includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * preprocess()
  24  *   removes HTML comments and expands templates
  25  * cleanSig()
  26  *   Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *   Extracts sections from an article for section editing
  29  *
  30  * Globals used:
  31  *    objects:   $wgLang, $wgContLang
  32  *
  33  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  34  *
  35  * settings:
  36  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  37  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  38  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  39  *  $wgMaxArticleSize*
  40  *
  41  *  * only within ParserOptions
  42  * </pre>
  43  *
  44  * @ingroup Parser
  45  */
  46 class Parser
  47 {
  48         /**
  49          * Update this version number when the ParserOutput format
  50          * changes in an incompatible way, so the parser cache
  51          * can automatically discard old data.
  52          */
  53         const VERSION = '1.6.4';
  54
  55         # Flags for Parser::setFunctionHook
  56         # Also available as global constants from Defines.php
  57         const SFH_NO_HASH = 1;
  58         const SFH_OBJECT_ARGS = 2;
  59
  60         # Constants needed for external link processing
  61         # Everything except bracket, space, or control characters
  62         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  63         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  64                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  65
  66         // State constants for the definition list colon extraction
  67         const COLON_STATE_TEXT = 0;
  68         const COLON_STATE_TAG = 1;
  69         const COLON_STATE_TAGSTART = 2;
  70         const COLON_STATE_CLOSETAG = 3;
  71         const COLON_STATE_TAGSLASH = 4;
  72         const COLON_STATE_COMMENT = 5;
  73         const COLON_STATE_COMMENTDASH = 6;
  74         const COLON_STATE_COMMENTDASHDASH = 7;
  75
  76         // Flags for preprocessToDom
  77         const PTD_FOR_INCLUSION = 1;
  78
  79         // Allowed values for $this->mOutputType
  80         // Parameter to startExternalParse().
  81         const OT_HTML = 1;
  82         const OT_WIKI = 2;
  83         const OT_PREPROCESS = 3;
  84         const OT_MSG = 3;
  85
  86         // Marker Suffix needs to be accessible staticly.
  87         const MARKER_SUFFIX = "-QINU\x7f";
  88
  89         /**#@+
  90          * @private
  91          */
  92         # Persistent:
  93         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  94                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex, $mPreprocessor,
  95                 $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList, $mVarCache, $mConf;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mLinkHolders, $mLinkID;
 102         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 103         var $mTplExpandCache; // empty-frame expansion cache
 104         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 105         var $mExpensiveFunctionCount; // number of expensive parser function calls
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions,      // ParserOptions object
 110                 $mTitle,        // Title context, used for self-link rendering and similar things
 111                 $mOutputType,   // Output type, one of the OT_xxx constants
 112                 $ot,            // Shortcut alias, see setOutputType()
 113                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 114                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 115                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 116
 117         /**#@-*/
 118
 119         /**
 120          * Constructor
 121          *
 122          * @public
 123          */
 124         function __construct( $conf = array() ) {
 125                 $this->mConf = $conf;
 126                 $this->mTagHooks = array();
 127                 $this->mTransparentTagHooks = array();
 128                 $this->mFunctionHooks = array();
 129                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 130                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 131                 $this->mUrlProtocols = wfUrlProtocols();
 132                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 133                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 134                 $this->mVarCache = array();
 135                 if ( isset( $conf['preprocessorClass'] ) ) {
 136                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 137                 } elseif ( extension_loaded( 'domxml' ) ) {
 138                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 139                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 140                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 141                 } elseif ( extension_loaded( 'dom' ) ) {
 142                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 143                 } else {
 144                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 145                 }
 146                 $this->mMarkerIndex = 0;
 147                 $this->mFirstCall = true;
 148         }
 149
 150         /**
 151          * Reduce memory usage to reduce the impact of circular references
 152          */
 153         function __destruct() {
 154                 if ( isset( $this->mLinkHolders ) ) {
 155                         $this->mLinkHolders->__destruct();
 156                 }
 157                 foreach ( $this as $name => $value ) {
 158                         unset( $this->$name );
 159                 }
 160         }
 161
 162         /**
 163          * Do various kinds of initialisation on the first call of the parser
 164          */
 165         function firstCallInit() {
 166                 if ( !$this->mFirstCall ) {
 167                         return;
 168                 }
 169                 $this->mFirstCall = false;
 170
 171                 wfProfileIn( __METHOD__ );
 172
 173                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 174                 CoreParserFunctions::register( $this );
 175                 $this->initialiseVariables();
 176
 177                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 178                 wfProfileOut( __METHOD__ );
 179         }
 180
 181         /**
 182          * Clear Parser state
 183          *
 184          * @private
 185          */
 186         function clearState() {
 187                 wfProfileIn( __METHOD__ );
 188                 if ( $this->mFirstCall ) {
 189                         $this->firstCallInit();
 190                 }
 191                 $this->mOutput = new ParserOutput;
 192                 $this->mAutonumber = 0;
 193                 $this->mLastSection = '';
 194                 $this->mDTopen = false;
 195                 $this->mIncludeCount = array();
 196                 $this->mStripState = new StripState;
 197                 $this->mArgStack = false;
 198                 $this->mInPre = false;
 199                 $this->mLinkHolders = new LinkHolderArray( $this );
 200                 $this->mLinkID = 0;
 201                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 202
 203                 /**
 204                  * Prefix for temporary replacement strings for the multipass parser.
 205                  * \x07 should never appear in input as it's disallowed in XML.
 206                  * Using it at the front also gives us a little extra robustness
 207                  * since it shouldn't match when butted up against identifier-like
 208                  * string constructs.
 209                  *
 210                  * Must not consist of all title characters, or else it will change
 211                  * the behaviour of <nowiki> in a link.
