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', 'poem' );
 132                 $this->mUrlProtocols = wfUrlProtocols();
 133                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 134                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 135                 $this->mVarCache = array();
 136                 if ( isset( $conf['preprocessorClass'] ) ) {
 137                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 138                 } elseif ( extension_loaded( 'domxml' ) ) {
 139                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 140                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 141                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 142                 } elseif ( extension_loaded( 'dom' ) ) {
 143                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 144                 } else {
 145                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 146                 }
 147                 $this->mMarkerIndex = 0;
 148                 $this->mFirstCall = true;
 149         }
 150
 151         /**
 152          * Reduce memory usage to reduce the impact of circular references
 153          */
 154         function __destruct() {
 155                 if ( isset( $this->mLinkHolders ) ) {
 156                         $this->mLinkHolders->__destruct();
 157                 }
 158                 foreach ( $this as $name => $value ) {
 159                         unset( $this->$name );
 160                 }
 161         }
 162
 163         /**
 164          * Do various kinds of initialisation on the first call of the parser
 165          */
 166         function firstCallInit() {
 167                 if ( !$this->mFirstCall ) {
 168                         return;
 169                 }
 170                 $this->mFirstCall = false;
 171
 172                 wfProfileIn( __METHOD__ );
 173
 174                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 175                 CoreParserFunctions::register( $this );
 176                 $this->initialiseVariables();
 177
 178                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 179                 wfProfileOut( __METHOD__ );
 180         }
 181
 182         /**
 183          * Clear Parser state
 184          *
 185          * @private
 186          */
 187         function clearState() {
 188                 wfProfileIn( __METHOD__ );
 189                 if ( $this->mFirstCall ) {
 190                         $this->firstCallInit();
 191                 }
 192                 $this->mOutput = new ParserOutput;
 193                 $this->mAutonumber = 0;
 194                 $this->mLastSection = '';
 195                 $this->mDTopen = false;
 196                 $this->mIncludeCount = array();
 197                 $this->mStripState = new StripState;
 198                 $this->mArgStack = false;
 199                 $this->mInPre = false;
 200                 $this->mLinkHolders = new LinkHolderArray( $this );
 201                 $this->mLinkID = 0;
 202                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 203
 204                 /**
 205                  * Prefix for temporary replacement strings for the multipass parser.
 206                  * \x07 should never appear in input as it's disallowed in XML.
 207                  * Using it at the front also gives us a little extra robustness
 208                  * since it shouldn't match when butted up against identifier-like
 209                  * string constructs.
 210                  *
 211                  * Must not consist of all title characters, or else it will change
 212                  * the behaviour of <nowiki> in a link.
 213                  */
 214                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 215                 # Changed to \x7f to allow XML double-parsing -- TS
 216                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 217
 218
 219                 # Clear these on every parse, bug 4549
 220                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 221
 222                 $this->mShowToc = true;
 223                 $this->mForceTocPosition = false;
 224                 $this->mIncludeSizes = array(
 225                         'post-expand' => 0,
 226                         'arg' => 0,
 227                 );
 228                 $this->mPPNodeCount = 0;
 229                 $this->mDefaultSort = false;
 230                 $this->mHeadings = array();
 231                 $this->mDoubleUnderscores = array();
 232                 $this->mExpensiveFunctionCount = 0;
 233                 $this->mFileCache = array();
 234
 235                 # Fix cloning
 236                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 237                         $this->mPreprocessor = null;
 238                 }
 239
 240                 wfRunHooks( 'ParserClearState', array( &$this ) );
 241                 wfProfileOut( __METHOD__ );
 242         }
 243
 244         function setOutputType( $ot ) {
 245                 $this->mOutputType = $ot;
 246                 // Shortcut alias
 247                 $this->ot = array(
 248                         'html' => $ot == self::OT_HTML,
 249                         'wiki' => $ot == self::OT_WIKI,
 250                         'pre' => $ot == self::OT_PREPROCESS,
 251                 );
 252         }
 253
 254         /**
 255          * Set the context title
 256          */
 257         function setTitle( $t ) {
 258                 if ( !$t || $t instanceof FakeTitle ) {
 259                         $t = Title::newFromText( 'NO TITLE' );
 260                 }
 261                 if ( strval( $t->getFragment() ) !== '' ) {
 262                         # Strip the fragment to avoid various odd effects
 263                         $this->mTitle = clone $t;
 264                         $this->mTitle->setFragment( '' );
 265                 } else {
 266                         $this->mTitle = $t;
 267                 }
 268         }
 269
 270         /**
 271          * Accessor for mUniqPrefix.
 272          *
 273          * @public
 274          */
 275         function uniqPrefix() {
 276                 if( !isset( $this->mUniqPrefix ) ) {
 277                         // @fixme this is probably *horribly wrong*
 278                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 279                         // if this is called for a parser cache hit, the parser may not
 280                         // have ever been initialized in the first place.
 281                         // Not really sure what the heck is supposed to be going on here.
 282                         return '';
 283                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 284                 }
 285                 return $this->mUniqPrefix;
 286         }
 287
 288         /**
 289          * Convert wikitext to HTML
 290          * Do not call this function recursively.
 291          *
 292          * @param $text String: text we want to parse
 293          * @param $title A title object
 294          * @param $options ParserOptions
 295          * @param $linestart boolean
 296          * @param $clearState boolean
 297          * @param $revid Int: number to pass in {{REVISIONID}}
 298          * @return ParserOutput a ParserOutput
 299          */
 300         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 301                 /**
 302                  * First pass--just handle <nowiki> sections, pass the rest off
 303                  * to internalParse() which does all the real work.
 304                  */
 305
 306                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 307                 $fname = __METHOD__.'-' . wfGetCaller();
 308                 wfProfileIn( __METHOD__ );
 309                 wfProfileIn( $fname );
 310
 311                 if ( $clearState ) {
 312                         $this->clearState();
 313                 }
 314
 315                 $this->mOptions = $options;
 316                 $this->setTitle( $title );
 317                 $oldRevisionId = $this->mRevisionId;
 318                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 319                 if( $revid !== null ) {
 320                         $this->mRevisionId = $revid;
 321                         $this->mRevisionTimestamp = null;
 322                 }
 323                 $this->setOutputType( self::OT_HTML );
 324                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 325                 # No more strip!
 326                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 327                 $text = $this->internalParse( $text );
 328                 $text = $this->mStripState->unstripGeneral( $text );
 329
 330                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 331                 $fixtags = array(
 332                         # french spaces, last one Guillemet-left
 333                         # only if there is something before the space
 334                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 335                         # french spaces, Guillemet-right
 336                         '/(\\302\\253) /' => '\\1&nbsp;',
 337                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 338                 );
 339                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 340
 341                 $text = $this->doBlockLevels( $text, $linestart );
 342
 343                 $this->replaceLinkHolders( $text );
 344
 345                 # the position of the parserConvert() call should not be changed. it
 346                 # assumes that the links are all replaced and the only thing left
 347                 # is the <nowiki> mark.
 348                 # Side-effects: this calls $this->mOutput->setTitleText()
 349                 $text = $wgContLang->parserConvert( $text, $this );
 350
 351                 $text = $this->mStripState->unstripNoWiki( $text );
 352
 353                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 354
 355 //!JF Move to its own function
 356
 357                 $uniq_prefix = $this->mUniqPrefix;
 358                 $matches = array();
 359                 $elements = array_keys( $this->mTransparentTagHooks );
 360                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 361
 362                 foreach( $matches as $marker => $data ) {
 363                         list( $element, $content, $params, $tag ) = $data;
 364                         $tagName = strtolower( $element );
 365                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 366                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 367                                         array( $content, $params, $this ) );
 368                         } else {
 369                                 $output = $tag;
 370                         }
 371                         $this->mStripState->general->setPair( $marker, $output );
 372                 }
 373                 $text = $this->mStripState->unstripGeneral( $text );
 374
 375                 $text = Sanitizer::normalizeCharReferences( $text );
 376
 377                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 378                         $text = self::tidy($text);
 379                 } else {
 380                         # attempt to sanitize at least some nesting problems
 381                         # (bug #2702 and quite a few others)
 382                         $tidyregs = array(
 383                                 # ''Something [http://www.cool.com cool''] -->
 384                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 385                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 386                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 387                                 # fix up an anchor inside another anchor, only
 388                                 # at least for a single single nested link (bug 3695)
 389                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 390                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 391                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 392                                 # contains a div, so fix it up here; replace
 393                                 # div with escaped text
 394                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 395                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 396                                 # remove empty italic or bold tag pairs, some
 397                                 # introduced by rules above
 398                                 '/<([bi])><\/\\1>/' => '',
 399                         );
 400
 401                         $text = preg_replace(
 402                                 array_keys( $tidyregs ),
 403                                 array_values( $tidyregs ),
 404                                 $text );
 405                 }
 406                 global $wgExpensiveParserFunctionLimit;
 407                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 408                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 409                 }
 410
 411                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 412
 413                 # Information on include size limits, for the benefit of users who try to skirt them
 414                 if ( $this->mOptions->getEnableLimitReport() ) {
 415                         global $wgExpensiveParserFunctionLimit;
 416                         $max = $this->mOptions->getMaxIncludeSize();
 417                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 418                         $limitReport =
 419                                 "NewPP limit report\n" .
 420                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 421                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 422                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 423                                 $PFreport;
 424                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 425                         $text .= "\n<!-- \n$limitReport-->\n";
 426                 }
 427                 $this->mOutput->setText( $text );
 428                 $this->mRevisionId = $oldRevisionId;
 429                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 430                 wfProfileOut( $fname );
 431                 wfProfileOut( __METHOD__ );
 432
 433                 return $this->mOutput;
 434         }
 435
 436         /**
 437          * Recursive parser entry point that can be called from an extension tag
 438          * hook.
 439          */
 440         function recursiveTagParse( $text ) {
 441                 wfProfileIn( __METHOD__ );
 442                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 443                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 444                 $text = $this->internalParse( $text );
 445                 wfProfileOut( __METHOD__ );
 446                 return $text;
 447         }
 448
 449         /**
 450          * Expand templates and variables in the text, producing valid, static wikitext.
 451          * Also removes comments.
 452          */
 453         function preprocess( $text, $title, $options, $revid = null ) {
 454                 wfProfileIn( __METHOD__ );
 455                 $this->clearState();
 456                 $this->setOutputType( self::OT_PREPROCESS );
 457                 $this->mOptions = $options;
 458                 $this->setTitle( $title );
 459                 if( $revid !== null ) {
 460                         $this->mRevisionId = $revid;
 461                 }
 462                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 463                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 464                 $text = $this->replaceVariables( $text );
 465                 $text = $this->mStripState->unstripBoth( $text );
 466                 wfProfileOut( __METHOD__ );
 467                 return $text;
 468         }
 469
 470         /**
 471          * Get a random string
 472          *
 473          * @private
 474          * @static
 475          */
 476         function getRandomString() {
 477                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 478         }
 479
 480         function &getTitle() { return $this->mTitle; }
 481         function getOptions() { return $this->mOptions; }
 482         function getRevisionId() { return $this->mRevisionId; }
 483         function getOutput() { return $this->mOutput; }
 484         function nextLinkID() { return $this->mLinkID++; }
 485
 486         function getFunctionLang() {
 487                 global $wgLang, $wgContLang;
 488
 489                 $target = $this->mOptions->getTargetLanguage();
 490                 if ( $target !== null ) {
 491                         return $target;
 492                 } else {
 493                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 494                 }
 495         }
 496
 497         /**
 498          * Get a preprocessor object
 499          */
 500         function getPreprocessor() {
 501                 if ( !isset( $this->mPreprocessor ) ) {
 502                         $class = $this->mPreprocessorClass;
 503                         $this->mPreprocessor = new $class( $this );
 504                 }
 505                 return $this->mPreprocessor;
 506         }
 507
 508         /**
 509          * Replaces all occurrences of HTML-style comments and the given tags
 510          * in the text with a random marker and returns the next text. The output
 511          * parameter $matches will be an associative array filled with data in
 512          * the form:
 513          *   'UNIQ-xxxxx' => array(
 514          *     'element',
 515          *     'tag content',
 516          *     array( 'param' => 'x' ),
 517          *     '<element param="x">tag content</element>' ) )
 518          *
 519          * @param $elements list of element names. Comments are always extracted.
 520          * @param $text Source text string.
 521          * @param $uniq_prefix
 522          *
 523          * @public
 524          * @static
 525          */
 526         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 527                 static $n = 1;
 528                 $stripped = '';
 529                 $matches = array();
 530
 531                 $taglist = implode( '|', $elements );
 532                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 533
 534                 while ( '' != $text ) {
 535                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 536                         $stripped .= $p[0];
 537                         if( count( $p ) < 5 ) {
 538                                 break;
 539                         }
 540                         if( count( $p ) > 5 ) {
 541                                 // comment
 542                                 $element    = $p[4];
 543                                 $attributes = '';
 544                                 $close      = '';
 545                                 $inside     = $p[5];
 546                         } else {
 547                                 // tag
 548                                 $element    = $p[1];
 549                                 $attributes = $p[2];
 550                                 $close      = $p[3];
 551                                 $inside     = $p[4];
 552                         }
 553
 554                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 555                         $stripped .= $marker;
 556
 557                         if ( $close === '/>' ) {
 558                                 // Empty element tag, <tag />
 559                                 $content = null;
 560                                 $text = $inside;
 561                                 $tail = null;
 562                         } else {
 563                                 if( $element === '!--' ) {
 564                                         $end = '/(-->)/';
 565                                 } else {
 566                                         $end = "/(<\\/$element\\s*>)/i";
 567                                 }
 568                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 569                                 $content = $q[0];
 570                                 if( count( $q ) < 3 ) {
 571                                         # No end tag -- let it run out to the end of the text.
 572                                         $tail = '';
 573                                         $text = '';
 574                                 } else {
 575                                         $tail = $q[1];
 576                                         $text = $q[2];
 577                                 }
 578                         }
 579
 580                         $matches[$marker] = array( $element,
 581                                 $content,
 582                                 Sanitizer::decodeTagAttributes( $attributes ),
 583                                 "<$element$attributes$close$content$tail" );
 584                 }
 585                 return $stripped;
 586         }
 587
 588         /**
 589          * Get a list of strippable XML-like elements
 590          */
 591         function getStripList() {
 592                 global $wgRawHtml;
 593                 $elements = $this->mStripList;
 594                 if( $wgRawHtml ) {
 595                         $elements[] = 'html';
 596                 }
 597                 if( $this->mOptions->getUseTeX() ) {
 598                         $elements[] = 'math';
 599                 }
 600                 return $elements;
 601         }
 602
 603         /**
 604          * @deprecated use replaceVariables
 605          */
 606         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 607                 return $text;
 608         }
 609
 610         /**
 611          * Restores pre, math, and other extensions removed by strip()
 612          *
 613          * always call unstripNoWiki() after this one
 614          * @private
 615          * @deprecated use $this->mStripState->unstrip()
 616          */
 617         function unstrip( $text, $state ) {
 618                 return $state->unstripGeneral( $text );
 619         }
 620
 621         /**
 622          * Always call this after unstrip() to preserve the order
 623          *
 624          * @private
 625          * @deprecated use $this->mStripState->unstrip()
 626          */
 627         function unstripNoWiki( $text, $state ) {
 628                 return $state->unstripNoWiki( $text );
 629         }
 630
 631         /**
 632          * @deprecated use $this->mStripState->unstripBoth()
 633          */
 634         function unstripForHTML( $text ) {
 635                 return $this->mStripState->unstripBoth( $text );
 636         }
 637
 638         /**
 639          * Add an item to the strip state
 640          * Returns the unique tag which must be inserted into the stripped text
 641          * The tag will be replaced with the original text in unstrip()
 642          *
 643          * @private
 644          */
 645         function insertStripItem( $text ) {
 646                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 647                 $this->mMarkerIndex++;
 648                 $this->mStripState->general->setPair( $rnd, $text );
 649                 return $rnd;
 650         }
 651
 652         /**
 653          * Interface with html tidy, used if $wgUseTidy = true.
 654          * If tidy isn't able to correct the markup, the original will be
 655          * returned in all its glory with a warning comment appended.
 656          *
 657          * Either the external tidy program or the in-process tidy extension
 658          * will be used depending on availability. Override the default
 659          * $wgTidyInternal setting to disable the internal if it's not working.
 660          *
 661          * @param string $text Hideous HTML input
 662          * @return string Corrected HTML output
 663          * @public
 664          * @static
 665          */
 666         function tidy( $text ) {
 667                 global $wgTidyInternal;
 668                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 669 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 670 '<head><title>test</title></head><body>'.$text.'</body></html>';
 671                 if( $wgTidyInternal ) {
 672                         $correctedtext = self::internalTidy( $wrappedtext );
 673                 } else {
 674                         $correctedtext = self::externalTidy( $wrappedtext );
 675                 }
 676                 if( is_null( $correctedtext ) ) {
 677                         wfDebug( "Tidy error detected!\n" );
 678                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 679                 }
 680                 return $correctedtext;
 681         }
 682
 683         /**
 684          * Spawn an external HTML tidy process and get corrected markup back from it.
 685          *
 686          * @private
 687          * @static
 688          */
 689         function externalTidy( $text ) {
 690                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 691                 wfProfileIn( __METHOD__ );
 692
 693                 $cleansource = '';
 694                 $opts = ' -utf8';
 695
 696                 $descriptorspec = array(
 697                         0 => array('pipe', 'r'),
 698                         1 => array('pipe', 'w'),
 699                         2 => array('file', wfGetNull(), 'a')
 700                 );
 701                 $pipes = array();
 702                 if( function_exists('proc_open') ) {
 703                         $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 704                         if (is_resource($process)) {
 705                                 // Theoretically, this style of communication could cause a deadlock
 706                                 // here. If the stdout buffer fills up, then writes to stdin could
 707                                 // block. This doesn't appear to happen with tidy, because tidy only
 708                                 // writes to stdout after it's finished reading from stdin. Search
 709                                 // for tidyParseStdin and tidySaveStdout in console/tidy.c
 710                                 fwrite($pipes[0], $text);
 711                                 fclose($pipes[0]);
 712                                 while (!feof($pipes[1])) {
 713                                         $cleansource .= fgets($pipes[1], 1024);
 714                                 }
 715                                 fclose($pipes[1]);
 716                                 proc_close($process);
 717                         }
 718                 }
 719
 720                 wfProfileOut( __METHOD__ );
 721
 722                 if( $cleansource == '' && $text != '') {
 723                         // Some kind of error happened, so we couldn't get the corrected text.
 724                         // Just give up; we'll use the source text and append a warning.
 725                         return null;
 726                 } else {
 727                         return $cleansource;
 728                 }
 729         }
 730
 731         /**
 732          * Use the HTML tidy PECL extension to use the tidy library in-process,
 733          * saving the overhead of spawning a new process.
 734          *
 735          * 'pear install tidy' should be able to compile the extension module.
 736          *
 737          * @private
 738          * @static
 739          */
 740         function internalTidy( $text ) {
 741                 global $wgTidyConf, $IP, $wgDebugTidy;
 742                 wfProfileIn( __METHOD__ );
 743
 744                 $tidy = new tidy;
 745                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 746                 $tidy->cleanRepair();
 747                 if( $tidy->getStatus() == 2 ) {
 748                         // 2 is magic number for fatal error
 749                         // http://www.php.net/manual/en/function.tidy-get-status.php
 750                         $cleansource = null;
 751                 } else {
 752                         $cleansource = tidy_get_output( $tidy );
 753                 }
 754                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 755                         $cleansource .= "<!--\nTidy reports:\n" .
