                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3013                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3014                 }
3015
3016                 return array( $dom, $title );
3017         }
3018
3019         /**
3020          * Fetch the unparsed text of a template and register a reference to it.
3021          */
3022         function fetchTemplateAndTitle( $title ) {
3023                 $templateCb = $this->mOptions->getTemplateCallback();
3024                 $stuff = call_user_func( $templateCb, $title, $this );
3025                 $text = $stuff['text'];
3026                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3027                 if ( isset( $stuff['deps'] ) ) {
3028                         foreach ( $stuff['deps'] as $dep ) {
3029                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3030                         }
3031                 }
3032                 return array($text,$finalTitle);
3033         }
3034
3035         function fetchTemplate( $title ) {
3036                 $rv = $this->fetchTemplateAndTitle($title);
3037                 return $rv[0];
3038         }
3039
3040         /**
3041          * Static function to get a template
3042          * Can be overridden via ParserOptions::setTemplateCallback().
3043          */
3044         static function statelessFetchTemplate( $title, $parser=false ) {
3045                 $text = $skip = false;
3046                 $finalTitle = $title;
3047                 $deps = array();
3048
3049                 // Loop to fetch the article, with up to 1 redirect
3050                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3051                         # Give extensions a chance to select the revision instead
3052                         $id = false; // Assume current
3053                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3054
3055                         if( $skip ) {
3056                                 $text = false;
3057                                 $deps[] = array(
3058                                         'title' => $title,
3059                                         'page_id' => $title->getArticleID(),
3060                                         'rev_id' => null );
3061                                 break;
3062                         }
3063                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3064                         $rev_id = $rev ? $rev->getId() : 0;
3065                         // If there is no current revision, there is no page
3066                         if( $id === false && !$rev ) {
3067                                 $linkCache = LinkCache::singleton();
3068                                 $linkCache->addBadLinkObj( $title );
3069                         }
3070
3071                         $deps[] = array(
3072                                 'title' => $title,
3073                                 'page_id' => $title->getArticleID(),
3074                                 'rev_id' => $rev_id );
3075
3076                         if( $rev ) {
3077                                 $text = $rev->getText();
3078                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3079                                 global $wgContLang;
3080                                 $message = $wgContLang->lcfirst( $title->getText() );
3081                                 $text = wfMsgForContentNoTrans( $message );
3082                                 if( wfEmptyMsg( $message, $text ) ) {
3083                                         $text = false;
3084                                         break;
3085                                 }
3086                         } else {
3087                                 break;
3088                         }
3089                         if ( $text === false ) {
3090                                 break;
3091                         }
3092                         // Redirect?
3093                         $finalTitle = $title;
3094                         $title = Title::newFromRedirect( $text );
3095                 }
3096                 return array(
3097                         'text' => $text,
3098                         'finalTitle' => $finalTitle,
3099                         'deps' => $deps );
3100         }
3101
3102         /**
3103          * Transclude an interwiki link.
3104          */
3105         function interwikiTransclude( $title, $action ) {
3106                 global $wgEnableScaryTranscluding;
3107
3108                 if (!$wgEnableScaryTranscluding)
3109                         return wfMsg('scarytranscludedisabled');
3110
3111                 $url = $title->getFullUrl( "action=$action" );
3112
3113                 if (strlen($url) > 255)
3114                         return wfMsg('scarytranscludetoolong');
3115                 return $this->fetchScaryTemplateMaybeFromCache($url);
3116         }
3117
3118         function fetchScaryTemplateMaybeFromCache($url) {
3119                 global $wgTranscludeCacheExpiry;
3120                 $dbr = wfGetDB(DB_SLAVE);
3121                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3122                                 array('tc_url' => $url));
3123                 if ($obj) {
3124                         $time = $obj->tc_time;
3125                         $text = $obj->tc_contents;
3126                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3127                                 return $text;
3128                         }
3129                 }
3130
3131                 $text = Http::get($url);
3132                 if (!$text)
3133                         return wfMsg('scarytranscludefailed', $url);
3134
3135                 $dbw = wfGetDB(DB_MASTER);
3136                 $dbw->replace('transcache', array('tc_url'), array(
3137                         'tc_url' => $url,
3138                         'tc_time' => time(),
3139                         'tc_contents' => $text));
3140                 return $text;
3141         }
3142
3143
3144         /**
3145          * Triple brace replacement -- used for template arguments
3146          * @private
3147          */
3148         function argSubstitution( $piece, $frame ) {
3149                 wfProfileIn( __METHOD__ );
3150
3151                 $error = false;
3152                 $parts = $piece['parts'];
3153                 $nameWithSpaces = $frame->expand( $piece['title'] );
3154                 $argName = trim( $nameWithSpaces );
3155                 $object = false;
3156                 $text = $frame->getArgument( $argName );
3157                 if (  $text === false && $parts->getLength() > 0
3158                   && (
3159                     $this->ot['html']
3160                     || $this->ot['pre']
3161                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3162                   )
3163                 ) {
3164                         # No match in frame, use the supplied default
3165                         $object = $parts->item( 0 )->getChildren();
3166                 }
3167                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3168                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3169                         $this->limitationWarn( 'post-expand-template-argument' );
3170                 }
3171
3172                 if ( $text === false && $object === false ) {
3173                         # No match anywhere
3174                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3175                 }
3176                 if ( $error !== false ) {
3177                         $text .= $error;
3178                 }
3179                 if ( $object !== false ) {
3180                         $ret = array( 'object' => $object );
3181                 } else {
3182                         $ret = array( 'text' => $text );
3183                 }
3184
3185                 wfProfileOut( __METHOD__ );
3186                 return $ret;
3187         }
3188
3189         /**
3190          * Return the text to be used for a given extension tag.
3191          * This is the ghost of strip().
3192          *
3193          * @param array $params Associative array of parameters:
3194          *     name       PPNode for the tag name
3195          *     attr       PPNode for unparsed text where tag attributes are thought to be
3196          *     attributes Optional associative array of parsed attributes
3197          *     inner      Contents of extension element
3198          *     noClose    Original text did not have a close tag
3199          * @param PPFrame $frame
3200          */
3201         function extensionSubstitution( $params, $frame ) {
3202                 global $wgRawHtml, $wgContLang;
3203
3204                 $name = $frame->expand( $params['name'] );
3205                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3206                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3207
3208                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3209
3210                 if ( $this->ot['html'] ) {
3211                         $name = strtolower( $name );
3212
3213                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3214                         if ( isset( $params['attributes'] ) ) {
3215                                 $attributes = $attributes + $params['attributes'];
3216                         }
3217                         switch ( $name ) {
3218                                 case 'html':
3219                                         if( $wgRawHtml ) {
3220                                                 $output = $content;
3221                                                 break;
3222                                         } else {
3223                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3224                                         }
3225                                 case 'nowiki':
3226                                         $content = strtr($content, array('-{' => '-&#123;', '}-' => '&#125;-'));
3227                                         $output = Xml::escapeTagsOnly( $content );
3228                                         break;
3229                                 case 'math':
3230                                         $output = $wgContLang->armourMath(
3231                                                 MathRenderer::renderMath( $content, $attributes ) );
3232                                         break;
3233                                 case 'gallery':
3234                                         $output = $this->renderImageGallery( $content, $attributes );
3235                                         break;
3236                                 default:
3237                                         if( isset( $this->mTagHooks[$name] ) ) {
3238                                                 # Workaround for PHP bug 35229 and similar
3239                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3240                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3241                                                 }
3242                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3243                                                         array( $content, $attributes, $this ) );
3244                                         } else {
3245                                                 $output = '<span class="error">Invalid tag extension name: ' .