 212                  */
 213                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 214                 # Changed to \x7f to allow XML double-parsing -- TS
 215                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 216
 217
 218                 # Clear these on every parse, bug 4549
 219                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 220
 221                 $this->mShowToc = true;
 222                 $this->mForceTocPosition = false;
 223                 $this->mIncludeSizes = array(
 224                         'post-expand' => 0,
 225                         'arg' => 0,
 226                 );
 227                 $this->mPPNodeCount = 0;
 228                 $this->mDefaultSort = false;
 229                 $this->mHeadings = array();
 230                 $this->mDoubleUnderscores = array();
 231                 $this->mExpensiveFunctionCount = 0;
 232
 233                 # Fix cloning
 234                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 235                         $this->mPreprocessor = null;
 236                 }
 237
 238                 wfRunHooks( 'ParserClearState', array( &$this ) );
 239                 wfProfileOut( __METHOD__ );
 240         }
 241
 242         function setOutputType( $ot ) {
 243                 $this->mOutputType = $ot;
 244                 // Shortcut alias
 245                 $this->ot = array(
 246                         'html' => $ot == self::OT_HTML,
 247                         'wiki' => $ot == self::OT_WIKI,
 248                         'pre' => $ot == self::OT_PREPROCESS,
 249                 );
 250         }
 251
 252         /**
 253          * Set the context title
 254          */
 255         function setTitle( $t ) {
 256                 if ( !$t || $t instanceof FakeTitle ) {
 257                         $t = Title::newFromText( 'NO TITLE' );
 258                 }
 259                 if ( strval( $t->getFragment() ) !== '' ) {
 260                         # Strip the fragment to avoid various odd effects
 261                         $this->mTitle = clone $t;
 262                         $this->mTitle->setFragment( '' );
 263                 } else {
 264                         $this->mTitle = $t;
 265                 }
 266         }
 267
 268         /**
 269          * Accessor for mUniqPrefix.
 270          *
 271          * @public
 272          */
 273         function uniqPrefix() {
 274                 if( !isset( $this->mUniqPrefix ) ) {
 275                         // @fixme this is probably *horribly wrong*
 276                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 277                         // if this is called for a parser cache hit, the parser may not
 278                         // have ever been initialized in the first place.
 279                         // Not really sure what the heck is supposed to be going on here.
 280                         return '';
 281                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 282                 }
 283                 return $this->mUniqPrefix;
 284         }
 285
 286         /**
 287          * Convert wikitext to HTML
 288          * Do not call this function recursively.
 289          *
 290          * @param $text String: text we want to parse
 291          * @param $title A title object
 292          * @param $options ParserOptions
 293          * @param $linestart boolean
 294          * @param $clearState boolean
 295          * @param $revid Int: number to pass in {{REVISIONID}}
 296          * @return ParserOutput a ParserOutput
 297          */
 298         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 299                 /**
 300                  * First pass--just handle <nowiki> sections, pass the rest off
 301                  * to internalParse() which does all the real work.
 302                  */
 303
 304                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 305                 $fname = __METHOD__.'-' . wfGetCaller();
 306                 wfProfileIn( __METHOD__ );
 307                 wfProfileIn( $fname );
 308
 309                 if ( $clearState ) {
 310                         $this->clearState();
 311                 }
 312
 313                 $this->mOptions = $options;
 314                 $this->setTitle( $title );
 315                 $oldRevisionId = $this->mRevisionId;
 316                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 317                 if( $revid !== null ) {
 318                         $this->mRevisionId = $revid;
 319                         $this->mRevisionTimestamp = null;
 320                 }
 321                 $this->setOutputType( self::OT_HTML );
 322                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 323                 # No more strip!
 324                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 325                 $text = $this->internalParse( $text );
 326                 $text = $this->mStripState->unstripGeneral( $text );
 327
 328                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 329                 $fixtags = array(
 330                         # french spaces, last one Guillemet-left
 331                         # only if there is something before the space
 332                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 333                         # french spaces, Guillemet-right
 334                         '/(\\302\\253) /' => '\\1&nbsp;',
 335                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 336                 );
 337                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 338
 339                 $text = $this->doBlockLevels( $text, $linestart );
 340
 341                 $this->replaceLinkHolders( $text );
 342
 343                 # the position of the parserConvert() call should not be changed. it
 344                 # assumes that the links are all replaced and the only thing left
 345                 # is the <nowiki> mark.
 346                 # Side-effects: this calls $this->mOutput->setTitleText()
 347                 $text = $wgContLang->parserConvert( $text, $this );
 348
 349                 $text = $this->mStripState->unstripNoWiki( $text );
 350
 351                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 352
 353 //!JF Move to its own function
 354
 355                 $uniq_prefix = $this->mUniqPrefix;
 356                 $matches = array();
 357                 $elements = array_keys( $this->mTransparentTagHooks );
 358                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 359
 360                 foreach( $matches as $marker => $data ) {
 361                         list( $element, $content, $params, $tag ) = $data;
 362                         $tagName = strtolower( $element );
 363                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 364                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 365                                         array( $content, $params, $this ) );
 366                         } else {
 367                                 $output = $tag;
 368                         }
 369                         $this->mStripState->general->setPair( $marker, $output );
 370                 }
 371                 $text = $this->mStripState->unstripGeneral( $text );
 372
 373                 $text = Sanitizer::normalizeCharReferences( $text );
 374
 375                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 376                         $text = self::tidy($text);
 377                 } else {
 378                         # attempt to sanitize at least some nesting problems
 379                         # (bug #2702 and quite a few others)
 380                         $tidyregs = array(
 381                                 # ''Something [http://www.cool.com cool''] -->
 382                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 383                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 384                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 385                                 # fix up an anchor inside another anchor, only
 386                                 # at least for a single single nested link (bug 3695)
 387                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 388                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 389                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 390                                 # contains a div, so fix it up here; replace
 391                                 # div with escaped text
 392                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 393                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 394                                 # remove empty italic or bold tag pairs, some
 395                                 # introduced by rules above
 396                                 '/<([bi])><\/\\1>/' => '',
 397                         );
 398
 399                         $text = preg_replace(
 400                                 array_keys( $tidyregs ),
 401                                 array_values( $tidyregs ),
 402                                 $text );
 403                 }
 404                 global $wgExpensiveParserFunctionLimit;
 405                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 406                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 407                 }
 408
 409                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 410
 411                 # Information on include size limits, for the benefit of users who try to skirt them
 412                 if ( $this->mOptions->getEnableLimitReport() ) {
 413                         global $wgExpensiveParserFunctionLimit;
 414                         $max = $this->mOptions->getMaxIncludeSize();
 415                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 416                         $limitReport =
 417                                 "NewPP limit report\n" .