 756                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 757                                 "\n-->";
 758                 }
 759
 760                 wfProfileOut( __METHOD__ );
 761                 return $cleansource;
 762         }
 763
 764         /**
 765          * parse the wiki syntax used to render tables
 766          *
 767          * @private
 768          */
 769         function doTableStuff ( $text ) {
 770                 wfProfileIn( __METHOD__ );
 771
 772                 $lines = StringUtils::explode( "\n", $text );
 773                 $out = '';
 774                 $td_history = array (); // Is currently a td tag open?
 775                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 776                 $tr_history = array (); // Is currently a tr tag open?
 777                 $tr_attributes = array (); // history of tr attributes
 778                 $has_opened_tr = array(); // Did this table open a <tr> element?
 779                 $indent_level = 0; // indent level of the table
 780
 781                 foreach ( $lines as $outLine ) {
 782                         $line = trim( $outLine );
 783
 784                         if( $line == '' ) { // empty line, go to next line
 785                                 $out .= $outLine."\n";
 786                                 continue;
 787                         }
 788                         $first_character = $line[0];
 789                         $matches = array();
 790
 791                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 792                                 // First check if we are starting a new table
 793                                 $indent_level = strlen( $matches[1] );
 794
 795                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 796                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 797
 798                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 799                                 array_push ( $td_history , false );
 800                                 array_push ( $last_tag_history , '' );
 801                                 array_push ( $tr_history , false );
 802                                 array_push ( $tr_attributes , '' );
 803                                 array_push ( $has_opened_tr , false );
 804                         } else if ( count ( $td_history ) == 0 ) {
 805                                 // Don't do any of the following
 806                                 $out .= $outLine."\n";
 807                                 continue;
 808                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 809                                 // We are ending a table
 810                                 $line = '</table>' . substr ( $line , 2 );
 811                                 $last_tag = array_pop ( $last_tag_history );
 812
 813                                 if ( !array_pop ( $has_opened_tr ) ) {
 814                                         $line = "<tr><td></td></tr>{$line}";
 815                                 }
 816
 817                                 if ( array_pop ( $tr_history ) ) {
 818                                         $line = "</tr>{$line}";
 819                                 }
 820
 821                                 if ( array_pop ( $td_history ) ) {
 822                                         $line = "</{$last_tag}>{$line}";
 823                                 }
 824                                 array_pop ( $tr_attributes );
 825                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 826                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 827                                 // Now we have a table row
 828                                 $line = preg_replace( '#^\|-+#', '', $line );
 829
 830                                 // Whats after the tag is now only attributes
 831                                 $attributes = $this->mStripState->unstripBoth( $line );
 832                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 833                                 array_pop ( $tr_attributes );
 834                                 array_push ( $tr_attributes , $attributes );
 835
 836                                 $line = '';
 837                                 $last_tag = array_pop ( $last_tag_history );
 838                                 array_pop ( $has_opened_tr );
 839                                 array_push ( $has_opened_tr , true );
 840
 841                                 if ( array_pop ( $tr_history ) ) {
 842                                         $line = '</tr>';
 843                                 }
 844
 845                                 if ( array_pop ( $td_history ) ) {
 846                                         $line = "</{$last_tag}>{$line}";
 847                                 }
 848
 849                                 $outLine = $line;
 850                                 array_push ( $tr_history , false );
 851                                 array_push ( $td_history , false );
 852                                 array_push ( $last_tag_history , '' );
 853                         }
 854                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 855                                 // This might be cell elements, td, th or captions
 856                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 857                                         $first_character = '+';
 858                                         $line = substr ( $line , 1 );
 859                                 }
 860
 861                                 $line = substr ( $line , 1 );
 862
 863                                 if ( $first_character === '!' ) {
 864                                         $line = str_replace ( '!!' , '||' , $line );
 865                                 }
 866
 867                                 // Split up multiple cells on the same line.
 868                                 // FIXME : This can result in improper nesting of tags processed
 869                                 // by earlier parser steps, but should avoid splitting up eg
 870                                 // attribute values containing literal "||".
 871                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 872
 873                                 $outLine = '';
 874
 875                                 // Loop through each table cell
 876                                 foreach ( $cells as $cell )
 877                                 {
 878                                         $previous = '';
 879                                         if ( $first_character !== '+' )
 880                                         {
 881                                                 $tr_after = array_pop ( $tr_attributes );
 882                                                 if ( !array_pop ( $tr_history ) ) {
 883                                                         $previous = "<tr{$tr_after}>\n";
 884                                                 }
 885                                                 array_push ( $tr_history , true );
 886                                                 array_push ( $tr_attributes , '' );
 887                                                 array_pop ( $has_opened_tr );
 888                                                 array_push ( $has_opened_tr , true );
 889                                         }
 890
 891                                         $last_tag = array_pop ( $last_tag_history );
 892
 893                                         if ( array_pop ( $td_history ) ) {
 894                                                 $previous = "</{$last_tag}>{$previous}";
 895                                         }
 896
 897                                         if ( $first_character === '|' ) {
 898                                                 $last_tag = 'td';
 899                                         } else if ( $first_character === '!' ) {
 900                                                 $last_tag = 'th';
 901                                         } else if ( $first_character === '+' ) {
 902                                                 $last_tag = 'caption';
 903                                         } else {
 904                                                 $last_tag = '';
 905                                         }
 906
 907                                         array_push ( $last_tag_history , $last_tag );
 908
 909                                         // A cell could contain both parameters and data
 910                                         $cell_data = explode ( '|' , $cell , 2 );
 911
 912                                         // Bug 553: Note that a '|' inside an invalid link should not
 913                                         // be mistaken as delimiting cell parameters
 914                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 915                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 916                                         } else if ( count ( $cell_data ) == 1 )
 917                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 918                                         else {
 919                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 920                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 921                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 922                                         }
 923
 924                                         $outLine .= $cell;
 925                                         array_push ( $td_history , true );
 926                                 }
 927                         }
 928                         $out .= $outLine . "\n";
 929                 }
 930
 931                 // Closing open td, tr && table
 932                 while ( count ( $td_history ) > 0 )
 933                 {
 934                         if ( array_pop ( $td_history ) ) {
 935                                 $out .= "</td>\n";
 936                         }
 937                         if ( array_pop ( $tr_history ) ) {
 938                                 $out .= "</tr>\n";
 939                         }
 940                         if ( !array_pop ( $has_opened_tr ) ) {
 941                                 $out .= "<tr><td></td></tr>\n" ;
 942                         }
 943
 944                         $out .= "</table>\n";
 945                 }
 946
 947                 // Remove trailing line-ending (b/c)
 948                 if ( substr( $out, -1 ) === "\n" ) {
 949                         $out = substr( $out, 0, -1 );
 950                 }
 951
 952                 // special case: don't return empty table
 953                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 954                         $out = '';
 955                 }
 956
 957                 wfProfileOut( __METHOD__ );
 958
 959                 return $out;
 960         }
 961
 962         /**
 963          * Helper function for parse() that transforms wiki markup into
 964          * HTML. Only called for $mOutputType == self::OT_HTML.
 965          *
 966          * @private
 967          */
 968         function internalParse( $text ) {
 969                 $isMain = true;
 970                 wfProfileIn( __METHOD__ );
 971
 972                 # Hook to suspend the parser in this state
 973                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 974                         wfProfileOut( __METHOD__ );
 975                         return $text ;
 976                 }
 977
 978                 $text = $this->replaceVariables( $text );
 979                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 980                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 981
 982                 // Tables need to come after variable replacement for things to work
 983                 // properly; putting them before other transformations should keep
 984                 // exciting things like link expansions from showing up in surprising
 985                 // places.
 986                 $text = $this->doTableStuff( $text );
 987
 988                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 989
 990                 $text = $this->doDoubleUnderscore( $text );
 991                 $text = $this->doHeadings( $text );
 992                 if($this->mOptions->getUseDynamicDates()) {
 993                         $df = DateFormatter::getInstance();
 994                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 995                 }
 996                 $text = $this->doAllQuotes( $text );
 997                 $text = $this->replaceInternalLinks( $text );
 998                 $text = $this->replaceExternalLinks( $text );
 999
1000                 # replaceInternalLinks may sometimes leave behind
1001                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1002                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1003
1004                 $text = $this->doMagicLinks( $text );
1005                 $text = $this->formatHeadings( $text, $isMain );
1006
1007                 wfProfileOut( __METHOD__ );
1008                 return $text;
1009         }
1010
1011         /**
1012          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1013          * magic external links.
1014          *
1015          * DML
1016          * @private
1017          */
1018         function doMagicLinks( $text ) {
1019                 wfProfileIn( __METHOD__ );
1020                 $prots = $this->mUrlProtocols;
1021                 $urlChar = self::EXT_LINK_URL_CLASS;
1022                 $text = preg_replace_callback(
1023                         '!(?:                           # Start cases
1024                                 (<a.*?</a>) |               # m[1]: Skip link text
1025                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1026                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1027                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1028                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1029                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1030                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1031                                     [0-9Xx]                 # check digit
1032                                     \b)
1033                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1034                 wfProfileOut( __METHOD__ );
1035                 return $text;
1036         }
1037
1038         function magicLinkCallback( $m ) {
1039                 if ( isset( $m[1] ) && strval( $m[1] ) !== '' ) {
1040                         # Skip anchor
1041                         return $m[0];
1042                 } elseif ( isset( $m[2] ) && strval( $m[2] ) !== '' ) {
1043                         # Skip HTML element
1044                         return $m[0];
1045                 } elseif ( isset( $m[3] ) && strval( $m[3] ) !== '' ) {
1046                         # Free external link
1047                         return $this->makeFreeExternalLink( $m[0] );
1048                 } elseif ( isset( $m[4] ) && strval( $m[4] ) !== '' ) {
1049                         # RFC or PMID
1050                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1051                                 $keyword = 'RFC';
1052                                 $urlmsg = 'rfcurl';
1053                                 $id = $m[4];
1054                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1055                                 $keyword = 'PMID';
1056                                 $urlmsg = 'pubmedurl';
1057                                 $id = $m[4];
1058                         } else {
1059                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1060                                         substr($m[0], 0, 20 ) . '"' );
1061                         }
1062                         $url = wfMsg( $urlmsg, $id);
1063                         $sk = $this->mOptions->getSkin();
1064                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1065                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1066                 } elseif ( isset( $m[5] ) && strval( $m[5] ) !== '' ) {
1067                         # ISBN
1068                         $isbn = $m[5];
1069                         $num = strtr( $isbn, array(
1070                                 '-' => '',
1071                                 ' ' => '',
1072                                 'x' => 'X',
1073                         ));
1074                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1075                         return'<a href="' .
1076                                 $titleObj->escapeLocalUrl() .
1077                                 "\" class=\"internal\">ISBN $isbn</a>";
1078                 } else {
1079                         return $m[0];
1080                 }
1081         }
1082
1083         /**
1084          * Make a free external link, given a user-supplied URL
1085          * @return HTML
1086          * @private
1087          */
1088         function makeFreeExternalLink( $url ) {
1089                 global $wgContLang;
1090                 wfProfileIn( __METHOD__ );
1091
1092                 $sk = $this->mOptions->getSkin();
1093                 $trail = '';
1094
1095                 # The characters '<' and '>' (which were escaped by
1096                 # removeHTMLtags()) should not be included in
1097                 # URLs, per RFC 2396.
1098                 $m2 = array();
1099                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1100                         $trail = substr($url, $m2[0][1]) . $trail;
1101                         $url = substr($url, 0, $m2[0][1]);
1102                 }
1103
1104                 # Move trailing punctuation to $trail
1105                 $sep = ',;\.:!?';
1106                 # If there is no left bracket, then consider right brackets fair game too
1107                 if ( strpos( $url, '(' ) === false ) {
1108                         $sep .= ')';
1109                 }
1110
1111                 $numSepChars = strspn( strrev( $url ), $sep );
1112                 if ( $numSepChars ) {
1113                         $trail = substr( $url, -$numSepChars ) . $trail;
1114                         $url = substr( $url, 0, -$numSepChars );
1115                 }
1116
1117                 $url = Sanitizer::cleanUrl( $url );
1118
1119                 # Is this an external image?
1120                 $text = $this->maybeMakeExternalImage( $url );
1121                 if ( $text === false ) {
1122                         # Not an image, make a link
1123                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1124                                 $this->getExternalLinkAttribs() );
1125                         # Register it in the output object...
1126                         # Replace unnecessary URL escape codes with their equivalent characters
1127                         $pasteurized = self::replaceUnusualEscapes( $url );
1128                         $this->mOutput->addExternalLink( $pasteurized );
1129                 }
1130                 wfProfileOut( __METHOD__ );
1131                 return $text . $trail;
1132         }
1133
1134
1135         /**
1136          * Parse headers and return html
1137          *
1138          * @private
1139          */
1140         function doHeadings( $text ) {
1141                 wfProfileIn( __METHOD__ );
1142                 for ( $i = 6; $i >= 1; --$i ) {
1143                         $h = str_repeat( '=', $i );
1144                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1145                           "<h$i>\\1</h$i>", $text );
1146                 }
1147                 wfProfileOut( __METHOD__ );
1148                 return $text;
1149         }
1150
1151         /**
1152          * Replace single quotes with HTML markup
1153          * @private
1154          * @return string the altered text
1155          */
1156         function doAllQuotes( $text ) {
1157                 wfProfileIn( __METHOD__ );
1158                 $outtext = '';
1159                 $lines = StringUtils::explode( "\n", $text );
1160                 foreach ( $lines as $line ) {
1161                         $outtext .= $this->doQuotes( $line ) . "\n";
1162                 }
1163                 $outtext = substr($outtext, 0,-1);
1164                 wfProfileOut( __METHOD__ );
1165                 return $outtext;
1166         }
1167
1168         /**
1169          * Helper function for doAllQuotes()
1170          */
1171         public function doQuotes( $text ) {
1172                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1173                 if ( count( $arr ) == 1 )
1174                         return $text;
1175                 else
1176                 {
1177                         # First, do some preliminary work. This may shift some apostrophes from
1178                         # being mark-up to being text. It also counts the number of occurrences
1179                         # of bold and italics mark-ups.
1180                         $i = 0;
1181                         $numbold = 0;
1182                         $numitalics = 0;
1183                         foreach ( $arr as $r )
1184                         {
1185                                 if ( ( $i % 2 ) == 1 )
1186                                 {
1187                                         # If there are ever four apostrophes, assume the first is supposed to
1188                                         # be text, and the remaining three constitute mark-up for bold text.
1189                                         if ( strlen( $arr[$i] ) == 4 )
1190                                         {
1191                                                 $arr[$i-1] .= "'";
1192                                                 $arr[$i] = "'''";
1193                                         }
1194                                         # If there are more than 5 apostrophes in a row, assume they're all
1195                                         # text except for the last 5.
1196                                         else if ( strlen( $arr[$i] ) > 5 )
1197                                         {
1198                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1199                                                 $arr[$i] = "'''''";
1200                                         }
1201                                         # Count the number of occurrences of bold and italics mark-ups.
1202                                         # We are not counting sequences of five apostrophes.
1203                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1204                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1205                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1206                                 }
1207                                 $i++;
1208                         }
1209
1210                         # If there is an odd number of both bold and italics, it is likely
1211                         # that one of the bold ones was meant to be an apostrophe followed
1212                         # by italics. Which one we cannot know for certain, but it is more
1213                         # likely to be one that has a single-letter word before it.
1214                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1215                         {
1216                                 $i = 0;
1217                                 $firstsingleletterword = -1;
1218                                 $firstmultiletterword = -1;
1219                                 $firstspace = -1;
1220                                 foreach ( $arr as $r )
1221                                 {
1222                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1223                                         {
1224                                                 $x1 = substr ($arr[$i-1], -1);
1225                                                 $x2 = substr ($arr[$i-1], -2, 1);
1226                                                 if ($x1 === ' ') {
1227                                                         if ($firstspace == -1) $firstspace = $i;
1228                                                 } else if ($x2 === ' ') {
1229                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1230                                                 } else {
1231                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1232                                                 }
1233                                         }
1234                                         $i++;
1235                                 }
1236
1237                                 # If there is a single-letter word, use it!
1238                                 if ($firstsingleletterword > -1)
1239                                 {
1240                                         $arr [ $firstsingleletterword ] = "''";
1241                                         $arr [ $firstsingleletterword-1 ] .= "'";
1242                                 }
1243                                 # If not, but there's a multi-letter word, use that one.
1244                                 else if ($firstmultiletterword > -1)
1245                                 {
1246                                         $arr [ $firstmultiletterword ] = "''";
1247                                         $arr [ $firstmultiletterword-1 ] .= "'";
1248                                 }
1249                                 # ... otherwise use the first one that has neither.
1250                                 # (notice that it is possible for all three to be -1 if, for example,
1251                                 # there is only one pentuple-apostrophe in the line)
1252                                 else if ($firstspace > -1)
1253                                 {
1254                                         $arr [ $firstspace ] = "''";
1255                                         $arr [ $firstspace-1 ] .= "'";
1256                                 }
1257                         }
1258
1259                         # Now let's actually convert our apostrophic mush to HTML!
1260                         $output = '';
1261                         $buffer = '';
1262                         $state = '';
1263                         $i = 0;
1264                         foreach ($arr as $r)
1265                         {
1266                                 if (($i % 2) == 0)
1267                                 {
1268                                         if ($state === 'both')
1269                                                 $buffer .= $r;
1270                                         else
1271                                                 $output .= $r;
1272                                 }
1273                                 else
1274                                 {
1275                                         if (strlen ($r) == 2)
1276                                         {
1277                                                 if ($state === 'i')
1278                                                 { $output .= '</i>'; $state = ''; }
1279                                                 else if ($state === 'bi')
1280                                                 { $output .= '</i>'; $state = 'b'; }
1281                                                 else if ($state === 'ib')
1282                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1283                                                 else if ($state === 'both')
1284                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1285                                                 else # $state can be 'b' or ''
1286                                                 { $output .= '<i>'; $state .= 'i'; }
1287                                         }
1288                                         else if (strlen ($r) == 3)
1289                                         {
1290                                                 if ($state === 'b')
1291                                                 { $output .= '</b>'; $state = ''; }
1292                                                 else if ($state === 'bi')
1293                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1294                                                 else if ($state === 'ib')
1295                                                 { $output .= '</b>'; $state = 'i'; }
1296                                                 else if ($state === 'both')
1297                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1298                                                 else # $state can be 'i' or ''
1299                                                 { $output .= '<b>'; $state .= 'b'; }
1300                                         }
1301                                         else if (strlen ($r) == 5)
1302                                         {
1303                                                 if ($state === 'b')
1304                                                 { $output .= '</b><i>'; $state = 'i'; }
1305                                                 else if ($state === 'i')
1306                                                 { $output .= '</i><b>'; $state = 'b'; }
1307                                                 else if ($state === 'bi')
1308                                                 { $output .= '</i></b>'; $state = ''; }
1309                                                 else if ($state === 'ib')
1310                                                 { $output .= '</b></i>'; $state = ''; }
1311                                                 else if ($state === 'both')
1312                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1313                                                 else # ($state == '')
1314                                                 { $buffer = ''; $state = 'both'; }
1315                                         }
1316                                 }
1317                                 $i++;
1318                         }
1319                         # Now close all remaining tags.  Notice that the order is important.
