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                 # To make wikitext easier to read where significant
 331                 # whitespace is a problem, '__\' will be eaten up along with
 332                 # all whitespace and newlines that follow it, up to the next
 333                 # non-whitespace character. (Compare 'dnl' from m4.)
 334                 $text = preg_replace('/__\\\\\s*/', '', $text);
 335
 336                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 337                 $fixtags = array(
 338                         # french spaces, last one Guillemet-left
 339                         # only if there is something before the space
 340                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 341                         # french spaces, Guillemet-right
 342                         '/(\\302\\253) /' => '\\1&nbsp;',
 343                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 344                 );
 345                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 346
 347                 $text = $this->doBlockLevels( $text, $linestart );
 348
 349                 $this->replaceLinkHolders( $text );
 350
 351                 # the position of the parserConvert() call should not be changed. it
 352                 # assumes that the links are all replaced and the only thing left
 353                 # is the <nowiki> mark.
 354                 # Side-effects: this calls $this->mOutput->setTitleText()
 355                 $text = $wgContLang->parserConvert( $text, $this );
 356
 357                 $text = $this->mStripState->unstripNoWiki( $text );
 358
 359                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 360
 361 //!JF Move to its own function
 362
 363                 $uniq_prefix = $this->mUniqPrefix;
 364                 $matches = array();
 365                 $elements = array_keys( $this->mTransparentTagHooks );
 366                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 367
 368                 foreach( $matches as $marker => $data ) {
 369                         list( $element, $content, $params, $tag ) = $data;
 370                         $tagName = strtolower( $element );
 371                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 372                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 373                                         array( $content, $params, $this ) );
 374                         } else {
 375                                 $output = $tag;
 376                         }
 377                         $this->mStripState->general->setPair( $marker, $output );
 378                 }
 379                 $text = $this->mStripState->unstripGeneral( $text );
 380
 381                 $text = Sanitizer::normalizeCharReferences( $text );
 382
 383                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 384                         $text = self::tidy($text);
 385                 } else {
 386                         # attempt to sanitize at least some nesting problems
 387                         # (bug #2702 and quite a few others)
 388                         $tidyregs = array(
 389                                 # ''Something [http://www.cool.com cool''] -->
 390                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 391                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 392                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 393                                 # fix up an anchor inside another anchor, only
 394                                 # at least for a single single nested link (bug 3695)
 395                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 396                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 397                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 398                                 # contains a div, so fix it up here; replace
 399                                 # div with escaped text
 400                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 401                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 402                                 # remove empty italic or bold tag pairs, some
 403                                 # introduced by rules above
 404                                 '/<([bi])><\/\\1>/' => '',
 405                         );
 406
 407                         $text = preg_replace(
 408                                 array_keys( $tidyregs ),
 409                                 array_values( $tidyregs ),
 410                                 $text );
 411                 }
 412                 global $wgExpensiveParserFunctionLimit;
 413                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 414                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 415                 }
 416
 417                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 418
 419                 # Information on include size limits, for the benefit of users who try to skirt them
 420                 if ( $this->mOptions->getEnableLimitReport() ) {
 421                         global $wgExpensiveParserFunctionLimit;
 422                         $max = $this->mOptions->getMaxIncludeSize();
 423                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 424                         $limitReport =
 425                                 "NewPP limit report\n" .
 426                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 427                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 428                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 429                                 $PFreport;
 430                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 431                         $text .= "\n<!-- \n$limitReport-->\n";
 432                 }
 433                 $this->mOutput->setText( $text );
 434                 $this->mRevisionId = $oldRevisionId;
 435                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 436                 wfProfileOut( $fname );
 437                 wfProfileOut( __METHOD__ );
 438
 439                 return $this->mOutput;
 440         }
 441
 442         /**
 443          * Recursive parser entry point that can be called from an extension tag
 444          * hook.
 445          */
 446         function recursiveTagParse( $text ) {
 447                 wfProfileIn( __METHOD__ );
 448                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 449                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 450                 $text = $this->internalParse( $text );
 451                 wfProfileOut( __METHOD__ );
 452                 return $text;
 453         }
 454
 455         /**
 456          * Expand templates and variables in the text, producing valid, static wikitext.
 457          * Also removes comments.
 458          */
 459         function preprocess( $text, $title, $options, $revid = null ) {
 460                 wfProfileIn( __METHOD__ );
 461                 $this->clearState();
 462                 $this->setOutputType( self::OT_PREPROCESS );
 463                 $this->mOptions = $options;
 464                 $this->setTitle( $title );
 465                 if( $revid !== null ) {
 466                         $this->mRevisionId = $revid;
 467                 }
 468                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 469                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 470                 $text = $this->replaceVariables( $text );
 471                 $text = $this->mStripState->unstripBoth( $text );
 472                 wfProfileOut( __METHOD__ );
 473                 return $text;
 474         }
 475
 476         /**
 477          * Get a random string
 478          *
 479          * @private
 480          * @static
 481          */
 482         function getRandomString() {
 483                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 484         }
 485
 486         function &getTitle() { return $this->mTitle; }
 487         function getOptions() { return $this->mOptions; }
 488         function getRevisionId() { return $this->mRevisionId; }
 489         function getOutput() { return $this->mOutput; }
 490         function nextLinkID() { return $this->mLinkID++; }
 491
 492         function getFunctionLang() {
 493                 global $wgLang, $wgContLang;
 494
 495                 $target = $this->mOptions->getTargetLanguage();
 496                 if ( $target !== null ) {
 497                         return $target;
 498                 } else {
 499                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 500                 }
 501         }
 502
 503         /**
 504          * Get a preprocessor object
 505          */
 506         function getPreprocessor() {
 507                 if ( !isset( $this->mPreprocessor ) ) {
 508                         $class = $this->mPreprocessorClass;
 509                         $this->mPreprocessor = new $class( $this );
 510                 }
 511                 return $this->mPreprocessor;
 512         }
 513
 514         /**
 515          * Replaces all occurrences of HTML-style comments and the given tags
 516          * in the text with a random marker and returns the next text. The output
 517          * parameter $matches will be an associative array filled with data in
 518          * the form:
 519          *   'UNIQ-xxxxx' => array(
 520          *     'element',
 521          *     'tag content',
 522          *     array( 'param' => 'x' ),
 523          *     '<element param="x">tag content</element>' ) )
 524          *
 525          * @param $elements list of element names. Comments are always extracted.
 526          * @param $text Source text string.
 527          * @param $uniq_prefix
 528          *
 529          * @public
 530          * @static
 531          */
 532         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 533                 static $n = 1;
 534                 $stripped = '';
 535                 $matches = array();
 536
 537                 $taglist = implode( '|', $elements );
 538                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 539
 540                 while ( '' != $text ) {
 541                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 542                         $stripped .= $p[0];
 543                         if( count( $p ) < 5 ) {
 544                                 break;
 545                         }
 546                         if( count( $p ) > 5 ) {
 547                                 // comment
 548                                 $element    = $p[4];
 549                                 $attributes = '';
 550                                 $close      = '';
 551                                 $inside     = $p[5];
 552                         } else {
 553                                 // tag
 554                                 $element    = $p[1];
 555                                 $attributes = $p[2];
 556                                 $close      = $p[3];
 557                                 $inside     = $p[4];
 558                         }
 559
 560                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 561                         $stripped .= $marker;
 562
 563                         if ( $close === '/>' ) {
 564                                 // Empty element tag, <tag />
 565                                 $content = null;
 566                                 $text = $inside;
 567                                 $tail = null;
 568                         } else {
 569                                 if( $element === '!--' ) {
 570                                         $end = '/(-->)/';
 571                                 } else {
 572                                         $end = "/(<\\/$element\\s*>)/i";
 573                                 }
 574                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 575                                 $content = $q[0];
 576                                 if( count( $q ) < 3 ) {
 577                                         # No end tag -- let it run out to the end of the text.
 578                                         $tail = '';
 579                                         $text = '';
 580                                 } else {
 581                                         $tail = $q[1];
 582                                         $text = $q[2];
 583                                 }
 584                         }
 585
 586                         $matches[$marker] = array( $element,
 587                                 $content,
 588                                 Sanitizer::decodeTagAttributes( $attributes ),
 589                                 "<$element$attributes$close$content$tail" );
 590                 }
 591                 return $stripped;
 592         }
 593
 594         /**
 595          * Get a list of strippable XML-like elements
 596          */
 597         function getStripList() {
 598                 global $wgRawHtml;
 599                 $elements = $this->mStripList;
 600                 if( $wgRawHtml ) {
 601                         $elements[] = 'html';
 602                 }
 603                 if( $this->mOptions->getUseTeX() ) {
 604                         $elements[] = 'math';
 605                 }
 606                 return $elements;
 607         }
 608
 609         /**
 610          * @deprecated use replaceVariables
 611          */
 612         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 613                 return $text;
 614         }
 615
 616         /**
 617          * Restores pre, math, and other extensions removed by strip()
 618          *
 619          * always call unstripNoWiki() after this one
 620          * @private
 621          * @deprecated use $this->mStripState->unstrip()
 622          */
 623         function unstrip( $text, $state ) {
 624                 return $state->unstripGeneral( $text );
 625         }
 626
 627         /**
 628          * Always call this after unstrip() to preserve the order
 629          *
 630          * @private
 631          * @deprecated use $this->mStripState->unstrip()
 632          */
 633         function unstripNoWiki( $text, $state ) {
 634                 return $state->unstripNoWiki( $text );
 635         }
 636
 637         /**
 638          * @deprecated use $this->mStripState->unstripBoth()
 639          */
 640         function unstripForHTML( $text ) {
 641                 return $this->mStripState->unstripBoth( $text );
 642         }
 643
 644         /**
 645          * Add an item to the strip state
 646          * Returns the unique tag which must be inserted into the stripped text
 647          * The tag will be replaced with the original text in unstrip()
 648          *
 649          * @private
 650          */
 651         function insertStripItem( $text ) {
 652                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 653                 $this->mMarkerIndex++;
 654                 $this->mStripState->general->setPair( $rnd, $text );
 655                 return $rnd;
 656         }
 657
 658         /**
 659          * Interface with html tidy, used if $wgUseTidy = true.
 660          * If tidy isn't able to correct the markup, the original will be
 661          * returned in all its glory with a warning comment appended.
 662          *
 663          * Either the external tidy program or the in-process tidy extension
 664          * will be used depending on availability. Override the default
 665          * $wgTidyInternal setting to disable the internal if it's not working.
 666          *
 667          * @param string $text Hideous HTML input
 668          * @return string Corrected HTML output
 669          * @public
 670          * @static
 671          */
 672         function tidy( $text ) {
 673                 global $wgTidyInternal;
 674
 675                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 676 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 677 '<head><title>test</title></head><body>'.$text.'</body></html>';
 678
 679                 # Tidy is known to clobber tabs; convert 'em to entities
 680                 $wrappedtext = str_replace("\t", '&#9;', $wrappedtext);
 681
 682                 if( $wgTidyInternal ) {
 683                         $correctedtext = self::internalTidy( $wrappedtext );
 684                 } else {
 685                         $correctedtext = self::externalTidy( $wrappedtext );
 686                 }
 687                 if( is_null( $correctedtext ) ) {
 688                         wfDebug( "Tidy error detected!\n" );
 689                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 690                 }
 691
 692                 # Convert the tabs back from entities
 693                 $correctedtext = str_replace('&#9;', "\t", $correctedtext);
 694
 695                 return $correctedtext;
 696         }
 697
 698         /**
 699          * Spawn an external HTML tidy process and get corrected markup back from it.
 700          *
 701          * @private
 702          * @static
 703          */
 704         function externalTidy( $text ) {
 705                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 706                 wfProfileIn( __METHOD__ );
 707
 708                 $cleansource = '';
 709                 $opts = ' -utf8';
 710
 711                 $descriptorspec = array(
 712                         0 => array('pipe', 'r'),
 713                         1 => array('pipe', 'w'),
 714                         2 => array('file', wfGetNull(), 'a')
 715                 );
 716                 $pipes = array();
 717                 if( function_exists('proc_open') ) {
 718                         $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 719                         if (is_resource($process)) {
 720                                 // Theoretically, this style of communication could cause a deadlock
 721                                 // here. If the stdout buffer fills up, then writes to stdin could
 722                                 // block. This doesn't appear to happen with tidy, because tidy only
 723                                 // writes to stdout after it's finished reading from stdin. Search
 724                                 // for tidyParseStdin and tidySaveStdout in console/tidy.c
 725                                 fwrite($pipes[0], $text);
 726                                 fclose($pipes[0]);
 727                                 while (!feof($pipes[1])) {
 728                                         $cleansource .= fgets($pipes[1], 1024);
 729                                 }
 730                                 fclose($pipes[1]);
 731                                 proc_close($process);
 732                         }
 733                 }
 734
 735                 wfProfileOut( __METHOD__ );
 736
 737                 if( $cleansource == '' && $text != '') {
 738                         // Some kind of error happened, so we couldn't get the corrected text.
 739                         // Just give up; we'll use the source text and append a warning.
 740                         return null;
 741                 } else {
 742                         return $cleansource;
 743                 }
 744         }
 745
 746         /**
 747          * Use the HTML tidy PECL extension to use the tidy library in-process,
 748          * saving the overhead of spawning a new process.
 749          *
 750          * 'pear install tidy' should be able to compile the extension module.
 751          *
 752          * @private
 753          * @static
 754          */
 755         function internalTidy( $text ) {
 756                 global $wgTidyConf, $IP, $wgDebugTidy;
 757                 wfProfileIn( __METHOD__ );
 758
 759                 $tidy = new tidy;
 760                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 761                 $tidy->cleanRepair();
 762                 if( $tidy->getStatus() == 2 ) {
 763                         // 2 is magic number for fatal error
 764                         // http://www.php.net/manual/en/function.tidy-get-status.php
 765                         $cleansource = null;
 766                 } else {
 767                         $cleansource = tidy_get_output( $tidy );
 768                 }
 769                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 770                         $cleansource .= "<!--\nTidy reports:\n" .
 771                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 772                                 "\n-->";
 773                 }
 774
 775                 wfProfileOut( __METHOD__ );
 776                 return $cleansource;
 777         }
 778
 779         /**
 780          * parse the wiki syntax used to render tables
 781          *
 782          * @private
 783          */
 784         function doTableStuff ( $text ) {
 785                 wfProfileIn( __METHOD__ );
 786
 787                 $lines = StringUtils::explode( "\n", $text );
 788                 $out = '';
 789                 $td_history = array (); // Is currently a td tag open?
 790                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 791                 $tr_history = array (); // Is currently a tr tag open?
 792                 $tr_attributes = array (); // history of tr attributes
 793                 $has_opened_tr = array(); // Did this table open a <tr> element?
 794                 $indent_level = 0; // indent level of the table
 795
 796                 foreach ( $lines as $outLine ) {
 797                         $line = trim( $outLine );
 798
 799                         if( $line == '' ) { // empty line, go to next line
 800                                 $out .= $outLine."\n";
 801                                 continue;
 802                         }
 803                         $first_character = $line[0];
 804                         $matches = array();
 805
 806                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 807                                 // First check if we are starting a new table
 808                                 $indent_level = strlen( $matches[1] );
 809
 810                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 811                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 812
 813                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 814                                 array_push ( $td_history , false );
 815                                 array_push ( $last_tag_history , '' );
 816                                 array_push ( $tr_history , false );
 817                                 array_push ( $tr_attributes , '' );
 818                                 array_push ( $has_opened_tr , false );
 819                         } else if ( count ( $td_history ) == 0 ) {
 820                                 // Don't do any of the following
 821                                 $out .= $outLine."\n";
 822                                 continue;
 823                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 824                                 // We are ending a table
 825                                 $line = '</table>' . substr ( $line , 2 );
 826                                 $last_tag = array_pop ( $last_tag_history );
 827
 828                                 if ( !array_pop ( $has_opened_tr ) ) {
 829                                         $line = "<tr><td></td></tr>{$line}";
 830                                 }
 831
 832                                 if ( array_pop ( $tr_history ) ) {
 833                                         $line = "</tr>{$line}";
 834                                 }
 835
 836                                 if ( array_pop ( $td_history ) ) {
 837                                         $line = "</{$last_tag}>{$line}";
 838                                 }
 839                                 array_pop ( $tr_attributes );
 840                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 841                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 842                                 // Now we have a table row
 843                                 $line = preg_replace( '#^\|-+#', '', $line );
 844
 845                                 // Whats after the tag is now only attributes
 846                                 $attributes = $this->mStripState->unstripBoth( $line );
 847                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 848                                 array_pop ( $tr_attributes );
 849                                 array_push ( $tr_attributes , $attributes );
 850
 851                                 $line = '';
 852                                 $last_tag = array_pop ( $last_tag_history );
 853                                 array_pop ( $has_opened_tr );
 854                                 array_push ( $has_opened_tr , true );
 855
 856                                 if ( array_pop ( $tr_history ) ) {
 857                                         $line = '</tr>';
 858                                 }
 859
 860                                 if ( array_pop ( $td_history ) ) {
 861                                         $line = "</{$last_tag}>{$line}";
 862                                 }
 863
 864                                 $outLine = $line;
 865                                 array_push ( $tr_history , false );
 866                                 array_push ( $td_history , false );
 867                                 array_push ( $last_tag_history , '' );
 868                         }
 869                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 870                                 // This might be cell elements, td, th or captions
 871                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 872                                         $first_character = '+';
 873                                         $line = substr ( $line , 1 );
 874                                 }
 875
 876                                 $line = substr ( $line , 1 );
 877
 878                                 if ( $first_character === '!' ) {
 879                                         $line = str_replace ( '!!' , '||' , $line );
 880                                 }
 881
 882                                 // Split up multiple cells on the same line.
 883                                 // FIXME : This can result in improper nesting of tags processed
 884                                 // by earlier parser steps, but should avoid splitting up eg
 885                                 // attribute values containing literal "||".