 418                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 419                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 420                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 421                                 $PFreport;
 422                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 423                         $text .= "\n<!-- \n$limitReport-->\n";
 424                 }
 425                 $this->mOutput->setText( $text );
 426                 $this->mRevisionId = $oldRevisionId;
 427                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 428                 wfProfileOut( $fname );
 429                 wfProfileOut( __METHOD__ );
 430
 431                 return $this->mOutput;
 432         }
 433
 434         /**
 435          * Recursive parser entry point that can be called from an extension tag
 436          * hook.
 437          */
 438         function recursiveTagParse( $text ) {
 439                 wfProfileIn( __METHOD__ );
 440                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 441                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 442                 $text = $this->internalParse( $text );
 443                 wfProfileOut( __METHOD__ );
 444                 return $text;
 445         }
 446
 447         /**
 448          * Expand templates and variables in the text, producing valid, static wikitext.
 449          * Also removes comments.
 450          */
 451         function preprocess( $text, $title, $options, $revid = null ) {
 452                 wfProfileIn( __METHOD__ );
 453                 $this->clearState();
 454                 $this->setOutputType( self::OT_PREPROCESS );
 455                 $this->mOptions = $options;
 456                 $this->setTitle( $title );
 457                 if( $revid !== null ) {
 458                         $this->mRevisionId = $revid;
 459                 }
 460                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 461                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 462                 $text = $this->replaceVariables( $text );
 463                 $text = $this->mStripState->unstripBoth( $text );
 464                 wfProfileOut( __METHOD__ );
 465                 return $text;
 466         }
 467
 468         /**
 469          * Get a random string
 470          *
 471          * @private
 472          * @static
 473          */
 474         function getRandomString() {
 475                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 476         }
 477
 478         function &getTitle() { return $this->mTitle; }
 479         function getOptions() { return $this->mOptions; }
 480         function getRevisionId() { return $this->mRevisionId; }
 481         function getOutput() { return $this->mOutput; }
 482         function nextLinkID() { return $this->mLinkID++; }
 483
 484         function getFunctionLang() {
 485                 global $wgLang, $wgContLang;
 486
 487                 $target = $this->mOptions->getTargetLanguage();
 488                 if ( $target !== null ) {
 489                         return $target;
 490                 } else {
 491                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 492                 }
 493         }
 494
 495         /**
 496          * Get a preprocessor object
 497          */
 498         function getPreprocessor() {
 499                 if ( !isset( $this->mPreprocessor ) ) {
 500                         $class = $this->mPreprocessorClass;
 501                         $this->mPreprocessor = new $class( $this );
 502                 }
 503                 return $this->mPreprocessor;
 504         }
 505
 506         /**
 507          * Replaces all occurrences of HTML-style comments and the given tags
 508          * in the text with a random marker and returns the next text. The output
 509          * parameter $matches will be an associative array filled with data in
 510          * the form:
 511          *   'UNIQ-xxxxx' => array(
 512          *     'element',
 513          *     'tag content',
 514          *     array( 'param' => 'x' ),
 515          *     '<element param="x">tag content</element>' ) )
 516          *
 517          * @param $elements list of element names. Comments are always extracted.
 518          * @param $text Source text string.
 519          * @param $uniq_prefix
 520          *
 521          * @public
 522          * @static
 523          */
 524         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 525                 static $n = 1;
 526                 $stripped = '';
 527                 $matches = array();
 528
 529                 $taglist = implode( '|', $elements );
 530                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 531
 532                 while ( '' != $text ) {
 533                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 534                         $stripped .= $p[0];
 535                         if( count( $p ) < 5 ) {
 536                                 break;
 537                         }
 538                         if( count( $p ) > 5 ) {
 539                                 // comment
 540                                 $element    = $p[4];
 541                                 $attributes = '';
 542                                 $close      = '';
 543                                 $inside     = $p[5];
 544                         } else {
 545                                 // tag
 546                                 $element    = $p[1];
 547                                 $attributes = $p[2];
 548                                 $close      = $p[3];
 549                                 $inside     = $p[4];
 550                         }
 551
 552                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 553                         $stripped .= $marker;
 554
 555                         if ( $close === '/>' ) {
 556                                 // Empty element tag, <tag />
 557                                 $content = null;
 558                                 $text = $inside;
 559                                 $tail = null;
 560                         } else {
 561                                 if( $element === '!--' ) {
 562                                         $end = '/(-->)/';
 563                                 } else {
 564                                         $end = "/(<\\/$element\\s*>)/i";
 565                                 }
 566                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 567                                 $content = $q[0];
 568                                 if( count( $q ) < 3 ) {
 569                                         # No end tag -- let it run out to the end of the text.