1320                         if ($state === 'b' || $state === 'ib')
1321                                 $output .= '</b>';
1322                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1323                                 $output .= '</i>';
1324                         if ($state === 'bi')
1325                                 $output .= '</b>';
1326                         # There might be lonely ''''', so make sure we have a buffer
1327                         if ($state === 'both' && $buffer)
1328                                 $output .= '<b><i>'.$buffer.'</i></b>';
1329                         return $output;
1330                 }
1331         }
1332
1333         /**
1334          * Replace external links (REL)
1335          *
1336          * Note: this is all very hackish and the order of execution matters a lot.
1337          * Make sure to run maintenance/parserTests.php if you change this code.
1338          *
1339          * @private
1340          */
1341         function replaceExternalLinks( $text ) {
1342                 global $wgContLang;
1343                 wfProfileIn( __METHOD__ );
1344
1345                 $sk = $this->mOptions->getSkin();
1346
1347                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1348                 $s = array_shift( $bits );
1349
1350                 $i = 0;
1351                 while ( $i<count( $bits ) ) {
1352                         $url = $bits[$i++];
1353                         $protocol = $bits[$i++];
1354                         $text = $bits[$i++];
1355                         $trail = $bits[$i++];
1356
1357                         # The characters '<' and '>' (which were escaped by
1358                         # removeHTMLtags()) should not be included in
1359                         # URLs, per RFC 2396.
1360                         $m2 = array();
1361                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1362                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1363                                 $url = substr($url, 0, $m2[0][1]);
1364                         }
1365
1366                         # If the link text is an image URL, replace it with an <img> tag
1367                         # This happened by accident in the original parser, but some people used it extensively
1368                         $img = $this->maybeMakeExternalImage( $text );
1369                         if ( $img !== false ) {
1370                                 $text = $img;
1371                         }
1372
1373                         $dtrail = '';
1374
1375                         # Set linktype for CSS - if URL==text, link is essentially free
1376                         $linktype = ($text === $url) ? 'free' : 'text';
1377
1378                         # No link text, e.g. [http://domain.tld/some.link]
1379                         if ( $text == '' ) {
1380                                 # Autonumber if allowed. See bug #5918
1381                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1382                                         $langObj = $this->getFunctionLang();
1383                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1384                                         $linktype = 'autonumber';
1385                                 } else {
1386                                         # Otherwise just use the URL
1387                                         $text = htmlspecialchars( $url );
1388                                         $linktype = 'free';
1389                                 }
1390                         } else {
1391                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1392                                 # Check for trail
1393                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1394                         }
1395
1396                         $text = $wgContLang->markNoConversion($text);
1397
1398                         $url = Sanitizer::cleanUrl( $url );
1399
1400                         if ( $this->mOptions->mExternalLinkTarget ) {
1401                                 $attribs = array( 'target' => $this->mOptions->mExternalLinkTarget );
1402                         } else {
1403                                 $attribs = array();
1404                         }
1405
1406                         # Use the encoded URL
1407                         # This means that users can paste URLs directly into the text
1408                         # Funny characters like &ouml; aren't valid in URLs anyway
1409                         # This was changed in August 2004
1410                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->getExternalLinkAttribs() )
1411                                 . $dtrail . $trail;
1412
1413                         # Register link in the output object.
1414                         # Replace unnecessary URL escape codes with the referenced character
1415                         # This prevents spammers from hiding links from the filters
1416                         $pasteurized = self::replaceUnusualEscapes( $url );
1417                         $this->mOutput->addExternalLink( $pasteurized );
1418                 }
1419
1420                 wfProfileOut( __METHOD__ );
1421                 return $s;
1422         }
1423
1424         function getExternalLinkAttribs() {
1425                 $attribs = array();
1426                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1427                 $ns = $this->mTitle->getNamespace();
1428                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1429                         $attribs['rel'] = 'nofollow';
1430                 }
1431                 if ( $this->mOptions->getExternalLinkTarget() ) {
1432                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1433                 }
1434                 return $attribs;
1435         }
1436
1437
1438         /**
1439          * Replace unusual URL escape codes with their equivalent characters
1440          * @param string
1441          * @return string
1442          * @static
1443          * @todo  This can merge genuinely required bits in the path or query string,
1444          *        breaking legit URLs. A proper fix would treat the various parts of
1445          *        the URL differently; as a workaround, just use the output for
1446          *        statistical records, not for actual linking/output.
1447          */
1448         static function replaceUnusualEscapes( $url ) {
1449                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1450                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1451         }
1452
1453         /**
1454          * Callback function used in replaceUnusualEscapes().
1455          * Replaces unusual URL escape codes with their equivalent character
1456          * @static
1457          * @private
1458          */
1459         private static function replaceUnusualEscapesCallback( $matches ) {
1460                 $char = urldecode( $matches[0] );
1461                 $ord = ord( $char );
1462                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1463                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1464                         // No, shouldn't be escaped
1465                         return $char;
1466                 } else {
1467                         // Yes, leave it escaped
1468                         return $matches[0];
1469                 }
1470         }
1471
1472         /**
1473          * make an image if it's allowed, either through the global
1474          * option, through the exception, or through the on-wiki whitelist
1475          * @private
1476          */
1477         function maybeMakeExternalImage( $url ) {
1478                 $sk = $this->mOptions->getSkin();
1479                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1480                 $imagesexception = !empty($imagesfrom);
1481                 $text = false;
1482                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1483                 if( $imagesexception && is_array( $imagesfrom ) ) {
1484                         $imagematch = false;
1485                         foreach( $imagesfrom as $match ) {
1486                                 if( strpos( $url, $match ) === 0 ) {
1487                                         $imagematch = true;
1488                                         break;
1489                                 }
1490                         }
1491                 } elseif( $imagesexception ) {
1492                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1493                 } else {
1494                         $imagematch = false;
1495                 }
1496                 if ( $this->mOptions->getAllowExternalImages()
1497                      || ( $imagesexception && $imagematch ) ) {
1498                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1499                                 # Image found
1500                                 $text = $sk->makeExternalImage( $url );
1501                         }
1502                 }
1503                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1504                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1505                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1506                         foreach( $whitelist as $entry ) {
1507                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1508                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1509                                         continue;
1510                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1511                                         # Image matches a whitelist entry
1512                                         $text = $sk->makeExternalImage( $url );
1513                                         break;
1514                                 }
1515                         }
1516                 }
1517                 return $text;
1518         }
1519
1520         /**
1521          * Process [[ ]] wikilinks
1522          * @return processed text
1523          *
1524          * @private
1525          */
1526         function replaceInternalLinks( $s ) {
1527                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1528                 return $s;
1529         }
1530
1531         /**
1532          * Process [[ ]] wikilinks (RIL)
1533          * @return LinkHolderArray
1534          *
1535          * @private
1536          */
1537         function replaceInternalLinks2( &$s ) {
1538                 global $wgContLang;
1539
1540                 wfProfileIn( __METHOD__ );
1541
1542                 wfProfileIn( __METHOD__.'-setup' );
1543                 static $tc = FALSE, $e1, $e1_img;
1544                 # the % is needed to support urlencoded titles as well
1545                 if ( !$tc ) {
1546                         $tc = Title::legalChars() . '#%';
1547                         # Match a link having the form [[namespace:link|alternate]]trail
1548                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1549                         # Match cases where there is no "]]", which might still be images
1550                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1551                 }
1552
1553                 $sk = $this->mOptions->getSkin();
1554                 $holders = new LinkHolderArray( $this );
1555
1556                 #split the entire text string on occurences of [[
1557                 $a = StringUtils::explode( '[[', ' ' . $s );
1558                 #get the first element (all text up to first [[), and remove the space we added
1559                 $s = $a->current();
1560                 $a->next();
1561                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1562                 $s = substr( $s, 1 );
1563
1564                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1565                 $e2 = null;
1566                 if ( $useLinkPrefixExtension ) {
1567                         # Match the end of a line for a word that's not followed by whitespace,
1568                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1569                         $e2 = wfMsgForContent( 'linkprefix' );
1570                 }
1571
1572                 if( is_null( $this->mTitle ) ) {
1573                         wfProfileOut( __METHOD__.'-setup' );
1574                         wfProfileOut( __METHOD__ );
1575                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1576                 }
1577                 $nottalk = !$this->mTitle->isTalkPage();
1578
1579                 if ( $useLinkPrefixExtension ) {
1580                         $m = array();
1581                         if ( preg_match( $e2, $s, $m ) ) {
1582                                 $first_prefix = $m[2];
1583                         } else {
1584                                 $first_prefix = false;
1585                         }
1586                 } else {
1587                         $prefix = '';
1588                 }
1589
1590                 if($wgContLang->hasVariants()) {
1591                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1592                 } else {
1593                         $selflink = array($this->mTitle->getPrefixedText());
1594                 }
1595                 $useSubpages = $this->areSubpagesAllowed();
1596                 wfProfileOut( __METHOD__.'-setup' );
1597
1598                 # Loop for each link
1599                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1600                         # Check for excessive memory usage
1601                         if ( $holders->isBig() ) {
1602                                 # Too big
1603                                 # Do the existence check, replace the link holders and clear the array
1604                                 $holders->replace( $s );
1605                                 $holders->clear();
1606                         }
1607
1608                         if ( $useLinkPrefixExtension ) {
1609                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1610                                 if ( preg_match( $e2, $s, $m ) ) {
1611                                         $prefix = $m[2];
1612                                         $s = $m[1];
1613                                 } else {
1614                                         $prefix='';
1615                                 }
1616                                 # first link
1617                                 if($first_prefix) {
1618                                         $prefix = $first_prefix;
1619                                         $first_prefix = false;
1620                                 }
1621                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1622                         }
1623
1624                         $might_be_img = false;
1625
1626                         wfProfileIn( __METHOD__."-e1" );
1627                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1628                                 $text = $m[2];
1629                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1630                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1631                                 # the real problem is with the $e1 regex
1632                                 # See bug 1300.
1633                                 #
1634                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1635                                 # and no image is in sight. See bug 2095.
1636                                 #
1637                                 if( $text !== '' &&
1638                                         substr( $m[3], 0, 1 ) === ']' &&
1639                                         strpos($text, '[') !== false
1640                                 )
1641                                 {
1642                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1643                                         $m[3] = substr( $m[3], 1 );
1644                                 }
1645                                 # fix up urlencoded title texts
1646                                 if( strpos( $m[1], '%' ) !== false ) {
1647                                         # Should anchors '#' also be rejected?
1648                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1649                                 }
1650                                 $trail = $m[3];
1651                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1652                                 $might_be_img = true;
1653                                 $text = $m[2];
1654                                 if ( strpos( $m[1], '%' ) !== false ) {
1655                                         $m[1] = urldecode($m[1]);
1656                                 }
1657                                 $trail = "";
1658                         } else { # Invalid form; output directly
1659                                 $s .= $prefix . '[[' . $line ;
1660                                 wfProfileOut( __METHOD__."-e1" );
1661                                 continue;
1662                         }
1663                         wfProfileOut( __METHOD__."-e1" );
1664                         wfProfileIn( __METHOD__."-misc" );
1665
1666                         # Don't allow internal links to pages containing
1667                         # PROTO: where PROTO is a valid URL protocol; these
1668                         # should be external links.
1669                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1670                                 $s .= $prefix . '[[' . $line ;
1671                                 wfProfileOut( __METHOD__."-misc" );
1672                                 continue;
1673                         }
1674
1675                         # Make subpage if necessary
1676                         if( $useSubpages ) {
1677                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1678                         } else {
1679                                 $link = $m[1];
1680                         }
1681
1682                         $noforce = (substr($m[1], 0, 1) !== ':');
1683                         if (!$noforce) {
1684                                 # Strip off leading ':'
1685                                 $link = substr($link, 1);
1686                         }
1687
1688                         wfProfileOut( __METHOD__."-misc" );
1689                         wfProfileIn( __METHOD__."-title" );
1690                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1691                         if( !$nt ) {
1692                                 $s .= $prefix . '[[' . $line;
1693                                 wfProfileOut( __METHOD__."-title" );
1694                                 continue;
1695                         }
1696
1697                         $ns = $nt->getNamespace();
1698                         $iw = $nt->getInterWiki();
1699                         wfProfileOut( __METHOD__."-title" );
1700
1701                         if ($might_be_img) { # if this is actually an invalid link
1702                                 wfProfileIn( __METHOD__."-might_be_img" );
1703                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1704                                         $found = false;
1705                                         while ( true ) {
1706                                                 #look at the next 'line' to see if we can close it there
1707                                                 $a->next();
1708                                                 $next_line = $a->current();
1709                                                 if ( $next_line === false || $next_line === null ) {
1710                                                         break;
1711                                                 }
1712                                                 $m = explode( ']]', $next_line, 3 );
1713                                                 if ( count( $m ) == 3 ) {
1714                                                         # the first ]] closes the inner link, the second the image
1715                                                         $found = true;
1716                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1717                                                         $trail = $m[2];
1718                                                         break;
1719                                                 } elseif ( count( $m ) == 2 ) {
1720                                                         #if there's exactly one ]] that's fine, we'll keep looking
1721                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1722                                                 } else {
1723                                                         #if $next_line is invalid too, we need look no further
1724                                                         $text .= '[[' . $next_line;
1725                                                         break;
1726                                                 }
1727                                         }
1728                                         if ( !$found ) {
1729                                                 # we couldn't find the end of this imageLink, so output it raw
1730                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1731                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1732                                                 $s .= "{$prefix}[[$link|$text";
1733                                                 # note: no $trail, because without an end, there *is* no trail
1734                                                 wfProfileOut( __METHOD__."-might_be_img" );
1735                                                 continue;
1736                                         }
1737                                 } else { #it's not an image, so output it raw
1738                                         $s .= "{$prefix}[[$link|$text";
1739                                         # note: no $trail, because without an end, there *is* no trail
1740                                         wfProfileOut( __METHOD__."-might_be_img" );
1741                                         continue;
1742                                 }
1743                                 wfProfileOut( __METHOD__."-might_be_img" );
1744                         }
1745
1746                         $wasblank = ( '' == $text );
1747                         if( $wasblank ) $text = $link;
1748
1749                         # Link not escaped by : , create the various objects
1750                         if( $noforce ) {
1751
1752                                 # Interwikis
1753                                 wfProfileIn( __METHOD__."-interwiki" );
1754                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1755                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1756                                         $s = rtrim($s . $prefix);
1757                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1758                                         wfProfileOut( __METHOD__."-interwiki" );
1759                                         continue;
1760                                 }
1761                                 wfProfileOut( __METHOD__."-interwiki" );
1762
1763                                 if ( $ns == NS_IMAGE ) {
1764                                         wfProfileIn( __METHOD__."-image" );
1765                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1766                                                 # recursively parse links inside the image caption
1767                                                 # actually, this will parse them in any other parameters, too,
1768                                                 # but it might be hard to fix that, and it doesn't matter ATM
1769                                                 $text = $this->replaceExternalLinks($text);
1770                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1771
1772                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1773                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1774                                         }
1775                                         $this->mOutput->addImage( $nt->getDBkey() );
1776                                         wfProfileOut( __METHOD__."-image" );
1777                                         continue;
1778
1779                                 }
1780
1781                                 if ( $ns == NS_CATEGORY ) {
1782                                         wfProfileIn( __METHOD__."-category" );
1783                                         $s = rtrim($s . "\n"); # bug 87
1784
1785                                         if ( $wasblank ) {
1786                                                 $sortkey = $this->getDefaultSort();
1787                                         } else {
1788                                                 $sortkey = $text;
1789                                         }
1790                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1791                                         $sortkey = str_replace( "\n", '', $sortkey );
1792                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1793                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1794
1795                                         /**
1796                                          * Strip the whitespace Category links produce, see bug 87
1797                                          * @todo We might want to use trim($tmp, "\n") here.
1798                                          */
1799                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1800
1801                                         wfProfileOut( __METHOD__."-category" );
1802                                         continue;
1803                                 }
1804                         }
1805
1806                         # Self-link checking
1807                         if( $nt->getFragment() === '' && $nt->getNamespace() != NS_SPECIAL ) {
1808                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1809                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1810                                         continue;
1811                                 }
1812                         }
1813
1814                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1815                         if( $ns == NS_MEDIA ) {
1816                                 # Give extensions a chance to select the file revision for us
1817                                 $skip = $time = false;
1818                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1819                                 if ( $skip ) {
1820                                         $link = $sk->link( $nt );
1821                                 } else {
1822                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1823                                 }
1824                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1825                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1826                                 $this->mOutput->addImage( $nt->getDBkey() );
1827                                 continue;
1828                         } elseif( $ns == NS_SPECIAL ) {
1829                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1830                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1831                                 } else {
1832                                         $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1833                                 }
1834                                 continue;
1835                         } elseif( $ns == NS_IMAGE ) {
1836                                 $img = wfFindFile( $nt );
1837                                 if( $img ) {
1838                                         // Force a blue link if the file exists; may be a remote
1839                                         // upload on the shared repository, and we want to see its
1840                                         // auto-generated page.
1841                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1842                                         $this->mOutput->addLink( $nt );
1843                                         continue;
1844                                 }
1845                         }
1846                         $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1847                 }
1848                 wfProfileOut( __METHOD__ );
1849                 return $holders;
1850         }
1851
1852         /**
1853          * Make a link placeholder. The text returned can be later resolved to a real link with
1854          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1855          * parsing of interwiki links, and secondly to allow all existence checks and
1856          * article length checks (for stub links) to be bundled into a single query.
1857          *
1858          * @deprecated
1859          */
1860         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1861                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1862         }
1863
1864         /**
1865          * Render a forced-blue link inline; protect against double expansion of
1866          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1867          * Since this little disaster has to split off the trail text to avoid
1868          * breaking URLs in the following text without breaking trails on the
1869          * wiki links, it's been made into a horrible function.
1870          *
1871          * @param Title $nt
1872          * @param string $text
1873          * @param string $query
1874          * @param string $trail
1875          * @param string $prefix
1876          * @return string HTML-wikitext mix oh yuck
1877          */
1878         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1879                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1880                 $sk = $this->mOptions->getSkin();
1881                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1882                 return $this->armorLinks( $link ) . $trail;
1883         }
1884
1885         /**
1886          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1887          * going to go through further parsing steps before inline URL expansion.
1888          *
1889          * Not needed quite as much as it used to be since free links are a bit
1890          * more sensible these days. But bracketed links are still an issue.
1891          *
1892          * @param string more-or-less HTML
1893          * @return string less-or-more HTML with NOPARSE bits
1894          */
1895         function armorLinks( $text ) {
1896                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1897                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1898         }
1899
1900         /**
1901          * Return true if subpage links should be expanded on this page.