 886                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 887
 888                                 $outLine = '';
 889
 890                                 // Loop through each table cell
 891                                 foreach ( $cells as $cell )
 892                                 {
 893                                         $previous = '';
 894                                         if ( $first_character !== '+' )
 895                                         {
 896                                                 $tr_after = array_pop ( $tr_attributes );
 897                                                 if ( !array_pop ( $tr_history ) ) {
 898                                                         $previous = "<tr{$tr_after}>\n";
 899                                                 }
 900                                                 array_push ( $tr_history , true );
 901                                                 array_push ( $tr_attributes , '' );
 902                                                 array_pop ( $has_opened_tr );
 903                                                 array_push ( $has_opened_tr , true );
 904                                         }
 905
 906                                         $last_tag = array_pop ( $last_tag_history );
 907
 908                                         if ( array_pop ( $td_history ) ) {
 909                                                 $previous = "</{$last_tag}>{$previous}";
 910                                         }
 911
 912                                         if ( $first_character === '|' ) {
 913                                                 $last_tag = 'td';
 914                                         } else if ( $first_character === '!' ) {
 915                                                 $last_tag = 'th';
 916                                         } else if ( $first_character === '+' ) {
 917                                                 $last_tag = 'caption';
 918                                         } else {
 919                                                 $last_tag = '';
 920                                         }
 921
 922                                         array_push ( $last_tag_history , $last_tag );
 923
 924                                         // A cell could contain both parameters and data
 925                                         $cell_data = explode ( '|' , $cell , 2 );
 926
 927                                         // Bug 553: Note that a '|' inside an invalid link should not
 928                                         // be mistaken as delimiting cell parameters
 929                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 930                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 931                                         } else if ( count ( $cell_data ) == 1 )
 932                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 933                                         else {
 934                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 935                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 936                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 937                                         }
 938
 939                                         $outLine .= $cell;
 940                                         array_push ( $td_history , true );
 941                                 }
 942                         }
 943                         $out .= $outLine . "\n";
 944                 }
 945
 946                 // Closing open td, tr && table
 947                 while ( count ( $td_history ) > 0 )
 948                 {
 949                         if ( array_pop ( $td_history ) ) {
 950                                 $out .= "</td>\n";
 951                         }
 952                         if ( array_pop ( $tr_history ) ) {
 953                                 $out .= "</tr>\n";
 954                         }
 955                         if ( !array_pop ( $has_opened_tr ) ) {
 956                                 $out .= "<tr><td></td></tr>\n" ;
 957                         }
 958
 959                         $out .= "</table>\n";
 960                 }
 961
 962                 // Remove trailing line-ending (b/c)
 963                 if ( substr( $out, -1 ) === "\n" ) {
 964                         $out = substr( $out, 0, -1 );
 965                 }
 966
 967                 // special case: don't return empty table
 968                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 969                         $out = '';
 970                 }
 971
 972                 wfProfileOut( __METHOD__ );
 973
 974                 return $out;
 975         }
 976
 977         /**
 978          * Helper function for parse() that transforms wiki markup into
 979          * HTML. Only called for $mOutputType == self::OT_HTML.
 980          *
 981          * @private
 982          */
 983         function internalParse( $text ) {
 984                 $isMain = true;
 985                 wfProfileIn( __METHOD__ );
 986
 987                 # Hook to suspend the parser in this state
 988                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 989                         wfProfileOut( __METHOD__ );
 990                         return $text ;
 991                 }
 992
 993                 $text = $this->replaceVariables( $text );
 994                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 995                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 996
 997                 // Tables need to come after variable replacement for things to work
 998                 // properly; putting them before other transformations should keep
 999                 // exciting things like link expansions from showing up in surprising
1000                 // places.
1001                 $text = $this->doTableStuff( $text );
1002
1003                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1004
1005                 $text = $this->doDoubleUnderscore( $text );
1006                 $text = $this->doHeadings( $text );
1007                 if($this->mOptions->getUseDynamicDates()) {
1008                         $df = DateFormatter::getInstance();
1009                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1010                 }
1011                 $text = $this->doAllQuotes( $text );
1012                 $text = $this->replaceInternalLinks( $text );
1013                 $text = $this->replaceExternalLinks( $text );
1014
1015                 # replaceInternalLinks may sometimes leave behind
1016                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1017                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1018
1019                 $text = $this->doMagicLinks( $text );
1020                 $text = $this->formatHeadings( $text, $isMain );
1021
1022                 wfProfileOut( __METHOD__ );
1023                 return $text;
1024         }
1025
1026         /**
1027          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1028          * magic external links.
1029          *
1030          * DML
1031          * @private
1032          */
1033         function doMagicLinks( $text ) {
1034                 wfProfileIn( __METHOD__ );
1035                 $prots = $this->mUrlProtocols;
1036                 $urlChar = self::EXT_LINK_URL_CLASS;
1037                 $text = preg_replace_callback(
1038                         '!(?:                           # Start cases
1039                                 (<a.*?</a>) |               # m[1]: Skip link text
1040                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1041                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1042                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1043                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1044                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1045                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1046                                     [0-9Xx]                 # check digit
1047                                     \b)
1048                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1049                 wfProfileOut( __METHOD__ );
1050                 return $text;
1051         }
1052
1053         function magicLinkCallback( $m ) {
1054                 if ( isset( $m[1] ) && strval( $m[1] ) !== '' ) {
1055                         # Skip anchor
1056                         return $m[0];
1057                 } elseif ( isset( $m[2] ) && strval( $m[2] ) !== '' ) {
1058                         # Skip HTML element
1059                         return $m[0];
1060                 } elseif ( isset( $m[3] ) && strval( $m[3] ) !== '' ) {
1061                         # Free external link
1062                         return $this->makeFreeExternalLink( $m[0] );
1063                 } elseif ( isset( $m[4] ) && strval( $m[4] ) !== '' ) {
1064                         # RFC or PMID
1065                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1066                                 $keyword = 'RFC';
1067                                 $urlmsg = 'rfcurl';
1068                                 $id = $m[4];
1069                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1070                                 $keyword = 'PMID';
1071                                 $urlmsg = 'pubmedurl';
1072                                 $id = $m[4];
1073                         } else {
1074                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1075                                         substr($m[0], 0, 20 ) . '"' );
1076                         }
1077                         $url = wfMsg( $urlmsg, $id);
1078                         $sk = $this->mOptions->getSkin();
1079                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1080                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1081                 } elseif ( isset( $m[5] ) && strval( $m[5] ) !== '' ) {
1082                         # ISBN
1083                         $isbn = $m[5];
1084                         $num = strtr( $isbn, array(
1085                                 '-' => '',
1086                                 ' ' => '',
1087                                 'x' => 'X',
1088                         ));
1089                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1090                         return'<a href="' .
1091                                 $titleObj->escapeLocalUrl() .
1092                                 "\" class=\"internal\">ISBN $isbn</a>";
1093                 } else {
1094                         return $m[0];
1095                 }
1096         }
1097
1098         /**
1099          * Make a free external link, given a user-supplied URL
1100          * @return HTML
1101          * @private
1102          */
1103         function makeFreeExternalLink( $url ) {
1104                 global $wgContLang;
1105                 wfProfileIn( __METHOD__ );
1106
1107                 $sk = $this->mOptions->getSkin();
1108                 $trail = '';
1109
1110                 # The characters '<' and '>' (which were escaped by
1111                 # removeHTMLtags()) should not be included in
1112                 # URLs, per RFC 2396.
1113                 $m2 = array();
1114                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1115                         $trail = substr($url, $m2[0][1]) . $trail;
1116                         $url = substr($url, 0, $m2[0][1]);
1117                 }
1118
1119                 # Move trailing punctuation to $trail
1120                 $sep = ',;\.:!?';
1121                 # If there is no left bracket, then consider right brackets fair game too
1122                 if ( strpos( $url, '(' ) === false ) {
1123                         $sep .= ')';
1124                 }
1125
1126                 $numSepChars = strspn( strrev( $url ), $sep );
1127                 if ( $numSepChars ) {
1128                         $trail = substr( $url, -$numSepChars ) . $trail;
1129                         $url = substr( $url, 0, -$numSepChars );
1130                 }
1131
1132                 $url = Sanitizer::cleanUrl( $url );
1133
1134                 # Is this an external image?
1135                 $text = $this->maybeMakeExternalImage( $url );
1136                 if ( $text === false ) {
1137                         # Not an image, make a link
1138                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1139                                 $this->getExternalLinkAttribs() );
1140                         # Register it in the output object...
1141                         # Replace unnecessary URL escape codes with their equivalent characters
1142                         $pasteurized = self::replaceUnusualEscapes( $url );
1143                         $this->mOutput->addExternalLink( $pasteurized );
1144                 }
1145                 wfProfileOut( __METHOD__ );
1146                 return $text . $trail;
1147         }
1148
1149
1150         /**
1151          * Parse headers and return html
1152          *
1153          * @private
1154          */
1155         function doHeadings( $text ) {
1156                 wfProfileIn( __METHOD__ );
1157                 for ( $i = 6; $i >= 1; --$i ) {
1158                         $h = str_repeat( '=', $i );
1159                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1160                           "<h$i>\\1</h$i>", $text );
1161                 }
1162                 wfProfileOut( __METHOD__ );
1163                 return $text;
1164         }
1165
1166         /**
1167          * Replace single quotes with HTML markup
1168          * @private
1169          * @return string the altered text
1170          */
1171         function doAllQuotes( $text ) {
1172                 wfProfileIn( __METHOD__ );
1173                 $outtext = '';
1174                 $lines = StringUtils::explode( "\n", $text );
1175                 foreach ( $lines as $line ) {
1176                         $outtext .= $this->doQuotes( $line ) . "\n";
1177                 }
1178                 $outtext = substr($outtext, 0,-1);
1179                 wfProfileOut( __METHOD__ );
1180                 return $outtext;
1181         }
1182
1183         /**
1184          * Helper function for doAllQuotes()
1185          */
1186         public function doQuotes( $text ) {
1187                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1188                 if ( count( $arr ) == 1 )
1189                         return $text;
1190                 else
1191                 {
1192                         # First, do some preliminary work. This may shift some apostrophes from
1193                         # being mark-up to being text. It also counts the number of occurrences
1194                         # of bold and italics mark-ups.
1195                         $i = 0;
1196                         $numbold = 0;
1197                         $numitalics = 0;
1198                         foreach ( $arr as $r )
1199                         {
1200                                 if ( ( $i % 2 ) == 1 )
1201                                 {
1202                                         # If there are ever four apostrophes, assume the first is supposed to
1203                                         # be text, and the remaining three constitute mark-up for bold text.
1204                                         if ( strlen( $arr[$i] ) == 4 )
1205                                         {
1206                                                 $arr[$i-1] .= "'";
1207                                                 $arr[$i] = "'''";
1208                                         }
1209                                         # If there are more than 5 apostrophes in a row, assume they're all
1210                                         # text except for the last 5.
1211                                         else if ( strlen( $arr[$i] ) > 5 )
1212                                         {
1213                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1214                                                 $arr[$i] = "'''''";
1215                                         }
1216                                         # Count the number of occurrences of bold and italics mark-ups.
1217                                         # We are not counting sequences of five apostrophes.
1218                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1219                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1220                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1221                                 }
1222                                 $i++;
1223                         }
1224
1225                         # If there is an odd number of both bold and italics, it is likely
1226                         # that one of the bold ones was meant to be an apostrophe followed
1227                         # by italics. Which one we cannot know for certain, but it is more
1228                         # likely to be one that has a single-letter word before it.
1229                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1230                         {
1231                                 $i = 0;
1232                                 $firstsingleletterword = -1;
1233                                 $firstmultiletterword = -1;
1234                                 $firstspace = -1;
1235                                 foreach ( $arr as $r )
1236                                 {
1237                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1238                                         {
1239                                                 $x1 = substr ($arr[$i-1], -1);
1240                                                 $x2 = substr ($arr[$i-1], -2, 1);
1241                                                 if ($x1 === ' ') {
1242                                                         if ($firstspace == -1) $firstspace = $i;
1243                                                 } else if ($x2 === ' ') {
1244                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1245                                                 } else {
1246                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1247                                                 }
1248                                         }
1249                                         $i++;
1250                                 }
1251
1252                                 # If there is a single-letter word, use it!
1253                                 if ($firstsingleletterword > -1)
1254                                 {
1255                                         $arr [ $firstsingleletterword ] = "''";
1256                                         $arr [ $firstsingleletterword-1 ] .= "'";
1257                                 }
1258                                 # If not, but there's a multi-letter word, use that one.
1259                                 else if ($firstmultiletterword > -1)
1260                                 {
1261                                         $arr [ $firstmultiletterword ] = "''";
1262                                         $arr [ $firstmultiletterword-1 ] .= "'";
1263                                 }
1264                                 # ... otherwise use the first one that has neither.
1265                                 # (notice that it is possible for all three to be -1 if, for example,
1266                                 # there is only one pentuple-apostrophe in the line)
1267                                 else if ($firstspace > -1)
1268                                 {
1269                                         $arr [ $firstspace ] = "''";
1270                                         $arr [ $firstspace-1 ] .= "'";
1271                                 }
1272                         }
1273
1274                         # Now let's actually convert our apostrophic mush to HTML!
1275                         $output = '';
1276                         $buffer = '';
1277                         $state = '';
1278                         $i = 0;
1279                         foreach ($arr as $r)
1280                         {
1281                                 if (($i % 2) == 0)
1282                                 {
1283                                         if ($state === 'both')
1284                                                 $buffer .= $r;
1285                                         else
1286                                                 $output .= $r;
1287                                 }
1288                                 else
1289                                 {
1290                                         if (strlen ($r) == 2)
1291                                         {
1292                                                 if ($state === 'i')
1293                                                 { $output .= '</i>'; $state = ''; }
1294                                                 else if ($state === 'bi')
1295                                                 { $output .= '</i>'; $state = 'b'; }
1296                                                 else if ($state === 'ib')
1297                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1298                                                 else if ($state === 'both')
1299                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1300                                                 else # $state can be 'b' or ''
1301                                                 { $output .= '<i>'; $state .= 'i'; }
1302                                         }
1303                                         else if (strlen ($r) == 3)
1304                                         {
1305                                                 if ($state === 'b')
1306                                                 { $output .= '</b>'; $state = ''; }
1307                                                 else if ($state === 'bi')
1308                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1309                                                 else if ($state === 'ib')
1310                                                 { $output .= '</b>'; $state = 'i'; }
1311                                                 else if ($state === 'both')
1312                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1313                                                 else # $state can be 'i' or ''
1314                                                 { $output .= '<b>'; $state .= 'b'; }
1315                                         }
1316                                         else if (strlen ($r) == 5)
1317                                         {
1318                                                 if ($state === 'b')
1319                                                 { $output .= '</b><i>'; $state = 'i'; }
1320                                                 else if ($state === 'i')
1321                                                 { $output .= '</i><b>'; $state = 'b'; }
1322                                                 else if ($state === 'bi')
1323                                                 { $output .= '</i></b>'; $state = ''; }
1324                                                 else if ($state === 'ib')
1325                                                 { $output .= '</b></i>'; $state = ''; }
1326                                                 else if ($state === 'both')
1327                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1328                                                 else # ($state == '')
1329                                                 { $buffer = ''; $state = 'both'; }
1330                                         }
1331                                 }
1332                                 $i++;
1333                         }
1334                         # Now close all remaining tags.  Notice that the order is important.
1335                         if ($state === 'b' || $state === 'ib')
1336                                 $output .= '</b>';
1337                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1338                                 $output .= '</i>';
1339                         if ($state === 'bi')
1340                                 $output .= '</b>';
1341                         # There might be lonely ''''', so make sure we have a buffer
1342                         if ($state === 'both' && $buffer)
1343                                 $output .= '<b><i>'.$buffer.'</i></b>';
1344                         return $output;
1345                 }
1346         }
1347
1348         /**
1349          * Replace external links (REL)
1350          *
1351          * Note: this is all very hackish and the order of execution matters a lot.
1352          * Make sure to run maintenance/parserTests.php if you change this code.
1353          *
1354          * @private
1355          */
1356         function replaceExternalLinks( $text ) {
1357                 global $wgContLang;
1358                 wfProfileIn( __METHOD__ );
1359
1360                 $sk = $this->mOptions->getSkin();
1361
1362                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1363                 $s = array_shift( $bits );
1364
1365                 $i = 0;
1366                 while ( $i<count( $bits ) ) {
1367                         $url = $bits[$i++];
1368                         $protocol = $bits[$i++];
1369                         $text = $bits[$i++];
1370                         $trail = $bits[$i++];
1371
1372                         # The characters '<' and '>' (which were escaped by
1373                         # removeHTMLtags()) should not be included in
1374                         # URLs, per RFC 2396.
1375                         $m2 = array();
1376                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1377                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1378                                 $url = substr($url, 0, $m2[0][1]);
1379                         }
1380
1381                         # If the link text is an image URL, replace it with an <img> tag
1382                         # This happened by accident in the original parser, but some people used it extensively
1383                         $img = $this->maybeMakeExternalImage( $text );
1384                         if ( $img !== false ) {
1385                                 $text = $img;
1386                         }
1387
1388                         $dtrail = '';
1389
1390                         # Set linktype for CSS - if URL==text, link is essentially free
1391                         $linktype = ($text === $url) ? 'free' : 'text';
1392
1393                         # No link text, e.g. [http://domain.tld/some.link]
1394                         if ( $text == '' ) {
1395                                 # Autonumber if allowed. See bug #5918
1396                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1397                                         $langObj = $this->getFunctionLang();
1398                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1399                                         $linktype = 'autonumber';
1400                                 } else {
1401                                         # Otherwise just use the URL
1402                                         $text = htmlspecialchars( $url );
1403                                         $linktype = 'free';
1404                                 }
1405                         } else {
1406                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1407                                 # Check for trail
1408                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1409                         }
1410
1411                         $text = $wgContLang->markNoConversion($text);
1412
1413                         $url = Sanitizer::cleanUrl( $url );
1414
1415                         if ( $this->mOptions->mExternalLinkTarget ) {
1416                                 $attribs = array( 'target' => $this->mOptions->mExternalLinkTarget );
1417                         } else {
1418                                 $attribs = array();
1419                         }
1420
1421                         # Use the encoded URL
1422                         # This means that users can paste URLs directly into the text
1423                         # Funny characters like &ouml; aren't valid in URLs anyway
1424                         # This was changed in August 2004
1425                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->getExternalLinkAttribs() )
1426                                 . $dtrail . $trail;
1427
1428                         # Register link in the output object.
1429                         # Replace unnecessary URL escape codes with the referenced character
1430                         # This prevents spammers from hiding links from the filters
1431                         $pasteurized = self::replaceUnusualEscapes( $url );
1432                         $this->mOutput->addExternalLink( $pasteurized );
1433                 }
1434
1435                 wfProfileOut( __METHOD__ );
1436                 return $s;
1437         }
1438
1439         function getExternalLinkAttribs() {
1440                 $attribs = array();
1441                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1442                 $ns = $this->mTitle->getNamespace();
1443                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1444                         $attribs['rel'] = 'nofollow';
1445                 }
1446                 if ( $this->mOptions->getExternalLinkTarget() ) {
1447                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1448                 }
1449                 return $attribs;
1450         }
1451
1452
1453         /**
1454          * Replace unusual URL escape codes with their equivalent characters
1455          * @param string
1456          * @return string
1457          * @static
1458          * @todo  This can merge genuinely required bits in the path or query string,
1459          *        breaking legit URLs. A proper fix would treat the various parts of
1460          *        the URL differently; as a workaround, just use the output for
1461          *        statistical records, not for actual linking/output.