 570                                         $tail = '';
 571                                         $text = '';
 572                                 } else {
 573                                         $tail = $q[1];
 574                                         $text = $q[2];
 575                                 }
 576                         }
 577
 578                         $matches[$marker] = array( $element,
 579                                 $content,
 580                                 Sanitizer::decodeTagAttributes( $attributes ),
 581                                 "<$element$attributes$close$content$tail" );
 582                 }
 583                 return $stripped;
 584         }
 585
 586         /**
 587          * Get a list of strippable XML-like elements
 588          */
 589         function getStripList() {
 590                 global $wgRawHtml;
 591                 $elements = $this->mStripList;
 592                 if( $wgRawHtml ) {
 593                         $elements[] = 'html';
 594                 }
 595                 if( $this->mOptions->getUseTeX() ) {
 596                         $elements[] = 'math';
 597                 }
 598                 return $elements;
 599         }
 600
 601         /**
 602          * @deprecated use replaceVariables
 603          */
 604         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 605                 return $text;
 606         }
 607
 608         /**
 609          * Restores pre, math, and other extensions removed by strip()
 610          *
 611          * always call unstripNoWiki() after this one
 612          * @private
 613          * @deprecated use $this->mStripState->unstrip()
 614          */
 615         function unstrip( $text, $state ) {
 616                 return $state->unstripGeneral( $text );
 617         }
 618
 619         /**
 620          * Always call this after unstrip() to preserve the order
 621          *
 622          * @private
 623          * @deprecated use $this->mStripState->unstrip()
 624          */
 625         function unstripNoWiki( $text, $state ) {
 626                 return $state->unstripNoWiki( $text );
 627         }
 628
 629         /**
 630          * @deprecated use $this->mStripState->unstripBoth()
 631          */
 632         function unstripForHTML( $text ) {
 633                 return $this->mStripState->unstripBoth( $text );
 634         }
 635
 636         /**
 637          * Add an item to the strip state
 638          * Returns the unique tag which must be inserted into the stripped text
 639          * The tag will be replaced with the original text in unstrip()
 640          *
 641          * @private
 642          */
 643         function insertStripItem( $text ) {
 644                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 645                 $this->mMarkerIndex++;
 646                 $this->mStripState->general->setPair( $rnd, $text );
 647                 return $rnd;
 648         }
 649
 650         /**
 651          * Interface with html tidy, used if $wgUseTidy = true.
 652          * If tidy isn't able to correct the markup, the original will be
 653          * returned in all its glory with a warning comment appended.
 654          *
 655          * Either the external tidy program or the in-process tidy extension
 656          * will be used depending on availability. Override the default
 657          * $wgTidyInternal setting to disable the internal if it's not working.
 658          *
 659          * @param string $text Hideous HTML input
 660          * @return string Corrected HTML output
 661          * @public
 662          * @static
 663          */
 664         function tidy( $text ) {
 665                 global $wgTidyInternal;
 666                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 667 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 668 '<head><title>test</title></head><body>'.$text.'</body></html>';
 669                 if( $wgTidyInternal ) {
 670                         $correctedtext = self::internalTidy( $wrappedtext );
 671                 } else {
 672                         $correctedtext = self::externalTidy( $wrappedtext );
 673                 }
 674                 if( is_null( $correctedtext ) ) {
 675                         wfDebug( "Tidy error detected!\n" );
 676                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 677                 }
 678                 return $correctedtext;
 679         }
 680
 681         /**
 682          * Spawn an external HTML tidy process and get corrected markup back from it.
 683          *
 684          * @private
 685          * @static
 686          */
 687         function externalTidy( $text ) {
 688                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 689                 wfProfileIn( __METHOD__ );
 690
 691                 $cleansource = '';
 692                 $opts = ' -utf8';
 693
 694                 $descriptorspec = array(
 695                         0 => array('pipe', 'r'),
 696                         1 => array('pipe', 'w'),
 697                         2 => array('file', wfGetNull(), 'a')
 698                 );
 699                 $pipes = array();
 700                 if( function_exists('proc_open') ) {
 701                         $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 702                         if (is_resource($process)) {
 703                                 // Theoretically, this style of communication could cause a deadlock
 704                                 // here. If the stdout buffer fills up, then writes to stdin could
 705                                 // block. This doesn't appear to happen with tidy, because tidy only
 706                                 // writes to stdout after it's finished reading from stdin. Search
 707                                 // for tidyParseStdin and tidySaveStdout in console/tidy.c
 708                                 fwrite($pipes[0], $text);
 709                                 fclose($pipes[0]);
 710                                 while (!feof($pipes[1])) {
 711                                         $cleansource .= fgets($pipes[1], 1024);
 712                                 }
 713                                 fclose($pipes[1]);
 714                                 proc_close($process);
 715                         }
 716                 }
 717
 718                 wfProfileOut( __METHOD__ );
 719
 720                 if( $cleansource == '' && $text != '') {
 721                         // Some kind of error happened, so we couldn't get the corrected text.
 722                         // Just give up; we'll use the source text and append a warning.
 723                         return null;
 724                 } else {
 725                         return $cleansource;
 726                 }
 727         }
 728
 729         /**
 730          * Use the HTML tidy PECL extension to use the tidy library in-process,
 731          * saving the overhead of spawning a new process.
 732          *
 733          * 'pear install tidy' should be able to compile the extension module.
 734          *
 735          * @private
 736          * @static
 737          */
 738         function internalTidy( $text ) {
 739                 global $wgTidyConf, $IP, $wgDebugTidy;
 740                 wfProfileIn( __METHOD__ );
 741
 742                 $tidy = new tidy;
 743                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 744                 $tidy->cleanRepair();
 745                 if( $tidy->getStatus() == 2 ) {
 746                         // 2 is magic number for fatal error
 747                         // http://www.php.net/manual/en/function.tidy-get-status.php
 748                         $cleansource = null;
 749                 } else {
 750                         $cleansource = tidy_get_output( $tidy );
 751                 }
 752                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 753                         $cleansource .= "<!--\nTidy reports:\n" .
 754                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 755                                 "\n-->";
 756                 }
 757
 758                 wfProfileOut( __METHOD__ );
 759                 return $cleansource;
 760         }
 761
 762         /**
 763          * parse the wiki syntax used to render tables
 764          *
 765          * @private
 766          */
 767         function doTableStuff ( $text ) {
 768                 wfProfileIn( __METHOD__ );
 769
 770                 $lines = StringUtils::explode( "\n", $text );
 771                 $out = '';
 772                 $td_history = array (); // Is currently a td tag open?
 773                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 774                 $tr_history = array (); // Is currently a tr tag open?
 775                 $tr_attributes = array (); // history of tr attributes
 776                 $has_opened_tr = array(); // Did this table open a <tr> element?