3246                                                         htmlspecialchars( $name ) . '</span>';
3247                                         }
3248                         }
3249                 } else {
3250                         if ( is_null( $attrText ) ) {
3251                                 $attrText = '';
3252                         }
3253                         if ( isset( $params['attributes'] ) ) {
3254                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3255                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3256                                                 htmlspecialchars( $attrValue ) . '"';
3257                                 }
3258                         }
3259                         if ( $content === null ) {
3260                                 $output = "<$name$attrText/>";
3261                         } else {
3262                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3263                                 $output = "<$name$attrText>$content$close";
3264                         }
3265                 }
3266
3267                 if ( $name === 'html' || $name === 'nowiki' ) {
3268                         $this->mStripState->nowiki->setPair( $marker, $output );
3269                 } else {
3270                         $this->mStripState->general->setPair( $marker, $output );
3271                 }
3272                 return $marker;
3273         }
3274
3275         /**
3276          * Increment an include size counter
3277          *
3278          * @param string $type The type of expansion
3279          * @param integer $size The size of the text
3280          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3281          */
3282         function incrementIncludeSize( $type, $size ) {
3283                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3284                         return false;
3285                 } else {
3286                         $this->mIncludeSizes[$type] += $size;
3287                         return true;
3288                 }
3289         }
3290
3291         /**
3292          * Increment the expensive function count
3293          *
3294          * @return boolean False if the limit has been exceeded
3295          */
3296         function incrementExpensiveFunctionCount() {
3297                 global $wgExpensiveParserFunctionLimit;
3298                 $this->mExpensiveFunctionCount++;
3299                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3300                         return true;
3301                 }
3302                 return false;
3303         }
3304
3305         /**
3306          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3307          * Fills $this->mDoubleUnderscores, returns the modified text
3308          */
3309         function doDoubleUnderscore( $text ) {
3310                 wfProfileIn( __METHOD__ );
3311                 // The position of __TOC__ needs to be recorded
3312                 $mw = MagicWord::get( 'toc' );
3313                 if( $mw->match( $text ) ) {
3314                         $this->mShowToc = true;
3315                         $this->mForceTocPosition = true;
3316
3317                         // Set a placeholder. At the end we'll fill it in with the TOC.
3318                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3319
3320                         // Only keep the first one.
3321                         $text = $mw->replace( '', $text );
3322                 }
3323
3324                 // Now match and remove the rest of them
3325                 $mwa = MagicWord::getDoubleUnderscoreArray();
3326                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3327
3328                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3329                         $this->mOutput->mNoGallery = true;
3330                 }
3331                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3332                         $this->mShowToc = false;
3333                 }
3334                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3335                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3336
3337                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3338                         if ( $containerCategory ) {
3339                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3340                         } else {
3341                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3342                         }
3343                 }
3344                 # (bug 8068) Allow control over whether robots index a page.
3345                 #
3346                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3347                 # is not desirable, the last one on the page should win.
3348                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3349                         $this->mOutput->setIndexPolicy( 'noindex' );
3350                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3351                         $this->mOutput->setIndexPolicy( 'index' );
3352                 }
3353                 wfProfileOut( __METHOD__ );
3354                 return $text;
3355         }
3356
3357         /**
3358          * This function accomplishes several tasks:
3359          * 1) Auto-number headings if that option is enabled
3360          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3361          * 3) Add a Table of contents on the top for users who have enabled the option
3362          * 4) Auto-anchor headings
3363          *
3364          * It loops through all headlines, collects the necessary data, then splits up the
3365          * string and re-inserts the newly formatted headlines.
3366          *
3367          * @param string $text
3368          * @param boolean $isMain
3369          * @private
3370          */
3371         function formatHeadings( $text, $isMain=true ) {
3372                 global $wgMaxTocLevel, $wgContLang, $wgEnforceHtmlIds;
3373
3374                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3375                 $showEditLink = $this->mOptions->getEditSection();
3376
3377                 // Do not call quickUserCan unless necessary
3378                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3379                         $showEditLink = 0;
3380                 }
3381
3382                 # Inhibit editsection links if requested in the page
3383                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3384                         $showEditLink = 0;
3385                 }
3386
3387                 # Get all headlines for numbering them and adding funky stuff like [edit]
3388                 # links - this is for later, but we need the number of headlines right now
3389                 $matches = array();
3390                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3391
3392                 # if there are fewer than 4 headlines in the article, do not show TOC
3393                 # unless it's been explicitly enabled.
3394                 $enoughToc = $this->mShowToc &&
3395                         (($numMatches >= 4) || $this->mForceTocPosition);
3396
3397                 # Allow user to stipulate that a page should have a "new section"
3398                 # link added via __NEWSECTIONLINK__
3399                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3400                         $this->mOutput->setNewSection( true );
3401                 }
3402
3403                 # Allow user to remove the "new section"
3404                 # link via __NONEWSECTIONLINK__
3405                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3406                         $this->mOutput->hideNewSection( true );
3407                 }
3408
3409                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3410                 # override above conditions and always show TOC above first header
3411                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3412                         $this->mShowToc = true;
3413                         $enoughToc = true;
3414                 }
3415
3416                 # We need this to perform operations on the HTML
3417                 $sk = $this->mOptions->getSkin();
3418
3419                 # headline counter
3420                 $headlineCount = 0;
3421                 $numVisible = 0;
3422
3423                 # Ugh .. the TOC should have neat indentation levels which can be
3424                 # passed to the skin functions. These are determined here
3425                 $toc = '';
3426                 $full = '';
3427                 $head = array();
3428                 $sublevelCount = array();
3429                 $levelCount = array();
3430                 $toclevel = 0;
3431                 $level = 0;
3432                 $prevlevel = 0;
3433                 $toclevel = 0;
3434                 $prevtoclevel = 0;
3435                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3436                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3437                 $tocraw = array();
3438
3439                 foreach( $matches[3] as $headline ) {
3440                         $isTemplate = false;
3441                         $titleText = false;
3442                         $sectionIndex = false;
3443                         $numbering = '';
3444                         $markerMatches = array();
3445                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3446                                 $serial = $markerMatches[1];
3447                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3448                                 $isTemplate = ($titleText != $baseTitleText);
3449                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3450                         }
3451
3452                         if( $toclevel ) {
3453                                 $prevlevel = $level;
3454                                 $prevtoclevel = $toclevel;
3455                         }
3456                         $level = $matches[1][$headlineCount];
3457
3458                         if( $doNumberHeadings || $enoughToc ) {
3459
3460                                 if ( $level > $prevlevel ) {
3461                                         # Increase TOC level
3462                                         $toclevel++;
3463                                         $sublevelCount[$toclevel] = 0;
3464                                         if( $toclevel<$wgMaxTocLevel ) {
3465                                                 $prevtoclevel = $toclevel;
3466                                                 $toc .= $sk->tocIndent();
3467                                                 $numVisible++;
3468                                         }
3469                                 }
3470                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3471                                         # Decrease TOC level, find level to jump to
3472
3473                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3474                                                 # Can only go down to level 1
3475                                                 $toclevel = 1;
3476                                         } else {
3477                                                 for ($i = $toclevel; $i > 0; $i--) {
3478                                                         if ( $levelCount[$i] == $level ) {
3479                                                                 # Found last matching level
3480                                                                 $toclevel = $i;
3481                                                                 break;
3482                                                         }
3483                                                         elseif ( $levelCount[$i] < $level ) {
3484                                                                 # Found first matching level below current level
3485                                                                 $toclevel = $i + 1;
3486                                                                 break;
3487                                                         }
3488                                                 }
3489                                         }
3490                                         if( $toclevel<$wgMaxTocLevel ) {
3491                                                 if($prevtoclevel < $wgMaxTocLevel) {
3492                                                         # Unindent only if the previous toc level was shown :p
3493                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3494                                                         $prevtoclevel = $toclevel;
3495                                                 } else {
3496                                                         $toc .= $sk->tocLineEnd();
3497                                                 }
3498                                         }
3499                                 }
3500                                 else {
3501                                         # No change in level, end TOC line
3502                                         if( $toclevel<$wgMaxTocLevel ) {
3503                                                 $toc .= $sk->tocLineEnd();
3504                                         }
3505                                 }
3506
3507                                 $levelCount[$toclevel] = $level;
3508
3509                                 # count number of headlines for each level
3510                                 @$sublevelCount[$toclevel]++;
3511                                 $dot = 0;
3512                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3513                                         if( !empty( $sublevelCount[$i] ) ) {
3514                                                 if( $dot ) {
3515                                                         $numbering .= '.';
3516                                                 }
3517                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3518                                                 $dot = 1;
3519                                         }
3520                                 }
3521                         }
3522
3523                         # The safe header is a version of the header text safe to use for links
3524                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3525                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3526
3527                         # Remove link placeholders by the link text.