1462          */
1463         static function replaceUnusualEscapes( $url ) {
1464                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1465                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1466         }
1467
1468         /**
1469          * Callback function used in replaceUnusualEscapes().
1470          * Replaces unusual URL escape codes with their equivalent character
1471          * @static
1472          * @private
1473          */
1474         private static function replaceUnusualEscapesCallback( $matches ) {
1475                 $char = urldecode( $matches[0] );
1476                 $ord = ord( $char );
1477                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1478                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1479                         // No, shouldn't be escaped
1480                         return $char;
1481                 } else {
1482                         // Yes, leave it escaped
1483                         return $matches[0];
1484                 }
1485         }
1486
1487         /**
1488          * make an image if it's allowed, either through the global
1489          * option, through the exception, or through the on-wiki whitelist
1490          * @private
1491          */
1492         function maybeMakeExternalImage( $url ) {
1493                 $sk = $this->mOptions->getSkin();
1494                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1495                 $imagesexception = !empty($imagesfrom);
1496                 $text = false;
1497                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1498                 if( $imagesexception && is_array( $imagesfrom ) ) {
1499                         $imagematch = false;
1500                         foreach( $imagesfrom as $match ) {
1501                                 if( strpos( $url, $match ) === 0 ) {
1502                                         $imagematch = true;
1503                                         break;
1504                                 }
1505                         }
1506                 } elseif( $imagesexception ) {
1507                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1508                 } else {
1509                         $imagematch = false;
1510                 }
1511                 if ( $this->mOptions->getAllowExternalImages()
1512                      || ( $imagesexception && $imagematch ) ) {
1513                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1514                                 # Image found
1515                                 $text = $sk->makeExternalImage( $url );
1516                         }
1517                 }
1518                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1519                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1520                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1521                         foreach( $whitelist as $entry ) {
1522                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1523                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1524                                         continue;
1525                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1526                                         # Image matches a whitelist entry
1527                                         $text = $sk->makeExternalImage( $url );
1528                                         break;
1529                                 }
1530                         }
1531                 }
1532                 return $text;
1533         }
1534
1535         /**
1536          * Process [[ ]] wikilinks
1537          * @return processed text
1538          *
1539          * @private
1540          */
1541         function replaceInternalLinks( $s ) {
1542                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1543                 return $s;
1544         }
1545
1546         /**
1547          * Process [[ ]] wikilinks (RIL)
1548          * @return LinkHolderArray
1549          *
1550          * @private
1551          */
1552         function replaceInternalLinks2( &$s ) {
1553                 global $wgContLang;
1554
1555                 wfProfileIn( __METHOD__ );
1556
1557                 wfProfileIn( __METHOD__.'-setup' );
1558                 static $tc = FALSE, $e1, $e1_img;
1559                 # the % is needed to support urlencoded titles as well
1560                 if ( !$tc ) {
1561                         $tc = Title::legalChars() . '#%';
1562                         # Match a link having the form [[namespace:link|alternate]]trail
1563                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1564                         # Match cases where there is no "]]", which might still be images
1565                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1566                 }
1567
1568                 $sk = $this->mOptions->getSkin();
1569                 $holders = new LinkHolderArray( $this );
1570
1571                 #split the entire text string on occurences of [[
1572                 $a = StringUtils::explode( '[[', ' ' . $s );
1573                 #get the first element (all text up to first [[), and remove the space we added
1574                 $s = $a->current();
1575                 $a->next();
1576                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1577                 $s = substr( $s, 1 );
1578
1579                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1580                 $e2 = null;
1581                 if ( $useLinkPrefixExtension ) {
1582                         # Match the end of a line for a word that's not followed by whitespace,
1583                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1584                         $e2 = wfMsgForContent( 'linkprefix' );
1585                 }
1586
1587                 if( is_null( $this->mTitle ) ) {
1588                         wfProfileOut( __METHOD__.'-setup' );
1589                         wfProfileOut( __METHOD__ );
1590                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1591                 }
1592                 $nottalk = !$this->mTitle->isTalkPage();
1593
1594                 if ( $useLinkPrefixExtension ) {
1595                         $m = array();
1596                         if ( preg_match( $e2, $s, $m ) ) {
1597                                 $first_prefix = $m[2];
1598                         } else {
1599                                 $first_prefix = false;
1600                         }
1601                 } else {
1602                         $prefix = '';
1603                 }
1604
1605                 if($wgContLang->hasVariants()) {
1606                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1607                 } else {
1608                         $selflink = array($this->mTitle->getPrefixedText());
1609                 }
1610                 $useSubpages = $this->areSubpagesAllowed();
1611                 wfProfileOut( __METHOD__.'-setup' );
1612
1613                 # Loop for each link
1614                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1615                         # Check for excessive memory usage
1616                         if ( $holders->isBig() ) {
1617                                 # Too big
1618                                 # Do the existence check, replace the link holders and clear the array
1619                                 $holders->replace( $s );
1620                                 $holders->clear();
1621                         }
1622
1623                         if ( $useLinkPrefixExtension ) {
1624                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1625                                 if ( preg_match( $e2, $s, $m ) ) {
1626                                         $prefix = $m[2];
1627                                         $s = $m[1];
1628                                 } else {
1629                                         $prefix='';
1630                                 }
1631                                 # first link
1632                                 if($first_prefix) {
1633                                         $prefix = $first_prefix;
1634                                         $first_prefix = false;
1635                                 }
1636                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1637                         }
1638
1639                         $might_be_img = false;
1640
1641                         wfProfileIn( __METHOD__."-e1" );
1642                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1643                                 $text = $m[2];
1644                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1645                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1646                                 # the real problem is with the $e1 regex
1647                                 # See bug 1300.
1648                                 #
1649                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1650                                 # and no image is in sight. See bug 2095.
1651                                 #
1652                                 if( $text !== '' &&
1653                                         substr( $m[3], 0, 1 ) === ']' &&
1654                                         strpos($text, '[') !== false
1655                                 )
1656                                 {
1657                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1658                                         $m[3] = substr( $m[3], 1 );
1659                                 }
1660                                 # fix up urlencoded title texts
1661                                 if( strpos( $m[1], '%' ) !== false ) {
1662                                         # Should anchors '#' also be rejected?
1663                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1664                                 }
1665                                 $trail = $m[3];
1666                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1667                                 $might_be_img = true;
1668                                 $text = $m[2];
1669                                 if ( strpos( $m[1], '%' ) !== false ) {
1670                                         $m[1] = urldecode($m[1]);
1671                                 }
1672                                 $trail = "";
1673                         } else { # Invalid form; output directly
1674                                 $s .= $prefix . '[[' . $line ;
1675                                 wfProfileOut( __METHOD__."-e1" );
1676                                 continue;
1677                         }
1678                         wfProfileOut( __METHOD__."-e1" );
1679                         wfProfileIn( __METHOD__."-misc" );
1680
1681                         # Don't allow internal links to pages containing
1682                         # PROTO: where PROTO is a valid URL protocol; these
1683                         # should be external links.
1684                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1685                                 $s .= $prefix . '[[' . $line ;
1686                                 wfProfileOut( __METHOD__."-misc" );
1687                                 continue;
1688                         }
1689
1690                         # Make subpage if necessary
1691                         if( $useSubpages ) {
1692                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1693                         } else {
1694                                 $link = $m[1];
1695                         }
1696
1697                         $noforce = (substr($m[1], 0, 1) !== ':');
1698                         if (!$noforce) {
1699                                 # Strip off leading ':'
1700                                 $link = substr($link, 1);
1701                         }
1702
1703                         wfProfileOut( __METHOD__."-misc" );
1704                         wfProfileIn( __METHOD__."-title" );
1705                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1706                         if( !$nt ) {
1707                                 $s .= $prefix . '[[' . $line;
1708                                 wfProfileOut( __METHOD__."-title" );
1709                                 continue;
1710                         }
1711
1712                         $ns = $nt->getNamespace();
1713                         $iw = $nt->getInterWiki();
1714                         wfProfileOut( __METHOD__."-title" );
1715
1716                         if ($might_be_img) { # if this is actually an invalid link
1717                                 wfProfileIn( __METHOD__."-might_be_img" );
1718                                 if ($ns == NS_FILE && $noforce) { #but might be an image
1719                                         $found = false;
1720                                         while ( true ) {
1721                                                 #look at the next 'line' to see if we can close it there
1722                                                 $a->next();
1723                                                 $next_line = $a->current();
1724                                                 if ( $next_line === false || $next_line === null ) {
1725                                                         break;
1726                                                 }
1727                                                 $m = explode( ']]', $next_line, 3 );
1728                                                 if ( count( $m ) == 3 ) {
1729                                                         # the first ]] closes the inner link, the second the image
1730                                                         $found = true;
1731                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1732                                                         $trail = $m[2];
1733                                                         break;
1734                                                 } elseif ( count( $m ) == 2 ) {
1735                                                         #if there's exactly one ]] that's fine, we'll keep looking
1736                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1737                                                 } else {
1738                                                         #if $next_line is invalid too, we need look no further
1739                                                         $text .= '[[' . $next_line;
1740                                                         break;
1741                                                 }
1742                                         }
1743                                         if ( !$found ) {
1744                                                 # we couldn't find the end of this imageLink, so output it raw
1745                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1746                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1747                                                 $s .= "{$prefix}[[$link|$text";
1748                                                 # note: no $trail, because without an end, there *is* no trail
1749                                                 wfProfileOut( __METHOD__."-might_be_img" );
1750                                                 continue;
1751                                         }
1752                                 } else { #it's not an image, so output it raw
1753                                         $s .= "{$prefix}[[$link|$text";
1754                                         # note: no $trail, because without an end, there *is* no trail
1755                                         wfProfileOut( __METHOD__."-might_be_img" );
1756                                         continue;
1757                                 }
1758                                 wfProfileOut( __METHOD__."-might_be_img" );
1759                         }
1760
1761                         $wasblank = ( '' == $text );
1762                         if( $wasblank ) $text = $link;
1763
1764                         # Link not escaped by : , create the various objects
1765                         if( $noforce ) {
1766
1767                                 # Interwikis
1768                                 wfProfileIn( __METHOD__."-interwiki" );
1769                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1770                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1771                                         $s = rtrim($s . $prefix);
1772                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1773                                         wfProfileOut( __METHOD__."-interwiki" );
1774                                         continue;
1775                                 }
1776                                 wfProfileOut( __METHOD__."-interwiki" );
1777
1778                                 if ( $ns == NS_FILE ) {
1779                                         wfProfileIn( __METHOD__."-image" );
1780                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1781                                                 # recursively parse links inside the image caption
1782                                                 # actually, this will parse them in any other parameters, too,
1783                                                 # but it might be hard to fix that, and it doesn't matter ATM
1784                                                 $text = $this->replaceExternalLinks($text);
1785                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1786
1787                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1788                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1789                                         }
1790                                         $this->mOutput->addImage( $nt->getDBkey() );
1791                                         wfProfileOut( __METHOD__."-image" );
1792                                         continue;
1793
1794                                 }
1795
1796                                 if ( $ns == NS_CATEGORY ) {
1797                                         wfProfileIn( __METHOD__."-category" );
1798                                         $s = rtrim($s . "\n"); # bug 87
1799
1800                                         if ( $wasblank ) {
1801                                                 $sortkey = $this->getDefaultSort();
1802                                         } else {
1803                                                 $sortkey = $text;
1804                                         }
1805                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1806                                         $sortkey = str_replace( "\n", '', $sortkey );
1807                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1808                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1809
1810                                         /**
1811                                          * Strip the whitespace Category links produce, see bug 87
1812                                          * @todo We might want to use trim($tmp, "\n") here.
1813                                          */
1814                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1815
1816                                         wfProfileOut( __METHOD__."-category" );
1817                                         continue;
1818                                 }
1819                         }
1820
1821                         # Self-link checking
1822                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1823                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1824                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1825                                         continue;
1826                                 }
1827                         }
1828
1829                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1830                         # FIXME: Should do batch file existence checks, see comment below
1831                         if( $ns == NS_MEDIA ) {
1832                                 # Give extensions a chance to select the file revision for us
1833                                 $skip = $time = false;
1834                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1835                                 if ( $skip ) {
1836                                         $link = $sk->link( $nt );
1837                                 } else {
1838                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1839                                 }
1840                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1841                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1842                                 $this->mOutput->addImage( $nt->getDBkey() );
1843                                 continue;
1844                         }
1845
1846                         # Some titles, such as valid special pages or files in foreign repos, should
1847                         # be shown as bluelinks even though they're not included in the page table
1848                         #
1849                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1850                         # batch file existence checks for NS_FILE and NS_MEDIA
1851                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1852                                 $this->mOutput->addLink( $nt );
1853                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1854                         } else {
1855                                 # Links will be added to the output link list after checking
1856                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1857                         }
1858                 }
1859                 wfProfileOut( __METHOD__ );
1860                 return $holders;
1861         }
1862
1863         /**
1864          * Make a link placeholder. The text returned can be later resolved to a real link with
1865          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1866          * parsing of interwiki links, and secondly to allow all existence checks and
1867          * article length checks (for stub links) to be bundled into a single query.
1868          *
1869          * @deprecated
1870          */
1871         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1872                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1873         }
1874
1875         /**
1876          * Render a forced-blue link inline; protect against double expansion of
1877          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1878          * Since this little disaster has to split off the trail text to avoid
1879          * breaking URLs in the following text without breaking trails on the
1880          * wiki links, it's been made into a horrible function.
1881          *
1882          * @param Title $nt
1883          * @param string $text
1884          * @param string $query
1885          * @param string $trail
1886          * @param string $prefix
1887          * @return string HTML-wikitext mix oh yuck
1888          */
1889         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1890                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1891                 $sk = $this->mOptions->getSkin();
1892                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1893                 return $this->armorLinks( $link ) . $trail;
1894         }
1895
1896         /**
1897          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1898          * going to go through further parsing steps before inline URL expansion.
1899          *
1900          * Not needed quite as much as it used to be since free links are a bit
1901          * more sensible these days. But bracketed links are still an issue.
1902          *
1903          * @param string more-or-less HTML
1904          * @return string less-or-more HTML with NOPARSE bits
1905          */
1906         function armorLinks( $text ) {
1907                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1908                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1909         }
1910
1911         /**
1912          * Return true if subpage links should be expanded on this page.
1913          * @return bool
1914          */
1915         function areSubpagesAllowed() {
1916                 # Some namespaces don't allow subpages
1917                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1918         }
1919
1920         /**
1921          * Handle link to subpage if necessary
1922          * @param string $target the source of the link
1923          * @param string &$text the link text, modified as necessary
1924          * @return string the full name of the link
1925          * @private
1926          */
1927         function maybeDoSubpageLink($target, &$text) {
1928                 # Valid link forms:
1929                 # Foobar -- normal
1930                 # :Foobar -- override special treatment of prefix (images, language links)
1931                 # /Foobar -- convert to CurrentPage/Foobar
1932                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1933                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1934                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1935
1936                 wfProfileIn( __METHOD__ );
1937                 $ret = $target; # default return value is no change
1938
1939                 # Some namespaces don't allow subpages,
1940                 # so only perform processing if subpages are allowed
1941                 if( $this->areSubpagesAllowed() ) {
1942                         $hash = strpos( $target, '#' );
1943                         if( $hash !== false ) {
1944                                 $suffix = substr( $target, $hash );
1945                                 $target = substr( $target, 0, $hash );
1946                         } else {
1947                                 $suffix = '';
1948                         }
1949                         # bug 7425
1950                         $target = trim( $target );
1951                         # Look at the first character
1952                         if( $target != '' && $target{0} === '/' ) {
1953                                 # / at end means we don't want the slash to be shown
1954                                 $m = array();
1955                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1956                                 if( $trailingSlashes ) {
1957                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1958                                 } else {
1959                                         $noslash = substr( $target, 1 );
1960                                 }
1961
1962                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1963                                 if( '' === $text ) {
1964                                         $text = $target . $suffix;
1965                                 } # this might be changed for ugliness reasons
1966                         } else {
1967                                 # check for .. subpage backlinks
1968                                 $dotdotcount = 0;
1969                                 $nodotdot = $target;
1970                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1971                                         ++$dotdotcount;
1972                                         $nodotdot = substr( $nodotdot, 3 );
1973                                 }
1974                                 if($dotdotcount > 0) {
1975                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1976                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1977                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1978                                                 # / at the end means don't show full path
1979                                                 if( substr( $nodotdot, -1, 1 ) === '/' ) {
1980                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1981                                                         if( '' === $text ) {
1982                                                                 $text = $nodotdot . $suffix;
1983                                                         }
1984                                                 }
1985                                                 $nodotdot = trim( $nodotdot );
1986                                                 if( $nodotdot != '' ) {
1987                                                         $ret .= '/' . $nodotdot;
1988                                                 }
1989                                                 $ret .= $suffix;
1990                                         }
1991                                 }
1992                         }
1993                 }
1994
1995                 wfProfileOut( __METHOD__ );
1996                 return $ret;
1997         }
1998
1999         /**#@+
2000          * Used by doBlockLevels()
2001          * @private
2002          */
2003         /* private */ function closeParagraph() {
2004                 $result = '';
2005                 if ( '' != $this->mLastSection ) {
2006                         $result = '</' . $this->mLastSection  . ">\n";
2007                 }
2008                 $this->mInPre = false;
2009                 $this->mLastSection = '';
2010                 return $result;
2011         }
2012         # getCommon() returns the length of the longest common substring
2013         # of both arguments, starting at the beginning of both.
2014         #
2015         /* private */ function getCommon( $st1, $st2 ) {
2016                 $fl = strlen( $st1 );
2017                 $shorter = strlen( $st2 );
2018                 if ( $fl < $shorter ) { $shorter = $fl; }
2019
2020                 for ( $i = 0; $i < $shorter; ++$i ) {
2021                         if ( $st1{$i} != $st2{$i} ) { break; }
2022                 }
2023                 return $i;
2024         }
2025         # These next three functions open, continue, and close the list
2026         # element appropriate to the prefix character passed into them.