 777                 $indent_level = 0; // indent level of the table
 778
 779                 foreach ( $lines as $outLine ) {
 780                         $line = trim( $outLine );
 781
 782                         if( $line == '' ) { // empty line, go to next line
 783                                 $out .= $outLine."\n";
 784                                 continue;
 785                         }
 786                         $first_character = $line[0];
 787                         $matches = array();
 788
 789                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 790                                 // First check if we are starting a new table
 791                                 $indent_level = strlen( $matches[1] );
 792
 793                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 794                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 795
 796                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 797                                 array_push ( $td_history , false );
 798                                 array_push ( $last_tag_history , '' );
 799                                 array_push ( $tr_history , false );
 800                                 array_push ( $tr_attributes , '' );
 801                                 array_push ( $has_opened_tr , false );
 802                         } else if ( count ( $td_history ) == 0 ) {
 803                                 // Don't do any of the following
 804                                 $out .= $outLine."\n";
 805                                 continue;
 806                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 807                                 // We are ending a table
 808                                 $line = '</table>' . substr ( $line , 2 );
 809                                 $last_tag = array_pop ( $last_tag_history );
 810
 811                                 if ( !array_pop ( $has_opened_tr ) ) {
 812                                         $line = "<tr><td></td></tr>{$line}";
 813                                 }
 814
 815                                 if ( array_pop ( $tr_history ) ) {
 816                                         $line = "</tr>{$line}";
 817                                 }
 818
 819                                 if ( array_pop ( $td_history ) ) {
 820                                         $line = "</{$last_tag}>{$line}";
 821                                 }
 822                                 array_pop ( $tr_attributes );
 823                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 824                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 825                                 // Now we have a table row
 826                                 $line = preg_replace( '#^\|-+#', '', $line );
 827
 828                                 // Whats after the tag is now only attributes
 829                                 $attributes = $this->mStripState->unstripBoth( $line );
 830                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 831                                 array_pop ( $tr_attributes );
 832                                 array_push ( $tr_attributes , $attributes );
 833
 834                                 $line = '';
 835                                 $last_tag = array_pop ( $last_tag_history );
 836                                 array_pop ( $has_opened_tr );
 837                                 array_push ( $has_opened_tr , true );
 838
 839                                 if ( array_pop ( $tr_history ) ) {
 840                                         $line = '</tr>';
 841                                 }
 842
 843                                 if ( array_pop ( $td_history ) ) {
 844                                         $line = "</{$last_tag}>{$line}";
 845                                 }
 846
 847                                 $outLine = $line;
 848                                 array_push ( $tr_history , false );
 849                                 array_push ( $td_history , false );
 850                                 array_push ( $last_tag_history , '' );
 851                         }
 852                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 853                                 // This might be cell elements, td, th or captions
 854                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 855                                         $first_character = '+';
 856                                         $line = substr ( $line , 1 );
 857                                 }
 858
 859                                 $line = substr ( $line , 1 );
 860
 861                                 if ( $first_character === '!' ) {
 862                                         $line = str_replace ( '!!' , '||' , $line );
 863                                 }
 864
 865                                 // Split up multiple cells on the same line.
 866                                 // FIXME : This can result in improper nesting of tags processed
 867                                 // by earlier parser steps, but should avoid splitting up eg
 868                                 // attribute values containing literal "||".
 869                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 870
 871                                 $outLine = '';
 872
 873                                 // Loop through each table cell
 874                                 foreach ( $cells as $cell )
 875                                 {
 876                                         $previous = '';
 877                                         if ( $first_character !== '+' )
 878                                         {
 879                                                 $tr_after = array_pop ( $tr_attributes );
 880                                                 if ( !array_pop ( $tr_history ) ) {
 881                                                         $previous = "<tr{$tr_after}>\n";
 882                                                 }
 883                                                 array_push ( $tr_history , true );
 884                                                 array_push ( $tr_attributes , '' );
 885                                                 array_pop ( $has_opened_tr );
 886                                                 array_push ( $has_opened_tr , true );
 887                                         }
 888
 889                                         $last_tag = array_pop ( $last_tag_history );
 890
 891                                         if ( array_pop ( $td_history ) ) {
 892                                                 $previous = "</{$last_tag}>{$previous}";
 893                                         }
 894
 895                                         if ( $first_character === '|' ) {
 896                                                 $last_tag = 'td';
 897                                         } else if ( $first_character === '!' ) {
 898                                                 $last_tag = 'th';
 899                                         } else if ( $first_character === '+' ) {
 900                                                 $last_tag = 'caption';
 901                                         } else {
 902                                                 $last_tag = '';
 903                                         }
 904
 905                                         array_push ( $last_tag_history , $last_tag );
 906
 907                                         // A cell could contain both parameters and data
 908                                         $cell_data = explode ( '|' , $cell , 2 );
 909
 910                                         // Bug 553: Note that a '|' inside an invalid link should not
 911                                         // be mistaken as delimiting cell parameters
 912                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 913                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 914                                         } else if ( count ( $cell_data ) == 1 )
 915                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 916                                         else {
 917                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 918                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 919                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 920                                         }
 921
 922                                         $outLine .= $cell;
 923                                         array_push ( $td_history , true );
 924                                 }
 925                         }
 926                         $out .= $outLine . "\n";
 927                 }
 928
 929                 // Closing open td, tr && table
 930                 while ( count ( $td_history ) > 0 )
 931                 {
 932                         if ( array_pop ( $td_history ) ) {
 933                                 $out .= "</td>\n";
 934                         }
 935                         if ( array_pop ( $tr_history ) ) {
 936                                 $out .= "</tr>\n";
 937                         }
 938                         if ( !array_pop ( $has_opened_tr ) ) {
 939                                 $out .= "<tr><td></td></tr>\n" ;
 940                         }
 941
 942                         $out .= "</table>\n";
 943                 }
 944
 945                 // Remove trailing line-ending (b/c)
 946                 if ( substr( $out, -1 ) === "\n" ) {
 947                         $out = substr( $out, 0, -1 );
 948                 }
 949
 950                 // special case: don't return empty table
 951                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 952                         $out = '';
 953                 }
 954
 955                 wfProfileOut( __METHOD__ );
 956
 957                 return $out;
 958         }
 959
 960         /**
 961          * Helper function for parse() that transforms wiki markup into
 962          * HTML. Only called for $mOutputType == self::OT_HTML.