1902          * @return bool
1903          */
1904         function areSubpagesAllowed() {
1905                 # Some namespaces don't allow subpages
1906                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1907         }
1908
1909         /**
1910          * Handle link to subpage if necessary
1911          * @param string $target the source of the link
1912          * @param string &$text the link text, modified as necessary
1913          * @return string the full name of the link
1914          * @private
1915          */
1916         function maybeDoSubpageLink($target, &$text) {
1917                 # Valid link forms:
1918                 # Foobar -- normal
1919                 # :Foobar -- override special treatment of prefix (images, language links)
1920                 # /Foobar -- convert to CurrentPage/Foobar
1921                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1922                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1923                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1924
1925                 wfProfileIn( __METHOD__ );
1926                 $ret = $target; # default return value is no change
1927
1928                 # Some namespaces don't allow subpages,
1929                 # so only perform processing if subpages are allowed
1930                 if( $this->areSubpagesAllowed() ) {
1931                         $hash = strpos( $target, '#' );
1932                         if( $hash !== false ) {
1933                                 $suffix = substr( $target, $hash );
1934                                 $target = substr( $target, 0, $hash );
1935                         } else {
1936                                 $suffix = '';
1937                         }
1938                         # bug 7425
1939                         $target = trim( $target );
1940                         # Look at the first character
1941                         if( $target != '' && $target{0} === '/' ) {
1942                                 # / at end means we don't want the slash to be shown
1943                                 $m = array();
1944                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1945                                 if( $trailingSlashes ) {
1946                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1947                                 } else {
1948                                         $noslash = substr( $target, 1 );
1949                                 }
1950
1951                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1952                                 if( '' === $text ) {
1953                                         $text = $target . $suffix;
1954                                 } # this might be changed for ugliness reasons
1955                         } else {
1956                                 # check for .. subpage backlinks
1957                                 $dotdotcount = 0;
1958                                 $nodotdot = $target;
1959                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1960                                         ++$dotdotcount;
1961                                         $nodotdot = substr( $nodotdot, 3 );
1962                                 }
1963                                 if($dotdotcount > 0) {
1964                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1965                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1966                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1967                                                 # / at the end means don't show full path
1968                                                 if( substr( $nodotdot, -1, 1 ) === '/' ) {
1969                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1970                                                         if( '' === $text ) {
1971                                                                 $text = $nodotdot . $suffix;
1972                                                         }
1973                                                 }
1974                                                 $nodotdot = trim( $nodotdot );
1975                                                 if( $nodotdot != '' ) {
1976                                                         $ret .= '/' . $nodotdot;
1977                                                 }
1978                                                 $ret .= $suffix;
1979                                         }
1980                                 }
1981                         }
1982                 }
1983
1984                 wfProfileOut( __METHOD__ );
1985                 return $ret;
1986         }
1987
1988         /**#@+
1989          * Used by doBlockLevels()
1990          * @private
1991          */
1992         /* private */ function closeParagraph() {
1993                 $result = '';
1994                 if ( '' != $this->mLastSection ) {
1995                         $result = '</' . $this->mLastSection  . ">\n";
1996                 }
1997                 $this->mInPre = false;
1998                 $this->mLastSection = '';
1999                 return $result;
2000         }
2001         # getCommon() returns the length of the longest common substring
2002         # of both arguments, starting at the beginning of both.
2003         #
2004         /* private */ function getCommon( $st1, $st2 ) {
2005                 $fl = strlen( $st1 );
2006                 $shorter = strlen( $st2 );
2007                 if ( $fl < $shorter ) { $shorter = $fl; }
2008
2009                 for ( $i = 0; $i < $shorter; ++$i ) {
2010                         if ( $st1{$i} != $st2{$i} ) { break; }
2011                 }
2012                 return $i;
2013         }
2014         # These next three functions open, continue, and close the list
2015         # element appropriate to the prefix character passed into them.
2016         #
2017         /* private */ function openList( $char ) {
2018                 $result = $this->closeParagraph();
2019
2020                 if ( '*' === $char ) { $result .= '<ul><li>'; }
2021                 else if ( '#' === $char ) { $result .= '<ol><li>'; }
2022                 else if ( ':' === $char ) { $result .= '<dl><dd>'; }
2023                 else if ( ';' === $char ) {
2024                         $result .= '<dl><dt>';
2025                         $this->mDTopen = true;
2026                 }
2027                 else { $result = '<!-- ERR 1 -->'; }
2028
2029                 return $result;
2030         }
2031
2032         /* private */ function nextItem( $char ) {
2033                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
2034                 else if ( ':' === $char || ';' === $char ) {
2035                         $close = '</dd>';
2036                         if ( $this->mDTopen ) { $close = '</dt>'; }
2037                         if ( ';' === $char ) {
2038                                 $this->mDTopen = true;
2039                                 return $close . '<dt>';
2040                         } else {
2041                                 $this->mDTopen = false;
2042                                 return $close . '<dd>';
2043                         }
2044                 }
2045                 return '<!-- ERR 2 -->';
2046         }
2047
2048         /* private */ function closeList( $char ) {
2049                 if ( '*' === $char ) { $text = '</li></ul>'; }
2050                 else if ( '#' === $char ) { $text = '</li></ol>'; }
2051                 else if ( ':' === $char ) {
2052                         if ( $this->mDTopen ) {
2053                                 $this->mDTopen = false;
2054                                 $text = '</dt></dl>';
2055                         } else {
2056                                 $text = '</dd></dl>';
2057                         }
2058                 }
2059                 else {  return '<!-- ERR 3 -->'; }
2060                 return $text."\n";
2061         }
2062         /**#@-*/
2063
2064         /**
2065          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2066          *
2067          * @private
2068          * @return string the lists rendered as HTML
2069          */
2070         function doBlockLevels( $text, $linestart ) {
2071                 wfProfileIn( __METHOD__ );
2072
2073                 # Parsing through the text line by line.  The main thing
2074                 # happening here is handling of block-level elements p, pre,
2075                 # and making lists from lines starting with * # : etc.
2076                 #
2077                 $textLines = StringUtils::explode( "\n", $text );
2078
2079                 $lastPrefix = $output = '';
2080                 $this->mDTopen = $inBlockElem = false;
2081                 $prefixLength = 0;
2082                 $paragraphStack = false;
2083
2084                 foreach ( $textLines as $oLine ) {
2085                         # Fix up $linestart
2086                         if ( !$linestart ) {
2087                                 $output .= $oLine;
2088                                 $linestart = true;
2089                                 continue;
2090                         }
2091
2092                         $lastPrefixLength = strlen( $lastPrefix );
2093                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2094                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2095                         if ( !$this->mInPre ) {
2096                                 # Multiple prefixes may abut each other for nested lists.
2097                                 $prefixLength = strspn( $oLine, '*#:;' );
2098                                 $prefix = substr( $oLine, 0, $prefixLength );
2099
2100                                 # eh?
2101                                 $prefix2 = str_replace( ';', ':', $prefix );
2102                                 $t = substr( $oLine, $prefixLength );
2103                                 $this->mInPre = (bool)$preOpenMatch;
2104                         } else {
2105                                 # Don't interpret any other prefixes in preformatted text
2106                                 $prefixLength = 0;
2107                                 $prefix = $prefix2 = '';
2108                                 $t = $oLine;
2109                         }
2110
2111                         # List generation
2112                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2113                                 # Same as the last item, so no need to deal with nesting or opening stuff
2114                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2115                                 $paragraphStack = false;
2116
2117                                 if ( substr( $prefix, -1 ) === ';') {
2118                                         # The one nasty exception: definition lists work like this:
2119                                         # ; title : definition text
2120                                         # So we check for : in the remainder text to split up the
2121                                         # title and definition, without b0rking links.
2122                                         $term = $t2 = '';
2123                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2124                                                 $t = $t2;
2125                                                 $output .= $term . $this->nextItem( ':' );
2126                                         }
2127                                 }
2128                         } elseif( $prefixLength || $lastPrefixLength ) {
2129                                 # Either open or close a level...
2130                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2131                                 $paragraphStack = false;
2132
2133                                 while( $commonPrefixLength < $lastPrefixLength ) {
2134                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2135                                         --$lastPrefixLength;
2136                                 }
2137                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2138                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2139                                 }
2140                                 while ( $prefixLength > $commonPrefixLength ) {
2141                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2142                                         $output .= $this->openList( $char );
2143
2144                                         if ( ';' === $char ) {
2145                                                 # FIXME: This is dupe of code above
2146                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2147                                                         $t = $t2;
2148                                                         $output .= $term . $this->nextItem( ':' );
2149                                                 }
2150                                         }
2151                                         ++$commonPrefixLength;
2152                                 }
2153                                 $lastPrefix = $prefix2;
2154                         }
2155                         if( 0 == $prefixLength ) {
2156                                 wfProfileIn( __METHOD__."-paragraph" );
2157                                 # No prefix (not in list)--go to paragraph mode
2158                                 // XXX: use a stack for nestable elements like span, table and div
2159                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2160                                 $closematch = preg_match(
2161                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2162                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2163                                 if ( $openmatch or $closematch ) {
2164                                         $paragraphStack = false;
2165                                         # TODO bug 5718: paragraph closed
2166                                         $output .= $this->closeParagraph();
2167                                         if ( $preOpenMatch and !$preCloseMatch ) {
2168                                                 $this->mInPre = true;
2169                                         }
2170                                         if ( $closematch ) {
2171                                                 $inBlockElem = false;
2172                                         } else {
2173                                                 $inBlockElem = true;
2174                                         }
2175                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2176                                         if ( ' ' == $t{0} and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2177                                                 // pre
2178                                                 if ($this->mLastSection !== 'pre') {
2179                                                         $paragraphStack = false;
2180                                                         $output .= $this->closeParagraph().'<pre>';
2181                                                         $this->mLastSection = 'pre';
2182                                                 }
2183                                                 $t = substr( $t, 1 );
2184                                         } else {
2185                                                 // paragraph
2186                                                 if ( '' == trim($t) ) {
2187                                                         if ( $paragraphStack ) {
2188                                                                 $output .= $paragraphStack.'<br />';
2189                                                                 $paragraphStack = false;
2190                                                                 $this->mLastSection = 'p';
2191                                                         } else {
2192                                                                 if ($this->mLastSection !== 'p' ) {
2193                                                                         $output .= $this->closeParagraph();
2194                                                                         $this->mLastSection = '';
2195                                                                         $paragraphStack = '<p>';
2196                                                                 } else {
2197                                                                         $paragraphStack = '</p><p>';
2198                                                                 }
2199                                                         }
2200                                                 } else {
2201                                                         if ( $paragraphStack ) {
2202                                                                 $output .= $paragraphStack;
2203                                                                 $paragraphStack = false;
2204                                                                 $this->mLastSection = 'p';
2205                                                         } else if ($this->mLastSection !== 'p') {
2206                                                                 $output .= $this->closeParagraph().'<p>';
2207                                                                 $this->mLastSection = 'p';
2208                                                         }
2209                                                 }
2210                                         }
2211                                 }
2212                                 wfProfileOut( __METHOD__."-paragraph" );
2213                         }
2214                         // somewhere above we forget to get out of pre block (bug 785)
2215                         if($preCloseMatch && $this->mInPre) {
2216                                 $this->mInPre = false;
2217                         }
2218                         if ($paragraphStack === false) {
2219                                 $output .= $t."\n";
2220                         }
2221                 }
2222                 while ( $prefixLength ) {
2223                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2224                         --$prefixLength;
2225                 }
2226                 if ( '' != $this->mLastSection ) {
2227                         $output .= '</' . $this->mLastSection . '>';
2228                         $this->mLastSection = '';
2229                 }
2230
2231                 wfProfileOut( __METHOD__ );
2232                 return $output;
2233         }
2234
2235         /**
2236          * Split up a string on ':', ignoring any occurences inside tags
2237          * to prevent illegal overlapping.
2238          * @param string $str the string to split
2239          * @param string &$before set to everything before the ':'
2240          * @param string &$after set to everything after the ':'
2241          * return string the position of the ':', or false if none found
2242          */
2243         function findColonNoLinks($str, &$before, &$after) {
2244                 wfProfileIn( __METHOD__ );
2245
2246                 $pos = strpos( $str, ':' );
2247                 if( $pos === false ) {
2248                         // Nothing to find!
2249                         wfProfileOut( __METHOD__ );
2250                         return false;
2251                 }
2252
2253                 $lt = strpos( $str, '<' );
2254                 if( $lt === false || $lt > $pos ) {
2255                         // Easy; no tag nesting to worry about
2256                         $before = substr( $str, 0, $pos );
2257                         $after = substr( $str, $pos+1 );
2258                         wfProfileOut( __METHOD__ );
2259                         return $pos;
2260                 }
2261
2262                 // Ugly state machine to walk through avoiding tags.
2263                 $state = self::COLON_STATE_TEXT;
2264                 $stack = 0;
2265                 $len = strlen( $str );
2266                 for( $i = 0; $i < $len; $i++ ) {
2267                         $c = $str{$i};
2268
2269                         switch( $state ) {
2270                         // (Using the number is a performance hack for common cases)
2271                         case 0: // self::COLON_STATE_TEXT:
2272                                 switch( $c ) {
2273                                 case "<":
2274                                         // Could be either a <start> tag or an </end> tag
2275                                         $state = self::COLON_STATE_TAGSTART;
2276                                         break;
2277                                 case ":":
2278                                         if( $stack == 0 ) {
2279                                                 // We found it!
2280                                                 $before = substr( $str, 0, $i );
2281                                                 $after = substr( $str, $i + 1 );
2282                                                 wfProfileOut( __METHOD__ );
2283                                                 return $i;
2284                                         }
2285                                         // Embedded in a tag; don't break it.
2286                                         break;
2287                                 default:
2288                                         // Skip ahead looking for something interesting
2289                                         $colon = strpos( $str, ':', $i );
2290                                         if( $colon === false ) {
2291                                                 // Nothing else interesting
2292                                                 wfProfileOut( __METHOD__ );
2293                                                 return false;
2294                                         }
2295                                         $lt = strpos( $str, '<', $i );
2296                                         if( $stack === 0 ) {
2297                                                 if( $lt === false || $colon < $lt ) {
2298                                                         // We found it!
2299                                                         $before = substr( $str, 0, $colon );
2300                                                         $after = substr( $str, $colon + 1 );
2301                                                         wfProfileOut( __METHOD__ );
2302                                                         return $i;
2303                                                 }
2304                                         }
2305                                         if( $lt === false ) {
2306                                                 // Nothing else interesting to find; abort!
2307                                                 // We're nested, but there's no close tags left. Abort!
2308                                                 break 2;
2309                                         }
2310                                         // Skip ahead to next tag start
2311                                         $i = $lt;
2312                                         $state = self::COLON_STATE_TAGSTART;
2313                                 }
2314                                 break;
2315                         case 1: // self::COLON_STATE_TAG:
2316                                 // In a <tag>
2317                                 switch( $c ) {
2318                                 case ">":
2319                                         $stack++;
2320                                         $state = self::COLON_STATE_TEXT;
2321                                         break;
2322                                 case "/":
2323                                         // Slash may be followed by >?
2324                                         $state = self::COLON_STATE_TAGSLASH;
2325                                         break;
2326                                 default:
2327                                         // ignore
2328                                 }
2329                                 break;
2330                         case 2: // self::COLON_STATE_TAGSTART:
2331                                 switch( $c ) {
2332                                 case "/":
2333                                         $state = self::COLON_STATE_CLOSETAG;
2334                                         break;
2335                                 case "!":
2336                                         $state = self::COLON_STATE_COMMENT;
2337                                         break;
2338                                 case ">":
2339                                         // Illegal early close? This shouldn't happen D:
2340                                         $state = self::COLON_STATE_TEXT;
2341                                         break;
2342                                 default:
2343                                         $state = self::COLON_STATE_TAG;
2344                                 }
2345                                 break;
2346                         case 3: // self::COLON_STATE_CLOSETAG:
2347                                 // In a </tag>
2348                                 if( $c === ">" ) {
2349                                         $stack--;
2350                                         if( $stack < 0 ) {
2351                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2352                                                 wfProfileOut( __METHOD__ );
2353                                                 return false;
2354                                         }
2355                                         $state = self::COLON_STATE_TEXT;
2356                                 }
2357                                 break;
2358                         case self::COLON_STATE_TAGSLASH:
2359                                 if( $c === ">" ) {
2360                                         // Yes, a self-closed tag <blah/>
2361                                         $state = self::COLON_STATE_TEXT;
2362                                 } else {
2363                                         // Probably we're jumping the gun, and this is an attribute
2364                                         $state = self::COLON_STATE_TAG;
2365                                 }
2366                                 break;
2367                         case 5: // self::COLON_STATE_COMMENT:
2368                                 if( $c === "-" ) {
2369                                         $state = self::COLON_STATE_COMMENTDASH;
2370                                 }
2371                                 break;
2372                         case self::COLON_STATE_COMMENTDASH:
2373                                 if( $c === "-" ) {
2374                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2375                                 } else {
2376                                         $state = self::COLON_STATE_COMMENT;
2377                                 }
2378                                 break;
2379                         case self::COLON_STATE_COMMENTDASHDASH:
2380                                 if( $c === ">" ) {
2381                                         $state = self::COLON_STATE_TEXT;
2382                                 } else {
2383                                         $state = self::COLON_STATE_COMMENT;
2384                                 }
2385                                 break;
2386                         default:
2387                                 throw new MWException( "State machine error in " . __METHOD__ );
2388                         }
2389                 }
2390                 if( $stack > 0 ) {
2391                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2392                         return false;
2393                 }
2394                 wfProfileOut( __METHOD__ );
2395                 return false;
2396         }
2397
2398         /**
2399          * Return value of a magic variable (like PAGENAME)
2400          *
2401          * @private
2402          */
2403         function getVariableValue( $index ) {
2404                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2405
2406                 /**
2407                  * Some of these require message or data lookups and can be
2408                  * expensive to check many times.
2409                  */
2410                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2411                         if ( isset( $this->mVarCache[$index] ) ) {
2412                                 return $this->mVarCache[$index];
2413                         }
2414                 }
2415
2416                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2417                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2418
2419                 # Use the time zone
2420                 global $wgLocaltimezone;
2421                 if ( isset( $wgLocaltimezone ) ) {
2422                         $oldtz = getenv( 'TZ' );
2423                         putenv( 'TZ='.$wgLocaltimezone );
2424                 }
2425
2426                 wfSuppressWarnings(); // E_STRICT system time bitching
2427                 $localTimestamp = date( 'YmdHis', $ts );
2428                 $localMonth = date( 'm', $ts );
2429                 $localMonthName = date( 'n', $ts );
2430                 $localDay = date( 'j', $ts );
2431                 $localDay2 = date( 'd', $ts );
2432                 $localDayOfWeek = date( 'w', $ts );
2433                 $localWeek = date( 'W', $ts );
2434                 $localYear = date( 'Y', $ts );
2435                 $localHour = date( 'H', $ts );
2436                 if ( isset( $wgLocaltimezone ) ) {
2437                         putenv( 'TZ='.$oldtz );
2438                 }
2439                 wfRestoreWarnings();
2440
2441                 switch ( $index ) {
2442                         case 'currentmonth':
2443                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2444                         case 'currentmonthname':
2445                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2446                         case 'currentmonthnamegen':
2447                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2448                         case 'currentmonthabbrev':
2449                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2450                         case 'currentday':
2451                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2452                         case 'currentday2':
2453                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2454                         case 'localmonth':
2455                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2456                         case 'localmonthname':
2457                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2458                         case 'localmonthnamegen':
2459                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2460                         case 'localmonthabbrev':
2461                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2462                         case 'localday':
2463                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2464                         case 'localday2':
2465                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2466                         case 'pagename':
2467                                 return wfEscapeWikiText( $this->mTitle->getText() );
2468                         case 'pagenamee':
2469                                 return $this->mTitle->getPartialURL();
2470                         case 'fullpagename':
2471                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2472                         case 'fullpagenamee':
2473                                 return $this->mTitle->getPrefixedURL();
2474                         case 'subpagename':
2475                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2476                         case 'subpagenamee':
2477                                 return $this->mTitle->getSubpageUrlForm();
2478                         case 'basepagename':
2479                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2480                         case 'basepagenamee':
2481                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2482                         case 'talkpagename':
2483                                 if( $this->mTitle->canTalk() ) {
2484                                         $talkPage = $this->mTitle->getTalkPage();
2485                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2486                                 } else {
2487                                         return '';
2488                                 }
2489                         case 'talkpagenamee':
2490                                 if( $this->mTitle->canTalk() ) {
2491                                         $talkPage = $this->mTitle->getTalkPage();
2492                                         return $talkPage->getPrefixedUrl();
2493                                 } else {
2494                                         return '';
2495                                 }
2496                         case 'subjectpagename':
2497                                 $subjPage = $this->mTitle->getSubjectPage();
2498                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2499                         case 'subjectpagenamee':
2500                                 $subjPage = $this->mTitle->getSubjectPage();
2501                                 return $subjPage->getPrefixedUrl();
2502                         case 'revisionid':
2503                                 // Let the edit saving system know we should parse the page
2504                                 // *after* a revision ID has been assigned.