2027         #
2028         /* private */ function openList( $char ) {
2029                 $result = $this->closeParagraph();
2030
2031                 if ( '*' === $char ) { $result .= '<ul><li>'; }
2032                 else if ( '#' === $char ) { $result .= '<ol><li>'; }
2033                 else if ( ':' === $char ) { $result .= '<dl><dd>'; }
2034                 else if ( ';' === $char ) {
2035                         $result .= '<dl><dt>';
2036                         $this->mDTopen = true;
2037                 }
2038                 else { $result = '<!-- ERR 1 -->'; }
2039
2040                 return $result;
2041         }
2042
2043         /* private */ function nextItem( $char ) {
2044                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
2045                 else if ( ':' === $char || ';' === $char ) {
2046                         $close = '</dd>';
2047                         if ( $this->mDTopen ) { $close = '</dt>'; }
2048                         if ( ';' === $char ) {
2049                                 $this->mDTopen = true;
2050                                 return $close . '<dt>';
2051                         } else {
2052                                 $this->mDTopen = false;
2053                                 return $close . '<dd>';
2054                         }
2055                 }
2056                 return '<!-- ERR 2 -->';
2057         }
2058
2059         /* private */ function closeList( $char ) {
2060                 if ( '*' === $char ) { $text = '</li></ul>'; }
2061                 else if ( '#' === $char ) { $text = '</li></ol>'; }
2062                 else if ( ':' === $char ) {
2063                         if ( $this->mDTopen ) {
2064                                 $this->mDTopen = false;
2065                                 $text = '</dt></dl>';
2066                         } else {
2067                                 $text = '</dd></dl>';
2068                         }
2069                 }
2070                 else {  return '<!-- ERR 3 -->'; }
2071                 return $text."\n";
2072         }
2073         /**#@-*/
2074
2075         /**
2076          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2077          *
2078          * @private
2079          * @return string the lists rendered as HTML
2080          */
2081         function doBlockLevels( $text, $linestart ) {
2082                 wfProfileIn( __METHOD__ );
2083
2084                 # Parsing through the text line by line.  The main thing
2085                 # happening here is handling of block-level elements p, pre,
2086                 # and making lists from lines starting with * # : etc.
2087                 #
2088                 $textLines = StringUtils::explode( "\n", $text );
2089
2090                 $lastPrefix = $output = '';
2091                 $this->mDTopen = $inBlockElem = false;
2092                 $prefixLength = 0;
2093                 $paragraphStack = false;
2094
2095                 foreach ( $textLines as $oLine ) {
2096                         # Fix up $linestart
2097                         if ( !$linestart ) {
2098                                 $output .= $oLine;
2099                                 $linestart = true;
2100                                 continue;
2101                         }
2102
2103                         $lastPrefixLength = strlen( $lastPrefix );
2104                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2105                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2106                         if ( !$this->mInPre ) {
2107                                 # Multiple prefixes may abut each other for nested lists.
2108                                 $prefixLength = strspn( $oLine, '*#:;' );
2109                                 $prefix = substr( $oLine, 0, $prefixLength );
2110
2111                                 # eh?
2112                                 $prefix2 = str_replace( ';', ':', $prefix );
2113                                 $t = substr( $oLine, $prefixLength );
2114                                 $this->mInPre = (bool)$preOpenMatch;
2115                         } else {
2116                                 # Don't interpret any other prefixes in preformatted text
2117                                 $prefixLength = 0;
2118                                 $prefix = $prefix2 = '';
2119                                 $t = $oLine;
2120                         }
2121
2122                         # List generation
2123                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2124                                 # Same as the last item, so no need to deal with nesting or opening stuff
2125                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2126                                 $paragraphStack = false;
2127
2128                                 if ( substr( $prefix, -1 ) === ';') {
2129                                         # The one nasty exception: definition lists work like this:
2130                                         # ; title : definition text
2131                                         # So we check for : in the remainder text to split up the
2132                                         # title and definition, without b0rking links.
2133                                         $term = $t2 = '';
2134                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2135                                                 $t = $t2;
2136                                                 $output .= $term . $this->nextItem( ':' );
2137                                         }
2138                                 }
2139                         } elseif( $prefixLength || $lastPrefixLength ) {
2140                                 # Either open or close a level...
2141                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2142                                 $paragraphStack = false;
2143
2144                                 while( $commonPrefixLength < $lastPrefixLength ) {
2145                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2146                                         --$lastPrefixLength;
2147                                 }
2148                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2149                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2150                                 }
2151                                 while ( $prefixLength > $commonPrefixLength ) {
2152                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2153                                         $output .= $this->openList( $char );
2154
2155                                         if ( ';' === $char ) {
2156                                                 # FIXME: This is dupe of code above
2157                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2158                                                         $t = $t2;
2159                                                         $output .= $term . $this->nextItem( ':' );
2160                                                 }
2161                                         }
2162                                         ++$commonPrefixLength;
2163                                 }
2164                                 $lastPrefix = $prefix2;
2165                         }
2166                         if( 0 == $prefixLength ) {
2167                                 wfProfileIn( __METHOD__."-paragraph" );
2168                                 # No prefix (not in list)--go to paragraph mode
2169                                 // XXX: use a stack for nestable elements like span, table and div
2170                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2171                                 $closematch = preg_match(
2172                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2173                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2174                                 if ( $openmatch or $closematch ) {
2175                                         $paragraphStack = false;
2176                                         # TODO bug 5718: paragraph closed
2177                                         $output .= $this->closeParagraph();
2178                                         if ( $preOpenMatch and !$preCloseMatch ) {
2179                                                 $this->mInPre = true;
2180                                         }
2181                                         if ( $closematch ) {
2182                                                 $inBlockElem = false;
2183                                         } else {
2184                                                 $inBlockElem = true;
2185                                         }
2186                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2187                                         if ( ' ' == $t{0} and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2188                                                 // pre
2189                                                 if ($this->mLastSection !== 'pre') {
2190                                                         $paragraphStack = false;
2191                                                         $output .= $this->closeParagraph().'<pre>';
2192                                                         $this->mLastSection = 'pre';
2193                                                 }
2194                                                 $t = substr( $t, 1 );
2195                                         } else {
2196                                                 // paragraph
2197                                                 if ( '' == trim($t) ) {
2198                                                         if ( $paragraphStack ) {
2199                                                                 $output .= $paragraphStack.'<br />';
2200                                                                 $paragraphStack = false;
2201                                                                 $this->mLastSection = 'p';
2202                                                         } else {
2203                                                                 if ($this->mLastSection !== 'p' ) {
2204                                                                         $output .= $this->closeParagraph();
2205                                                                         $this->mLastSection = '';
2206                                                                         $paragraphStack = '<p>';
2207                                                                 } else {
2208                                                                         $paragraphStack = '</p><p>';
2209                                                                 }
2210                                                         }
2211                                                 } else {
2212                                                         if ( $paragraphStack ) {
2213                                                                 $output .= $paragraphStack;
2214                                                                 $paragraphStack = false;
2215                                                                 $this->mLastSection = 'p';
2216                                                         } else if ($this->mLastSection !== 'p') {
2217                                                                 $output .= $this->closeParagraph().'<p>';
2218                                                                 $this->mLastSection = 'p';
2219                                                         }
2220                                                 }
2221                                         }
2222                                 }
2223                                 wfProfileOut( __METHOD__."-paragraph" );
2224                         }
2225                         // somewhere above we forget to get out of pre block (bug 785)
2226                         if($preCloseMatch && $this->mInPre) {
2227                                 $this->mInPre = false;
2228                         }
2229                         if ($paragraphStack === false) {
2230                                 $output .= $t."\n";
2231                         }
2232                 }
2233                 while ( $prefixLength ) {
2234                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2235                         --$prefixLength;
2236                 }
2237                 if ( '' != $this->mLastSection ) {
2238                         $output .= '</' . $this->mLastSection . '>';
2239                         $this->mLastSection = '';
2240                 }
2241
2242                 wfProfileOut( __METHOD__ );
2243                 return $output;
2244         }
2245
2246         /**
2247          * Split up a string on ':', ignoring any occurences inside tags
2248          * to prevent illegal overlapping.
2249          * @param string $str the string to split
2250          * @param string &$before set to everything before the ':'
2251          * @param string &$after set to everything after the ':'
2252          * return string the position of the ':', or false if none found
2253          */
2254         function findColonNoLinks($str, &$before, &$after) {
2255                 wfProfileIn( __METHOD__ );
2256
2257                 $pos = strpos( $str, ':' );
2258                 if( $pos === false ) {
2259                         // Nothing to find!
2260                         wfProfileOut( __METHOD__ );
2261                         return false;
2262                 }
2263
2264                 $lt = strpos( $str, '<' );
2265                 if( $lt === false || $lt > $pos ) {
2266                         // Easy; no tag nesting to worry about
2267                         $before = substr( $str, 0, $pos );
2268                         $after = substr( $str, $pos+1 );
2269                         wfProfileOut( __METHOD__ );
2270                         return $pos;
2271                 }
2272
2273                 // Ugly state machine to walk through avoiding tags.
2274                 $state = self::COLON_STATE_TEXT;
2275                 $stack = 0;
2276                 $len = strlen( $str );
2277                 for( $i = 0; $i < $len; $i++ ) {
2278                         $c = $str{$i};
2279
2280                         switch( $state ) {
2281                         // (Using the number is a performance hack for common cases)
2282                         case 0: // self::COLON_STATE_TEXT:
2283                                 switch( $c ) {
2284                                 case "<":
2285                                         // Could be either a <start> tag or an </end> tag
2286                                         $state = self::COLON_STATE_TAGSTART;
2287                                         break;
2288                                 case ":":
2289                                         if( $stack == 0 ) {
2290                                                 // We found it!
2291                                                 $before = substr( $str, 0, $i );
2292                                                 $after = substr( $str, $i + 1 );
2293                                                 wfProfileOut( __METHOD__ );
2294                                                 return $i;
2295                                         }
2296                                         // Embedded in a tag; don't break it.
2297                                         break;
2298                                 default:
2299                                         // Skip ahead looking for something interesting
2300                                         $colon = strpos( $str, ':', $i );
2301                                         if( $colon === false ) {
2302                                                 // Nothing else interesting
2303                                                 wfProfileOut( __METHOD__ );
2304                                                 return false;
2305                                         }
2306                                         $lt = strpos( $str, '<', $i );
2307                                         if( $stack === 0 ) {
2308                                                 if( $lt === false || $colon < $lt ) {
2309                                                         // We found it!
2310                                                         $before = substr( $str, 0, $colon );
2311                                                         $after = substr( $str, $colon + 1 );
2312                                                         wfProfileOut( __METHOD__ );
2313                                                         return $i;
2314                                                 }
2315                                         }
2316                                         if( $lt === false ) {
2317                                                 // Nothing else interesting to find; abort!
2318                                                 // We're nested, but there's no close tags left. Abort!
2319                                                 break 2;
2320                                         }
2321                                         // Skip ahead to next tag start
2322                                         $i = $lt;
2323                                         $state = self::COLON_STATE_TAGSTART;
2324                                 }
2325                                 break;
2326                         case 1: // self::COLON_STATE_TAG:
2327                                 // In a <tag>
2328                                 switch( $c ) {
2329                                 case ">":
2330                                         $stack++;
2331                                         $state = self::COLON_STATE_TEXT;
2332                                         break;
2333                                 case "/":
2334                                         // Slash may be followed by >?
2335                                         $state = self::COLON_STATE_TAGSLASH;
2336                                         break;
2337                                 default:
2338                                         // ignore
2339                                 }
2340                                 break;
2341                         case 2: // self::COLON_STATE_TAGSTART:
2342                                 switch( $c ) {
2343                                 case "/":
2344                                         $state = self::COLON_STATE_CLOSETAG;
2345                                         break;
2346                                 case "!":
2347                                         $state = self::COLON_STATE_COMMENT;
2348                                         break;
2349                                 case ">":
2350                                         // Illegal early close? This shouldn't happen D:
2351                                         $state = self::COLON_STATE_TEXT;
2352                                         break;
2353                                 default:
2354                                         $state = self::COLON_STATE_TAG;
2355                                 }
2356                                 break;
2357                         case 3: // self::COLON_STATE_CLOSETAG:
2358                                 // In a </tag>
2359                                 if( $c === ">" ) {
2360                                         $stack--;
2361                                         if( $stack < 0 ) {
2362                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2363                                                 wfProfileOut( __METHOD__ );
2364                                                 return false;
2365                                         }
2366                                         $state = self::COLON_STATE_TEXT;
2367                                 }
2368                                 break;
2369                         case self::COLON_STATE_TAGSLASH:
2370                                 if( $c === ">" ) {
2371                                         // Yes, a self-closed tag <blah/>
2372                                         $state = self::COLON_STATE_TEXT;
2373                                 } else {
2374                                         // Probably we're jumping the gun, and this is an attribute
2375                                         $state = self::COLON_STATE_TAG;
2376                                 }
2377                                 break;
2378                         case 5: // self::COLON_STATE_COMMENT:
2379                                 if( $c === "-" ) {
2380                                         $state = self::COLON_STATE_COMMENTDASH;
2381                                 }
2382                                 break;
2383                         case self::COLON_STATE_COMMENTDASH:
2384                                 if( $c === "-" ) {
2385                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2386                                 } else {
2387                                         $state = self::COLON_STATE_COMMENT;
2388                                 }
2389                                 break;
2390                         case self::COLON_STATE_COMMENTDASHDASH:
2391                                 if( $c === ">" ) {
2392                                         $state = self::COLON_STATE_TEXT;
2393                                 } else {
2394                                         $state = self::COLON_STATE_COMMENT;
2395                                 }
2396                                 break;
2397                         default:
2398                                 throw new MWException( "State machine error in " . __METHOD__ );
2399                         }
2400                 }
2401                 if( $stack > 0 ) {
2402                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2403                         return false;
2404                 }
2405                 wfProfileOut( __METHOD__ );
2406                 return false;
2407         }
2408
2409         /**
2410          * Return value of a magic variable (like PAGENAME)
2411          *
2412          * @private
2413          */
2414         function getVariableValue( $index ) {
2415                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2416
2417                 /**
2418                  * Some of these require message or data lookups and can be
2419                  * expensive to check many times.
2420                  */
2421                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2422                         if ( isset( $this->mVarCache[$index] ) ) {
2423                                 return $this->mVarCache[$index];
2424                         }
2425                 }
2426
2427                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2428                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2429
2430                 # Use the time zone
2431                 global $wgLocaltimezone;
2432                 if ( isset( $wgLocaltimezone ) ) {
2433                         $oldtz = getenv( 'TZ' );
2434                         putenv( 'TZ='.$wgLocaltimezone );
2435                 }
2436
2437                 wfSuppressWarnings(); // E_STRICT system time bitching
2438                 $localTimestamp = date( 'YmdHis', $ts );
2439                 $localMonth = date( 'm', $ts );
2440                 $localMonthName = date( 'n', $ts );
2441                 $localDay = date( 'j', $ts );
2442                 $localDay2 = date( 'd', $ts );
2443                 $localDayOfWeek = date( 'w', $ts );
2444                 $localWeek = date( 'W', $ts );
2445                 $localYear = date( 'Y', $ts );
2446                 $localHour = date( 'H', $ts );
2447                 if ( isset( $wgLocaltimezone ) ) {
2448                         putenv( 'TZ='.$oldtz );
2449                 }
2450                 wfRestoreWarnings();
2451
2452                 switch ( $index ) {
2453                         case 'currentmonth':
2454                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2455                         case 'currentmonthname':
2456                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2457                         case 'currentmonthnamegen':
2458                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2459                         case 'currentmonthabbrev':
2460                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2461                         case 'currentday':
2462                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2463                         case 'currentday2':
2464                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2465                         case 'localmonth':
2466                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2467                         case 'localmonthname':
2468                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2469                         case 'localmonthnamegen':
2470                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2471                         case 'localmonthabbrev':
2472                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2473                         case 'localday':
2474                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2475                         case 'localday2':
2476                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2477                         case 'pagename':
2478                                 return wfEscapeWikiText( $this->mTitle->getText() );
2479                         case 'pagenamee':
2480                                 return $this->mTitle->getPartialURL();
2481                         case 'fullpagename':
2482                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2483                         case 'fullpagenamee':
2484                                 return $this->mTitle->getPrefixedURL();
2485                         case 'subpagename':
2486                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2487                         case 'subpagenamee':
2488                                 return $this->mTitle->getSubpageUrlForm();
2489                         case 'basepagename':
2490                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2491                         case 'basepagenamee':
2492                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2493                         case 'talkpagename':
2494                                 if( $this->mTitle->canTalk() ) {
2495                                         $talkPage = $this->mTitle->getTalkPage();
2496                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2497                                 } else {
2498                                         return '';
2499                                 }
2500                         case 'talkpagenamee':
2501                                 if( $this->mTitle->canTalk() ) {
2502                                         $talkPage = $this->mTitle->getTalkPage();
2503                                         return $talkPage->getPrefixedUrl();
2504                                 } else {
2505                                         return '';
2506                                 }
2507                         case 'subjectpagename':
2508                                 $subjPage = $this->mTitle->getSubjectPage();
2509                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2510                         case 'subjectpagenamee':
2511                                 $subjPage = $this->mTitle->getSubjectPage();
2512                                 return $subjPage->getPrefixedUrl();
2513                         case 'revisionid':
2514                                 // Let the edit saving system know we should parse the page
2515                                 // *after* a revision ID has been assigned.
2516                                 $this->mOutput->setFlag( 'vary-revision' );
2517                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2518                                 return $this->mRevisionId;
2519                         case 'revisionday':
2520                                 // Let the edit saving system know we should parse the page
2521                                 // *after* a revision ID has been assigned. This is for null edits.
2522                                 $this->mOutput->setFlag( 'vary-revision' );
2523                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2524                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2525                         case 'revisionday2':
2526                                 // Let the edit saving system know we should parse the page
2527                                 // *after* a revision ID has been assigned. This is for null edits.
2528                                 $this->mOutput->setFlag( 'vary-revision' );
2529                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2530                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2531                         case 'revisionmonth':
2532                                 // Let the edit saving system know we should parse the page
2533                                 // *after* a revision ID has been assigned. This is for null edits.
2534                                 $this->mOutput->setFlag( 'vary-revision' );
2535                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2536                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2537                         case 'revisionyear':
2538                                 // Let the edit saving system know we should parse the page
2539                                 // *after* a revision ID has been assigned. This is for null edits.
2540                                 $this->mOutput->setFlag( 'vary-revision' );
2541                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2542                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2543                         case 'revisiontimestamp':
2544                                 // Let the edit saving system know we should parse the page
2545                                 // *after* a revision ID has been assigned. This is for null edits.