 963          *
 964          * @private
 965          */
 966         function internalParse( $text ) {
 967                 $isMain = true;
 968                 wfProfileIn( __METHOD__ );
 969
 970                 # Hook to suspend the parser in this state
 971                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 972                         wfProfileOut( __METHOD__ );
 973                         return $text ;
 974                 }
 975
 976                 $text = $this->replaceVariables( $text );
 977                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 978                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 979
 980                 // Tables need to come after variable replacement for things to work
 981                 // properly; putting them before other transformations should keep
 982                 // exciting things like link expansions from showing up in surprising
 983                 // places.
 984                 $text = $this->doTableStuff( $text );
 985
 986                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 987
 988                 $text = $this->doDoubleUnderscore( $text );
 989                 $text = $this->doHeadings( $text );
 990                 if($this->mOptions->getUseDynamicDates()) {
 991                         $df = DateFormatter::getInstance();
 992                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 993                 }
 994                 $text = $this->doAllQuotes( $text );
 995                 $text = $this->replaceInternalLinks( $text );
 996                 $text = $this->replaceExternalLinks( $text );
 997
 998                 # replaceInternalLinks may sometimes leave behind
 999                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1000                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1001
1002                 $text = $this->doMagicLinks( $text );
1003                 $text = $this->formatHeadings( $text, $isMain );
1004
1005                 wfProfileOut( __METHOD__ );
1006                 return $text;
1007         }
1008
1009         /**
1010          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1011          * magic external links.
1012          *
1013          * DML
1014          * @private
1015          */
1016         function doMagicLinks( $text ) {
1017                 wfProfileIn( __METHOD__ );
1018                 $prots = $this->mUrlProtocols;
1019                 $urlChar = self::EXT_LINK_URL_CLASS;
1020                 $text = preg_replace_callback(
1021                         '!(?:                           # Start cases
1022                                 (<a.*?</a>) |               # m[1]: Skip link text
1023                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1024                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1025                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1026                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1027                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1028                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1029                                     [0-9Xx]                 # check digit
1030                                     \b)
1031                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1032                 wfProfileOut( __METHOD__ );
1033                 return $text;
1034         }
1035
1036         function magicLinkCallback( $m ) {
1037                 if ( isset( $m[1] ) && strval( $m[1] ) !== '' ) {
1038                         # Skip anchor
1039                         return $m[0];
1040                 } elseif ( isset( $m[2] ) && strval( $m[2] ) !== '' ) {
1041                         # Skip HTML element
1042                         return $m[0];
1043                 } elseif ( isset( $m[3] ) && strval( $m[3] ) !== '' ) {
1044                         # Free external link
1045                         return $this->makeFreeExternalLink( $m[0] );
1046                 } elseif ( isset( $m[4] ) && strval( $m[4] ) !== '' ) {
1047                         # RFC or PMID
1048                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1049                                 $keyword = 'RFC';
1050                                 $urlmsg = 'rfcurl';
1051                                 $id = $m[4];
1052                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1053                                 $keyword = 'PMID';
1054                                 $urlmsg = 'pubmedurl';
1055                                 $id = $m[4];
1056                         } else {
1057                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1058                                         substr($m[0], 0, 20 ) . '"' );
1059                         }
1060                         $url = wfMsg( $urlmsg, $id);
1061                         $sk = $this->mOptions->getSkin();
1062                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1063                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1064                 } elseif ( isset( $m[5] ) && strval( $m[5] ) !== '' ) {
1065                         # ISBN
1066                         $isbn = $m[5];
1067                         $num = strtr( $isbn, array(
1068                                 '-' => '',
1069                                 ' ' => '',
1070                                 'x' => 'X',
1071                         ));
1072                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1073                         return'<a href="' .
1074                                 $titleObj->escapeLocalUrl() .
1075                                 "\" class=\"internal\">ISBN $isbn</a>";
1076                 } else {
1077                         return $m[0];
1078                 }
1079         }
1080
1081         /**
1082          * Make a free external link, given a user-supplied URL
1083          * @return HTML
1084          * @private
1085          */
1086         function makeFreeExternalLink( $url ) {
1087                 global $wgContLang;
1088                 wfProfileIn( __METHOD__ );
1089
1090                 $sk = $this->mOptions->getSkin();
1091                 $trail = '';
1092
1093                 # The characters '<' and '>' (which were escaped by
1094                 # removeHTMLtags()) should not be included in
1095                 # URLs, per RFC 2396.
1096                 $m2 = array();
1097                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1098                         $trail = substr($url, $m2[0][1]) . $trail;
1099                         $url = substr($url, 0, $m2[0][1]);
1100                 }
1101
1102                 # Move trailing punctuation to $trail
1103                 $sep = ',;\.:!?';
1104                 # If there is no left bracket, then consider right brackets fair game too
1105                 if ( strpos( $url, '(' ) === false ) {
1106                         $sep .= ')';
1107                 }
1108
1109                 $numSepChars = strspn( strrev( $url ), $sep );
1110                 if ( $numSepChars ) {
1111                         $trail = substr( $url, -$numSepChars ) . $trail;
1112                         $url = substr( $url, 0, -$numSepChars );
1113                 }
1114
1115                 $url = Sanitizer::cleanUrl( $url );
1116
1117                 # Is this an external image?
1118                 $text = $this->maybeMakeExternalImage( $url );
1119                 if ( $text === false ) {
1120                         # Not an image, make a link
1121                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1122                         # Register it in the output object...