3528                         #     <!--LINK number-->
3529                         # turns into
3530                         #     link text with suffix
3531                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3532
3533                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3534                         $tocline = preg_replace(
3535                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3536                                 array( '',                          '<$1>'),
3537                                 $safeHeadline
3538                         );
3539                         $tocline = trim( $tocline );
3540
3541                         # For the anchor, strip out HTML-y stuff period
3542                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3543                         $safeHeadline = trim( $safeHeadline );
3544
3545                         # Save headline for section edit hint before it's escaped
3546                         $headlineHint = $safeHeadline;
3547
3548                         if ( $wgEnforceHtmlIds ) {
3549                                 $legacyHeadline = false;
3550                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3551                                         'noninitial' );
3552                         } else {
3553                                 # For reverse compatibility, provide an id that's
3554                                 # HTML4-compatible, like we used to.
3555                                 #
3556                                 # It may be worth noting, academically, that it's possible for
3557                                 # the legacy anchor to conflict with a non-legacy headline
3558                                 # anchor on the page.  In this case likely the "correct" thing
3559                                 # would be to either drop the legacy anchors or make sure
3560                                 # they're numbered first.  However, this would require people
3561                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3562                                 # manually, so let's not bother worrying about it.
3563                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3564                                         'noninitial' );
3565                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline, 'xml' );
3566
3567                                 if ( $legacyHeadline == $safeHeadline ) {
3568                                         # No reason to have both (in fact, we can't)
3569                                         $legacyHeadline = false;
3570                                 } elseif ( $legacyHeadline != Sanitizer::escapeId(
3571                                 $legacyHeadline, 'xml' ) ) {
3572                                         # The legacy id is invalid XML.  We used to allow this, but
3573                                         # there's no reason to do so anymore.  Backward
3574                                         # compatibility will fail slightly in this case, but it's
3575                                         # no big deal.
3576                                         $legacyHeadline = false;
3577                                 }
3578                         }
3579
3580                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3581                         # Does this apply to Unicode characters?  Because we aren't
3582                         # handling those here.
3583                         $arrayKey = strtolower( $safeHeadline );
3584                         if ( $legacyHeadline === false ) {
3585                                 $legacyArrayKey = false;
3586                         } else {
3587                                 $legacyArrayKey = strtolower( $legacyHeadline );
3588                         }
3589
3590                         # count how many in assoc. array so we can track dupes in anchors
3591                         if ( isset( $refers[$arrayKey] ) ) {
3592                                 $refers[$arrayKey]++;
3593                         } else {
3594                                 $refers[$arrayKey] = 1;
3595                         }
3596                         if ( isset( $refers[$legacyArrayKey] ) ) {
3597                                 $refers[$legacyArrayKey]++;
3598                         } else {
3599                                 $refers[$legacyArrayKey] = 1;
3600                         }
3601
3602                         # Don't number the heading if it is the only one (looks silly)
3603                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3604                                 # the two are different if the line contains a link
3605                                 $headline=$numbering . ' ' . $headline;
3606                         }
3607
3608                         # Create the anchor for linking from the TOC to the section
3609                         $anchor = $safeHeadline;
3610                         $legacyAnchor = $legacyHeadline;
3611                         if ( $refers[$arrayKey] > 1 ) {
3612                                 $anchor .= '_' . $refers[$arrayKey];
3613                         }
3614                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3615                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3616                         }
3617                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3618                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3619                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3620                         }
3621                         # give headline the correct <h#> tag
3622                         if( $showEditLink && $sectionIndex !== false ) {
3623                                 if( $isTemplate ) {
3624                                         # Put a T flag in the section identifier, to indicate to extractSections()
3625                                         # that sections inside <includeonly> should be counted.
3626                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3627                                 } else {
3628                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3629                                 }
3630                         } else {
3631                                 $editlink = '';
3632                         }
3633                         $head[$headlineCount] = $sk->makeHeadline( $level,
3634                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3635                                 $editlink, $legacyAnchor );
3636
3637                         $headlineCount++;
3638                 }
3639
3640                 $this->mOutput->setSections( $tocraw );
3641
3642                 # Never ever show TOC if no headers
3643                 if( $numVisible < 1 ) {
3644                         $enoughToc = false;
3645                 }
3646
3647                 if( $enoughToc ) {
3648                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3649                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3650                         }
3651                         $toc = $sk->tocList( $toc );
3652                 }
3653
3654                 # split up and insert constructed headlines
3655
3656                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3657                 $i = 0;
3658
3659                 foreach( $blocks as $block ) {
3660                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3661                                 # This is the [edit] link that appears for the top block of text when
3662                                 # section editing is enabled
3663
3664                                 # Disabled because it broke block formatting
3665                                 # For example, a bullet point in the top line
3666                                 # $full .= $sk->editSectionLink(0);
3667                         }
3668                         $full .= $block;
3669                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3670                                 # Top anchor now in skin
3671                                 $full = $full.$toc;
3672                         }
3673
3674                         if( !empty( $head[$i] ) ) {
3675                                 $full .= $head[$i];
3676                         }
3677                         $i++;
3678                 }
3679                 if( $this->mForceTocPosition ) {
3680                         return str_replace( '<!--MWTOC-->', $toc, $full );
3681                 } else {
3682                         return $full;
3683                 }
3684         }
3685
3686         /**
3687          * Transform wiki markup when saving a page by doing \r\n -> \n
3688          * conversion, substitting signatures, {{subst:}} templates, etc.
3689          *
3690          * @param string $text the text to transform
3691          * @param Title &$title the Title object for the current article
3692          * @param User $user the User object describing the current user
3693          * @param ParserOptions $options parsing options
3694          * @param bool $clearState whether to clear the parser state first
3695          * @return string the altered wiki markup
3696          * @public
3697          */
3698         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3699                 $this->mOptions = $options;
3700                 $this->setTitle( $title );
3701                 $this->setOutputType( self::OT_WIKI );
3702
3703                 if ( $clearState ) {
3704                         $this->clearState();
3705                 }
3706
3707                 $pairs = array(
3708                         "\r\n" => "\n",
3709                 );
3710                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3711                 $text = $this->pstPass2( $text, $user );
3712                 $text = $this->mStripState->unstripBoth( $text );
3713                 return $text;
3714         }
3715
3716         /**
3717          * Pre-save transform helper function
3718          * @private
3719          */
3720         function pstPass2( $text, $user ) {
3721                 global $wgContLang, $wgLocaltimezone;
3722
3723                 /* Note: This is the timestamp saved as hardcoded wikitext to
3724                  * the database, we use $wgContLang here in order to give
3725                  * everyone the same signature and use the default one rather
3726                  * than the one selected in each user's preferences.
3727                  *
3728                  * (see also bug 12815)
3729                  */
3730                 $ts = $this->mOptions->getTimestamp();
3731                 $tz = wfMsgForContent( 'timezone-utc' );
3732                 if ( isset( $wgLocaltimezone ) ) {
3733                         $unixts = wfTimestamp( TS_UNIX, $ts );
3734                         $oldtz = getenv( 'TZ' );
3735                         putenv( 'TZ='.$wgLocaltimezone );
3736                         $ts = date( 'YmdHis', $unixts );
3737                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3738
3739                         /* Allow translation of timezones trough wiki. date() can return
3740                          * whatever crap the system uses, localised or not, so we cannot
3741                          * ship premade translations.