2505                                 $this->mOutput->setFlag( 'vary-revision' );
2506                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2507                                 return $this->mRevisionId;
2508                         case 'revisionday':
2509                                 // Let the edit saving system know we should parse the page
2510                                 // *after* a revision ID has been assigned. This is for null edits.
2511                                 $this->mOutput->setFlag( 'vary-revision' );
2512                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2513                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2514                         case 'revisionday2':
2515                                 // Let the edit saving system know we should parse the page
2516                                 // *after* a revision ID has been assigned. This is for null edits.
2517                                 $this->mOutput->setFlag( 'vary-revision' );
2518                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2519                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2520                         case 'revisionmonth':
2521                                 // Let the edit saving system know we should parse the page
2522                                 // *after* a revision ID has been assigned. This is for null edits.
2523                                 $this->mOutput->setFlag( 'vary-revision' );
2524                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2525                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2526                         case 'revisionyear':
2527                                 // Let the edit saving system know we should parse the page
2528                                 // *after* a revision ID has been assigned. This is for null edits.
2529                                 $this->mOutput->setFlag( 'vary-revision' );
2530                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2531                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2532                         case 'revisiontimestamp':
2533                                 // Let the edit saving system know we should parse the page
2534                                 // *after* a revision ID has been assigned. This is for null edits.
2535                                 $this->mOutput->setFlag( 'vary-revision' );
2536                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2537                                 return $this->getRevisionTimestamp();
2538                         case 'namespace':
2539                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2540                         case 'namespacee':
2541                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2542                         case 'talkspace':
2543                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2544                         case 'talkspacee':
2545                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2546                         case 'subjectspace':
2547                                 return $this->mTitle->getSubjectNsText();
2548                         case 'subjectspacee':
2549                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2550                         case 'currentdayname':
2551                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2552                         case 'currentyear':
2553                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2554                         case 'currenttime':
2555                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2556                         case 'currenthour':
2557                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2558                         case 'currentweek':
2559                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2560                                 // int to remove the padding
2561                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2562                         case 'currentdow':
2563                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2564                         case 'localdayname':
2565                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2566                         case 'localyear':
2567                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2568                         case 'localtime':
2569                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2570                         case 'localhour':
2571                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2572                         case 'localweek':
2573                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2574                                 // int to remove the padding
2575                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2576                         case 'localdow':
2577                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2578                         case 'numberofarticles':
2579                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2580                         case 'numberoffiles':
2581                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2582                         case 'numberofusers':
2583                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2584                         case 'numberofpages':
2585                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2586                         case 'numberofadmins':
2587                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2588                         case 'numberofedits':
2589                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2590                         case 'currenttimestamp':
2591                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2592                         case 'localtimestamp':
2593                                 return $this->mVarCache[$index] = $localTimestamp;
2594                         case 'currentversion':
2595                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2596                         case 'sitename':
2597                                 return $wgSitename;
2598                         case 'server':
2599                                 return $wgServer;
2600                         case 'servername':
2601                                 return $wgServerName;
2602                         case 'scriptpath':
2603                                 return $wgScriptPath;
2604                         case 'directionmark':
2605                                 return $wgContLang->getDirMark();
2606                         case 'contentlanguage':
2607                                 global $wgContLanguageCode;
2608                                 return $wgContLanguageCode;
2609                         default:
2610                                 $ret = null;
2611                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2612                                         return $ret;
2613                                 else
2614                                         return null;
2615                 }
2616         }
2617
2618         /**
2619          * initialise the magic variables (like CURRENTMONTHNAME)
2620          *
2621          * @private
2622          */
2623         function initialiseVariables() {
2624                 wfProfileIn( __METHOD__ );
2625                 $variableIDs = MagicWord::getVariableIDs();
2626
2627                 $this->mVariables = new MagicWordArray( $variableIDs );
2628                 wfProfileOut( __METHOD__ );
2629         }
2630
2631         /**
2632          * Preprocess some wikitext and return the document tree.
2633          * This is the ghost of replace_variables().
2634          *
2635          * @param string $text The text to parse
2636          * @param integer flags Bitwise combination of:
2637          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2638          *                                     included. Default is to assume a direct page view.
2639          *
2640          * The generated DOM tree must depend only on the input text and the flags.
2641          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2642          *
2643          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2644          * change in the DOM tree for a given text, must be passed through the section identifier
2645          * in the section edit link and thus back to extractSections().
2646          *
2647          * The output of this function is currently only cached in process memory, but a persistent
2648          * cache may be implemented at a later date which takes further advantage of these strict
2649          * dependency requirements.
2650          *
2651          * @private
2652          */
2653         function preprocessToDom ( $text, $flags = 0 ) {
2654                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2655                 return $dom;
2656         }
2657
2658         /*
2659          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2660          */
2661         public static function splitWhitespace( $s ) {
2662                 $ltrimmed = ltrim( $s );
2663                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2664                 $trimmed = rtrim( $ltrimmed );
2665                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2666                 if ( $diff > 0 ) {
2667                         $w2 = substr( $ltrimmed, -$diff );
2668                 } else {
2669                         $w2 = '';
2670                 }
2671                 return array( $w1, $trimmed, $w2 );
2672         }
2673
2674         /**
2675          * Replace magic variables, templates, and template arguments
2676          * with the appropriate text. Templates are substituted recursively,
2677          * taking care to avoid infinite loops.
2678          *
2679          * Note that the substitution depends on value of $mOutputType:
2680          *  self::OT_WIKI: only {{subst:}} templates
2681          *  self::OT_PREPROCESS: templates but not extension tags
2682          *  self::OT_HTML: all templates and extension tags
2683          *
2684          * @param string $tex The text to transform
2685          * @param PPFrame $frame Object describing the arguments passed to the template.
2686          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2687          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2688          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2689          * @private
2690          */
2691         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2692                 # Prevent too big inclusions
2693                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2694                         return $text;
2695                 }
2696
2697                 wfProfileIn( __METHOD__ );
2698
2699                 if ( $frame === false ) {
2700                         $frame = $this->getPreprocessor()->newFrame();
2701                 } elseif ( !( $frame instanceof PPFrame ) ) {
2702                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2703                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2704                 }
2705
2706                 $dom = $this->preprocessToDom( $text );
2707                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2708                 $text = $frame->expand( $dom, $flags );
2709
2710                 wfProfileOut( __METHOD__ );
2711                 return $text;
2712         }
2713
2714         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2715         static function createAssocArgs( $args ) {
2716                 $assocArgs = array();
2717                 $index = 1;
2718                 foreach( $args as $arg ) {
2719                         $eqpos = strpos( $arg, '=' );
2720                         if ( $eqpos === false ) {
2721                                 $assocArgs[$index++] = $arg;
2722                         } else {
2723                                 $name = trim( substr( $arg, 0, $eqpos ) );
2724                                 $value = trim( substr( $arg, $eqpos+1 ) );
2725                                 if ( $value === false ) {
2726                                         $value = '';
2727                                 }
2728                                 if ( $name !== false ) {
2729                                         $assocArgs[$name] = $value;
2730                                 }
2731                         }
2732                 }
2733
2734                 return $assocArgs;
2735         }
2736
2737         /**
2738          * Warn the user when a parser limitation is reached
2739          * Will warn at most once the user per limitation type
2740          *
2741          * @param string $limitationType, should be one of:
2742          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2743          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2744          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2745          * @params int $current, $max When an explicit limit has been
2746          *       exceeded, provide the values (optional)
2747          */
2748         function limitationWarn( $limitationType, $current=null, $max=null) {
2749                 $msgName = $limitationType . '-warning';
2750                 //does no harm if $current and $max are present but are unnecessary for the message
2751                 $warning = wfMsgExt( $msgName, array( 'parsemag', 'escape' ), $current, $max );
2752                 $this->mOutput->addWarning( $warning );
2753                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2754                 if ( $cat ) {
2755                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2756                 }
2757         }
2758
2759         /**
2760          * Return the text of a template, after recursively
2761          * replacing any variables or templates within the template.
2762          *
2763          * @param array $piece The parts of the template
2764          *  $piece['title']: the title, i.e. the part before the |
2765          *  $piece['parts']: the parameter array
2766          *  $piece['lineStart']: whether the brace was at the start of a line
2767          * @param PPFrame The current frame, contains template arguments
2768          * @return string the text of the template
2769          * @private
2770          */
2771         function braceSubstitution( $piece, $frame ) {
2772                 global $wgContLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2773                 wfProfileIn( __METHOD__ );
2774                 wfProfileIn( __METHOD__.'-setup' );
2775
2776                 # Flags
2777                 $found = false;             # $text has been filled
2778                 $nowiki = false;            # wiki markup in $text should be escaped
2779                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2780                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2781                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2782                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2783
2784                 # Title object, where $text came from
2785                 $title = NULL;
2786
2787                 # $part1 is the bit before the first |, and must contain only title characters.
2788                 # Various prefixes will be stripped from it later.
2789                 $titleWithSpaces = $frame->expand( $piece['title'] );
2790                 $part1 = trim( $titleWithSpaces );
2791                 $titleText = false;
2792
2793                 # Original title text preserved for various purposes
2794                 $originalTitle = $part1;
2795
2796                 # $args is a list of argument nodes, starting from index 0, not including $part1
2797                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2798                 wfProfileOut( __METHOD__.'-setup' );
2799
2800                 # SUBST
2801                 wfProfileIn( __METHOD__.'-modifiers' );
2802                 if ( !$found ) {
2803                         $mwSubst = MagicWord::get( 'subst' );
2804                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2805                                 # One of two possibilities is true:
2806                                 # 1) Found SUBST but not in the PST phase
2807                                 # 2) Didn't find SUBST and in the PST phase
2808                                 # In either case, return without further processing
2809                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2810                                 $isLocalObj = true;
2811                                 $found = true;
2812                         }
2813                 }
2814
2815                 # Variables
2816                 if ( !$found && $args->getLength() == 0 ) {
2817                         $id = $this->mVariables->matchStartToEnd( $part1 );
2818                         if ( $id !== false ) {
2819                                 $text = $this->getVariableValue( $id );
2820                                 if (MagicWord::getCacheTTL($id)>-1)
2821                                         $this->mOutput->mContainsOldMagic = true;
2822                                 $found = true;
2823                         }
2824                 }
2825
2826                 # MSG, MSGNW and RAW
2827                 if ( !$found ) {
2828                         # Check for MSGNW:
2829                         $mwMsgnw = MagicWord::get( 'msgnw' );
2830                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2831                                 $nowiki = true;
2832                         } else {
2833                                 # Remove obsolete MSG:
2834                                 $mwMsg = MagicWord::get( 'msg' );
2835                                 $mwMsg->matchStartAndRemove( $part1 );
2836                         }
2837
2838                         # Check for RAW:
2839                         $mwRaw = MagicWord::get( 'raw' );
2840                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2841                                 $forceRawInterwiki = true;
2842                         }
2843                 }
2844                 wfProfileOut( __METHOD__.'-modifiers' );
2845
2846                 # Parser functions
2847                 if ( !$found ) {
2848                         wfProfileIn( __METHOD__ . '-pfunc' );
2849
2850                         $colonPos = strpos( $part1, ':' );
2851                         if ( $colonPos !== false ) {
2852                                 # Case sensitive functions
2853                                 $function = substr( $part1, 0, $colonPos );
2854                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2855                                         $function = $this->mFunctionSynonyms[1][$function];
2856                                 } else {
2857                                         # Case insensitive functions
2858                                         $function = strtolower( $function );
2859                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2860                                                 $function = $this->mFunctionSynonyms[0][$function];
2861                                         } else {
2862                                                 $function = false;
2863                                         }
2864                                 }
2865                                 if ( $function ) {
2866                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2867                                         $initialArgs = array( &$this );
2868                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2869                                         if ( $flags & SFH_OBJECT_ARGS ) {
2870                                                 # Add a frame parameter, and pass the arguments as an array
2871                                                 $allArgs = $initialArgs;
2872                                                 $allArgs[] = $frame;
2873                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2874                                                         $funcArgs[] = $args->item( $i );
2875                                                 }
2876                                                 $allArgs[] = $funcArgs;
2877                                         } else {
2878                                                 # Convert arguments to plain text
2879                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2880                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2881                                                 }
2882                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2883                                         }
2884
2885                                         # Workaround for PHP bug 35229 and similar
2886                                         if ( !is_callable( $callback ) ) {
2887                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2888                                         }
2889                                         $result = call_user_func_array( $callback, $allArgs );
2890                                         $found = true;
2891                                         $noparse = true;
2892                                         $preprocessFlags = 0;
2893
2894                                         if ( is_array( $result ) ) {
2895                                                 if ( isset( $result[0] ) ) {
2896                                                         $text = $result[0];
2897                                                         unset( $result[0] );
2898                                                 }
2899
2900                                                 // Extract flags into the local scope
2901                                                 // This allows callers to set flags such as nowiki, found, etc.
2902                                                 extract( $result );
2903                                         } else {
2904                                                 $text = $result;
2905                                         }
2906                                         if ( !$noparse ) {
2907                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2908                                                 $isChildObj = true;
2909                                         }
2910                                 }
2911                         }
2912                         wfProfileOut( __METHOD__ . '-pfunc' );
2913                 }
2914
2915                 # Finish mangling title and then check for loops.
2916                 # Set $title to a Title object and $titleText to the PDBK
2917                 if ( !$found ) {
2918                         $ns = NS_TEMPLATE;
2919                         # Split the title into page and subpage
2920                         $subpage = '';
2921                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2922                         if ($subpage !== '') {
2923                                 $ns = $this->mTitle->getNamespace();
2924                         }
2925                         $title = Title::newFromText( $part1, $ns );
2926                         if ( $title ) {
2927                                 $titleText = $title->getPrefixedText();
2928                                 # Check for language variants if the template is not found
2929                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2930                                         $wgContLang->findVariantLink( $part1, $title, true );
2931                                 }
2932                                 # Do infinite loop check
2933                                 if ( !$frame->loopCheck( $title ) ) {
2934                                         $found = true;
2935                                         $text = "<span class=\"error\">Template loop detected: [[$titleText]]</span>";
2936                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2937                                 }
2938                                 # Do recursion depth check
2939                                 $limit = $this->mOptions->getMaxTemplateDepth();
2940                                 if ( $frame->depth >= $limit ) {
2941                                         $found = true;
2942                                         $text = "<span class=\"error\">Template recursion depth limit exceeded ($limit)</span>";
2943                                 }
2944                         }
2945                 }
2946
2947                 # Load from database
2948                 if ( !$found && $title ) {
2949                         wfProfileIn( __METHOD__ . '-loadtpl' );
2950                         if ( !$title->isExternal() ) {
2951                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2952                                         $text = SpecialPage::capturePath( $title );
2953                                         if ( is_string( $text ) ) {
2954                                                 $found = true;
2955                                                 $isHTML = true;
2956                                                 $this->disableCache();
2957                                         }
2958                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2959                                         $found = false; //access denied
2960                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2961                                 } else {
2962                                         list( $text, $title ) = $this->getTemplateDom( $title );
2963                                         if ( $text !== false ) {
2964                                                 $found = true;
2965                                                 $isChildObj = true;
2966                                         }
2967                                 }
2968
2969                                 # If the title is valid but undisplayable, make a link to it
2970                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2971                                         $text = "[[:$titleText]]";
2972                                         $found = true;
2973                                 }
2974                         } elseif ( $title->isTrans() ) {
2975                                 // Interwiki transclusion
2976                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2977                                         $text = $this->interwikiTransclude( $title, 'render' );
2978                                         $isHTML = true;
2979                                 } else {
2980                                         $text = $this->interwikiTransclude( $title, 'raw' );
2981                                         // Preprocess it like a template
2982                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2983                                         $isChildObj = true;
2984                                 }
2985                                 $found = true;
2986                         }
2987                         wfProfileOut( __METHOD__ . '-loadtpl' );
2988                 }
2989
2990                 # If we haven't found text to substitute by now, we're done
2991                 # Recover the source wikitext and return it
2992                 if ( !$found ) {
2993                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2994                         wfProfileOut( __METHOD__ );
2995                         return array( 'object' => $text );
2996                 }
2997
2998                 # Expand DOM-style return values in a child frame
2999                 if ( $isChildObj ) {
3000                         # Clean up argument array
3001                         $newFrame = $frame->newChild( $args, $title );
3002
3003                         if ( $nowiki ) {
3004                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3005                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3006                                 # Expansion is eligible for the empty-frame cache
3007                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3008                                         $text = $this->mTplExpandCache[$titleText];
3009                                 } else {
3010                                         $text = $newFrame->expand( $text );
3011                                         $this->mTplExpandCache[$titleText] = $text;
3012                                 }
3013                         } else {
3014                                 # Uncached expansion
3015                                 $text = $newFrame->expand( $text );
3016                         }
3017                 }
3018                 if ( $isLocalObj && $nowiki ) {
3019                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3020                         $isLocalObj = false;
3021                 }
3022
3023                 # Replace raw HTML by a placeholder
3024                 # Add a blank line preceding, to prevent it from mucking up
3025                 # immediately preceding headings
3026                 if ( $isHTML ) {
3027                         $text = "\n\n" . $this->insertStripItem( $text );
3028                 }
3029                 # Escape nowiki-style return values
3030                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3031                         $text = wfEscapeWikiText( $text );
3032                 }
3033                 # Bug 529: if the template begins with a table or block-level
3034                 # element, it should be treated as beginning a new line.
3035                 # This behaviour is somewhat controversial.
3036                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3037                         $text = "\n" . $text;
3038                 }
3039
3040                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3041                         # Error, oversize inclusion
3042                         $text = "[[$originalTitle]]" .