1123                         # Replace unnecessary URL escape codes with their equivalent characters
1124                         $pasteurized = self::replaceUnusualEscapes( $url );
1125                         $this->mOutput->addExternalLink( $pasteurized );
1126                 }
1127                 wfProfileOut( __METHOD__ );
1128                 return $text . $trail;
1129         }
1130
1131
1132         /**
1133          * Parse headers and return html
1134          *
1135          * @private
1136          */
1137         function doHeadings( $text ) {
1138                 wfProfileIn( __METHOD__ );
1139                 for ( $i = 6; $i >= 1; --$i ) {
1140                         $h = str_repeat( '=', $i );
1141                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1142                           "<h$i>\\1</h$i>", $text );
1143                 }
1144                 wfProfileOut( __METHOD__ );
1145                 return $text;
1146         }
1147
1148         /**
1149          * Replace single quotes with HTML markup
1150          * @private
1151          * @return string the altered text
1152          */
1153         function doAllQuotes( $text ) {
1154                 wfProfileIn( __METHOD__ );
1155                 $outtext = '';
1156                 $lines = StringUtils::explode( "\n", $text );
1157                 foreach ( $lines as $line ) {
1158                         $outtext .= $this->doQuotes( $line ) . "\n";
1159                 }
1160                 $outtext = substr($outtext, 0,-1);
1161                 wfProfileOut( __METHOD__ );
1162                 return $outtext;
1163         }
1164
1165         /**
1166          * Helper function for doAllQuotes()
1167          */
1168         public function doQuotes( $text ) {
1169                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1170                 if ( count( $arr ) == 1 )
1171                         return $text;
1172                 else
1173                 {
1174                         # First, do some preliminary work. This may shift some apostrophes from
1175                         # being mark-up to being text. It also counts the number of occurrences
1176                         # of bold and italics mark-ups.
1177                         $i = 0;
1178                         $numbold = 0;
1179                         $numitalics = 0;
1180                         foreach ( $arr as $r )
1181                         {
1182                                 if ( ( $i % 2 ) == 1 )
1183                                 {
1184                                         # If there are ever four apostrophes, assume the first is supposed to
1185                                         # be text, and the remaining three constitute mark-up for bold text.
1186                                         if ( strlen( $arr[$i] ) == 4 )
1187                                         {
1188                                                 $arr[$i-1] .= "'";
1189                                                 $arr[$i] = "'''";
1190                                         }
1191                                         # If there are more than 5 apostrophes in a row, assume they're all
1192                                         # text except for the last 5.
1193                                         else if ( strlen( $arr[$i] ) > 5 )
1194                                         {
1195                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1196                                                 $arr[$i] = "'''''";
1197                                         }
1198                                         # Count the number of occurrences of bold and italics mark-ups.
1199                                         # We are not counting sequences of five apostrophes.
1200                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1201                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1202                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1203                                 }
1204                                 $i++;
1205                         }
1206
1207                         # If there is an odd number of both bold and italics, it is likely
1208                         # that one of the bold ones was meant to be an apostrophe followed
1209                         # by italics. Which one we cannot know for certain, but it is more
1210                         # likely to be one that has a single-letter word before it.
1211                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1212                         {
1213                                 $i = 0;
1214                                 $firstsingleletterword = -1;
1215                                 $firstmultiletterword = -1;
1216                                 $firstspace = -1;
1217                                 foreach ( $arr as $r )
1218                                 {
1219                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1220                                         {
1221                                                 $x1 = substr ($arr[$i-1], -1);
1222                                                 $x2 = substr ($arr[$i-1], -2, 1);
1223                                                 if ($x1 === ' ') {
1224                                                         if ($firstspace == -1) $firstspace = $i;
1225                                                 } else if ($x2 === ' ') {
1226                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1227                                                 } else {
1228                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1229                                                 }
1230                                         }
1231                                         $i++;
1232                                 }
1233
1234                                 # If there is a single-letter word, use it!
1235                                 if ($firstsingleletterword > -1)
1236                                 {
1237                                         $arr [ $firstsingleletterword ] = "''";
1238                                         $arr [ $firstsingleletterword-1 ] .= "'";
1239                                 }
1240                                 # If not, but there's a multi-letter word, use that one.
1241                                 else if ($firstmultiletterword > -1)
1242                                 {
1243                                         $arr [ $firstmultiletterword ] = "''";
1244                                         $arr [ $firstmultiletterword-1 ] .= "'";
1245                                 }
1246                                 # ... otherwise use the first one that has neither.
1247                                 # (notice that it is possible for all three to be -1 if, for example,
1248                                 # there is only one pentuple-apostrophe in the line)
1249                                 else if ($firstspace > -1)
1250                                 {
1251                                         $arr [ $firstspace ] = "''";
1252                                         $arr [ $firstspace-1 ] .= "'";
1253                                 }
1254                         }
1255
1256                         # Now let's actually convert our apostrophic mush to HTML!
1257                         $output = '';
1258                         $buffer = '';
1259                         $state = '';
1260                         $i = 0;
1261                         foreach ($arr as $r)
1262                         {
1263                                 if (($i % 2) == 0)
1264                                 {
1265                                         if ($state === 'both')
1266                                                 $buffer .= $r;
1267                                         else
1268                                                 $output .= $r;
1269                                 }
1270                                 else
1271                                 {
1272                                         if (strlen ($r) == 2)
1273                                         {
1274                                                 if ($state === 'i')
1275                                                 { $output .= '</i>'; $state = ''; }
1276                                                 else if ($state === 'bi')
1277                                                 { $output .= '</i>'; $state = 'b'; }
1278                                                 else if ($state === 'ib')
1279                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1280                                                 else if ($state === 'both')
1281                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1282                                                 else # $state can be 'b' or ''
1283                                                 { $output .= '<i>'; $state .= 'i'; }
1284                                         }
1285                                         else if (strlen ($r) == 3)
1286                                         {
1287                                                 if ($state === 'b')
1288                                                 { $output .= '</b>'; $state = ''; }
1289                                                 else if ($state === 'bi')
1290                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1291                                                 else if ($state === 'ib')
1292                                                 { $output .= '</b>'; $state = 'i'; }
1293                                                 else if ($state === 'both')
1294                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1295                                                 else # $state can be 'i' or ''
1296                                                 { $output .= '<b>'; $state .= 'b'; }
1297                                         }
1298                                         else if (strlen ($r) == 5)
1299                                         {
1300                                                 if ($state === 'b')
1301                                                 { $output .= '</b><i>'; $state = 'i'; }
1302                                                 else if ($state === 'i')
1303                                                 { $output .= '</i><b>'; $state = 'b'; }
1304                                                 else if ($state === 'bi')
1305                                                 { $output .= '</i></b>'; $state = ''; }
1306                                                 else if ($state === 'ib')
1307                                                 { $output .= '</b></i>'; $state = ''; }
1308                                                 else if ($state === 'both')
1309                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1310                                                 else # ($state == '')
1311                                                 { $buffer = ''; $state = 'both'; }
1312                                         }
1313                                 }
1314                                 $i++;
1315                         }
1316                         # Now close all remaining tags.  Notice that the order is important.