3742                          */
3743                         $key = 'timezone-' . strtolower( trim( $tz ) );
3744                         $value = wfMsgForContent( $key );
3745                         if ( !wfEmptyMsg( $key, $value ) ) $tz = $value;
3746
3747                         putenv( 'TZ='.$oldtz );
3748                 }
3749
3750                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3751
3752                 # Variable replacement
3753                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3754                 $text = $this->replaceVariables( $text );
3755
3756                 # Signatures
3757                 $sigText = $this->getUserSig( $user );
3758                 $text = strtr( $text, array(
3759                         '~~~~~' => $d,
3760                         '~~~~' => "$sigText $d",
3761                         '~~~' => $sigText
3762                 ) );
3763
3764                 # Context links: [[|name]] and [[name (context)|]]
3765                 #
3766                 global $wgLegalTitleChars;
3767                 $tc = "[$wgLegalTitleChars]";
3768                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3769
3770                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3771                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3772                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3773                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3774
3775                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3776                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3777                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3778                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3779
3780                 $t = $this->mTitle->getText();
3781                 $m = array();
3782                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3783                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3784                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3785                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3786                 } else {
3787                         # if there's no context, don't bother duplicating the title
3788                         $text = preg_replace( $p2, '[[\\1]]', $text );
3789                 }
3790
3791                 # Trim trailing whitespace
3792                 $text = rtrim( $text );
3793
3794                 return $text;
3795         }
3796
3797         /**
3798          * Fetch the user's signature text, if any, and normalize to
3799          * validated, ready-to-insert wikitext.
3800          *
3801          * @param User $user
3802          * @return string
3803          * @private
3804          */
3805         function getUserSig( &$user ) {
3806                 global $wgMaxSigChars;
3807
3808                 $username = $user->getName();
3809                 $nickname = $user->getOption( 'nickname' );
3810                 $nickname = $nickname === '' ? $username : $nickname;
3811
3812                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3813                         $nickname = $username;
3814                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3815                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3816                         # Sig. might contain markup; validate this
3817                         if( $this->validateSig( $nickname ) !== false ) {
3818                                 # Validated; clean up (if needed) and return it
3819                                 return $this->cleanSig( $nickname, true );
3820                         } else {
3821                                 # Failed to validate; fall back to the default
3822                                 $nickname = $username;
3823                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3824                         }
3825                 }
3826
3827                 // Make sure nickname doesnt get a sig in a sig
3828                 $nickname = $this->cleanSigInSig( $nickname );
3829
3830                 # If we're still here, make it a link to the user page
3831                 $userText = wfEscapeWikiText( $username );
3832                 $nickText = wfEscapeWikiText( $nickname );
3833                 if ( $user->isAnon() )  {
3834                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3835                 } else {
3836                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3837                 }
3838         }
3839
3840         /**
3841          * Check that the user's signature contains no bad XML
3842          *
3843          * @param string $text
3844          * @return mixed An expanded string, or false if invalid.
3845          */
3846         function validateSig( $text ) {
3847                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3848         }
3849
3850         /**
3851          * Clean up signature text
3852          *
3853          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3854          * 2) Substitute all transclusions
3855          *
3856          * @param string $text
3857          * @param $parsing Whether we're cleaning (preferences save) or parsing
3858          * @return string Signature text
3859          */
3860         function cleanSig( $text, $parsing = false ) {
3861                 if ( !$parsing ) {
3862                         global $wgTitle;
3863                         $this->clearState();
3864                         $this->setTitle( $wgTitle );
3865                         $this->mOptions = new ParserOptions;
3866                         $this->setOutputType = self::OT_PREPROCESS;
3867                 }
3868
3869                 # Option to disable this feature
3870                 if ( !$this->mOptions->getCleanSignatures() ) {
3871                         return $text;
3872                 }
3873
3874                 # FIXME: regex doesn't respect extension tags or nowiki
3875                 #  => Move this logic to braceSubstitution()
3876                 $substWord = MagicWord::get( 'subst' );
3877                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3878                 $substText = '{{' . $substWord->getSynonym( 0 );
3879
3880                 $text = preg_replace( $substRegex, $substText, $text );
3881                 $text = $this->cleanSigInSig( $text );
3882                 $dom = $this->preprocessToDom( $text );
3883                 $frame = $this->getPreprocessor()->newFrame();
3884                 $text = $frame->expand( $dom );
3885
3886                 if ( !$parsing ) {
3887                         $text = $this->mStripState->unstripBoth( $text );
3888                 }
3889
3890                 return $text;
3891         }
3892
3893         /**
3894          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3895          * @param string $text
3896          * @return string Signature text with /~{3,5}/ removed
3897          */
3898         function cleanSigInSig( $text ) {
3899                 $text = preg_replace( '/~{3,5}/', '', $text );
3900                 return $text;
3901         }
3902
3903         /**
3904          * Set up some variables which are usually set up in parse()
3905          * so that an external function can call some class members with confidence
3906          * @public
3907          */
3908         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3909                 $this->setTitle( $title );
3910                 $this->mOptions = $options;
3911                 $this->setOutputType( $outputType );
3912                 if ( $clearState ) {
3913                         $this->clearState();
3914                 }
3915         }
3916
3917         /**
3918          * Wrapper for preprocess()
3919          *
3920          * @param string $text the text to preprocess
3921          * @param ParserOptions $options  options
3922          * @return string
3923          * @public
3924          */
3925         function transformMsg( $text, $options ) {
3926                 global $wgTitle;
3927                 static $executing = false;
3928
3929                 # Guard against infinite recursion
3930                 if ( $executing ) {
3931                         return $text;
3932                 }
3933                 $executing = true;
3934
3935                 wfProfileIn(__METHOD__);
3936                 $text = $this->preprocess( $text, $wgTitle, $options );
3937
3938                 $executing = false;
3939                 wfProfileOut(__METHOD__);
3940                 return $text;
3941         }
3942
3943         /**
3944          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3945          * The callback should have the following form:
3946          *    function myParserHook( $text, $params, &$parser ) { ... }
3947          *
3948          * Transform and return $text. Use $parser for any required context, e.g. use
3949          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3950          *
3951          * @public
3952          *
3953          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3954          * @param mixed $callback The callback function (and object) to use for the tag
3955          *
3956          * @return The old value of the mTagHooks array associated with the hook
3957          */
3958         function setHook( $tag, $callback ) {
3959                 $tag = strtolower( $tag );
3960                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3961                 $this->mTagHooks[$tag] = $callback;
3962                 if( !in_array( $tag, $this->mStripList ) ) {
3963                         $this->mStripList[] = $tag;
3964                 }
3965
3966                 return $oldVal;
3967         }
3968
3969         function setTransparentTagHook( $tag, $callback ) {
3970                 $tag = strtolower( $tag );
3971                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3972                 $this->mTransparentTagHooks[$tag] = $callback;
3973
3974                 return $oldVal;
3975         }
3976
3977         /**
3978          * Remove all tag hooks
3979          */
3980         function clearTagHooks() {
3981                 $this->mTagHooks = array();
3982                 $this->mStripList = $this->mDefaultStripList;
3983         }
3984
3985         /**
3986          * Create a function, e.g. {{sum:1|2|3}}
3987          * The callback function should have the form:
3988          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3989          *
3990          * Or with SFH_OBJECT_ARGS:
3991          *    function myParserFunction( $parser, $frame, $args ) { ... }
3992          *
3993          * The callback may either return the text result of the function, or an array with the text
3994          * in element 0, and a number of flags in the other elements. The names of the flags are
3995          * specified in the keys. Valid flags are:
3996          *   found                     The text returned is valid, stop processing the template. This
3997          *                             is on by default.