3043                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3044                         $this->limitationWarn( 'post-expand-template-inclusion' );
3045                 }
3046
3047                 if ( $isLocalObj ) {
3048                         $ret = array( 'object' => $text );
3049                 } else {
3050                         $ret = array( 'text' => $text );
3051                 }
3052
3053                 wfProfileOut( __METHOD__ );
3054                 return $ret;
3055         }
3056
3057         /**
3058          * Get the semi-parsed DOM representation of a template with a given title,
3059          * and its redirect destination title. Cached.
3060          */
3061         function getTemplateDom( $title ) {
3062                 $cacheTitle = $title;
3063                 $titleText = $title->getPrefixedDBkey();
3064
3065                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3066                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3067                         $title = Title::makeTitle( $ns, $dbk );
3068                         $titleText = $title->getPrefixedDBkey();
3069                 }
3070                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3071                         return array( $this->mTplDomCache[$titleText], $title );
3072                 }
3073
3074                 // Cache miss, go to the database
3075                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3076
3077                 if ( $text === false ) {
3078                         $this->mTplDomCache[$titleText] = false;
3079                         return array( false, $title );
3080                 }
3081
3082                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3083                 $this->mTplDomCache[ $titleText ] = $dom;
3084
3085                 if (! $title->equals($cacheTitle)) {
3086                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3087                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3088                 }
3089
3090                 return array( $dom, $title );
3091         }
3092
3093         /**
3094          * Fetch the unparsed text of a template and register a reference to it.
3095          */
3096         function fetchTemplateAndTitle( $title ) {
3097                 $templateCb = $this->mOptions->getTemplateCallback();
3098                 $stuff = call_user_func( $templateCb, $title, $this );
3099                 $text = $stuff['text'];
3100                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3101                 if ( isset( $stuff['deps'] ) ) {
3102                         foreach ( $stuff['deps'] as $dep ) {
3103                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3104                         }
3105                 }
3106                 return array($text,$finalTitle);
3107         }
3108
3109         function fetchTemplate( $title ) {
3110                 $rv = $this->fetchTemplateAndTitle($title);
3111                 return $rv[0];
3112         }
3113
3114         /**
3115          * Static function to get a template
3116          * Can be overridden via ParserOptions::setTemplateCallback().
3117          */
3118         static function statelessFetchTemplate( $title, $parser=false ) {
3119                 $text = $skip = false;
3120                 $finalTitle = $title;
3121                 $deps = array();
3122
3123                 // Loop to fetch the article, with up to 1 redirect
3124                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3125                         # Give extensions a chance to select the revision instead
3126                         $id = false; // Assume current
3127                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3128
3129                         if( $skip ) {
3130                                 $text = false;
3131                                 $deps[] = array(
3132                                         'title' => $title,
3133                                         'page_id' => $title->getArticleID(),
3134                                         'rev_id' => null );
3135                                 break;
3136                         }
3137                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3138                         $rev_id = $rev ? $rev->getId() : 0;
3139                         // If there is no current revision, there is no page
3140                         if( $id === false && !$rev ) {
3141                                 $linkCache = LinkCache::singleton();
3142                                 $linkCache->addBadLinkObj( $title );
3143                         }
3144
3145                         $deps[] = array(
3146                                 'title' => $title,
3147                                 'page_id' => $title->getArticleID(),
3148                                 'rev_id' => $rev_id );
3149
3150                         if( $rev ) {
3151                                 $text = $rev->getText();
3152                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3153                                 global $wgContLang;
3154                                 $message = $wgContLang->lcfirst( $title->getText() );
3155                                 $text = wfMsgForContentNoTrans( $message );
3156                                 if( wfEmptyMsg( $message, $text ) ) {
3157                                         $text = false;
3158                                         break;
3159                                 }
3160                         } else {
3161                                 break;
3162                         }
3163                         if ( $text === false ) {
3164                                 break;
3165                         }
3166                         // Redirect?
3167                         $finalTitle = $title;
3168                         $title = Title::newFromRedirect( $text );
3169                 }
3170                 return array(
3171                         'text' => $text,
3172                         'finalTitle' => $finalTitle,
3173                         'deps' => $deps );
3174         }
3175
3176         /**
3177          * Transclude an interwiki link.
3178          */
3179         function interwikiTransclude( $title, $action ) {
3180                 global $wgEnableScaryTranscluding;
3181
3182                 if (!$wgEnableScaryTranscluding)
3183                         return wfMsg('scarytranscludedisabled');
3184
3185                 $url = $title->getFullUrl( "action=$action" );
3186
3187                 if (strlen($url) > 255)
3188                         return wfMsg('scarytranscludetoolong');
3189                 return $this->fetchScaryTemplateMaybeFromCache($url);
3190         }
3191
3192         function fetchScaryTemplateMaybeFromCache($url) {
3193                 global $wgTranscludeCacheExpiry;
3194                 $dbr = wfGetDB(DB_SLAVE);
3195                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3196                                 array('tc_url' => $url));
3197                 if ($obj) {
3198                         $time = $obj->tc_time;
3199                         $text = $obj->tc_contents;
3200                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3201                                 return $text;
3202                         }
3203                 }
3204
3205                 $text = Http::get($url);
3206                 if (!$text)
3207                         return wfMsg('scarytranscludefailed', $url);
3208
3209                 $dbw = wfGetDB(DB_MASTER);
3210                 $dbw->replace('transcache', array('tc_url'), array(
3211                         'tc_url' => $url,
3212                         'tc_time' => time(),
3213                         'tc_contents' => $text));
3214                 return $text;
3215         }
3216
3217
3218         /**
3219          * Triple brace replacement -- used for template arguments
3220          * @private
3221          */
3222         function argSubstitution( $piece, $frame ) {
3223                 wfProfileIn( __METHOD__ );
3224
3225                 $error = false;
3226                 $parts = $piece['parts'];
3227                 $nameWithSpaces = $frame->expand( $piece['title'] );
3228                 $argName = trim( $nameWithSpaces );
3229                 $object = false;
3230                 $text = $frame->getArgument( $argName );
3231                 if (  $text === false && $parts->getLength() > 0
3232                   && (
3233                     $this->ot['html']
3234                     || $this->ot['pre']
3235                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3236                   )
3237                 ) {
3238                         # No match in frame, use the supplied default
3239                         $object = $parts->item( 0 )->getChildren();
3240                 }
3241                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3242                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3243                         $this->limitationWarn( 'post-expand-template-argument' );
3244                 }
3245
3246                 if ( $text === false && $object === false ) {
3247                         # No match anywhere
3248                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3249                 }
3250                 if ( $error !== false ) {
3251                         $text .= $error;
3252                 }
3253                 if ( $object !== false ) {
3254                         $ret = array( 'object' => $object );
3255                 } else {
3256                         $ret = array( 'text' => $text );
3257                 }
3258
3259                 wfProfileOut( __METHOD__ );
3260                 return $ret;
3261         }
3262
3263         /**
3264          * Return the text to be used for a given extension tag.
3265          * This is the ghost of strip().
3266          *
3267          * @param array $params Associative array of parameters:
3268          *     name       PPNode for the tag name
3269          *     attr       PPNode for unparsed text where tag attributes are thought to be
3270          *     attributes Optional associative array of parsed attributes
3271          *     inner      Contents of extension element
3272          *     noClose    Original text did not have a close tag
3273          * @param PPFrame $frame
3274          */
3275         function extensionSubstitution( $params, $frame ) {
3276                 global $wgRawHtml, $wgContLang;
3277
3278                 $name = $frame->expand( $params['name'] );
3279                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3280                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3281
3282                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3283
3284                 if ( $this->ot['html'] ) {
3285                         $name = strtolower( $name );
3286
3287                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3288                         if ( isset( $params['attributes'] ) ) {
3289                                 $attributes = $attributes + $params['attributes'];
3290                         }
3291                         switch ( $name ) {
3292                                 case 'html':
3293                                         if( $wgRawHtml ) {
3294                                                 $output = $content;
3295                                                 break;
3296                                         } else {
3297                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3298                                         }
3299                                 case 'nowiki':
3300                                         $output = Xml::escapeTagsOnly( $content );
3301                                         break;
3302                                 case 'math':
3303                                         $output = $wgContLang->armourMath(
3304                                                 MathRenderer::renderMath( $content, $attributes ) );
3305                                         break;
3306                                 case 'gallery':
3307                                         $output = $this->renderImageGallery( $content, $attributes );
3308                                         break;
3309                                 case 'poem':
3310                                         $output = $this->renderPoem( $content, $attributes );
3311                                         break;
3312                                 default:
3313                                         if( isset( $this->mTagHooks[$name] ) ) {
3314                                                 # Workaround for PHP bug 35229 and similar
3315                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3316                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3317                                                 }
3318                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3319                                                         array( $content, $attributes, $this ) );
3320                                         } else {
3321                                                 $output = '<span class="error">Invalid tag extension name: ' .
3322                                                         htmlspecialchars( $name ) . '</span>';
3323                                         }
3324                         }
3325                 } else {
3326                         if ( is_null( $attrText ) ) {
3327                                 $attrText = '';
3328                         }
3329                         if ( isset( $params['attributes'] ) ) {
3330                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3331                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3332                                                 htmlspecialchars( $attrValue ) . '"';
3333                                 }
3334                         }
3335                         if ( $content === null ) {
3336                                 $output = "<$name$attrText/>";
3337                         } else {
3338                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3339                                 $output = "<$name$attrText>$content$close";
3340                         }
3341                 }
3342
3343                 if ( $name === 'html' || $name === 'nowiki' ) {
3344                         $this->mStripState->nowiki->setPair( $marker, $output );
3345                 } else {
3346                         $this->mStripState->general->setPair( $marker, $output );
3347                 }
3348                 return $marker;
3349         }
3350
3351         /**
3352          * Increment an include size counter
3353          *
3354          * @param string $type The type of expansion
3355          * @param integer $size The size of the text
3356          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3357          */
3358         function incrementIncludeSize( $type, $size ) {
3359                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3360                         return false;
3361                 } else {
3362                         $this->mIncludeSizes[$type] += $size;
3363                         return true;
3364                 }
3365         }
3366
3367         /**
3368          * Increment the expensive function count
3369          *
3370          * @return boolean False if the limit has been exceeded
3371          */
3372         function incrementExpensiveFunctionCount() {
3373                 global $wgExpensiveParserFunctionLimit;
3374                 $this->mExpensiveFunctionCount++;
3375                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3376                         return true;
3377                 }
3378                 return false;
3379         }
3380
3381         /**
3382          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3383          * Fills $this->mDoubleUnderscores, returns the modified text
3384          */
3385         function doDoubleUnderscore( $text ) {
3386                 // The position of __TOC__ needs to be recorded
3387                 $mw = MagicWord::get( 'toc' );
3388                 if( $mw->match( $text ) ) {
3389                         $this->mShowToc = true;
3390                         $this->mForceTocPosition = true;
3391
3392                         // Set a placeholder. At the end we'll fill it in with the TOC.
3393                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3394
3395                         // Only keep the first one.
3396                         $text = $mw->replace( '', $text );
3397                 }
3398
3399                 // Now match and remove the rest of them
3400                 $mwa = MagicWord::getDoubleUnderscoreArray();
3401                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3402
3403                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3404                         $this->mOutput->mNoGallery = true;
3405                 }
3406                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3407                         $this->mShowToc = false;
3408                 }
3409                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3410                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3411
3412                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3413                         if ( $containerCategory ) {
3414                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3415                         } else {
3416                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3417                         }
3418                 }
3419                 # (bug 8068) Allow control over whether robots index a page.
3420                 #
3421                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3422                 # is not desirable, the last one on the page should win.
3423                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3424                         $this->mOutput->setIndexPolicy( 'noindex' );
3425                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3426                         $this->mOutput->setIndexPolicy( 'index' );
3427                 }
3428
3429                 return $text;
3430         }
3431
3432         /**
3433          * This function accomplishes several tasks:
3434          * 1) Auto-number headings if that option is enabled
3435          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3436          * 3) Add a Table of contents on the top for users who have enabled the option
3437          * 4) Auto-anchor headings
3438          *
3439          * It loops through all headlines, collects the necessary data, then splits up the
3440          * string and re-inserts the newly formatted headlines.
3441          *
3442          * @param string $text
3443          * @param boolean $isMain
3444          * @private
3445          */
3446         function formatHeadings( $text, $isMain=true ) {
3447                 global $wgMaxTocLevel, $wgContLang;
3448
3449                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3450                 $showEditLink = $this->mOptions->getEditSection();
3451
3452                 // Do not call quickUserCan unless necessary
3453                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3454                         $showEditLink = 0;
3455                 }
3456
3457                 # Inhibit editsection links if requested in the page
3458                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3459                         $showEditLink = 0;
3460                 }
3461
3462                 # Get all headlines for numbering them and adding funky stuff like [edit]
3463                 # links - this is for later, but we need the number of headlines right now
3464                 $matches = array();
3465                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3466
3467                 # if there are fewer than 4 headlines in the article, do not show TOC
3468                 # unless it's been explicitly enabled.
3469                 $enoughToc = $this->mShowToc &&
3470                         (($numMatches >= 4) || $this->mForceTocPosition);
3471
3472                 # Allow user to stipulate that a page should have a "new section"
3473                 # link added via __NEWSECTIONLINK__
3474                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3475                         $this->mOutput->setNewSection( true );
3476                 }
3477
3478                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3479                 # override above conditions and always show TOC above first header
3480                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3481                         $this->mShowToc = true;
3482                         $enoughToc = true;
3483                 }
3484
3485                 # We need this to perform operations on the HTML
3486                 $sk = $this->mOptions->getSkin();
3487
3488                 # headline counter
3489                 $headlineCount = 0;
3490                 $numVisible = 0;
3491
3492                 # Ugh .. the TOC should have neat indentation levels which can be
3493                 # passed to the skin functions. These are determined here
3494                 $toc = '';
3495                 $full = '';
3496                 $head = array();
3497                 $sublevelCount = array();
3498                 $levelCount = array();
3499                 $toclevel = 0;
3500                 $level = 0;
3501                 $prevlevel = 0;
3502                 $toclevel = 0;
3503                 $prevtoclevel = 0;
3504                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3505                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3506                 $tocraw = array();
3507
3508                 foreach( $matches[3] as $headline ) {
3509                         $isTemplate = false;
3510                         $titleText = false;
3511                         $sectionIndex = false;
3512                         $numbering = '';
3513                         $markerMatches = array();
3514                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3515                                 $serial = $markerMatches[1];
3516                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3517                                 $isTemplate = ($titleText != $baseTitleText);
3518                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3519                         }
3520
3521                         if( $toclevel ) {
3522                                 $prevlevel = $level;
3523                                 $prevtoclevel = $toclevel;
3524                         }
3525                         $level = $matches[1][$headlineCount];
3526
3527                         if( $doNumberHeadings || $enoughToc ) {
3528
3529                                 if ( $level > $prevlevel ) {
3530                                         # Increase TOC level
3531                                         $toclevel++;
3532                                         $sublevelCount[$toclevel] = 0;
3533                                         if( $toclevel<$wgMaxTocLevel ) {
3534                                                 $prevtoclevel = $toclevel;
3535                                                 $toc .= $sk->tocIndent();
3536                                                 $numVisible++;
3537                                         }
3538                                 }
3539                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3540                                         # Decrease TOC level, find level to jump to
3541
3542                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3543                                                 # Can only go down to level 1
3544                                                 $toclevel = 1;
3545                                         } else {
3546                                                 for ($i = $toclevel; $i > 0; $i--) {
3547                                                         if ( $levelCount[$i] == $level ) {
3548                                                                 # Found last matching level
3549                                                                 $toclevel = $i;
3550                                                                 break;
3551                                                         }
3552                                                         elseif ( $levelCount[$i] < $level ) {
3553                                                                 # Found first matching level below current level
3554                                                                 $toclevel = $i + 1;
3555                                                                 break;
3556                                                         }
3557                                                 }
3558                                         }
3559                                         if( $toclevel<$wgMaxTocLevel ) {
3560                                                 if($prevtoclevel < $wgMaxTocLevel) {
3561                                                         # Unindent only if the previous toc level was shown :p
3562                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3563                                                         $prevtoclevel = $toclevel;
3564                                                 } else {
3565                                                         $toc .= $sk->tocLineEnd();
3566                                                 }
3567                                         }
3568                                 }
3569                                 else {
3570                                         # No change in level, end TOC line
3571                                         if( $toclevel<$wgMaxTocLevel ) {
3572                                                 $toc .= $sk->tocLineEnd();
3573                                         }
3574                                 }
3575
3576                                 $levelCount[$toclevel] = $level;
3577
3578                                 # count number of headlines for each level
3579                                 @$sublevelCount[$toclevel]++;
3580                                 $dot = 0;
3581                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3582                                         if( !empty( $sublevelCount[$i] ) ) {
3583                                                 if( $dot ) {
3584                                                         $numbering .= '.';
3585                                                 }
3586                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3587                                                 $dot = 1;
3588                                         }
3589                                 }
3590                         }
3591
3592                         # The safe header is a version of the header text safe to use for links
3593                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3594                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3595
3596                         # Remove link placeholders by the link text.
3597                         #     <!--LINK number-->
3598                         # turns into
3599                         #     link text with suffix
3600                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3601
3602                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3603                         $tocline = preg_replace(
3604                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3605                                 array( '',                          '<$1>'),
3606                                 $safeHeadline
3607                         );
3608                         $tocline = trim( $tocline );
3609
3610                         # For the anchor, strip out HTML-y stuff period
3611                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3612                         $safeHeadline = trim( $safeHeadline );
3613
3614                         # Save headline for section edit hint before it's escaped
3615                         $headlineHint = $safeHeadline;
3616                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3617                         # HTML names must be case-insensitively unique (bug 10721)
3618                         $arrayKey = strtolower( $safeHeadline );
3619
3620                         # count how many in assoc. array so we can track dupes in anchors
3621                         isset( $refers[$arrayKey] ) ? $refers[$arrayKey]++ : $refers[$arrayKey] = 1;
3622                         $refcount[$headlineCount] = $refers[$arrayKey];
3623
3624                         # Don't number the heading if it is the only one (looks silly)
3625                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3626                                 # the two are different if the line contains a link
3627                                 $headline=$numbering . ' ' . $headline;
3628                         }
3629
3630                         # Create the anchor for linking from the TOC to the section
3631                         $anchor = $safeHeadline;
3632                         if($refcount[$headlineCount] > 1 ) {
3633                                 $anchor .= '_' . $refcount[$headlineCount];
3634                         }
3635                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3636                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3637                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3638                         }
3639                         # give headline the correct <h#> tag
3640                         if( $showEditLink && $sectionIndex !== false ) {
3641                                 if( $isTemplate ) {
3642                                         # Put a T flag in the section identifier, to indicate to extractSections()
3643                                         # that sections inside <includeonly> should be counted.