2546                                 $this->mOutput->setFlag( 'vary-revision' );
2547                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2548                                 return $this->getRevisionTimestamp();
2549                         case 'namespace':
2550                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2551                         case 'namespacee':
2552                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2553                         case 'talkspace':
2554                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2555                         case 'talkspacee':
2556                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2557                         case 'subjectspace':
2558                                 return $this->mTitle->getSubjectNsText();
2559                         case 'subjectspacee':
2560                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2561                         case 'currentdayname':
2562                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2563                         case 'currentyear':
2564                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2565                         case 'currenttime':
2566                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2567                         case 'currenthour':
2568                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2569                         case 'currentweek':
2570                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2571                                 // int to remove the padding
2572                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2573                         case 'currentdow':
2574                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2575                         case 'localdayname':
2576                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2577                         case 'localyear':
2578                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2579                         case 'localtime':
2580                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2581                         case 'localhour':
2582                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2583                         case 'localweek':
2584                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2585                                 // int to remove the padding
2586                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2587                         case 'localdow':
2588                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2589                         case 'numberofarticles':
2590                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2591                         case 'numberoffiles':
2592                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2593                         case 'numberofusers':
2594                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2595                         case 'numberofpages':
2596                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2597                         case 'numberofadmins':
2598                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2599                         case 'numberofedits':
2600                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2601                         case 'numberofviews':
2602                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::views() );
2603                         case 'currenttimestamp':
2604                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2605                         case 'localtimestamp':
2606                                 return $this->mVarCache[$index] = $localTimestamp;
2607                         case 'currentversion':
2608                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2609                         case 'sitename':
2610                                 return $wgSitename;
2611                         case 'server':
2612                                 return $wgServer;
2613                         case 'servername':
2614                                 return $wgServerName;
2615                         case 'scriptpath':
2616                                 return $wgScriptPath;
2617                         case 'directionmark':
2618                                 return $wgContLang->getDirMark();
2619                         case 'contentlanguage':
2620                                 global $wgContLanguageCode;
2621                                 return $wgContLanguageCode;
2622                         default:
2623                                 $ret = null;
2624                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2625                                         return $ret;
2626                                 else
2627                                         return null;
2628                 }
2629         }
2630
2631         /**
2632          * initialise the magic variables (like CURRENTMONTHNAME)
2633          *
2634          * @private
2635          */
2636         function initialiseVariables() {
2637                 wfProfileIn( __METHOD__ );
2638                 $variableIDs = MagicWord::getVariableIDs();
2639
2640                 $this->mVariables = new MagicWordArray( $variableIDs );
2641                 wfProfileOut( __METHOD__ );
2642         }
2643
2644         /**
2645          * Preprocess some wikitext and return the document tree.
2646          * This is the ghost of replace_variables().
2647          *
2648          * @param string $text The text to parse
2649          * @param integer flags Bitwise combination of:
2650          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2651          *                                     included. Default is to assume a direct page view.
2652          *
2653          * The generated DOM tree must depend only on the input text and the flags.
2654          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2655          *
2656          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2657          * change in the DOM tree for a given text, must be passed through the section identifier
2658          * in the section edit link and thus back to extractSections().
2659          *
2660          * The output of this function is currently only cached in process memory, but a persistent
2661          * cache may be implemented at a later date which takes further advantage of these strict
2662          * dependency requirements.
2663          *
2664          * @private
2665          */
2666         function preprocessToDom ( $text, $flags = 0 ) {
2667                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2668                 return $dom;
2669         }
2670
2671         /*
2672          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2673          */
2674         public static function splitWhitespace( $s ) {
2675                 $ltrimmed = ltrim( $s );
2676                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2677                 $trimmed = rtrim( $ltrimmed );
2678                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2679                 if ( $diff > 0 ) {
2680                         $w2 = substr( $ltrimmed, -$diff );
2681                 } else {
2682                         $w2 = '';
2683                 }
2684                 return array( $w1, $trimmed, $w2 );
2685         }
2686
2687         /**
2688          * Replace magic variables, templates, and template arguments
2689          * with the appropriate text. Templates are substituted recursively,
2690          * taking care to avoid infinite loops.
2691          *
2692          * Note that the substitution depends on value of $mOutputType:
2693          *  self::OT_WIKI: only {{subst:}} templates
2694          *  self::OT_PREPROCESS: templates but not extension tags
2695          *  self::OT_HTML: all templates and extension tags
2696          *
2697          * @param string $tex The text to transform
2698          * @param PPFrame $frame Object describing the arguments passed to the template.
2699          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2700          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2701          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2702          * @private
2703          */
2704         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2705                 # Prevent too big inclusions
2706                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2707                         return $text;
2708                 }
2709
2710                 wfProfileIn( __METHOD__ );
2711
2712                 if ( $frame === false ) {
2713                         $frame = $this->getPreprocessor()->newFrame();
2714                 } elseif ( !( $frame instanceof PPFrame ) ) {
2715                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2716                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2717                 }
2718
2719                 $dom = $this->preprocessToDom( $text );
2720                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2721                 $text = $frame->expand( $dom, $flags );
2722
2723                 wfProfileOut( __METHOD__ );
2724                 return $text;
2725         }
2726
2727         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2728         static function createAssocArgs( $args ) {
2729                 $assocArgs = array();
2730                 $index = 1;
2731                 foreach( $args as $arg ) {
2732                         $eqpos = strpos( $arg, '=' );
2733                         if ( $eqpos === false ) {
2734                                 $assocArgs[$index++] = $arg;
2735                         } else {
2736                                 $name = trim( substr( $arg, 0, $eqpos ) );
2737                                 $value = trim( substr( $arg, $eqpos+1 ) );
2738                                 if ( $value === false ) {
2739                                         $value = '';
2740                                 }
2741                                 if ( $name !== false ) {
2742                                         $assocArgs[$name] = $value;
2743                                 }
2744                         }
2745                 }
2746
2747                 return $assocArgs;
2748         }
2749
2750         /**
2751          * Warn the user when a parser limitation is reached
2752          * Will warn at most once the user per limitation type
2753          *
2754          * @param string $limitationType, should be one of:
2755          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2756          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2757          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2758          * @params int $current, $max When an explicit limit has been
2759          *       exceeded, provide the values (optional)
2760          */
2761         function limitationWarn( $limitationType, $current=null, $max=null) {
2762                 $msgName = $limitationType . '-warning';
2763                 //does no harm if $current and $max are present but are unnecessary for the message
2764                 $warning = wfMsgExt( $msgName, array( 'parsemag', 'escape' ), $current, $max );
2765                 $this->mOutput->addWarning( $warning );
2766                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2767                 if ( $cat ) {
2768                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2769                 }
2770         }
2771
2772         /**
2773          * Return the text of a template, after recursively
2774          * replacing any variables or templates within the template.
2775          *
2776          * @param array $piece The parts of the template
2777          *  $piece['title']: the title, i.e. the part before the |
2778          *  $piece['parts']: the parameter array
2779          *  $piece['lineStart']: whether the brace was at the start of a line
2780          * @param PPFrame The current frame, contains template arguments
2781          * @return string the text of the template
2782          * @private
2783          */
2784         function braceSubstitution( $piece, $frame ) {
2785                 global $wgContLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2786                 wfProfileIn( __METHOD__ );
2787                 wfProfileIn( __METHOD__.'-setup' );
2788
2789                 # Flags
2790                 $found = false;             # $text has been filled
2791                 $nowiki = false;            # wiki markup in $text should be escaped
2792                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2793                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2794                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2795                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2796
2797                 # Title object, where $text came from
2798                 $title = NULL;
2799
2800                 # $part1 is the bit before the first |, and must contain only title characters.
2801                 # Various prefixes will be stripped from it later.
2802                 $titleWithSpaces = $frame->expand( $piece['title'] );
2803                 $part1 = trim( $titleWithSpaces );
2804                 $titleText = false;
2805
2806                 # Original title text preserved for various purposes
2807                 $originalTitle = $part1;
2808
2809                 # $args is a list of argument nodes, starting from index 0, not including $part1
2810                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2811                 wfProfileOut( __METHOD__.'-setup' );
2812
2813                 # SUBST
2814                 wfProfileIn( __METHOD__.'-modifiers' );
2815                 if ( !$found ) {
2816                         $mwSubst = MagicWord::get( 'subst' );
2817                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2818                                 # One of two possibilities is true:
2819                                 # 1) Found SUBST but not in the PST phase
2820                                 # 2) Didn't find SUBST and in the PST phase
2821                                 # In either case, return without further processing
2822                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2823                                 $isLocalObj = true;
2824                                 $found = true;
2825                         }
2826                 }
2827
2828                 # Variables
2829                 if ( !$found && $args->getLength() == 0 ) {
2830                         $id = $this->mVariables->matchStartToEnd( $part1 );
2831                         if ( $id !== false ) {
2832                                 $text = $this->getVariableValue( $id );
2833                                 if (MagicWord::getCacheTTL($id)>-1)
2834                                         $this->mOutput->mContainsOldMagic = true;
2835                                 $found = true;
2836                         }
2837                 }
2838
2839                 # MSG, MSGNW and RAW
2840                 if ( !$found ) {
2841                         # Check for MSGNW:
2842                         $mwMsgnw = MagicWord::get( 'msgnw' );
2843                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2844                                 $nowiki = true;
2845                         } else {
2846                                 # Remove obsolete MSG:
2847                                 $mwMsg = MagicWord::get( 'msg' );
2848                                 $mwMsg->matchStartAndRemove( $part1 );
2849                         }
2850
2851                         # Check for RAW:
2852                         $mwRaw = MagicWord::get( 'raw' );
2853                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2854                                 $forceRawInterwiki = true;
2855                         }
2856                 }
2857                 wfProfileOut( __METHOD__.'-modifiers' );
2858
2859                 # Parser functions
2860                 if ( !$found ) {
2861                         wfProfileIn( __METHOD__ . '-pfunc' );
2862
2863                         $colonPos = strpos( $part1, ':' );
2864                         if ( $colonPos !== false ) {
2865                                 # Case sensitive functions
2866                                 $function = substr( $part1, 0, $colonPos );
2867                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2868                                         $function = $this->mFunctionSynonyms[1][$function];
2869                                 } else {
2870                                         # Case insensitive functions
2871                                         $function = strtolower( $function );
2872                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2873                                                 $function = $this->mFunctionSynonyms[0][$function];
2874                                         } else {
2875                                                 $function = false;
2876                                         }
2877                                 }
2878                                 if ( $function ) {
2879                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2880                                         $initialArgs = array( &$this );
2881                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2882                                         if ( $flags & SFH_OBJECT_ARGS ) {
2883                                                 # Add a frame parameter, and pass the arguments as an array
2884                                                 $allArgs = $initialArgs;
2885                                                 $allArgs[] = $frame;
2886                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2887                                                         $funcArgs[] = $args->item( $i );
2888                                                 }
2889                                                 $allArgs[] = $funcArgs;
2890                                         } else {
2891                                                 # Convert arguments to plain text
2892                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2893                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2894                                                 }
2895                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2896                                         }
2897
2898                                         # Workaround for PHP bug 35229 and similar
2899                                         if ( !is_callable( $callback ) ) {
2900                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2901                                         }
2902                                         $result = call_user_func_array( $callback, $allArgs );
2903                                         $found = true;
2904                                         $noparse = true;
2905                                         $preprocessFlags = 0;
2906
2907                                         if ( is_array( $result ) ) {
2908                                                 if ( isset( $result[0] ) ) {
2909                                                         $text = $result[0];
2910                                                         unset( $result[0] );
2911                                                 }
2912
2913                                                 // Extract flags into the local scope
2914                                                 // This allows callers to set flags such as nowiki, found, etc.
2915                                                 extract( $result );
2916                                         } else {
2917                                                 $text = $result;
2918                                         }
2919                                         if ( !$noparse ) {
2920                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2921                                                 $isChildObj = true;
2922                                         }
2923                                 }
2924                         }
2925                         wfProfileOut( __METHOD__ . '-pfunc' );
2926                 }
2927
2928                 # Finish mangling title and then check for loops.
2929                 # Set $title to a Title object and $titleText to the PDBK
2930                 if ( !$found ) {
2931                         $ns = NS_TEMPLATE;
2932                         # Split the title into page and subpage
2933                         $subpage = '';
2934                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2935                         if ($subpage !== '') {
2936                                 $ns = $this->mTitle->getNamespace();
2937                         }
2938                         $title = Title::newFromText( $part1, $ns );
2939                         if ( $title ) {
2940                                 $titleText = $title->getPrefixedText();
2941                                 # Check for language variants if the template is not found
2942                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2943                                         $wgContLang->findVariantLink( $part1, $title, true );
2944                                 }
2945                                 # Do infinite loop check
2946                                 if ( !$frame->loopCheck( $title ) ) {
2947                                         $found = true;
2948                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2949                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2950                                 }
2951                                 # Do recursion depth check
2952                                 $limit = $this->mOptions->getMaxTemplateDepth();
2953                                 if ( $frame->depth >= $limit ) {
2954                                         $found = true;
2955                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2956                                 }
2957                         }
2958                 }
2959
2960                 # Load from database
2961                 if ( !$found && $title ) {
2962                         wfProfileIn( __METHOD__ . '-loadtpl' );
2963                         if ( !$title->isExternal() ) {
2964                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2965                                         $text = SpecialPage::capturePath( $title );
2966                                         if ( is_string( $text ) ) {
2967                                                 $found = true;
2968                                                 $isHTML = true;
2969                                                 $this->disableCache();
2970                                         }
2971                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2972                                         $found = false; //access denied
2973                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2974                                 } else {
2975                                         list( $text, $title ) = $this->getTemplateDom( $title );
2976                                         if ( $text !== false ) {
2977                                                 $found = true;
2978                                                 $isChildObj = true;
2979                                         }
2980                                 }
2981
2982                                 # If the title is valid but undisplayable, make a link to it
2983                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2984                                         $text = "[[:$titleText]]";
2985                                         $found = true;
2986                                 }
2987                         } elseif ( $title->isTrans() ) {
2988                                 // Interwiki transclusion
2989                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2990                                         $text = $this->interwikiTransclude( $title, 'render' );
2991                                         $isHTML = true;
2992                                 } else {
2993                                         $text = $this->interwikiTransclude( $title, 'raw' );
2994                                         // Preprocess it like a template
2995                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2996                                         $isChildObj = true;
2997                                 }
2998                                 $found = true;
2999                         }
3000                         wfProfileOut( __METHOD__ . '-loadtpl' );
3001                 }
3002
3003                 # If we haven't found text to substitute by now, we're done
3004                 # Recover the source wikitext and return it
3005                 if ( !$found ) {
3006                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3007                         wfProfileOut( __METHOD__ );
3008                         return array( 'object' => $text );
3009                 }
3010
3011                 # Expand DOM-style return values in a child frame
3012                 if ( $isChildObj ) {
3013                         # Clean up argument array
3014                         $newFrame = $frame->newChild( $args, $title );
3015
3016                         if ( $nowiki ) {
3017                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3018                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3019                                 # Expansion is eligible for the empty-frame cache
3020                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3021                                         $text = $this->mTplExpandCache[$titleText];
3022                                 } else {
3023                                         $text = $newFrame->expand( $text );
3024                                         $this->mTplExpandCache[$titleText] = $text;
3025                                 }
3026                         } else {
3027                                 # Uncached expansion
3028                                 $text = $newFrame->expand( $text );
3029                         }
3030                 }
3031                 if ( $isLocalObj && $nowiki ) {
3032                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3033                         $isLocalObj = false;
3034                 }
3035
3036                 # Replace raw HTML by a placeholder
3037                 # Add a blank line preceding, to prevent it from mucking up
3038                 # immediately preceding headings
3039                 if ( $isHTML ) {
3040                         $text = "\n\n" . $this->insertStripItem( $text );
3041                 }
3042                 # Escape nowiki-style return values
3043                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3044                         $text = wfEscapeWikiText( $text );
3045                 }
3046                 # Bug 529: if the template begins with a table or block-level
3047                 # element, it should be treated as beginning a new line.
3048                 # This behaviour is somewhat controversial.
3049                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3050                         $text = "\n" . $text;
3051                 }
3052
3053                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3054                         # Error, oversize inclusion
3055                         $text = "[[$originalTitle]]" .
3056                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3057                         $this->limitationWarn( 'post-expand-template-inclusion' );
3058                 }
3059
3060                 if ( $isLocalObj ) {
3061                         $ret = array( 'object' => $text );
3062                 } else {
3063                         $ret = array( 'text' => $text );
3064                 }
3065
3066                 wfProfileOut( __METHOD__ );
3067                 return $ret;
3068         }
3069
3070         /**
3071          * Get the semi-parsed DOM representation of a template with a given title,
3072          * and its redirect destination title. Cached.
3073          */
3074         function getTemplateDom( $title ) {
3075                 $cacheTitle = $title;
3076                 $titleText = $title->getPrefixedDBkey();
3077
3078                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3079                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3080                         $title = Title::makeTitle( $ns, $dbk );
3081                         $titleText = $title->getPrefixedDBkey();
3082                 }
3083                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3084                         return array( $this->mTplDomCache[$titleText], $title );
3085                 }
3086
3087                 // Cache miss, go to the database
3088                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3089
3090                 if ( $text === false ) {
3091                         $this->mTplDomCache[$titleText] = false;
3092                         return array( false, $title );
3093                 }
3094
3095                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3096                 $this->mTplDomCache[ $titleText ] = $dom;
3097
3098                 if (! $title->equals($cacheTitle)) {
3099                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3100                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3101                 }
3102
3103                 return array( $dom, $title );
3104         }
3105
3106         /**
3107          * Fetch the unparsed text of a template and register a reference to it.
3108          */
3109         function fetchTemplateAndTitle( $title ) {
3110                 $templateCb = $this->mOptions->getTemplateCallback();
3111                 $stuff = call_user_func( $templateCb, $title, $this );
3112                 $text = $stuff['text'];
3113                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3114                 if ( isset( $stuff['deps'] ) ) {
3115                         foreach ( $stuff['deps'] as $dep ) {
3116                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3117                         }
3118                 }
3119                 return array($text,$finalTitle);
3120         }
3121
3122         function fetchTemplate( $title ) {
3123                 $rv = $this->fetchTemplateAndTitle($title);
3124                 return $rv[0];
3125         }
3126
3127         /**
3128          * Static function to get a template
3129          * Can be overridden via ParserOptions::setTemplateCallback().