3998          *   nowiki                    Wiki markup in the return value should be escaped
3999          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4000          *
4001          * @public
4002          *
4003          * @param string $id The magic word ID
4004          * @param mixed $callback The callback function (and object) to use
4005          * @param integer $flags a combination of the following flags:
4006          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4007          *
4008          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4009          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4010          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4011          *     the arguments, and to control the way they are expanded.
4012          *
4013          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4014          *     arguments, for instance:
4015          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4016          *
4017          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4018          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4019          *     working if/when this is changed.
4020          *
4021          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4022          *     expansion.
4023          *
4024          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4025          *     about the methods available in PPFrame and PPNode.
4026          *
4027          * @return The old callback function for this name, if any
4028          */
4029         function setFunctionHook( $id, $callback, $flags = 0 ) {
4030                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4031                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4032
4033                 # Add to function cache
4034                 $mw = MagicWord::get( $id );
4035                 if( !$mw )
4036                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4037
4038                 $synonyms = $mw->getSynonyms();
4039                 $sensitive = intval( $mw->isCaseSensitive() );
4040
4041                 foreach ( $synonyms as $syn ) {
4042                         # Case
4043                         if ( !$sensitive ) {
4044                                 $syn = strtolower( $syn );
4045                         }
4046                         # Add leading hash
4047                         if ( !( $flags & SFH_NO_HASH ) ) {
4048                                 $syn = '#' . $syn;
4049                         }
4050                         # Remove trailing colon
4051                         if ( substr( $syn, -1, 1 ) === ':' ) {
4052                                 $syn = substr( $syn, 0, -1 );
4053                         }
4054                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4055                 }
4056                 return $oldVal;
4057         }
4058
4059         /**
4060          * Get all registered function hook identifiers
4061          *
4062          * @return array
4063          */
4064         function getFunctionHooks() {
4065                 return array_keys( $this->mFunctionHooks );
4066         }
4067
4068         /**
4069          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4070          * Placeholders created in Skin::makeLinkObj()
4071          * Returns an array of link CSS classes, indexed by PDBK.
4072          */
4073         function replaceLinkHolders( &$text, $options = 0 ) {
4074                 return $this->mLinkHolders->replace( $text );
4075         }
4076
4077         /**
4078          * Replace <!--LINK--> link placeholders with plain text of links
4079          * (not HTML-formatted).
4080          * @param string $text
4081          * @return string
4082          */
4083         function replaceLinkHoldersText( $text ) {
4084                 return $this->mLinkHolders->replaceText( $text );
4085         }
4086
4087         /**
4088          * Tag hook handler for 'pre'.
4089          */
4090         function renderPreTag( $text, $attribs ) {
4091                 // Backwards-compatibility hack
4092                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4093
4094                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4095                 return Xml::openElement( 'pre', $attribs ) .
4096                         Xml::escapeTagsOnly( $content ) .
4097                         '</pre>';
4098         }
4099
4100         /**
4101          * Renders an image gallery from a text with one line per image.
4102          * text labels may be given by using |-style alternative text. E.g.
4103          *   Image:one.jpg|The number "1"
4104          *   Image:tree.jpg|A tree
4105          * given as text will return the HTML of a gallery with two images,
4106          * labeled 'The number "1"' and
4107          * 'A tree'.
4108          */
4109         function renderImageGallery( $text, $params ) {
4110                 $ig = new ImageGallery();
4111                 $ig->setContextTitle( $this->mTitle );
4112                 $ig->setShowBytes( false );
4113                 $ig->setShowFilename( false );
4114                 $ig->setParser( $this );
4115                 $ig->setHideBadImages();
4116                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4117                 $ig->useSkin( $this->mOptions->getSkin() );
4118                 $ig->mRevisionId = $this->mRevisionId;
4119
4120                 if( isset( $params['caption'] ) ) {
4121                         $caption = $params['caption'];
4122                         $caption = htmlspecialchars( $caption );
4123                         $caption = $this->replaceInternalLinks( $caption );
4124                         $ig->setCaptionHtml( $caption );
4125                 }
4126                 if( isset( $params['perrow'] ) ) {
4127                         $ig->setPerRow( $params['perrow'] );
4128                 }
4129                 if( isset( $params['widths'] ) ) {
4130                         $ig->setWidths( $params['widths'] );
4131                 }
4132                 if( isset( $params['heights'] ) ) {
4133                         $ig->setHeights( $params['heights'] );
4134                 }
4135
4136                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4137
4138                 $lines = StringUtils::explode( "\n", $text );
4139                 foreach ( $lines as $line ) {
4140                         # match lines like these:
4141                         # Image:someimage.jpg|This is some image
4142                         $matches = array();
4143                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4144                         # Skip empty lines
4145                         if ( count( $matches ) == 0 ) {
4146                                 continue;
4147                         }
4148
4149                         if ( strpos( $matches[0], '%' ) !== false )
4150                                 $matches[1] = urldecode( $matches[1] );
4151                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4152                         $nt =& $tp;
4153                         if( is_null( $nt ) ) {
4154                                 # Bogus title. Ignore these so we don't bomb out later.
4155                                 continue;
4156                         }
4157                         if ( isset( $matches[3] ) ) {
4158                                 $label = $matches[3];
4159                         } else {
4160                                 $label = '';
4161                         }
4162
4163                         $html = $this->recursiveTagParse( trim( $label ) );
4164
4165                         $ig->add( $nt, $html );
4166
4167                         # Only add real images (bug #5586)
4168                         if ( $nt->getNamespace() == NS_FILE ) {
4169                                 $this->mOutput->addImage( $nt->getDBkey() );
4170                         }
4171                 }
4172                 return $ig->toHTML();
4173         }
4174
4175         function getImageParams( $handler ) {
4176                 if ( $handler ) {
4177                         $handlerClass = get_class( $handler );
4178                 } else {
4179                         $handlerClass = '';
4180                 }
4181                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4182                         // Initialise static lists
4183                         static $internalParamNames = array(
4184                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4185                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4186                                         'bottom', 'text-bottom' ),
4187                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4188                                         'upright', 'border', 'link', 'alt' ),
4189                         );
4190                         static $internalParamMap;
4191                         if ( !$internalParamMap ) {
4192                                 $internalParamMap = array();
4193                                 foreach ( $internalParamNames as $type => $names ) {
4194                                         foreach ( $names as $name ) {
4195                                                 $magicName = str_replace( '-', '_', "img_$name" );
4196                                                 $internalParamMap[$magicName] = array( $type, $name );
4197                                         }
4198                                 }
4199                         }
4200
4201                         // Add handler params
4202                         $paramMap = $internalParamMap;
4203                         if ( $handler ) {
4204                                 $handlerParamMap = $handler->getParamMap();
4205                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4206                                         $paramMap[$magic] = array( 'handler', $paramName );
4207                                 }
4208                         }
4209                         $this->mImageParams[$handlerClass] = $paramMap;
4210                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4211                 }
4212                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4213         }
4214
4215         /**
4216          * Parse image options text and use it to make an image
4217          * @param Title $title
4218          * @param string $options
4219          * @param LinkHolderArray $holders
4220          */
4221         function makeImage( $title, $options, $holders = false ) {
4222                 # Check if the options text is of the form "options|alt text"
4223                 # Options are:
4224                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4225                 #  * left       no resizing, just left align. label is used for alt= only
4226                 #  * right      same, but right aligned
4227                 #  * none       same, but not aligned
4228                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4229                 #  * center     center the image
4230                 #  * framed     Keep original image size, no magnify-button.