3644                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3645                                 } else {
3646                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3647                                 }
3648                         } else {
3649                                 $editlink = '';
3650                         }
3651                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3652
3653                         $headlineCount++;
3654                 }
3655
3656                 $this->mOutput->setSections( $tocraw );
3657
3658                 # Never ever show TOC if no headers
3659                 if( $numVisible < 1 ) {
3660                         $enoughToc = false;
3661                 }
3662
3663                 if( $enoughToc ) {
3664                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3665                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3666                         }
3667                         $toc = $sk->tocList( $toc );
3668                 }
3669
3670                 # split up and insert constructed headlines
3671
3672                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3673                 $i = 0;
3674
3675                 foreach( $blocks as $block ) {
3676                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3677                                 # This is the [edit] link that appears for the top block of text when
3678                                 # section editing is enabled
3679
3680                                 # Disabled because it broke block formatting
3681                                 # For example, a bullet point in the top line
3682                                 # $full .= $sk->editSectionLink(0);
3683                         }
3684                         $full .= $block;
3685                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3686                                 # Top anchor now in skin
3687                                 $full = $full.$toc;
3688                         }
3689
3690                         if( !empty( $head[$i] ) ) {
3691                                 $full .= $head[$i];
3692                         }
3693                         $i++;
3694                 }
3695                 if( $this->mForceTocPosition ) {
3696                         return str_replace( '<!--MWTOC-->', $toc, $full );
3697                 } else {
3698                         return $full;
3699                 }
3700         }
3701
3702         /**
3703          * Transform wiki markup when saving a page by doing \r\n -> \n
3704          * conversion, substitting signatures, {{subst:}} templates, etc.
3705          *
3706          * @param string $text the text to transform
3707          * @param Title &$title the Title object for the current article
3708          * @param User &$user the User object describing the current user
3709          * @param ParserOptions $options parsing options
3710          * @param bool $clearState whether to clear the parser state first
3711          * @return string the altered wiki markup
3712          * @public
3713          */
3714         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3715                 $this->mOptions = $options;
3716                 $this->setTitle( $title );
3717                 $this->setOutputType( self::OT_WIKI );
3718
3719                 if ( $clearState ) {
3720                         $this->clearState();
3721                 }
3722
3723                 $pairs = array(
3724                         "\r\n" => "\n",
3725                 );
3726                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3727                 $text = $this->pstPass2( $text, $user );
3728                 $text = $this->mStripState->unstripBoth( $text );
3729                 return $text;
3730         }
3731
3732         /**
3733          * Pre-save transform helper function
3734          * @private
3735          */
3736         function pstPass2( $text, $user ) {
3737                 global $wgContLang, $wgLocaltimezone;
3738
3739                 /* Note: This is the timestamp saved as hardcoded wikitext to
3740                  * the database, we use $wgContLang here in order to give
3741                  * everyone the same signature and use the default one rather
3742                  * than the one selected in each user's preferences.
3743                  *
3744                  * (see also bug 12815)
3745                  */
3746                 $ts = $this->mOptions->getTimestamp();
3747                 $tz = wfMsgForContent( 'timezone-utc' );
3748                 if ( isset( $wgLocaltimezone ) ) {
3749                         $unixts = wfTimestamp( TS_UNIX, $ts );
3750                         $oldtz = getenv( 'TZ' );
3751                         putenv( 'TZ='.$wgLocaltimezone );
3752                         $ts = date( 'YmdHis', $unixts );
3753                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3754                         putenv( 'TZ='.$oldtz );
3755                 }
3756
3757                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3758
3759                 # Variable replacement
3760                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3761                 $text = $this->replaceVariables( $text );
3762
3763                 # Signatures
3764                 $sigText = $this->getUserSig( $user );
3765                 $text = strtr( $text, array(
3766                         '~~~~~' => $d,
3767                         '~~~~' => "$sigText $d",
3768                         '~~~' => $sigText
3769                 ) );
3770
3771                 # Context links: [[|name]] and [[name (context)|]]
3772                 #
3773                 global $wgLegalTitleChars;
3774                 $tc = "[$wgLegalTitleChars]";
3775                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3776
3777                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3778                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3779                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3780                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3781
3782                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3783                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3784                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3785                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3786
3787                 $t = $this->mTitle->getText();
3788                 $m = array();
3789                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3790                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3791                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3792                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3793                 } else {
3794                         # if there's no context, don't bother duplicating the title
3795                         $text = preg_replace( $p2, '[[\\1]]', $text );
3796                 }
3797
3798                 # Trim trailing whitespace
3799                 $text = rtrim( $text );
3800
3801                 return $text;
3802         }
3803
3804         /**
3805          * Fetch the user's signature text, if any, and normalize to
3806          * validated, ready-to-insert wikitext.
3807          *
3808          * @param User $user
3809          * @return string
3810          * @private
3811          */
3812         function getUserSig( &$user ) {
3813                 global $wgMaxSigChars;
3814
3815                 $username = $user->getName();
3816                 $nickname = $user->getOption( 'nickname' );
3817                 $nickname = $nickname === '' ? $username : $nickname;
3818
3819                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3820                         $nickname = $username;
3821                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3822                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3823                         # Sig. might contain markup; validate this
3824                         if( $this->validateSig( $nickname ) !== false ) {
3825                                 # Validated; clean up (if needed) and return it
3826                                 return $this->cleanSig( $nickname, true );
3827                         } else {
3828                                 # Failed to validate; fall back to the default
3829                                 $nickname = $username;
3830                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3831                         }
3832                 }
3833
3834                 // Make sure nickname doesnt get a sig in a sig
3835                 $nickname = $this->cleanSigInSig( $nickname );
3836
3837                 # If we're still here, make it a link to the user page
3838                 $userText = wfEscapeWikiText( $username );
3839                 $nickText = wfEscapeWikiText( $nickname );
3840                 if ( $user->isAnon() )  {
3841                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3842                 } else {
3843                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3844                 }
3845         }
3846
3847         /**
3848          * Check that the user's signature contains no bad XML
3849          *
3850          * @param string $text
3851          * @return mixed An expanded string, or false if invalid.
3852          */
3853         function validateSig( $text ) {
3854                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3855         }
3856
3857         /**
3858          * Clean up signature text
3859          *
3860          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3861          * 2) Substitute all transclusions
3862          *
3863          * @param string $text
3864          * @param $parsing Whether we're cleaning (preferences save) or parsing
3865          * @return string Signature text
3866          */
3867         function cleanSig( $text, $parsing = false ) {
3868                 if ( !$parsing ) {
3869                         global $wgTitle;
3870                         $this->clearState();
3871                         $this->setTitle( $wgTitle );
3872                         $this->mOptions = new ParserOptions;
3873                         $this->setOutputType = self::OT_PREPROCESS;
3874                 }
3875
3876                 # Option to disable this feature
3877                 if ( !$this->mOptions->getCleanSignatures() ) {
3878                         return $text;
3879                 }
3880
3881                 # FIXME: regex doesn't respect extension tags or nowiki
3882                 #  => Move this logic to braceSubstitution()
3883                 $substWord = MagicWord::get( 'subst' );
3884                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3885                 $substText = '{{' . $substWord->getSynonym( 0 );
3886
3887                 $text = preg_replace( $substRegex, $substText, $text );
3888                 $text = $this->cleanSigInSig( $text );
3889                 $dom = $this->preprocessToDom( $text );
3890                 $frame = $this->getPreprocessor()->newFrame();
3891                 $text = $frame->expand( $dom );
3892
3893                 if ( !$parsing ) {
3894                         $text = $this->mStripState->unstripBoth( $text );
3895                 }
3896
3897                 return $text;
3898         }
3899
3900         /**
3901          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3902          * @param string $text
3903          * @return string Signature text with /~{3,5}/ removed
3904          */
3905         function cleanSigInSig( $text ) {
3906                 $text = preg_replace( '/~{3,5}/', '', $text );
3907                 return $text;
3908         }
3909
3910         /**
3911          * Set up some variables which are usually set up in parse()
3912          * so that an external function can call some class members with confidence
3913          * @public
3914          */
3915         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3916                 $this->setTitle( $title );
3917                 $this->mOptions = $options;
3918                 $this->setOutputType( $outputType );
3919                 if ( $clearState ) {
3920                         $this->clearState();
3921                 }
3922         }
3923
3924         /**
3925          * Wrapper for preprocess()
3926          *
3927          * @param string $text the text to preprocess
3928          * @param ParserOptions $options  options
3929          * @return string
3930          * @public
3931          */
3932         function transformMsg( $text, $options ) {
3933                 global $wgTitle;
3934                 static $executing = false;
3935
3936                 # Guard against infinite recursion
3937                 if ( $executing ) {
3938                         return $text;
3939                 }
3940                 $executing = true;
3941
3942                 wfProfileIn(__METHOD__);
3943                 $text = $this->preprocess( $text, $wgTitle, $options );
3944
3945                 $executing = false;
3946                 wfProfileOut(__METHOD__);
3947                 return $text;
3948         }
3949
3950         /**
3951          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3952          * The callback should have the following form:
3953          *    function myParserHook( $text, $params, &$parser ) { ... }
3954          *
3955          * Transform and return $text. Use $parser for any required context, e.g. use
3956          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3957          *
3958          * @public
3959          *
3960          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3961          * @param mixed $callback The callback function (and object) to use for the tag
3962          *
3963          * @return The old value of the mTagHooks array associated with the hook
3964          */
3965         function setHook( $tag, $callback ) {
3966                 $tag = strtolower( $tag );
3967                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3968                 $this->mTagHooks[$tag] = $callback;
3969                 if( !in_array( $tag, $this->mStripList ) ) {
3970                         $this->mStripList[] = $tag;
3971                 }
3972
3973                 return $oldVal;
3974         }
3975
3976         function setTransparentTagHook( $tag, $callback ) {
3977                 $tag = strtolower( $tag );
3978                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3979                 $this->mTransparentTagHooks[$tag] = $callback;
3980
3981                 return $oldVal;
3982         }
3983
3984         /**
3985          * Remove all tag hooks
3986          */
3987         function clearTagHooks() {
3988                 $this->mTagHooks = array();
3989                 $this->mStripList = $this->mDefaultStripList;
3990         }
3991
3992         /**
3993          * Create a function, e.g. {{sum:1|2|3}}
3994          * The callback function should have the form:
3995          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3996          *
3997          * Or with SFH_OBJECT_ARGS:
3998          *    function myParserFunction( $parser, $frame, $args ) { ... }
3999          *
4000          * The callback may either return the text result of the function, or an array with the text
4001          * in element 0, and a number of flags in the other elements. The names of the flags are
4002          * specified in the keys. Valid flags are:
4003          *   found                     The text returned is valid, stop processing the template. This
4004          *                             is on by default.
4005          *   nowiki                    Wiki markup in the return value should be escaped
4006          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4007          *
4008          * @public
4009          *
4010          * @param string $id The magic word ID
4011          * @param mixed $callback The callback function (and object) to use
4012          * @param integer $flags a combination of the following flags:
4013          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4014          *
4015          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4016          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4017          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4018          *     the arguments, and to control the way they are expanded.
4019          *
4020          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4021          *     arguments, for instance:
4022          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4023          *
4024          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4025          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4026          *     working if/when this is changed.
4027          *
4028          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4029          *     expansion.
4030          *
4031          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4032          *     about the methods available in PPFrame and PPNode.
4033          *
4034          * @return The old callback function for this name, if any
4035          */
4036         function setFunctionHook( $id, $callback, $flags = 0 ) {
4037                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4038                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4039
4040                 # Add to function cache
4041                 $mw = MagicWord::get( $id );
4042                 if( !$mw )
4043                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4044
4045                 $synonyms = $mw->getSynonyms();
4046                 $sensitive = intval( $mw->isCaseSensitive() );
4047
4048                 foreach ( $synonyms as $syn ) {
4049                         # Case
4050                         if ( !$sensitive ) {
4051                                 $syn = strtolower( $syn );
4052                         }
4053                         # Add leading hash
4054                         if ( !( $flags & SFH_NO_HASH ) ) {
4055                                 $syn = '#' . $syn;
4056                         }
4057                         # Remove trailing colon
4058                         if ( substr( $syn, -1, 1 ) === ':' ) {
4059                                 $syn = substr( $syn, 0, -1 );
4060                         }
4061                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4062                 }
4063                 return $oldVal;
4064         }
4065
4066         /**
4067          * Get all registered function hook identifiers
4068          *
4069          * @return array
4070          */
4071         function getFunctionHooks() {
4072                 return array_keys( $this->mFunctionHooks );
4073         }
4074
4075         /**
4076          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4077          * Placeholders created in Skin::makeLinkObj()
4078          * Returns an array of link CSS classes, indexed by PDBK.
4079          */
4080         function replaceLinkHolders( &$text, $options = 0 ) {
4081                 return $this->mLinkHolders->replace( $text );
4082         }
4083
4084         /**
4085          * Replace <!--LINK--> link placeholders with plain text of links
4086          * (not HTML-formatted).
4087          * @param string $text
4088          * @return string
4089          */
4090         function replaceLinkHoldersText( $text ) {
4091                 return $this->mLinkHolders->replaceText( $text );
4092         }
4093
4094         /**
4095          * Tag hook handler for 'pre'.
4096          */
4097         function renderPreTag( $text, $attribs ) {
4098                 // Backwards-compatibility hack
4099                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4100
4101                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4102                 return wfOpenElement( 'pre', $attribs ) .
4103                         Xml::escapeTagsOnly( $content ) .
4104                         '</pre>';
4105         }
4106
4107         /**
4108          * Renders an image gallery from a text with one line per image.
4109          * text labels may be given by using |-style alternative text. E.g.
4110          *   Image:one.jpg|The number "1"
4111          *   Image:tree.jpg|A tree
4112          * given as text will return the HTML of a gallery with two images,
4113          * labeled 'The number "1"' and
4114          * 'A tree'.
4115          */
4116         function renderImageGallery( $text, $params ) {
4117                 $ig = new ImageGallery();
4118                 $ig->setContextTitle( $this->mTitle );
4119                 $ig->setShowBytes( false );
4120                 $ig->setShowFilename( false );
4121                 $ig->setParser( $this );
4122                 $ig->setHideBadImages();
4123                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4124                 $ig->useSkin( $this->mOptions->getSkin() );
4125                 $ig->mRevisionId = $this->mRevisionId;
4126
4127                 if( isset( $params['caption'] ) ) {
4128                         $caption = $params['caption'];
4129                         $caption = htmlspecialchars( $caption );
4130                         $caption = $this->replaceInternalLinks( $caption );
4131                         $ig->setCaptionHtml( $caption );
4132                 }
4133                 if( isset( $params['perrow'] ) ) {
4134                         $ig->setPerRow( $params['perrow'] );
4135                 }
4136                 if( isset( $params['widths'] ) ) {
4137                         $ig->setWidths( $params['widths'] );
4138                 }
4139                 if( isset( $params['heights'] ) ) {
4140                         $ig->setHeights( $params['heights'] );
4141                 }
4142
4143                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4144
4145                 $lines = StringUtils::explode( "\n", $text );
4146                 foreach ( $lines as $line ) {
4147                         # match lines like these:
4148                         # Image:someimage.jpg|This is some image
4149                         $matches = array();
4150                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4151                         # Skip empty lines
4152                         if ( count( $matches ) == 0 ) {
4153                                 continue;
4154                         }
4155
4156                         if ( strpos( $matches[0], '%' ) !== false )
4157                                 $matches[1] = urldecode( $matches[1] );
4158                         $tp = Title::newFromText( $matches[1]/*, NS_IMAGE*/ );
4159                         $nt =& $tp;
4160                         if( is_null( $nt ) ) {
4161                                 # Bogus title. Ignore these so we don't bomb out later.
4162                                 continue;
4163                         }
4164                         if ( isset( $matches[3] ) ) {
4165                                 $label = $matches[3];
4166                         } else {
4167                                 $label = '';
4168                         }
4169
4170                         $html = $this->recursiveTagParse( trim( $label ) );
4171
4172                         $ig->add( $nt, $html );
4173
4174                         # Only add real images (bug #5586)
4175                         if ( $nt->getNamespace() == NS_IMAGE ) {
4176                                 $this->mOutput->addImage( $nt->getDBkey() );
4177                         }
4178                 }
4179                 return $ig->toHTML();
4180         }
4181
4182         /** Renders any text in between <poem></poem> tags
4183          * based on http://www.mediawiki.org/wiki/Extension:Poem
4184          */
4185
4186         function renderPoem( $in, $param = array() ) {
4187
4188                 /* using newlines in the text will cause the parser to add <p> tags,
4189                 * which may not be desired in some cases
4190                 */
4191                 $nl = array_key_exists( 'compact', $param ) ? '' : "\n";
4192
4193                 $replacer = new DoubleReplacer( ' ', '&nbsp;' );
4194                 $text = $this->recursiveTagParse( $in );
4195                 $text = $this->mStripState->unstripNoWiki( $text );
4196                 // Only strip the very first and very last \n (which trim cannot do)
4197                 if( substr( $text, 0, 1 ) == "\n" )
4198                         $text = substr( $text, 1 );
4199                 if( substr( $text, -1 ) == "\n" )
4200                         $text = substr( $text, 0, -1 );
4201
4202                 $text = str_replace( "\n", "<br />\n", $text );
4203                 $text = preg_replace_callback(
4204                         "/^( +)/m",
4205                         $replacer->cb(),
4206                         $text );
4207
4208                 // Pass HTML attributes through to the output.
4209                 $attribs = Sanitizer::validateTagAttributes( $param, 'div' );
4210
4211                 // Wrap output in a <div> with "poem" class.
4212                 if( array_key_exists( 'class', $attribs ) ) {
4213                         $attribs['class'] = 'poem ' . $attribs['class'];
4214                 } else {
4215                         $attribs['class'] = 'poem';
4216                 }
4217
4218                 return Xml::openElement( 'div', $attribs ) . $nl . trim( $text ) . $nl . Xml::closeElement( 'div' );
4219         }
4220
4221         function getImageParams( $handler ) {
4222                 if ( $handler ) {
4223                         $handlerClass = get_class( $handler );
4224                 } else {
4225                         $handlerClass = '';
4226                 }
4227                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4228                         // Initialise static lists
4229                         static $internalParamNames = array(
4230                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4231                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4232                                         'bottom', 'text-bottom' ),
4233                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4234                                         'upright', 'border', 'link', 'alt' ),
4235                         );
4236                         static $internalParamMap;
4237                         if ( !$internalParamMap ) {
4238                                 $internalParamMap = array();
4239                                 foreach ( $internalParamNames as $type => $names ) {
4240                                         foreach ( $names as $name ) {
4241                                                 $magicName = str_replace( '-', '_', "img_$name" );
4242                                                 $internalParamMap[$magicName] = array( $type, $name );
4243                                         }
4244                                 }
4245                         }
4246
4247                         // Add handler params
4248                         $paramMap = $internalParamMap;
4249                         if ( $handler ) {
4250                                 $handlerParamMap = $handler->getParamMap();
4251                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4252                                         $paramMap[$magic] = array( 'handler', $paramName );
4253                                 }
4254                         }
4255                         $this->mImageParams[$handlerClass] = $paramMap;
4256                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4257                 }
4258                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4259         }
4260
4261         /**
4262          * Parse image options text and use it to make an image
4263          * @param Title $title
4264          * @param string $options
4265          * @param LinkHolderArray $holders
4266          */
4267         function makeImage( $title, $options, $holders = false ) {
4268                 # Check if the options text is of the form "options|alt text"
4269                 # Options are:
4270                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4271                 #  * left       no resizing, just left align. label is used for alt= only
4272                 #  * right      same, but right aligned
4273                 #  * none       same, but not aligned
4274                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4275                 #  * center     center the image
4276                 #  * framed     Keep original image size, no magnify-button.