3130          */
3131         static function statelessFetchTemplate( $title, $parser=false ) {
3132                 $text = $skip = false;
3133                 $finalTitle = $title;
3134                 $deps = array();
3135
3136                 // Loop to fetch the article, with up to 1 redirect
3137                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3138                         # Give extensions a chance to select the revision instead
3139                         $id = false; // Assume current
3140                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3141
3142                         if( $skip ) {
3143                                 $text = false;
3144                                 $deps[] = array(
3145                                         'title' => $title,
3146                                         'page_id' => $title->getArticleID(),
3147                                         'rev_id' => null );
3148                                 break;
3149                         }
3150                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3151                         $rev_id = $rev ? $rev->getId() : 0;
3152                         // If there is no current revision, there is no page
3153                         if( $id === false && !$rev ) {
3154                                 $linkCache = LinkCache::singleton();
3155                                 $linkCache->addBadLinkObj( $title );
3156                         }
3157
3158                         $deps[] = array(
3159                                 'title' => $title,
3160                                 'page_id' => $title->getArticleID(),
3161                                 'rev_id' => $rev_id );
3162
3163                         if( $rev ) {
3164                                 $text = $rev->getText();
3165                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3166                                 global $wgContLang;
3167                                 $message = $wgContLang->lcfirst( $title->getText() );
3168                                 $text = wfMsgForContentNoTrans( $message );
3169                                 if( wfEmptyMsg( $message, $text ) ) {
3170                                         $text = false;
3171                                         break;
3172                                 }
3173                         } else {
3174                                 break;
3175                         }
3176                         if ( $text === false ) {
3177                                 break;
3178                         }
3179                         // Redirect?
3180                         $finalTitle = $title;
3181                         $title = Title::newFromRedirect( $text );
3182                 }
3183                 return array(
3184                         'text' => $text,
3185                         'finalTitle' => $finalTitle,
3186                         'deps' => $deps );
3187         }
3188
3189         /**
3190          * Transclude an interwiki link.
3191          */
3192         function interwikiTransclude( $title, $action ) {
3193                 global $wgEnableScaryTranscluding;
3194
3195                 if (!$wgEnableScaryTranscluding)
3196                         return wfMsg('scarytranscludedisabled');
3197
3198                 $url = $title->getFullUrl( "action=$action" );
3199
3200                 if (strlen($url) > 255)
3201                         return wfMsg('scarytranscludetoolong');
3202                 return $this->fetchScaryTemplateMaybeFromCache($url);
3203         }
3204
3205         function fetchScaryTemplateMaybeFromCache($url) {
3206                 global $wgTranscludeCacheExpiry;
3207                 $dbr = wfGetDB(DB_SLAVE);
3208                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3209                                 array('tc_url' => $url));
3210                 if ($obj) {
3211                         $time = $obj->tc_time;
3212                         $text = $obj->tc_contents;
3213                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3214                                 return $text;
3215                         }
3216                 }
3217
3218                 $text = Http::get($url);
3219                 if (!$text)
3220                         return wfMsg('scarytranscludefailed', $url);
3221
3222                 $dbw = wfGetDB(DB_MASTER);
3223                 $dbw->replace('transcache', array('tc_url'), array(
3224                         'tc_url' => $url,
3225                         'tc_time' => time(),
3226                         'tc_contents' => $text));
3227                 return $text;
3228         }
3229
3230
3231         /**
3232          * Triple brace replacement -- used for template arguments
3233          * @private
3234          */
3235         function argSubstitution( $piece, $frame ) {
3236                 wfProfileIn( __METHOD__ );
3237
3238                 $error = false;
3239                 $parts = $piece['parts'];
3240                 $nameWithSpaces = $frame->expand( $piece['title'] );
3241                 $argName = trim( $nameWithSpaces );
3242                 $object = false;
3243                 $text = $frame->getArgument( $argName );
3244                 if (  $text === false && $parts->getLength() > 0
3245                   && (
3246                     $this->ot['html']
3247                     || $this->ot['pre']
3248                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3249                   )
3250                 ) {
3251                         # No match in frame, use the supplied default
3252                         $object = $parts->item( 0 )->getChildren();
3253                 }
3254                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3255                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3256                         $this->limitationWarn( 'post-expand-template-argument' );
3257                 }
3258
3259                 if ( $text === false && $object === false ) {
3260                         # No match anywhere
3261                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3262                 }
3263                 if ( $error !== false ) {
3264                         $text .= $error;
3265                 }
3266                 if ( $object !== false ) {
3267                         $ret = array( 'object' => $object );
3268                 } else {
3269                         $ret = array( 'text' => $text );
3270                 }
3271
3272                 wfProfileOut( __METHOD__ );
3273                 return $ret;
3274         }
3275
3276         /**
3277          * Return the text to be used for a given extension tag.
3278          * This is the ghost of strip().
3279          *
3280          * @param array $params Associative array of parameters:
3281          *     name       PPNode for the tag name
3282          *     attr       PPNode for unparsed text where tag attributes are thought to be
3283          *     attributes Optional associative array of parsed attributes
3284          *     inner      Contents of extension element
3285          *     noClose    Original text did not have a close tag
3286          * @param PPFrame $frame
3287          */
3288         function extensionSubstitution( $params, $frame ) {
3289                 global $wgRawHtml, $wgContLang;
3290
3291                 $name = $frame->expand( $params['name'] );
3292                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3293                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3294
3295                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3296
3297                 if ( $this->ot['html'] ) {
3298                         $name = strtolower( $name );
3299
3300                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3301                         if ( isset( $params['attributes'] ) ) {
3302                                 $attributes = $attributes + $params['attributes'];
3303                         }
3304                         switch ( $name ) {
3305                                 case 'html':
3306                                         if( $wgRawHtml ) {
3307                                                 $output = $content;
3308                                                 break;
3309                                         } else {
3310                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3311                                         }
3312                                 case 'nowiki':
3313                                         $output = Xml::escapeTagsOnly( $content );
3314                                         break;
3315                                 case 'math':
3316                                         $output = $wgContLang->armourMath(
3317                                                 MathRenderer::renderMath( $content, $attributes ) );
3318                                         break;
3319                                 case 'gallery':
3320                                         $output = $this->renderImageGallery( $content, $attributes );
3321                                         break;
3322                                 default:
3323                                         if( isset( $this->mTagHooks[$name] ) ) {
3324                                                 # Workaround for PHP bug 35229 and similar
3325                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3326                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3327                                                 }
3328                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3329                                                         array( $content, $attributes, $this ) );
3330                                         } else {
3331                                                 $output = '<span class="error">Invalid tag extension name: ' .
3332                                                         htmlspecialchars( $name ) . '</span>';
3333                                         }
3334                         }
3335                 } else {
3336                         if ( is_null( $attrText ) ) {
3337                                 $attrText = '';
3338                         }
3339                         if ( isset( $params['attributes'] ) ) {
3340                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3341                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3342                                                 htmlspecialchars( $attrValue ) . '"';
3343                                 }
3344                         }
3345                         if ( $content === null ) {
3346                                 $output = "<$name$attrText/>";
3347                         } else {
3348                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3349                                 $output = "<$name$attrText>$content$close";
3350                         }
3351                 }
3352
3353                 if ( $name === 'html' || $name === 'nowiki' ) {
3354                         $this->mStripState->nowiki->setPair( $marker, $output );
3355                 } else {
3356                         $this->mStripState->general->setPair( $marker, $output );
3357                 }
3358                 return $marker;
3359         }
3360
3361         /**
3362          * Increment an include size counter
3363          *
3364          * @param string $type The type of expansion
3365          * @param integer $size The size of the text
3366          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3367          */
3368         function incrementIncludeSize( $type, $size ) {
3369                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3370                         return false;
3371                 } else {
3372                         $this->mIncludeSizes[$type] += $size;
3373                         return true;
3374                 }
3375         }
3376
3377         /**
3378          * Increment the expensive function count
3379          *
3380          * @return boolean False if the limit has been exceeded
3381          */
3382         function incrementExpensiveFunctionCount() {
3383                 global $wgExpensiveParserFunctionLimit;
3384                 $this->mExpensiveFunctionCount++;
3385                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3386                         return true;
3387                 }
3388                 return false;
3389         }
3390
3391         /**
3392          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3393          * Fills $this->mDoubleUnderscores, returns the modified text
3394          */
3395         function doDoubleUnderscore( $text ) {
3396                 // The position of __TOC__ needs to be recorded
3397                 $mw = MagicWord::get( 'toc' );
3398                 if( $mw->match( $text ) ) {
3399                         $this->mShowToc = true;
3400                         $this->mForceTocPosition = true;
3401
3402                         // Set a placeholder. At the end we'll fill it in with the TOC.
3403                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3404
3405                         // Only keep the first one.
3406                         $text = $mw->replace( '', $text );
3407                 }
3408
3409                 // Now match and remove the rest of them
3410                 $mwa = MagicWord::getDoubleUnderscoreArray();
3411                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3412
3413                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3414                         $this->mOutput->mNoGallery = true;
3415                 }
3416                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3417                         $this->mShowToc = false;
3418                 }
3419                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3420                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3421
3422                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3423                         if ( $containerCategory ) {
3424                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3425                         } else {
3426                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3427                         }
3428                 }
3429                 # (bug 8068) Allow control over whether robots index a page.
3430                 #
3431                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3432                 # is not desirable, the last one on the page should win.
3433                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3434                         $this->mOutput->setIndexPolicy( 'noindex' );
3435                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3436                         $this->mOutput->setIndexPolicy( 'index' );
3437                 }
3438
3439                 return $text;
3440         }
3441
3442         /**
3443          * This function accomplishes several tasks:
3444          * 1) Auto-number headings if that option is enabled
3445          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3446          * 3) Add a Table of contents on the top for users who have enabled the option
3447          * 4) Auto-anchor headings
3448          *
3449          * It loops through all headlines, collects the necessary data, then splits up the
3450          * string and re-inserts the newly formatted headlines.
3451          *
3452          * @param string $text
3453          * @param boolean $isMain
3454          * @private
3455          */
3456         function formatHeadings( $text, $isMain=true ) {
3457                 global $wgMaxTocLevel, $wgContLang, $wgEnforceHtmlIds;
3458
3459                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3460                 $showEditLink = $this->mOptions->getEditSection();
3461
3462                 // Do not call quickUserCan unless necessary
3463                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3464                         $showEditLink = 0;
3465                 }
3466
3467                 # Inhibit editsection links if requested in the page
3468                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3469                         $showEditLink = 0;
3470                 }
3471
3472                 # Get all headlines for numbering them and adding funky stuff like [edit]
3473                 # links - this is for later, but we need the number of headlines right now
3474                 $matches = array();
3475                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3476
3477                 # if there are fewer than 4 headlines in the article, do not show TOC
3478                 # unless it's been explicitly enabled.
3479                 $enoughToc = $this->mShowToc &&
3480                         (($numMatches >= 4) || $this->mForceTocPosition);
3481
3482                 # Allow user to stipulate that a page should have a "new section"
3483                 # link added via __NEWSECTIONLINK__
3484                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3485                         $this->mOutput->setNewSection( true );
3486                 }
3487
3488                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3489                 # override above conditions and always show TOC above first header
3490                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3491                         $this->mShowToc = true;
3492                         $enoughToc = true;
3493                 }
3494
3495                 # We need this to perform operations on the HTML
3496                 $sk = $this->mOptions->getSkin();
3497
3498                 # headline counter
3499                 $headlineCount = 0;
3500                 $numVisible = 0;
3501
3502                 # Ugh .. the TOC should have neat indentation levels which can be
3503                 # passed to the skin functions. These are determined here
3504                 $toc = '';
3505                 $full = '';
3506                 $head = array();
3507                 $sublevelCount = array();
3508                 $levelCount = array();
3509                 $toclevel = 0;
3510                 $level = 0;
3511                 $prevlevel = 0;
3512                 $toclevel = 0;
3513                 $prevtoclevel = 0;
3514                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3515                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3516                 $tocraw = array();
3517
3518                 foreach( $matches[3] as $headline ) {
3519                         $isTemplate = false;
3520                         $titleText = false;
3521                         $sectionIndex = false;
3522                         $numbering = '';
3523                         $markerMatches = array();
3524                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3525                                 $serial = $markerMatches[1];
3526                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3527                                 $isTemplate = ($titleText != $baseTitleText);
3528                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3529                         }
3530
3531                         if( $toclevel ) {
3532                                 $prevlevel = $level;
3533                                 $prevtoclevel = $toclevel;
3534                         }
3535                         $level = $matches[1][$headlineCount];
3536
3537                         if( $doNumberHeadings || $enoughToc ) {
3538
3539                                 if ( $level > $prevlevel ) {
3540                                         # Increase TOC level
3541                                         $toclevel++;
3542                                         $sublevelCount[$toclevel] = 0;
3543                                         if( $toclevel<$wgMaxTocLevel ) {
3544                                                 $prevtoclevel = $toclevel;
3545                                                 $toc .= $sk->tocIndent();
3546                                                 $numVisible++;
3547                                         }
3548                                 }
3549                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3550                                         # Decrease TOC level, find level to jump to
3551
3552                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3553                                                 # Can only go down to level 1
3554                                                 $toclevel = 1;
3555                                         } else {
3556                                                 for ($i = $toclevel; $i > 0; $i--) {
3557                                                         if ( $levelCount[$i] == $level ) {
3558                                                                 # Found last matching level
3559                                                                 $toclevel = $i;
3560                                                                 break;
3561                                                         }
3562                                                         elseif ( $levelCount[$i] < $level ) {
3563                                                                 # Found first matching level below current level
3564                                                                 $toclevel = $i + 1;
3565                                                                 break;
3566                                                         }
3567                                                 }
3568                                         }
3569                                         if( $toclevel<$wgMaxTocLevel ) {
3570                                                 if($prevtoclevel < $wgMaxTocLevel) {
3571                                                         # Unindent only if the previous toc level was shown :p
3572                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3573                                                         $prevtoclevel = $toclevel;
3574                                                 } else {
3575                                                         $toc .= $sk->tocLineEnd();
3576                                                 }
3577                                         }
3578                                 }
3579                                 else {
3580                                         # No change in level, end TOC line
3581                                         if( $toclevel<$wgMaxTocLevel ) {
3582                                                 $toc .= $sk->tocLineEnd();
3583                                         }
3584                                 }
3585
3586                                 $levelCount[$toclevel] = $level;
3587
3588                                 # count number of headlines for each level
3589                                 @$sublevelCount[$toclevel]++;
3590                                 $dot = 0;
3591                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3592                                         if( !empty( $sublevelCount[$i] ) ) {
3593                                                 if( $dot ) {
3594                                                         $numbering .= '.';
3595                                                 }
3596                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3597                                                 $dot = 1;
3598                                         }
3599                                 }
3600                         }
3601
3602                         # The safe header is a version of the header text safe to use for links
3603                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3604                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3605
3606                         # Remove link placeholders by the link text.
3607                         #     <!--LINK number-->
3608                         # turns into
3609                         #     link text with suffix
3610                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3611
3612                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3613                         $tocline = preg_replace(
3614                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3615                                 array( '',                          '<$1>'),
3616                                 $safeHeadline
3617                         );
3618                         $tocline = trim( $tocline );
3619
3620                         # For the anchor, strip out HTML-y stuff period
3621                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3622                         $safeHeadline = trim( $safeHeadline );
3623
3624                         # Save headline for section edit hint before it's escaped
3625                         $headlineHint = $safeHeadline;
3626
3627                         if ( $wgEnforceHtmlIds ) {
3628                                 $legacyHeadline = false;
3629                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3630                                         'noninitial' );
3631                         } else {
3632                                 # For reverse compatibility, provide an id that's
3633                                 # HTML4-compatible, like we used to.
3634                                 #
3635                                 # It may be worth noting, academically, that it's possible for
3636                                 # the legacy anchor to conflict with a non-legacy headline
3637                                 # anchor on the page.  In this case likely the "correct" thing
3638                                 # would be to either drop the legacy anchors or make sure
3639                                 # they're numbered first.  However, this would require people
3640                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3641                                 # manually, so let's not bother worrying about it.
3642                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3643                                         'noninitial' );
3644                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline, 'xml' );
3645
3646                                 if ( $legacyHeadline == $safeHeadline ) {
3647                                         # No reason to have both (in fact, we can't)
3648                                         $legacyHeadline = false;
3649                                 } elseif ( $legacyHeadline != Sanitizer::escapeId(
3650                                 $legacyHeadline, 'xml' ) ) {
3651                                         # The legacy id is invalid XML.  We used to allow this, but
3652                                         # there's no reason to do so anymore.  Backward
3653                                         # compatibility will fail slightly in this case, but it's
3654                                         # no big deal.
3655                                         $legacyHeadline = false;
3656                                 }
3657                         }
3658
3659                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3660                         # Does this apply to Unicode characters?  Because we aren't
3661                         # handling those here.
3662                         $arrayKey = strtolower( $safeHeadline );
3663                         if ( $legacyHeadline === false ) {
3664                                 $legacyArrayKey = false;
3665                         } else {
3666                                 $legacyArrayKey = strtolower( $legacyHeadline );
3667                         }
3668
3669                         # count how many in assoc. array so we can track dupes in anchors
3670                         if ( isset( $refers[$arrayKey] ) ) {
3671                                 $refers[$arrayKey]++;
3672                         } else {
3673                                 $refers[$arrayKey] = 1;
3674                         }
3675                         if ( isset( $refers[$legacyArrayKey] ) ) {
3676                                 $refers[$legacyArrayKey]++;
3677                         } else {
3678                                 $refers[$legacyArrayKey] = 1;
3679                         }
3680
3681                         # Don't number the heading if it is the only one (looks silly)
3682                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3683                                 # the two are different if the line contains a link
3684                                 $headline=$numbering . ' ' . $headline;
3685                         }
3686
3687                         # Create the anchor for linking from the TOC to the section
3688                         $anchor = $safeHeadline;
3689                         $legacyAnchor = $legacyHeadline;
3690                         if ( $refers[$arrayKey] > 1 ) {
3691                                 $anchor .= '_' . $refers[$arrayKey];
3692                         }
3693                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3694                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3695                         }
3696                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3697                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3698                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3699                         }
3700                         # give headline the correct <h#> tag
3701                         if( $showEditLink && $sectionIndex !== false ) {
3702                                 if( $isTemplate ) {
3703                                         # Put a T flag in the section identifier, to indicate to extractSections()
3704                                         # that sections inside <includeonly> should be counted.