4231                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4232                 #  * upright    reduce width for upright images, rounded to full __0 px
4233                 #  * border     draw a 1px border around the image
4234                 #  * alt        Text for HTML alt attribute (defaults to empty)
4235                 # vertical-align values (no % or length right now):
4236                 #  * baseline
4237                 #  * sub
4238                 #  * super
4239                 #  * top
4240                 #  * text-top
4241                 #  * middle
4242                 #  * bottom
4243                 #  * text-bottom
4244
4245                 $parts = StringUtils::explode( "|", $options );
4246                 $sk = $this->mOptions->getSkin();
4247
4248                 # Give extensions a chance to select the file revision for us
4249                 $skip = $time = $descQuery = false;
4250                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4251
4252                 if ( $skip ) {
4253                         return $sk->link( $title );
4254                 }
4255
4256                 # Get the file
4257                 $imagename = $title->getDBkey();
4258                 if ( isset( $this->mFileCache[$imagename][$time] ) ) {
4259                         $file = $this->mFileCache[$imagename][$time];
4260                 } else {
4261                         $file = wfFindFile( $title, $time );
4262                         if ( count( $this->mFileCache ) > 1000 ) {
4263                                 $this->mFileCache = array();
4264                         }
4265                         $this->mFileCache[$imagename][$time] = $file;
4266                 }
4267                 # Get parameter map
4268                 $handler = $file ? $file->getHandler() : false;
4269
4270                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4271
4272                 # Process the input parameters
4273                 $caption = '';
4274                 $params = array( 'frame' => array(), 'handler' => array(),
4275                         'horizAlign' => array(), 'vertAlign' => array() );
4276                 foreach( $parts as $part ) {
4277                         $part = trim( $part );
4278                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4279                         $validated = false;
4280                         if( isset( $paramMap[$magicName] ) ) {
4281                                 list( $type, $paramName ) = $paramMap[$magicName];
4282
4283                                 // Special case; width and height come in one variable together
4284                                 if( $type === 'handler' && $paramName === 'width' ) {
4285                                         $m = array();
4286                                         # (bug 13500) In both cases (width/height and width only),
4287                                         # permit trailing "px" for backward compatibility.
4288                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4289                                                 $width = intval( $m[1] );
4290                                                 $height = intval( $m[2] );
4291                                                 if ( $handler->validateParam( 'width', $width ) ) {
4292                                                         $params[$type]['width'] = $width;
4293                                                         $validated = true;
4294                                                 }
4295                                                 if ( $handler->validateParam( 'height', $height ) ) {
4296                                                         $params[$type]['height'] = $height;
4297                                                         $validated = true;
4298                                                 }
4299                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4300                                                 $width = intval( $value );
4301                                                 if ( $handler->validateParam( 'width', $width ) ) {
4302                                                         $params[$type]['width'] = $width;
4303                                                         $validated = true;
4304                                                 }
4305                                         } // else no validation -- bug 13436
4306                                 } else {
4307                                         if ( $type === 'handler' ) {
4308                                                 # Validate handler parameter
4309                                                 $validated = $handler->validateParam( $paramName, $value );
4310                                         } else {
4311                                                 # Validate internal parameters
4312                                                 switch( $paramName ) {
4313                                                 case 'manualthumb':
4314                                                 case 'alt':
4315                                                         // @fixme - possibly check validity here for
4316                                                         // manualthumb? downstream behavior seems odd with
4317                                                         // missing manual thumbs.
4318                                                         $validated = true;
4319                                                         $value = $this->stripAltText( $value, $holders );
4320                                                         break;
4321                                                 case 'link':
4322                                                         $chars = self::EXT_LINK_URL_CLASS;
4323                                                         $prots = $this->mUrlProtocols;
4324                                                         if ( $value === '' ) {
4325                                                                 $paramName = 'no-link';
4326                                                                 $value = true;
4327                                                                 $validated = true;
4328                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4329                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4330                                                                         $paramName = 'link-url';
4331                                                                         $this->mOutput->addExternalLink( $value );
4332                                                                         $validated = true;
4333                                                                 }
4334                                                         } else {
4335                                                                 $linkTitle = Title::newFromText( $value );
4336                                                                 if ( $linkTitle ) {
4337                                                                         $paramName = 'link-title';
4338                                                                         $value = $linkTitle;
4339                                                                         $this->mOutput->addLink( $linkTitle );
4340                                                                         $validated = true;
4341                                                                 }
4342                                                         }
4343                                                         break;
4344                                                 default:
4345                                                         // Most other things appear to be empty or numeric...
4346                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4347                                                 }
4348                                         }
4349
4350                                         if ( $validated ) {
4351                                                 $params[$type][$paramName] = $value;
4352                                         }
4353                                 }
4354                         }
4355                         if ( !$validated ) {
4356                                 $caption = $part;
4357                         }
4358                 }
4359
4360                 # Process alignment parameters
4361                 if ( $params['horizAlign'] ) {
4362                         $params['frame']['align'] = key( $params['horizAlign'] );
4363                 }
4364                 if ( $params['vertAlign'] ) {
4365                         $params['frame']['valign'] = key( $params['vertAlign'] );
4366                 }
4367
4368                 $params['frame']['caption'] = $caption;
4369
4370                 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4371
4372                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4373                 # came to also set the caption, ordinary text after the image -- which
4374                 # makes no sense, because that just repeats the text multiple times in
4375                 # screen readers.  It *also* came to set the title attribute.
4376                 #
4377                 # Now that we have an alt attribute, we should not set the alt text to
4378                 # equal the caption: that's worse than useless, it just repeats the
4379                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4380                 # use the unnamed parameter for alt text as well, just for the time be-
4381                 # ing, if the unnamed param is set and the alt param is not.
4382                 #
4383                 # For the future, we need to figure out if we want to tweak this more,
4384                 # e.g., introducing a title= parameter for the title; ignoring the un-
4385                 # named parameter entirely for images without a caption; adding an ex-
4386                 # plicit caption= parameter and preserving the old magic unnamed para-
4387                 # meter for BC; ...
4388                 if( $caption !== '' && !isset( $params['frame']['alt'] )
4389                 && !isset( $params['frame']['framed'] )
4390                 && !isset( $params['frame']['thumbnail'] )
4391                 && !isset( $params['frame']['manualthumb'] ) ) {
4392                         $params['frame']['alt'] = $params['frame']['title'];
4393                 }
4394
4395                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4396
4397                 # Linker does the rest
4398                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4399
4400                 # Give the handler a chance to modify the parser object
4401                 if ( $handler ) {
4402                         $handler->parserTransformHook( $this, $file );
4403                 }
4404
4405                 return $ret;
4406         }
4407
4408         protected function stripAltText( $caption, $holders ) {
4409                 # Strip bad stuff out of the title (tooltip).  We can't just use
4410                 # replaceLinkHoldersText() here, because if this function is called
4411                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4412                 if ( $holders ) {
4413                         $tooltip = $holders->replaceText( $caption );
4414                 } else {
4415                         $tooltip = $this->replaceLinkHoldersText( $caption );
4416                 }
4417
4418                 # make sure there are no placeholders in thumbnail attributes
4419                 # that are later expanded to html- so expand them now and
4420                 # remove the tags
4421                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4422                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4423
4424                 return $tooltip;
4425         }
4426
4427         /**
4428          * Set a flag in the output object indicating that the content is dynamic and
4429          * shouldn't be cached.
4430          */
4431         function disableCache() {
4432                 wfDebug( "Parser output marked as uncacheable.\n" );
4433                 $this->mOutput->mCacheTime = -1;
4434         }
4435
4436         /**#@+
4437          * Callback from the Sanitizer for expanding items found in HTML attribute
4438          * values, so they can be safely tested and escaped.