4277                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4278                 #  * upright    reduce width for upright images, rounded to full __0 px
4279                 #  * border     draw a 1px border around the image
4280                 #  * alt        Text for HTML alt attribute (defaults to empty)
4281                 # vertical-align values (no % or length right now):
4282                 #  * baseline
4283                 #  * sub
4284                 #  * super
4285                 #  * top
4286                 #  * text-top
4287                 #  * middle
4288                 #  * bottom
4289                 #  * text-bottom
4290
4291                 $parts = StringUtils::explode( "|", $options );
4292                 $sk = $this->mOptions->getSkin();
4293
4294                 # Give extensions a chance to select the file revision for us
4295                 $skip = $time = $descQuery = false;
4296                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4297
4298                 if ( $skip ) {
4299                         return $sk->link( $title );
4300                 }
4301
4302                 # Get the file
4303                 $imagename = $title->getDBkey();
4304                 if ( isset( $this->mFileCache[$imagename][$time] ) ) {
4305                         $file = $this->mFileCache[$imagename][$time];
4306                 } else {
4307                         $file = wfFindFile( $title, $time );
4308                         if ( !(count( $this->mFileCache ) <= 1000) ) {
4309                                 $this->mFileCache = array();
4310                         }
4311                         $this->mFileCache[$imagename][$time] = $file;
4312                 }
4313                 # Get parameter map
4314                 $handler = $file ? $file->getHandler() : false;
4315
4316                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4317
4318                 # Process the input parameters
4319                 $caption = '';
4320                 $params = array( 'frame' => array(), 'handler' => array(),
4321                         'horizAlign' => array(), 'vertAlign' => array() );
4322                 foreach( $parts as $part ) {
4323                         $part = trim( $part );
4324                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4325                         $validated = false;
4326                         if( isset( $paramMap[$magicName] ) ) {
4327                                 list( $type, $paramName ) = $paramMap[$magicName];
4328
4329                                 // Special case; width and height come in one variable together
4330                                 if( $type === 'handler' && $paramName === 'width' ) {
4331                                         $m = array();
4332                                         # (bug 13500) In both cases (width/height and width only),
4333                                         # permit trailing "px" for backward compatibility.
4334                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4335                                                 $width = intval( $m[1] );
4336                                                 $height = intval( $m[2] );
4337                                                 if ( $handler->validateParam( 'width', $width ) ) {
4338                                                         $params[$type]['width'] = $width;
4339                                                         $validated = true;
4340                                                 }
4341                                                 if ( $handler->validateParam( 'height', $height ) ) {
4342                                                         $params[$type]['height'] = $height;
4343                                                         $validated = true;
4344                                                 }
4345                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4346                                                 $width = intval( $value );
4347                                                 if ( $handler->validateParam( 'width', $width ) ) {
4348                                                         $params[$type]['width'] = $width;
4349                                                         $validated = true;
4350                                                 }
4351                                         } // else no validation -- bug 13436
4352                                 } else {
4353                                         if ( $type === 'handler' ) {
4354                                                 # Validate handler parameter
4355                                                 $validated = $handler->validateParam( $paramName, $value );
4356                                         } else {
4357                                                 # Validate internal parameters
4358                                                 switch( $paramName ) {
4359                                                 case 'manualthumb':
4360                                                 case 'alt':
4361                                                         // @fixme - possibly check validity here for
4362                                                         // manualthumb? downstream behavior seems odd with
4363                                                         // missing manual thumbs.
4364                                                         $validated = true;
4365                                                         break;
4366                                                 case 'link':
4367                                                         $chars = self::EXT_LINK_URL_CLASS;
4368                                                         $prots = $this->mUrlProtocols;
4369                                                         if ( $value === '' ) {
4370                                                                 $paramName = 'no-link';
4371                                                                 $value = true;
4372                                                                 $validated = true;
4373                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4374                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4375                                                                         $paramName = 'link-url';
4376                                                                         $this->mOutput->addExternalLink( $value );
4377                                                                         $validated = true;
4378                                                                 }
4379                                                         } else {
4380                                                                 $linkTitle = Title::newFromText( $value );
4381                                                                 if ( $linkTitle ) {
4382                                                                         $paramName = 'link-title';
4383                                                                         $value = $linkTitle;
4384                                                                         $this->mOutput->addLink( $linkTitle );
4385                                                                         $validated = true;
4386                                                                 }
4387                                                         }
4388                                                         break;
4389                                                 default:
4390                                                         // Most other things appear to be empty or numeric...
4391                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4392                                                 }
4393                                         }
4394
4395                                         if ( $validated ) {
4396                                                 $params[$type][$paramName] = $value;
4397                                         }
4398                                 }
4399                         }
4400                         if ( !$validated ) {
4401                                 $caption = $part;
4402                         }
4403                 }
4404
4405                 # Process alignment parameters
4406                 if ( $params['horizAlign'] ) {
4407                         $params['frame']['align'] = key( $params['horizAlign'] );
4408                 }
4409                 if ( $params['vertAlign'] ) {
4410                         $params['frame']['valign'] = key( $params['vertAlign'] );
4411                 }
4412
4413                 $params['frame']['caption'] = $caption;
4414
4415                 # Strip bad stuff out of the title (tooltip).  We can't just use
4416                 # replaceLinkHoldersText() here, because if this function is called
4417                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4418                 if ( $holders ) {
4419                         $tooltip = $holders->replaceText( $caption );
4420                 } else {
4421                         $tooltip = $this->replaceLinkHoldersText( $caption );
4422                 }
4423
4424                 # make sure there are no placeholders in thumbnail attributes
4425                 # that are later expanded to html- so expand them now and
4426                 # remove the tags
4427                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4428                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4429
4430                 $params['frame']['title'] = $tooltip;
4431
4432                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4433                 # came to also set the caption, ordinary text after the image -- which
4434                 # makes no sense, because that just repeats the text multiple times in
4435                 # screen readers.  It *also* came to set the title attribute.
4436                 #
4437                 # Now that we have an alt attribute, we should not set the alt text to
4438                 # equal the caption: that's worse than useless, it just repeats the
4439                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4440                 # use the unnamed parameter for alt text as well, just for the time be-
4441                 # ing, if the unnamed param is set and the alt param is not.
4442                 #
4443                 # For the future, we need to figure out if we want to tweak this more,
4444                 # e.g., introducing a title= parameter for the title; ignoring the un-
4445                 # named parameter entirely for images without a caption; adding an ex-
4446                 # plicit caption= parameter and preserving the old magic unnamed para-
4447                 # meter for BC; ...
4448                 if( $caption !== '' && !isset( $params['frame']['alt'] )
4449                 && !isset( $params['frame']['framed'] )
4450                 && !isset( $params['frame']['thumbnail'] )
4451                 && !isset( $params['frame']['manualthumb'] ) ) {
4452                         $params['frame']['alt'] = $tooltip;
4453                 }
4454
4455                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4456
4457                 # Linker does the rest
4458                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4459
4460                 # Give the handler a chance to modify the parser object
4461                 if ( $handler ) {
4462                         $handler->parserTransformHook( $this, $file );
4463                 }
4464
4465                 return $ret;
4466         }
4467
4468         /**
4469          * Set a flag in the output object indicating that the content is dynamic and
4470          * shouldn't be cached.
4471          */
4472         function disableCache() {
4473                 wfDebug( "Parser output marked as uncacheable.\n" );
4474                 $this->mOutput->mCacheTime = -1;
4475         }
4476
4477         /**#@+
4478          * Callback from the Sanitizer for expanding items found in HTML attribute
4479          * values, so they can be safely tested and escaped.
4480          * @param string $text
4481          * @param PPFrame $frame
4482          * @return string
4483          * @private
4484          */
4485         function attributeStripCallback( &$text, $frame = false ) {
4486                 $text = $this->replaceVariables( $text, $frame );
4487                 $text = $this->mStripState->unstripBoth( $text );
4488                 return $text;
4489         }
4490
4491         /**#@-*/
4492
4493         /**#@+
4494          * Accessor/mutator
4495          */
4496         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4497         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4498         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4499         /**#@-*/
4500
4501         /**#@+
4502          * Accessor
4503          */
4504         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4505         /**#@-*/
4506
4507
4508         /**
4509          * Break wikitext input into sections, and either pull or replace
4510          * some particular section's text.
4511          *
4512          * External callers should use the getSection and replaceSection methods.
4513          *
4514          * @param string $text Page wikitext
4515          * @param string $section A section identifier string of the form:
4516          *   <flag1> - <flag2> - ... - <section number>
4517          *
4518          * Currently the only recognised flag is "T", which means the target section number
4519          * was derived during a template inclusion parse, in other words this is a template
4520          * section edit link. If no flags are given, it was an ordinary section edit link.
4521          * This flag is required to avoid a section numbering mismatch when a section is
4522          * enclosed by <includeonly> (bug 6563).
4523          *
4524          * The section number 0 pulls the text before the first heading; other numbers will
4525          * pull the given section along with its lower-level subsections. If the section is
4526          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4527          *
4528          * @param string $mode One of "get" or "replace"
4529          * @param string $newText Replacement text for section data.
4530          * @return string for "get", the extracted section text.
4531          *                for "replace", the whole page with the section replaced.
4532          */
4533         private function extractSections( $text, $section, $mode, $newText='' ) {
4534                 global $wgTitle;
4535                 $this->clearState();
4536                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4537                 $this->mOptions = new ParserOptions;
4538                 $this->setOutputType( self::OT_WIKI );
4539                 $outText = '';
4540                 $frame = $this->getPreprocessor()->newFrame();
4541
4542                 // Process section extraction flags
4543                 $flags = 0;
4544                 $sectionParts = explode( '-', $section );
4545                 $sectionIndex = array_pop( $sectionParts );
4546                 foreach ( $sectionParts as $part ) {
4547                         if ( $part === 'T' ) {
4548                                 $flags |= self::PTD_FOR_INCLUSION;
4549                         }
4550                 }
4551                 // Preprocess the text
4552                 $root = $this->preprocessToDom( $text, $flags );
4553
4554                 // <h> nodes indicate section breaks
4555                 // They can only occur at the top level, so we can find them by iterating the root's children
4556                 $node = $root->getFirstChild();
4557
4558                 // Find the target section
4559                 if ( $sectionIndex == 0 ) {
4560                         // Section zero doesn't nest, level=big
4561                         $targetLevel = 1000;
4562                 } else {
4563             while ( $node ) {
4564                 if ( $node->getName() === 'h' ) {
4565                     $bits = $node->splitHeading();
4566                                         if ( $bits['i'] == $sectionIndex ) {
4567                                         $targetLevel = $bits['level'];
4568                                                 break;
4569                                         }
4570                                 }
4571                                 if ( $mode === 'replace' ) {
4572                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4573                                 }
4574                                 $node = $node->getNextSibling();
4575                         }
4576                 }
4577
4578                 if ( !$node ) {
4579                         // Not found
4580                         if ( $mode === 'get' ) {
4581                                 return $newText;
4582                         } else {
4583                                 return $text;
4584                         }
4585                 }
4586
4587                 // Find the end of the section, including nested sections
4588                 do {
4589                         if ( $node->getName() === 'h' ) {
4590                                 $bits = $node->splitHeading();
4591                                 $curLevel = $bits['level'];
4592                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4593                                         break;
4594                                 }
4595                         }
4596                         if ( $mode === 'get' ) {
4597                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4598                         }
4599                         $node = $node->getNextSibling();
4600                 } while ( $node );
4601
4602                 // Write out the remainder (in replace mode only)
4603                 if ( $mode === 'replace' ) {
4604                         // Output the replacement text
4605                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4606                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4607                         $outText .= $newText . "\n\n";
4608                         while ( $node ) {
4609                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4610                                 $node = $node->getNextSibling();
4611                         }
4612                 }
4613
4614                 if ( is_string( $outText ) ) {
4615                         // Re-insert stripped tags
4616                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4617                 }
4618
4619                 return $outText;
4620         }
4621
4622         /**
4623          * This function returns the text of a section, specified by a number ($section).
4624          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4625          * the first section before any such heading (section 0).
4626          *
4627          * If a section contains subsections, these are also returned.
4628          *
4629          * @param string $text text to look in
4630          * @param string $section section identifier
4631          * @param string $deftext default to return if section is not found
4632          * @return string text of the requested section
4633          */
4634         public function getSection( $text, $section, $deftext='' ) {
4635                 return $this->extractSections( $text, $section, "get", $deftext );
4636         }
4637
4638         public function replaceSection( $oldtext, $section, $text ) {
4639                 return $this->extractSections( $oldtext, $section, "replace", $text );
4640         }
4641
4642         /**
4643          * Get the timestamp associated with the current revision, adjusted for
4644          * the default server-local timestamp
4645          */
4646         function getRevisionTimestamp() {
4647                 if ( is_null( $this->mRevisionTimestamp ) ) {
4648                         wfProfileIn( __METHOD__ );
4649                         global $wgContLang;
4650                         $dbr = wfGetDB( DB_SLAVE );
4651                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4652                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4653
4654                         // Normalize timestamp to internal MW format for timezone processing.
4655                         // This has the added side-effect of replacing a null value with
4656                         // the current time, which gives us more sensible behavior for
4657                         // previews.
4658                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4659
4660                         // The cryptic '' timezone parameter tells to use the site-default
4661                         // timezone offset instead of the user settings.
4662                         //
4663                         // Since this value will be saved into the parser cache, served
4664                         // to other users, and potentially even used inside links and such,
4665                         // it needs to be consistent for all visitors.
4666                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4667
4668                         wfProfileOut( __METHOD__ );
4669                 }
4670                 return $this->mRevisionTimestamp;
4671         }
4672
4673         /**
4674          * Mutator for $mDefaultSort
4675          *
4676          * @param $sort New value
4677          */
4678         public function setDefaultSort( $sort ) {
4679                 $this->mDefaultSort = $sort;
4680         }
4681
4682         /**
4683          * Accessor for $mDefaultSort
4684          * Will use the title/prefixed title if none is set
4685          *
4686          * @return string
4687          */
4688         public function getDefaultSort() {
4689                 global $wgCategoryPrefixedDefaultSortkey;
4690                 if( $this->mDefaultSort !== false ) {
4691                         return $this->mDefaultSort;
4692                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4693                         !$wgCategoryPrefixedDefaultSortkey) {
4694                         return $this->mTitle->getText();
4695                 } else {
4696                         return $this->mTitle->getPrefixedText();
4697                 }
4698         }
4699
4700         /**
4701          * Try to guess the section anchor name based on a wikitext fragment
4702          * presumably extracted from a heading, for example "Header" from
4703          * "== Header ==".
4704          */
4705         public function guessSectionNameFromWikiText( $text ) {
4706                 # Strip out wikitext links(they break the anchor)
4707                 $text = $this->stripSectionName( $text );
4708                 $headline = Sanitizer::decodeCharReferences( $text );
4709                 # strip out HTML
4710                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4711                 $headline = trim( $headline );
4712                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4713                 $replacearray = array(
4714                         '%3A' => ':',
4715                         '%' => '.'
4716                 );
4717                 return str_replace(
4718                         array_keys( $replacearray ),
4719                         array_values( $replacearray ),
4720                         $sectionanchor );
4721         }
4722
4723         /**
4724          * Strips a text string of wikitext for use in a section anchor
4725          *
4726          * Accepts a text string and then removes all wikitext from the
4727          * string and leaves only the resultant text (i.e. the result of
4728          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4729          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4730          * to create valid section anchors by mimicing the output of the
4731          * parser when headings are parsed.
4732          *
4733          * @param $text string Text string to be stripped of wikitext
4734          * for use in a Section anchor
4735          * @return Filtered text string
4736          */
4737         public function stripSectionName( $text ) {
4738                 # Strip internal link markup
4739                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4740                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4741
4742                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4743                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4744                 # on how many empty links there are on the page - need to figure that out.
4745                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4746
4747                 # Parse wikitext quotes (italics & bold)
4748                 $text = $this->doQuotes($text);
4749
4750                 # Strip HTML tags
4751                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4752                 return $text;
4753         }
4754
4755         function srvus( $text ) {
4756                 return $this->testSrvus( $text, $this->mOutputType );
4757         }
4758
4759         /**
4760          * strip/replaceVariables/unstrip for preprocessor regression testing
4761          */
4762         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4763                 $this->clearState();
4764                 if ( ! ( $title instanceof Title ) ) {
4765                         $title = Title::newFromText( $title );
4766                 }
4767                 $this->mTitle = $title;
4768                 $this->mOptions = $options;
4769                 $this->setOutputType( $outputType );
4770                 $text = $this->replaceVariables( $text );
4771                 $text = $this->mStripState->unstripBoth( $text );
4772                 $text = Sanitizer::removeHTMLtags( $text );
4773                 return $text;
4774         }
4775
4776         function testPst( $text, $title, $options ) {
4777                 global $wgUser;
4778                 if ( ! ( $title instanceof Title ) ) {
4779                         $title = Title::newFromText( $title );
4780                 }
4781                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4782         }
4783
4784         function testPreprocess( $text, $title, $options ) {
4785                 if ( ! ( $title instanceof Title ) ) {
4786                         $title = Title::newFromText( $title );
4787                 }
4788                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4789         }
4790
4791         function markerSkipCallback( $s, $callback ) {
4792                 $i = 0;
4793                 $out = '';
4794                 while ( $i < strlen( $s ) ) {
4795                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4796                         if ( $markerStart === false ) {
4797                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4798                                 break;
4799                         } else {
4800                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4801                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4802                                 if ( $markerEnd === false ) {
4803                                         $out .= substr( $s, $markerStart );
4804                                         break;
4805                                 } else {
4806                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4807                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4808                                         $i = $markerEnd;
4809                                 }
4810                         }
4811                 }
4812                 return $out;
4813         }
4814 }
4815
4816 /**
4817  * @todo document, briefly.
4818  * @ingroup Parser
4819  */
4820 class StripState {
4821         var $general, $nowiki;
4822
4823         function __construct() {
4824                 $this->general = new ReplacementArray;
4825                 $this->nowiki = new ReplacementArray;
4826         }
4827
4828         function unstripGeneral( $text ) {
4829                 wfProfileIn( __METHOD__ );
4830                 do {
4831                         $oldText = $text;
4832                         $text = $this->general->replace( $text );
4833                 } while ( $text !== $oldText );
4834                 wfProfileOut( __METHOD__ );
4835                 return $text;
4836         }
4837
4838         function unstripNoWiki( $text ) {
4839                 wfProfileIn( __METHOD__ );
4840                 do {
4841                         $oldText = $text;
4842                         $text = $this->nowiki->replace( $text );
4843                 } while ( $text !== $oldText );
4844                 wfProfileOut( __METHOD__ );
4845                 return $text;
4846         }
4847
4848         function unstripBoth( $text ) {
4849                 wfProfileIn( __METHOD__ );
4850                 do {
4851                         $oldText = $text;
4852                         $text = $this->general->replace( $text );
4853                         $text = $this->nowiki->replace( $text );
4854                 } while ( $text !== $oldText );
4855                 wfProfileOut( __METHOD__ );
4856                 return $text;
4857         }
4858 }
4859
4860 /**
4861  * @todo document, briefly.
4862  * @ingroup Parser
4863  */
4864 class OnlyIncludeReplacer {
4865         var $output = '';
4866
4867         function replace( $matches ) {
4868                 if ( substr( $matches[1], -1 ) === "\n" ) {
4869                         $this->output .= substr( $matches[1], 0, -1 );
4870                 } else {
4871                         $this->output .= $matches[1];
4872                 }
4873         }
4874 }