3705                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3706                                 } else {
3707                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3708                                 }
3709                         } else {
3710                                 $editlink = '';
3711                         }
3712                         $head[$headlineCount] = $sk->makeHeadline( $level,
3713                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3714                                 $editlink, $legacyAnchor );
3715
3716                         $headlineCount++;
3717                 }
3718
3719                 $this->mOutput->setSections( $tocraw );
3720
3721                 # Never ever show TOC if no headers
3722                 if( $numVisible < 1 ) {
3723                         $enoughToc = false;
3724                 }
3725
3726                 if( $enoughToc ) {
3727                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3728                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3729                         }
3730                         $toc = $sk->tocList( $toc );
3731                 }
3732
3733                 # split up and insert constructed headlines
3734
3735                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3736                 $i = 0;
3737
3738                 foreach( $blocks as $block ) {
3739                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3740                                 # This is the [edit] link that appears for the top block of text when
3741                                 # section editing is enabled
3742
3743                                 # Disabled because it broke block formatting
3744                                 # For example, a bullet point in the top line
3745                                 # $full .= $sk->editSectionLink(0);
3746                         }
3747                         $full .= $block;
3748                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3749                                 # Top anchor now in skin
3750                                 $full = $full.$toc;
3751                         }
3752
3753                         if( !empty( $head[$i] ) ) {
3754                                 $full .= $head[$i];
3755                         }
3756                         $i++;
3757                 }
3758                 if( $this->mForceTocPosition ) {
3759                         return str_replace( '<!--MWTOC-->', $toc, $full );
3760                 } else {
3761                         return $full;
3762                 }
3763         }
3764
3765         /**
3766          * Transform wiki markup when saving a page by doing \r\n -> \n
3767          * conversion, substitting signatures, {{subst:}} templates, etc.
3768          *
3769          * @param string $text the text to transform
3770          * @param Title &$title the Title object for the current article
3771          * @param User &$user the User object describing the current user
3772          * @param ParserOptions $options parsing options
3773          * @param bool $clearState whether to clear the parser state first
3774          * @return string the altered wiki markup
3775          * @public
3776          */
3777         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3778                 $this->mOptions = $options;
3779                 $this->setTitle( $title );
3780                 $this->setOutputType( self::OT_WIKI );
3781
3782                 if ( $clearState ) {
3783                         $this->clearState();
3784                 }
3785
3786                 $pairs = array(
3787                         "\r\n" => "\n",
3788                 );
3789                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3790                 $text = $this->pstPass2( $text, $user );
3791                 $text = $this->mStripState->unstripBoth( $text );
3792                 return $text;
3793         }
3794
3795         /**
3796          * Pre-save transform helper function
3797          * @private
3798          */
3799         function pstPass2( $text, $user ) {
3800                 global $wgContLang, $wgLocaltimezone;
3801
3802                 /* Note: This is the timestamp saved as hardcoded wikitext to
3803                  * the database, we use $wgContLang here in order to give
3804                  * everyone the same signature and use the default one rather
3805                  * than the one selected in each user's preferences.
3806                  *
3807                  * (see also bug 12815)
3808                  */
3809                 $ts = $this->mOptions->getTimestamp();
3810                 $tz = wfMsgForContent( 'timezone-utc' );
3811                 if ( isset( $wgLocaltimezone ) ) {
3812                         $unixts = wfTimestamp( TS_UNIX, $ts );
3813                         $oldtz = getenv( 'TZ' );
3814                         putenv( 'TZ='.$wgLocaltimezone );
3815                         $ts = date( 'YmdHis', $unixts );
3816                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3817                         putenv( 'TZ='.$oldtz );
3818                 }
3819
3820                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3821
3822                 # Variable replacement
3823                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3824                 $text = $this->replaceVariables( $text );
3825
3826                 # Signatures
3827                 $sigText = $this->getUserSig( $user );
3828                 $text = strtr( $text, array(
3829                         '~~~~~' => $d,
3830                         '~~~~' => "$sigText $d",
3831                         '~~~' => $sigText
3832                 ) );
3833
3834                 # Context links: [[|name]] and [[name (context)|]]
3835                 #
3836                 global $wgLegalTitleChars;
3837                 $tc = "[$wgLegalTitleChars]";
3838                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3839
3840                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3841                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3842                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3843                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3844
3845                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3846                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3847                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3848                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3849
3850                 $t = $this->mTitle->getText();
3851                 $m = array();
3852                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3853                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3854                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3855                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3856                 } else {
3857                         # if there's no context, don't bother duplicating the title
3858                         $text = preg_replace( $p2, '[[\\1]]', $text );
3859                 }
3860
3861                 # Trim trailing whitespace
3862                 $text = rtrim( $text );
3863
3864                 return $text;
3865         }
3866
3867         /**
3868          * Fetch the user's signature text, if any, and normalize to
3869          * validated, ready-to-insert wikitext.
3870          *
3871          * @param User $user
3872          * @return string
3873          * @private
3874          */
3875         function getUserSig( &$user ) {
3876                 global $wgMaxSigChars;
3877
3878                 $username = $user->getName();
3879                 $nickname = $user->getOption( 'nickname' );
3880                 $nickname = $nickname === '' ? $username : $nickname;
3881
3882                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3883                         $nickname = $username;
3884                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3885                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3886                         # Sig. might contain markup; validate this
3887                         if( $this->validateSig( $nickname ) !== false ) {
3888                                 # Validated; clean up (if needed) and return it
3889                                 return $this->cleanSig( $nickname, true );
3890                         } else {
3891                                 # Failed to validate; fall back to the default
3892                                 $nickname = $username;
3893                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3894                         }
3895                 }
3896
3897                 // Make sure nickname doesnt get a sig in a sig
3898                 $nickname = $this->cleanSigInSig( $nickname );
3899
3900                 # If we're still here, make it a link to the user page
3901                 $userText = wfEscapeWikiText( $username );
3902                 $nickText = wfEscapeWikiText( $nickname );
3903                 if ( $user->isAnon() )  {
3904                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3905                 } else {
3906                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3907                 }
3908         }
3909
3910         /**
3911          * Check that the user's signature contains no bad XML
3912          *
3913          * @param string $text
3914          * @return mixed An expanded string, or false if invalid.
3915          */
3916         function validateSig( $text ) {
3917                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3918         }
3919
3920         /**
3921          * Clean up signature text
3922          *
3923          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3924          * 2) Substitute all transclusions
3925          *
3926          * @param string $text
3927          * @param $parsing Whether we're cleaning (preferences save) or parsing
3928          * @return string Signature text
3929          */
3930         function cleanSig( $text, $parsing = false ) {
3931                 if ( !$parsing ) {
3932                         global $wgTitle;
3933                         $this->clearState();
3934                         $this->setTitle( $wgTitle );
3935                         $this->mOptions = new ParserOptions;
3936                         $this->setOutputType = self::OT_PREPROCESS;
3937                 }
3938
3939                 # Option to disable this feature
3940                 if ( !$this->mOptions->getCleanSignatures() ) {
3941                         return $text;
3942                 }
3943
3944                 # FIXME: regex doesn't respect extension tags or nowiki
3945                 #  => Move this logic to braceSubstitution()
3946                 $substWord = MagicWord::get( 'subst' );
3947                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3948                 $substText = '{{' . $substWord->getSynonym( 0 );
3949
3950                 $text = preg_replace( $substRegex, $substText, $text );
3951                 $text = $this->cleanSigInSig( $text );
3952                 $dom = $this->preprocessToDom( $text );
3953                 $frame = $this->getPreprocessor()->newFrame();
3954                 $text = $frame->expand( $dom );
3955
3956                 if ( !$parsing ) {
3957                         $text = $this->mStripState->unstripBoth( $text );
3958                 }
3959
3960                 return $text;
3961         }
3962
3963         /**
3964          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3965          * @param string $text
3966          * @return string Signature text with /~{3,5}/ removed
3967          */
3968         function cleanSigInSig( $text ) {
3969                 $text = preg_replace( '/~{3,5}/', '', $text );
3970                 return $text;
3971         }
3972
3973         /**
3974          * Set up some variables which are usually set up in parse()
3975          * so that an external function can call some class members with confidence
3976          * @public
3977          */
3978         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3979                 $this->setTitle( $title );
3980                 $this->mOptions = $options;
3981                 $this->setOutputType( $outputType );
3982                 if ( $clearState ) {
3983                         $this->clearState();
3984                 }
3985         }
3986
3987         /**
3988          * Wrapper for preprocess()
3989          *
3990          * @param string $text the text to preprocess
3991          * @param ParserOptions $options  options
3992          * @return string
3993          * @public
3994          */
3995         function transformMsg( $text, $options ) {
3996                 global $wgTitle;
3997                 static $executing = false;
3998
3999                 # Guard against infinite recursion
4000                 if ( $executing ) {
4001                         return $text;
4002                 }
4003                 $executing = true;
4004
4005                 wfProfileIn(__METHOD__);
4006                 $text = $this->preprocess( $text, $wgTitle, $options );
4007
4008                 $executing = false;
4009                 wfProfileOut(__METHOD__);
4010                 return $text;
4011         }
4012
4013         /**
4014          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4015          * The callback should have the following form:
4016          *    function myParserHook( $text, $params, &$parser ) { ... }
4017          *
4018          * Transform and return $text. Use $parser for any required context, e.g. use
4019          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4020          *
4021          * @public
4022          *
4023          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4024          * @param mixed $callback The callback function (and object) to use for the tag
4025          *
4026          * @return The old value of the mTagHooks array associated with the hook
4027          */
4028         function setHook( $tag, $callback ) {
4029                 $tag = strtolower( $tag );
4030                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4031                 $this->mTagHooks[$tag] = $callback;
4032                 if( !in_array( $tag, $this->mStripList ) ) {
4033                         $this->mStripList[] = $tag;
4034                 }
4035
4036                 return $oldVal;
4037         }
4038
4039         function setTransparentTagHook( $tag, $callback ) {
4040                 $tag = strtolower( $tag );
4041                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4042                 $this->mTransparentTagHooks[$tag] = $callback;
4043
4044                 return $oldVal;
4045         }
4046
4047         /**
4048          * Remove all tag hooks
4049          */
4050         function clearTagHooks() {
4051                 $this->mTagHooks = array();
4052                 $this->mStripList = $this->mDefaultStripList;
4053         }
4054
4055         /**
4056          * Create a function, e.g. {{sum:1|2|3}}
4057          * The callback function should have the form:
4058          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4059          *
4060          * Or with SFH_OBJECT_ARGS:
4061          *    function myParserFunction( $parser, $frame, $args ) { ... }
4062          *
4063          * The callback may either return the text result of the function, or an array with the text
4064          * in element 0, and a number of flags in the other elements. The names of the flags are
4065          * specified in the keys. Valid flags are:
4066          *   found                     The text returned is valid, stop processing the template. This
4067          *                             is on by default.
4068          *   nowiki                    Wiki markup in the return value should be escaped
4069          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4070          *
4071          * @public
4072          *
4073          * @param string $id The magic word ID
4074          * @param mixed $callback The callback function (and object) to use
4075          * @param integer $flags a combination of the following flags:
4076          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4077          *
4078          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4079          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4080          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4081          *     the arguments, and to control the way they are expanded.
4082          *
4083          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4084          *     arguments, for instance:
4085          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4086          *
4087          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4088          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4089          *     working if/when this is changed.
4090          *
4091          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4092          *     expansion.
4093          *
4094          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4095          *     about the methods available in PPFrame and PPNode.
4096          *
4097          * @return The old callback function for this name, if any
4098          */
4099         function setFunctionHook( $id, $callback, $flags = 0 ) {
4100                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4101                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4102
4103                 # Add to function cache
4104                 $mw = MagicWord::get( $id );
4105                 if( !$mw )
4106                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4107
4108                 $synonyms = $mw->getSynonyms();
4109                 $sensitive = intval( $mw->isCaseSensitive() );
4110
4111                 foreach ( $synonyms as $syn ) {
4112                         # Case
4113                         if ( !$sensitive ) {
4114                                 $syn = strtolower( $syn );
4115                         }
4116                         # Add leading hash
4117                         if ( !( $flags & SFH_NO_HASH ) ) {
4118                                 $syn = '#' . $syn;
4119                         }
4120                         # Remove trailing colon
4121                         if ( substr( $syn, -1, 1 ) === ':' ) {
4122                                 $syn = substr( $syn, 0, -1 );
4123                         }
4124                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4125                 }
4126                 return $oldVal;
4127         }
4128
4129         /**
4130          * Get all registered function hook identifiers
4131          *
4132          * @return array
4133          */
4134         function getFunctionHooks() {
4135                 return array_keys( $this->mFunctionHooks );
4136         }
4137
4138         /**
4139          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4140          * Placeholders created in Skin::makeLinkObj()
4141          * Returns an array of link CSS classes, indexed by PDBK.
4142          */
4143         function replaceLinkHolders( &$text, $options = 0 ) {
4144                 return $this->mLinkHolders->replace( $text );
4145         }
4146
4147         /**
4148          * Replace <!--LINK--> link placeholders with plain text of links
4149          * (not HTML-formatted).
4150          * @param string $text
4151          * @return string
4152          */
4153         function replaceLinkHoldersText( $text ) {
4154                 return $this->mLinkHolders->replaceText( $text );
4155         }
4156
4157         /**
4158          * Tag hook handler for 'pre'.
4159          */
4160         function renderPreTag( $text, $attribs ) {
4161                 // Backwards-compatibility hack
4162                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4163
4164                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4165                 return Xml::openElement( 'pre', $attribs ) .
4166                         Xml::escapeTagsOnly( $content ) .
4167                         '</pre>';
4168         }
4169
4170         /**
4171          * Renders an image gallery from a text with one line per image.
4172          * text labels may be given by using |-style alternative text. E.g.
4173          *   Image:one.jpg|The number "1"
4174          *   Image:tree.jpg|A tree
4175          * given as text will return the HTML of a gallery with two images,
4176          * labeled 'The number "1"' and
4177          * 'A tree'.
4178          */
4179         function renderImageGallery( $text, $params ) {
4180                 $ig = new ImageGallery();
4181                 $ig->setContextTitle( $this->mTitle );
4182                 $ig->setShowBytes( false );
4183                 $ig->setShowFilename( false );
4184                 $ig->setParser( $this );
4185                 $ig->setHideBadImages();
4186                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4187                 $ig->useSkin( $this->mOptions->getSkin() );
4188                 $ig->mRevisionId = $this->mRevisionId;
4189
4190                 if( isset( $params['caption'] ) ) {
4191                         $caption = $params['caption'];
4192                         $caption = htmlspecialchars( $caption );
4193                         $caption = $this->replaceInternalLinks( $caption );
4194                         $ig->setCaptionHtml( $caption );
4195                 }
4196                 if( isset( $params['perrow'] ) ) {
4197                         $ig->setPerRow( $params['perrow'] );
4198                 }
4199                 if( isset( $params['widths'] ) ) {
4200                         $ig->setWidths( $params['widths'] );
4201                 }
4202                 if( isset( $params['heights'] ) ) {
4203                         $ig->setHeights( $params['heights'] );
4204                 }
4205
4206                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4207
4208                 $lines = StringUtils::explode( "\n", $text );
4209                 foreach ( $lines as $line ) {
4210                         # match lines like these:
4211                         # Image:someimage.jpg|This is some image
4212                         $matches = array();
4213                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4214                         # Skip empty lines
4215                         if ( count( $matches ) == 0 ) {
4216                                 continue;
4217                         }
4218
4219                         if ( strpos( $matches[0], '%' ) !== false )
4220                                 $matches[1] = urldecode( $matches[1] );
4221                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4222                         $nt =& $tp;
4223                         if( is_null( $nt ) ) {
4224                                 # Bogus title. Ignore these so we don't bomb out later.
4225                                 continue;
4226                         }
4227                         if ( isset( $matches[3] ) ) {
4228                                 $label = $matches[3];
4229                         } else {
4230                                 $label = '';
4231                         }
4232
4233                         $html = $this->recursiveTagParse( trim( $label ) );
4234
4235                         $ig->add( $nt, $html );
4236
4237                         # Only add real images (bug #5586)
4238                         if ( $nt->getNamespace() == NS_FILE ) {
4239                                 $this->mOutput->addImage( $nt->getDBkey() );
4240                         }
4241                 }
4242                 return $ig->toHTML();
4243         }
4244
4245         function getImageParams( $handler ) {
4246                 if ( $handler ) {
4247                         $handlerClass = get_class( $handler );
4248                 } else {
4249                         $handlerClass = '';
4250                 }
4251                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4252                         // Initialise static lists
4253                         static $internalParamNames = array(
4254                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4255                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4256                                         'bottom', 'text-bottom' ),
4257                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4258                                         'upright', 'border', 'link', 'alt' ),
4259                         );
4260                         static $internalParamMap;
4261                         if ( !$internalParamMap ) {
4262                                 $internalParamMap = array();
4263                                 foreach ( $internalParamNames as $type => $names ) {
4264                                         foreach ( $names as $name ) {
4265                                                 $magicName = str_replace( '-', '_', "img_$name" );
4266                                                 $internalParamMap[$magicName] = array( $type, $name );
4267                                         }
4268                                 }
4269                         }
4270
4271                         // Add handler params
4272                         $paramMap = $internalParamMap;
4273                         if ( $handler ) {
4274                                 $handlerParamMap = $handler->getParamMap();
4275                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4276                                         $paramMap[$magic] = array( 'handler', $paramName );
4277                                 }
4278                         }
4279                         $this->mImageParams[$handlerClass] = $paramMap;
4280                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4281                 }
4282                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4283         }
4284
4285         /**
4286          * Parse image options text and use it to make an image
4287          * @param Title $title
4288          * @param string $options
4289          * @param LinkHolderArray $holders
4290          */
4291         function makeImage( $title, $options, $holders = false ) {
4292                 # Check if the options text is of the form "options|alt text"
4293                 # Options are:
4294                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4295                 #  * left       no resizing, just left align. label is used for alt= only
4296                 #  * right      same, but right aligned
4297                 #  * none       same, but not aligned
4298                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4299                 #  * center     center the image
4300                 #  * framed     Keep original image size, no magnify-button.
4301                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4302                 #  * upright    reduce width for upright images, rounded to full __0 px
4303                 #  * border     draw a 1px border around the image
4304                 #  * alt        Text for HTML alt attribute (defaults to empty)
4305                 # vertical-align values (no % or length right now):
4306                 #  * baseline
4307                 #  * sub
4308                 #  * super
4309                 #  * top
4310                 #  * text-top
4311                 #  * middle
4312                 #  * bottom
4313                 #  * text-bottom
4314
4315                 $parts = StringUtils::explode( "|", $options );
4316                 $sk = $this->mOptions->getSkin();
4317
4318                 # Give extensions a chance to select the file revision for us
4319                 $skip = $time = $descQuery = false;
4320                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4321
4322                 if ( $skip ) {
4323                         return $sk->link( $title );
4324                 }
4325
4326                 # Get the file
4327                 $imagename = $title->getDBkey();
4328                 if ( isset( $this->mFileCache[$imagename][$time] ) ) {
4329                         $file = $this->mFileCache[$imagename][$time];
4330                 } else {
4331                         $file = wfFindFile( $title, $time );
4332                         if ( count( $this->mFileCache ) > 1000 ) {
4333                                 $this->mFileCache = array();
4334                         }
4335                         $this->mFileCache[$imagename][$time] = $file;
4336                 }
4337                 # Get parameter map
4338                 $handler = $file ? $file->getHandler() : false;
4339
4340                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4341
4342                 # Process the input parameters
4343                 $caption = '';
4344                 $params = array( 'frame' => array(), 'handler' => array(),
4345                         'horizAlign' => array(), 'vertAlign' => array() );
4346                 foreach( $parts as $part ) {
4347                         $part = trim( $part );
4348                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4349                         $validated = false;
4350                         if( isset( $paramMap[$magicName] ) ) {
4351                                 list( $type, $paramName ) = $paramMap[$magicName];
4352
4353                                 // Special case; width and height come in one variable together
4354                                 if( $type === 'handler' && $paramName === 'width' ) {
4355                                         $m = array();
4356                                         # (bug 13500) In both cases (width/height and width only),
4357                                         # permit trailing "px" for backward compatibility.