4439          * @param string $text
4440          * @param PPFrame $frame
4441          * @return string
4442          * @private
4443          */
4444         function attributeStripCallback( &$text, $frame = false ) {
4445                 $text = $this->replaceVariables( $text, $frame );
4446                 $text = $this->mStripState->unstripBoth( $text );
4447                 return $text;
4448         }
4449
4450         /**#@-*/
4451
4452         /**#@+
4453          * Accessor/mutator
4454          */
4455         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4456         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4457         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4458         /**#@-*/
4459
4460         /**#@+
4461          * Accessor
4462          */
4463         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4464         /**#@-*/
4465
4466
4467         /**
4468          * Break wikitext input into sections, and either pull or replace
4469          * some particular section's text.
4470          *
4471          * External callers should use the getSection and replaceSection methods.
4472          *
4473          * @param string $text Page wikitext
4474          * @param string $section A section identifier string of the form:
4475          *   <flag1> - <flag2> - ... - <section number>
4476          *
4477          * Currently the only recognised flag is "T", which means the target section number
4478          * was derived during a template inclusion parse, in other words this is a template
4479          * section edit link. If no flags are given, it was an ordinary section edit link.
4480          * This flag is required to avoid a section numbering mismatch when a section is
4481          * enclosed by <includeonly> (bug 6563).
4482          *
4483          * The section number 0 pulls the text before the first heading; other numbers will
4484          * pull the given section along with its lower-level subsections. If the section is
4485          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4486          *
4487          * @param string $mode One of "get" or "replace"
4488          * @param string $newText Replacement text for section data.
4489          * @return string for "get", the extracted section text.
4490          *                for "replace", the whole page with the section replaced.
4491          */
4492         private function extractSections( $text, $section, $mode, $newText='' ) {
4493                 global $wgTitle;
4494                 $this->clearState();
4495                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4496                 $this->mOptions = new ParserOptions;
4497                 $this->setOutputType( self::OT_WIKI );
4498                 $outText = '';
4499                 $frame = $this->getPreprocessor()->newFrame();
4500
4501                 // Process section extraction flags
4502                 $flags = 0;
4503                 $sectionParts = explode( '-', $section );
4504                 $sectionIndex = array_pop( $sectionParts );
4505                 foreach ( $sectionParts as $part ) {
4506                         if ( $part === 'T' ) {
4507                                 $flags |= self::PTD_FOR_INCLUSION;
4508                         }
4509                 }
4510                 // Preprocess the text
4511                 $root = $this->preprocessToDom( $text, $flags );
4512
4513                 // <h> nodes indicate section breaks
4514                 // They can only occur at the top level, so we can find them by iterating the root's children
4515                 $node = $root->getFirstChild();
4516
4517                 // Find the target section
4518                 if ( $sectionIndex == 0 ) {
4519                         // Section zero doesn't nest, level=big
4520                         $targetLevel = 1000;
4521                 } else {
4522             while ( $node ) {
4523                 if ( $node->getName() === 'h' ) {
4524                     $bits = $node->splitHeading();
4525                                         if ( $bits['i'] == $sectionIndex ) {
4526                                         $targetLevel = $bits['level'];
4527                                                 break;
4528                                         }
4529                                 }
4530                                 if ( $mode === 'replace' ) {
4531                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4532                                 }
4533                                 $node = $node->getNextSibling();
4534                         }
4535                 }
4536
4537                 if ( !$node ) {
4538                         // Not found
4539                         if ( $mode === 'get' ) {
4540                                 return $newText;
4541                         } else {
4542                                 return $text;
4543                         }
4544                 }
4545
4546                 // Find the end of the section, including nested sections
4547                 do {
4548                         if ( $node->getName() === 'h' ) {
4549                                 $bits = $node->splitHeading();
4550                                 $curLevel = $bits['level'];
4551                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4552                                         break;
4553                                 }
4554                         }
4555                         if ( $mode === 'get' ) {
4556                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4557                         }
4558                         $node = $node->getNextSibling();
4559                 } while ( $node );
4560
4561                 // Write out the remainder (in replace mode only)
4562                 if ( $mode === 'replace' ) {
4563                         // Output the replacement text
4564                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4565                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4566                         // Only add trailing whitespace if there is newText
4567                         if($newText != "") {
4568                                 $outText .= $newText . "\n\n";
4569                         }
4570
4571                         while ( $node ) {
4572                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4573                                 $node = $node->getNextSibling();
4574                         }
4575                 }
4576
4577                 if ( is_string( $outText ) ) {
4578                         // Re-insert stripped tags
4579                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4580                 }
4581
4582                 return $outText;
4583         }
4584
4585         /**
4586          * This function returns the text of a section, specified by a number ($section).
4587          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4588          * the first section before any such heading (section 0).
4589          *
4590          * If a section contains subsections, these are also returned.
4591          *
4592          * @param string $text text to look in
4593          * @param string $section section identifier
4594          * @param string $deftext default to return if section is not found
4595          * @return string text of the requested section
4596          */
4597         public function getSection( $text, $section, $deftext='' ) {
4598                 return $this->extractSections( $text, $section, "get", $deftext );
4599         }
4600
4601         public function replaceSection( $oldtext, $section, $text ) {
4602                 return $this->extractSections( $oldtext, $section, "replace", $text );
4603         }
4604
4605         /**
4606          * Get the timestamp associated with the current revision, adjusted for
4607          * the default server-local timestamp
4608          */
4609         function getRevisionTimestamp() {
4610                 if ( is_null( $this->mRevisionTimestamp ) ) {
4611                         wfProfileIn( __METHOD__ );
4612                         global $wgContLang;
4613                         $dbr = wfGetDB( DB_SLAVE );
4614                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4615                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4616
4617                         // Normalize timestamp to internal MW format for timezone processing.
4618                         // This has the added side-effect of replacing a null value with
4619                         // the current time, which gives us more sensible behavior for
4620                         // previews.
4621                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4622
4623                         // The cryptic '' timezone parameter tells to use the site-default
4624                         // timezone offset instead of the user settings.
4625                         //
4626                         // Since this value will be saved into the parser cache, served
4627                         // to other users, and potentially even used inside links and such,
4628                         // it needs to be consistent for all visitors.
4629                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4630
4631                         wfProfileOut( __METHOD__ );
4632                 }
4633                 return $this->mRevisionTimestamp;
4634         }
4635
4636         /**
4637          * Get the name of the user that edited the last revision
4638          */
4639         function getRevisionUser() {
4640                 // if this template is subst: the revision id will be blank,
4641                 // so just use the current user's name
4642                 if( $this->mRevisionId ) {
4643                         $revision = Revision::newFromId( $this->mRevisionId );
4644                         $revuser = $revision->getUserText();
4645                 } else {
4646                         global $wgUser;
4647                         $revuser = $wgUser->getName();
4648                 }
4649                 return $revuser;
4650         }
4651
4652         /**
4653          * Mutator for $mDefaultSort
4654          *
4655          * @param $sort New value
4656          */
4657         public function setDefaultSort( $sort ) {
4658                 $this->mDefaultSort = $sort;
4659         }
4660
4661         /**
4662          * Accessor for $mDefaultSort
4663          * Will use the title/prefixed title if none is set
4664          *
4665          * @return string
4666          */
4667         public function getDefaultSort() {
4668                 global $wgCategoryPrefixedDefaultSortkey;
4669                 if( $this->mDefaultSort !== false ) {
4670                         return $this->mDefaultSort;
4671                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4672                         !$wgCategoryPrefixedDefaultSortkey) {
4673                         return $this->mTitle->getText();
4674                 } else {
4675                         return $this->mTitle->getPrefixedText();
4676                 }
4677         }
4678
4679         /**
4680          * Accessor for $mDefaultSort
4681          * Unlike getDefaultSort(), will return false if none is set
4682          *
4683          * @return string or false
4684          */
4685         public function getCustomDefaultSort() {
4686                 return $this->mDefaultSort;
4687         }
4688
4689         /**
4690          * Try to guess the section anchor name based on a wikitext fragment
4691          * presumably extracted from a heading, for example "Header" from
4692          * "== Header ==".
4693          */
4694         public function guessSectionNameFromWikiText( $text ) {
4695                 # Strip out wikitext links(they break the anchor)
4696                 $text = $this->stripSectionName( $text );
4697                 $headline = Sanitizer::decodeCharReferences( $text );
4698                 # strip out HTML
4699                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4700                 $headline = trim( $headline );
4701                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4702                 $replacearray = array(
4703                         '%3A' => ':',
4704                         '%' => '.'
4705                 );
4706                 return str_replace(
4707                         array_keys( $replacearray ),
4708                         array_values( $replacearray ),
4709                         $sectionanchor );
4710         }
4711
4712         /**
4713          * Strips a text string of wikitext for use in a section anchor
4714          *
4715          * Accepts a text string and then removes all wikitext from the
4716          * string and leaves only the resultant text (i.e. the result of
4717          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4718          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4719          * to create valid section anchors by mimicing the output of the
4720          * parser when headings are parsed.
4721          *
4722          * @param $text string Text string to be stripped of wikitext
4723          * for use in a Section anchor
4724          * @return Filtered text string
4725          */
4726         public function stripSectionName( $text ) {
4727                 # Strip internal link markup
4728                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4729                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4730
4731                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4732                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4733                 # on how many empty links there are on the page - need to figure that out.
4734                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4735
4736                 # Parse wikitext quotes (italics & bold)
4737                 $text = $this->doQuotes($text);
4738
4739                 # Strip HTML tags
4740                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4741                 return $text;
4742         }
4743
4744         function srvus( $text ) {
4745                 return $this->testSrvus( $text, $this->mOutputType );
4746         }
4747
4748         /**
4749          * strip/replaceVariables/unstrip for preprocessor regression testing
4750          */
4751         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4752                 $this->clearState();
4753                 if ( ! ( $title instanceof Title ) ) {
4754                         $title = Title::newFromText( $title );
4755                 }
4756                 $this->mTitle = $title;
4757                 $this->mOptions = $options;
4758                 $this->setOutputType( $outputType );
4759                 $text = $this->replaceVariables( $text );
4760                 $text = $this->mStripState->unstripBoth( $text );
4761                 $text = Sanitizer::removeHTMLtags( $text );
4762                 return $text;
4763         }
4764
4765         function testPst( $text, $title, $options ) {
4766                 global $wgUser;
4767                 if ( ! ( $title instanceof Title ) ) {
4768                         $title = Title::newFromText( $title );
4769                 }
4770                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4771         }
4772
4773         function testPreprocess( $text, $title, $options ) {
4774                 if ( ! ( $title instanceof Title ) ) {
4775                         $title = Title::newFromText( $title );
4776                 }
4777                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4778         }
4779
4780         function markerSkipCallback( $s, $callback ) {
4781                 $i = 0;
4782                 $out = '';
4783                 while ( $i < strlen( $s ) ) {
4784                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4785                         if ( $markerStart === false ) {
4786                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4787                                 break;
4788                         } else {
4789                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4790                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4791                                 if ( $markerEnd === false ) {
4792                                         $out .= substr( $s, $markerStart );
4793                                         break;
4794                                 } else {
4795                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4796                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4797                                         $i = $markerEnd;
4798                                 }
4799                         }
4800                 }
4801                 return $out;
4802         }
4803
4804         function serialiseHalfParsedText( $text ) {
4805                 $data = array();
4806                 $data['text'] = $text;
4807
4808                 // First, find all strip markers, and store their
4809                 //  data in an array.
4810                 $stripState = new StripState;
4811                 $pos = 0;
4812                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
4813                         $end_pos += strlen( self::MARKER_SUFFIX );
4814                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
4815
4816                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
4817                                 $replaceArray = $stripState->general;
4818                                 $stripText = $this->mStripState->general->data[$marker];
4819                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
4820                                 $replaceArray = $stripState->nowiki;
4821                                 $stripText = $this->mStripState->nowiki->data[$marker];
4822                         } else {
4823                                 throw new MWException( "Hanging strip marker: '$marker'." );
4824                         }
4825
4826                         $replaceArray->setPair( $marker, $stripText );
4827                         $pos = $end_pos;
4828                 }
4829                 $data['stripstate'] = $stripState;
4830
4831                 // Now, find all of our links, and store THEIR
4832                 //  data in an array! :)
4833                 $links = array( 'internal' => array(), 'interwiki' => array() );
4834                 $pos = 0;
4835
4836                 // Internal links
4837                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
4838                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
4839
4840                         $ns = trim($ns);
4841                         if (empty( $links['internal'][$ns] )) {
4842                                 $links['internal'][$ns] = array();
4843                         }
4844
4845                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
4846                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
4847                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
4848                 }
4849
4850                 $pos = 0;
4851
4852                 // Interwiki links
4853                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
4854                         $data = substr( $text, $start_pos );
4855                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
4856                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
4857                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
4858                 }
4859
4860                 $data['linkholder'] = $links;
4861
4862                 return $data;
4863         }
4864
4865         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
4866                 if (!$intPrefix)
4867                         $intPrefix = $this->getRandomString();
4868
4869                 // First, extract the strip state.
4870                 $stripState = $data['stripstate'];
4871                 $this->mStripState->general->merge( $stripState->general );
4872                 $this->mStripState->nowiki->merge( $stripState->nowiki );
4873
4874                 // Now, extract the text, and renumber links
4875                 $text = $data['text'];
4876                 $links = $data['linkholder'];
4877
4878                 // Internal...
4879                 foreach( $links['internal'] as $ns => $nsLinks ) {
4880                         foreach( $nsLinks as $key => $entry ) {
4881                                 $newKey = $intPrefix . '-' . $key;
4882                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
4883
4884                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
4885                         }
4886                 }
4887
4888                 // Interwiki...
4889                 foreach( $links['interwiki'] as $key => $entry ) {
4890                         $newKey = "$intPrefix-$key";
4891                         $this->mLinkHolders->interwikis[$newKey] = $entry;
4892
4893                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
4894                 }
4895
4896                 // Should be good to go.
4897                 return $text;
4898         }
4899 }
4900
4901 /**
4902  * @todo document, briefly.
4903  * @ingroup Parser
4904  */
4905 class StripState {
4906         var $general, $nowiki;
4907
4908         function __construct() {
4909                 $this->general = new ReplacementArray;
4910                 $this->nowiki = new ReplacementArray;
4911         }
4912
4913         function unstripGeneral( $text ) {
4914                 wfProfileIn( __METHOD__ );
4915                 do {
4916                         $oldText = $text;
4917                         $text = $this->general->replace( $text );
4918                 } while ( $text !== $oldText );
4919                 wfProfileOut( __METHOD__ );
4920                 return $text;
4921         }
4922
4923         function unstripNoWiki( $text ) {
4924                 wfProfileIn( __METHOD__ );
4925                 do {
4926                         $oldText = $text;
4927                         $text = $this->nowiki->replace( $text );
4928                 } while ( $text !== $oldText );
4929                 wfProfileOut( __METHOD__ );
4930                 return $text;
4931         }
4932
4933         function unstripBoth( $text ) {
4934                 wfProfileIn( __METHOD__ );
4935                 do {
4936                         $oldText = $text;
4937                         $text = $this->general->replace( $text );
4938                         $text = $this->nowiki->replace( $text );
4939                 } while ( $text !== $oldText );
4940                 wfProfileOut( __METHOD__ );
4941                 return $text;
4942         }
4943 }
4944
4945 /**
4946  * @todo document, briefly.
4947  * @ingroup Parser
4948  */
4949 class OnlyIncludeReplacer {
4950         var $output = '';
4951
4952         function replace( $matches ) {
4953                 if ( substr( $matches[1], -1 ) === "\n" ) {
4954                         $this->output .= substr( $matches[1], 0, -1 );
4955                 } else {
4956                         $this->output .= $matches[1];
4957                 }
4958         }
4959 }