4358                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4359                                                 $width = intval( $m[1] );
4360                                                 $height = intval( $m[2] );
4361                                                 if ( $handler->validateParam( 'width', $width ) ) {
4362                                                         $params[$type]['width'] = $width;
4363                                                         $validated = true;
4364                                                 }
4365                                                 if ( $handler->validateParam( 'height', $height ) ) {
4366                                                         $params[$type]['height'] = $height;
4367                                                         $validated = true;
4368                                                 }
4369                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4370                                                 $width = intval( $value );
4371                                                 if ( $handler->validateParam( 'width', $width ) ) {
4372                                                         $params[$type]['width'] = $width;
4373                                                         $validated = true;
4374                                                 }
4375                                         } // else no validation -- bug 13436
4376                                 } else {
4377                                         if ( $type === 'handler' ) {
4378                                                 # Validate handler parameter
4379                                                 $validated = $handler->validateParam( $paramName, $value );
4380                                         } else {
4381                                                 # Validate internal parameters
4382                                                 switch( $paramName ) {
4383                                                 case 'manualthumb':
4384                                                 case 'alt':
4385                                                         // @fixme - possibly check validity here for
4386                                                         // manualthumb? downstream behavior seems odd with
4387                                                         // missing manual thumbs.
4388                                                         $validated = true;
4389                                                         $value = $this->stripAltText( $value, $holders );
4390                                                         break;
4391                                                 case 'link':
4392                                                         $chars = self::EXT_LINK_URL_CLASS;
4393                                                         $prots = $this->mUrlProtocols;
4394                                                         if ( $value === '' ) {
4395                                                                 $paramName = 'no-link';
4396                                                                 $value = true;
4397                                                                 $validated = true;
4398                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4399                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4400                                                                         $paramName = 'link-url';
4401                                                                         $this->mOutput->addExternalLink( $value );
4402                                                                         $validated = true;
4403                                                                 }
4404                                                         } else {
4405                                                                 $linkTitle = Title::newFromText( $value );
4406                                                                 if ( $linkTitle ) {
4407                                                                         $paramName = 'link-title';
4408                                                                         $value = $linkTitle;
4409                                                                         $this->mOutput->addLink( $linkTitle );
4410                                                                         $validated = true;
4411                                                                 }
4412                                                         }
4413                                                         break;
4414                                                 default:
4415                                                         // Most other things appear to be empty or numeric...
4416                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4417                                                 }
4418                                         }
4419
4420                                         if ( $validated ) {
4421                                                 $params[$type][$paramName] = $value;
4422                                         }
4423                                 }
4424                         }
4425                         if ( !$validated ) {
4426                                 $caption = $part;
4427                         }
4428                 }
4429
4430                 # Process alignment parameters
4431                 if ( $params['horizAlign'] ) {
4432                         $params['frame']['align'] = key( $params['horizAlign'] );
4433                 }
4434                 if ( $params['vertAlign'] ) {
4435                         $params['frame']['valign'] = key( $params['vertAlign'] );
4436                 }
4437
4438                 $params['frame']['caption'] = $caption;
4439
4440                 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4441
4442                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4443                 # came to also set the caption, ordinary text after the image -- which
4444                 # makes no sense, because that just repeats the text multiple times in
4445                 # screen readers.  It *also* came to set the title attribute.
4446                 #
4447                 # Now that we have an alt attribute, we should not set the alt text to
4448                 # equal the caption: that's worse than useless, it just repeats the
4449                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4450                 # use the unnamed parameter for alt text as well, just for the time be-
4451                 # ing, if the unnamed param is set and the alt param is not.
4452                 #
4453                 # For the future, we need to figure out if we want to tweak this more,
4454                 # e.g., introducing a title= parameter for the title; ignoring the un-
4455                 # named parameter entirely for images without a caption; adding an ex-
4456                 # plicit caption= parameter and preserving the old magic unnamed para-
4457                 # meter for BC; ...
4458                 if( $caption !== '' && !isset( $params['frame']['alt'] )
4459                 && !isset( $params['frame']['framed'] )
4460                 && !isset( $params['frame']['thumbnail'] )
4461                 && !isset( $params['frame']['manualthumb'] ) ) {
4462                         $params['frame']['alt'] = $params['frame']['title'];
4463                 }
4464
4465                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4466
4467                 # Linker does the rest
4468                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4469
4470                 # Give the handler a chance to modify the parser object
4471                 if ( $handler ) {
4472                         $handler->parserTransformHook( $this, $file );
4473                 }
4474
4475                 return $ret;
4476         }
4477
4478         protected function stripAltText( $caption, $holders ) {
4479                 # Strip bad stuff out of the title (tooltip).  We can't just use
4480                 # replaceLinkHoldersText() here, because if this function is called
4481                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4482                 if ( $holders ) {
4483                         $tooltip = $holders->replaceText( $caption );
4484                 } else {
4485                         $tooltip = $this->replaceLinkHoldersText( $caption );
4486                 }
4487
4488                 # make sure there are no placeholders in thumbnail attributes
4489                 # that are later expanded to html- so expand them now and
4490                 # remove the tags
4491                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4492                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4493
4494                 return $tooltip;
4495         }
4496
4497         /**
4498          * Set a flag in the output object indicating that the content is dynamic and
4499          * shouldn't be cached.
4500          */
4501         function disableCache() {
4502                 wfDebug( "Parser output marked as uncacheable.\n" );
4503                 $this->mOutput->mCacheTime = -1;
4504         }
4505
4506         /**#@+
4507          * Callback from the Sanitizer for expanding items found in HTML attribute
4508          * values, so they can be safely tested and escaped.
4509          * @param string $text
4510          * @param PPFrame $frame
4511          * @return string
4512          * @private
4513          */
4514         function attributeStripCallback( &$text, $frame = false ) {
4515                 $text = $this->replaceVariables( $text, $frame );
4516                 $text = $this->mStripState->unstripBoth( $text );
4517                 return $text;
4518         }
4519
4520         /**#@-*/
4521
4522         /**#@+
4523          * Accessor/mutator
4524          */
4525         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4526         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4527         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4528         /**#@-*/
4529
4530         /**#@+
4531          * Accessor
4532          */
4533         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4534         /**#@-*/
4535
4536
4537         /**
4538          * Break wikitext input into sections, and either pull or replace
4539          * some particular section's text.
4540          *
4541          * External callers should use the getSection and replaceSection methods.
4542          *
4543          * @param string $text Page wikitext
4544          * @param string $section A section identifier string of the form:
4545          *   <flag1> - <flag2> - ... - <section number>
4546          *
4547          * Currently the only recognised flag is "T", which means the target section number
4548          * was derived during a template inclusion parse, in other words this is a template
4549          * section edit link. If no flags are given, it was an ordinary section edit link.
4550          * This flag is required to avoid a section numbering mismatch when a section is
4551          * enclosed by <includeonly> (bug 6563).
4552          *
4553          * The section number 0 pulls the text before the first heading; other numbers will
4554          * pull the given section along with its lower-level subsections. If the section is
4555          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4556          *
4557          * @param string $mode One of "get" or "replace"
4558          * @param string $newText Replacement text for section data.
4559          * @return string for "get", the extracted section text.
4560          *                for "replace", the whole page with the section replaced.
4561          */
4562         private function extractSections( $text, $section, $mode, $newText='' ) {
4563                 global $wgTitle;
4564                 $this->clearState();
4565                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4566                 $this->mOptions = new ParserOptions;
4567                 $this->setOutputType( self::OT_WIKI );
4568                 $outText = '';
4569                 $frame = $this->getPreprocessor()->newFrame();
4570
4571                 // Process section extraction flags
4572                 $flags = 0;
4573                 $sectionParts = explode( '-', $section );
4574                 $sectionIndex = array_pop( $sectionParts );
4575                 foreach ( $sectionParts as $part ) {
4576                         if ( $part === 'T' ) {
4577                                 $flags |= self::PTD_FOR_INCLUSION;
4578                         }
4579                 }
4580                 // Preprocess the text
4581                 $root = $this->preprocessToDom( $text, $flags );
4582
4583                 // <h> nodes indicate section breaks
4584                 // They can only occur at the top level, so we can find them by iterating the root's children
4585                 $node = $root->getFirstChild();
4586
4587                 // Find the target section
4588                 if ( $sectionIndex == 0 ) {
4589                         // Section zero doesn't nest, level=big
4590                         $targetLevel = 1000;
4591                 } else {
4592             while ( $node ) {
4593                 if ( $node->getName() === 'h' ) {
4594                     $bits = $node->splitHeading();
4595                                         if ( $bits['i'] == $sectionIndex ) {
4596                                         $targetLevel = $bits['level'];
4597                                                 break;
4598                                         }
4599                                 }
4600                                 if ( $mode === 'replace' ) {
4601                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4602                                 }
4603                                 $node = $node->getNextSibling();
4604                         }
4605                 }
4606
4607                 if ( !$node ) {
4608                         // Not found
4609                         if ( $mode === 'get' ) {
4610                                 return $newText;
4611                         } else {
4612                                 return $text;
4613                         }
4614                 }
4615
4616                 // Find the end of the section, including nested sections
4617                 do {
4618                         if ( $node->getName() === 'h' ) {
4619                                 $bits = $node->splitHeading();
4620                                 $curLevel = $bits['level'];
4621                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4622                                         break;
4623                                 }
4624                         }
4625                         if ( $mode === 'get' ) {
4626                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4627                         }
4628                         $node = $node->getNextSibling();
4629                 } while ( $node );
4630
4631                 // Write out the remainder (in replace mode only)
4632                 if ( $mode === 'replace' ) {
4633                         // Output the replacement text
4634                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4635                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4636                         $outText .= $newText . "\n\n";
4637                         while ( $node ) {
4638                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4639                                 $node = $node->getNextSibling();
4640                         }
4641                 }
4642
4643                 if ( is_string( $outText ) ) {
4644                         // Re-insert stripped tags
4645                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4646                 }
4647
4648                 return $outText;
4649         }
4650
4651         /**
4652          * This function returns the text of a section, specified by a number ($section).
4653          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4654          * the first section before any such heading (section 0).
4655          *
4656          * If a section contains subsections, these are also returned.
4657          *
4658          * @param string $text text to look in
4659          * @param string $section section identifier
4660          * @param string $deftext default to return if section is not found
4661          * @return string text of the requested section
4662          */
4663         public function getSection( $text, $section, $deftext='' ) {
4664                 return $this->extractSections( $text, $section, "get", $deftext );
4665         }
4666
4667         public function replaceSection( $oldtext, $section, $text ) {
4668                 return $this->extractSections( $oldtext, $section, "replace", $text );
4669         }
4670
4671         /**
4672          * Get the timestamp associated with the current revision, adjusted for
4673          * the default server-local timestamp
4674          */
4675         function getRevisionTimestamp() {
4676                 if ( is_null( $this->mRevisionTimestamp ) ) {
4677                         wfProfileIn( __METHOD__ );
4678                         global $wgContLang;
4679                         $dbr = wfGetDB( DB_SLAVE );
4680                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4681                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4682
4683                         // Normalize timestamp to internal MW format for timezone processing.
4684                         // This has the added side-effect of replacing a null value with
4685                         // the current time, which gives us more sensible behavior for
4686                         // previews.
4687                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4688
4689                         // The cryptic '' timezone parameter tells to use the site-default
4690                         // timezone offset instead of the user settings.
4691                         //
4692                         // Since this value will be saved into the parser cache, served
4693                         // to other users, and potentially even used inside links and such,
4694                         // it needs to be consistent for all visitors.
4695                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4696
4697                         wfProfileOut( __METHOD__ );
4698                 }
4699                 return $this->mRevisionTimestamp;
4700         }
4701
4702         /**
4703          * Mutator for $mDefaultSort
4704          *
4705          * @param $sort New value
4706          */
4707         public function setDefaultSort( $sort ) {
4708                 $this->mDefaultSort = $sort;
4709         }
4710
4711         /**
4712          * Accessor for $mDefaultSort
4713          * Will use the title/prefixed title if none is set
4714          *
4715          * @return string
4716          */
4717         public function getDefaultSort() {
4718                 global $wgCategoryPrefixedDefaultSortkey;
4719                 if( $this->mDefaultSort !== false ) {
4720                         return $this->mDefaultSort;
4721                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4722                         !$wgCategoryPrefixedDefaultSortkey) {
4723                         return $this->mTitle->getText();
4724                 } else {
4725                         return $this->mTitle->getPrefixedText();
4726                 }
4727         }
4728
4729         /**
4730          * Accessor for $mDefaultSort
4731          * Unlike getDefaultSort(), will return false if none is set
4732          *
4733          * @return string or false
4734          */
4735         public function getCustomDefaultSort() {
4736                 return $this->mDefaultSort;
4737         }
4738
4739         /**
4740          * Try to guess the section anchor name based on a wikitext fragment
4741          * presumably extracted from a heading, for example "Header" from
4742          * "== Header ==".
4743          */
4744         public function guessSectionNameFromWikiText( $text ) {
4745                 # Strip out wikitext links(they break the anchor)
4746                 $text = $this->stripSectionName( $text );
4747                 $headline = Sanitizer::decodeCharReferences( $text );
4748                 # strip out HTML
4749                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4750                 $headline = trim( $headline );
4751                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4752                 $replacearray = array(
4753                         '%3A' => ':',
4754                         '%' => '.'
4755                 );
4756                 return str_replace(
4757                         array_keys( $replacearray ),
4758                         array_values( $replacearray ),
4759                         $sectionanchor );
4760         }
4761
4762         /**
4763          * Strips a text string of wikitext for use in a section anchor
4764          *
4765          * Accepts a text string and then removes all wikitext from the
4766          * string and leaves only the resultant text (i.e. the result of
4767          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4768          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4769          * to create valid section anchors by mimicing the output of the
4770          * parser when headings are parsed.
4771          *
4772          * @param $text string Text string to be stripped of wikitext
4773          * for use in a Section anchor
4774          * @return Filtered text string
4775          */
4776         public function stripSectionName( $text ) {
4777                 # Strip internal link markup
4778                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4779                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4780
4781                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4782                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4783                 # on how many empty links there are on the page - need to figure that out.
4784                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4785
4786                 # Parse wikitext quotes (italics & bold)
4787                 $text = $this->doQuotes($text);
4788
4789                 # Strip HTML tags
4790                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4791                 return $text;
4792         }
4793
4794         function srvus( $text ) {
4795                 return $this->testSrvus( $text, $this->mOutputType );
4796         }
4797
4798         /**
4799          * strip/replaceVariables/unstrip for preprocessor regression testing
4800          */
4801         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4802                 $this->clearState();
4803                 if ( ! ( $title instanceof Title ) ) {
4804                         $title = Title::newFromText( $title );
4805                 }
4806                 $this->mTitle = $title;
4807                 $this->mOptions = $options;
4808                 $this->setOutputType( $outputType );
4809                 $text = $this->replaceVariables( $text );
4810                 $text = $this->mStripState->unstripBoth( $text );
4811                 $text = Sanitizer::removeHTMLtags( $text );
4812                 return $text;
4813         }
4814
4815         function testPst( $text, $title, $options ) {
4816                 global $wgUser;
4817                 if ( ! ( $title instanceof Title ) ) {
4818                         $title = Title::newFromText( $title );
4819                 }
4820                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4821         }
4822
4823         function testPreprocess( $text, $title, $options ) {
4824                 if ( ! ( $title instanceof Title ) ) {
4825                         $title = Title::newFromText( $title );
4826                 }
4827                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4828         }
4829
4830         function markerSkipCallback( $s, $callback ) {
4831                 $i = 0;
4832                 $out = '';
4833                 while ( $i < strlen( $s ) ) {
4834                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4835                         if ( $markerStart === false ) {
4836                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4837                                 break;
4838                         } else {
4839                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4840                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4841                                 if ( $markerEnd === false ) {
4842                                         $out .= substr( $s, $markerStart );
4843                                         break;
4844                                 } else {
4845                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4846                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4847                                         $i = $markerEnd;
4848                                 }
4849                         }
4850                 }
4851                 return $out;
4852         }
4853 }
4854
4855 /**
4856  * @todo document, briefly.
4857  * @ingroup Parser
4858  */
4859 class StripState {
4860         var $general, $nowiki;
4861
4862         function __construct() {
4863                 $this->general = new ReplacementArray;
4864                 $this->nowiki = new ReplacementArray;
4865         }
4866
4867         function unstripGeneral( $text ) {
4868                 wfProfileIn( __METHOD__ );
4869                 do {
4870                         $oldText = $text;
4871                         $text = $this->general->replace( $text );
4872                 } while ( $text !== $oldText );
4873                 wfProfileOut( __METHOD__ );
4874                 return $text;
4875         }
4876
4877         function unstripNoWiki( $text ) {
4878                 wfProfileIn( __METHOD__ );
4879                 do {
4880                         $oldText = $text;
4881                         $text = $this->nowiki->replace( $text );
4882                 } while ( $text !== $oldText );
4883                 wfProfileOut( __METHOD__ );
4884                 return $text;
4885         }
4886
4887         function unstripBoth( $text ) {
4888                 wfProfileIn( __METHOD__ );
4889                 do {
4890                         $oldText = $text;
4891                         $text = $this->general->replace( $text );
4892                         $text = $this->nowiki->replace( $text );
4893                 } while ( $text !== $oldText );
4894                 wfProfileOut( __METHOD__ );
4895                 return $text;
4896         }
4897 }
4898
4899 /**
4900  * @todo document, briefly.
4901  * @ingroup Parser
4902  */
4903 class OnlyIncludeReplacer {
4904         var $output = '';
4905
4906         function replace( $matches ) {
4907                 if ( substr( $matches[1], -1 ) === "\n" ) {
4908                         $this->output .= substr( $matches[1], 0, -1 );
4909                 } else {
4910                         $this->output .= $matches[1];
4911                 }
4912         }
4913 }