1317                         if ($state === 'b' || $state === 'ib')
1318                                 $output .= '</b>';
1319                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1320                                 $output .= '</i>';
1321                         if ($state === 'bi')
1322                                 $output .= '</b>';
1323                         # There might be lonely ''''', so make sure we have a buffer
1324                         if ($state === 'both' && $buffer)
1325                                 $output .= '<b><i>'.$buffer.'</i></b>';
1326                         return $output;
1327                 }
1328         }
1329
1330         /**
1331          * Replace external links (REL)
1332          *
1333          * Note: this is all very hackish and the order of execution matters a lot.
1334          * Make sure to run maintenance/parserTests.php if you change this code.
1335          *
1336          * @private
1337          */
1338         function replaceExternalLinks( $text ) {
1339                 global $wgContLang;
1340                 wfProfileIn( __METHOD__ );
1341
1342                 $sk = $this->mOptions->getSkin();
1343
1344                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1345                 $s = array_shift( $bits );
1346
1347                 $i = 0;
1348                 while ( $i<count( $bits ) ) {
1349                         $url = $bits[$i++];
1350                         $protocol = $bits[$i++];
1351                         $text = $bits[$i++];
1352                         $trail = $bits[$i++];
1353
1354                         # The characters '<' and '>' (which were escaped by
1355                         # removeHTMLtags()) should not be included in
1356                         # URLs, per RFC 2396.
1357                         $m2 = array();
1358                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1359                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1360                                 $url = substr($url, 0, $m2[0][1]);
1361                         }
1362
1363                         # If the link text is an image URL, replace it with an <img> tag
1364                         # This happened by accident in the original parser, but some people used it extensively
1365                         $img = $this->maybeMakeExternalImage( $text );
1366                         if ( $img !== false ) {
1367                                 $text = $img;
1368                         }
1369
1370                         $dtrail = '';
1371
1372                         # Set linktype for CSS - if URL==text, link is essentially free
1373                         $linktype = ($text === $url) ? 'free' : 'text';
1374
1375                         # No link text, e.g. [http://domain.tld/some.link]
1376                         if ( $text == '' ) {
1377                                 # Autonumber if allowed. See bug #5918
1378                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1379                                         $langObj = $this->getFunctionLang();
1380                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1381                                         $linktype = 'autonumber';
1382                                 } else {
1383                                         # Otherwise just use the URL
1384                                         $text = htmlspecialchars( $url );
1385                                         $linktype = 'free';
1386                                 }
1387                         } else {
1388                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1389                                 # Check for trail
1390                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1391                         }
1392
1393                         $text = $wgContLang->markNoConversion($text);
1394
1395                         $url = Sanitizer::cleanUrl( $url );
1396
1397                         # Use the encoded URL
1398                         # This means that users can paste URLs directly into the text
1399                         # Funny characters like &ouml; aren't valid in URLs anyway
1400                         # This was changed in August 2004
1401                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1402
1403                         # Register link in the output object.
1404                         # Replace unnecessary URL escape codes with the referenced character
1405                         # This prevents spammers from hiding links from the filters
1406                         $pasteurized = self::replaceUnusualEscapes( $url );
1407                         $this->mOutput->addExternalLink( $pasteurized );
1408                 }
1409
1410                 wfProfileOut( __METHOD__ );
1411                 return $s;
1412         }
1413
1414         /**
1415          * Replace unusual URL escape codes with their equivalent characters
1416          * @param string
1417          * @return string
1418          * @static
1419          * @todo  This can merge genuinely required bits in the path or query string,
1420          *        breaking legit URLs. A proper fix would treat the various parts of
1421          *        the URL differently; as a workaround, just use the output for
1422          *        statistical records, not for actual linking/output.
1423          */
1424         static function replaceUnusualEscapes( $url ) {
1425                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1426                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1427         }
1428
1429         /**
1430          * Callback function used in replaceUnusualEscapes().
1431          * Replaces unusual URL escape codes with their equivalent character
1432          * @static
1433          * @private
1434          */
1435         private static function replaceUnusualEscapesCallback( $matches ) {
1436                 $char = urldecode( $matches[0] );
1437                 $ord = ord( $char );
1438                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1439                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1440                         // No, shouldn't be escaped
1441                         return $char;
1442                 } else {
1443                         // Yes, leave it escaped
1444                         return $matches[0];
1445                 }
1446         }
1447
1448         /**
1449          * make an image if it's allowed, either through the global
1450          * option, through the exception, or through the on-wiki whitelist
1451          * @private
1452          */
1453         function maybeMakeExternalImage( $url ) {
1454                 $sk = $this->mOptions->getSkin();
1455                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1456                 $imagesexception = !empty($imagesfrom);
1457                 $text = false;
1458                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1459                 if( $imagesexception && is_array( $imagesfrom ) ) {
1460                         $imagematch = false;
1461                         foreach( $imagesfrom as $match ) {
1462                                 if( strpos( $url, $match ) === 0 ) {
1463                                         $imagematch = true;
1464                                         break;
1465                                 }
1466                         }
1467                 } elseif( $imagesexception ) {
1468                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1469                 } else {
1470                         $imagematch = false;
1471                 }
1472                 if ( $this->mOptions->getAllowExternalImages()
1473                      || ( $imagesexception && $imagematch ) ) {
1474                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1475                                 # Image found
1476                                 $text = $sk->makeExternalImage( $url );
1477                         }
1478                 }
1479                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1480                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1481                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1482                         foreach( $whitelist as $entry ) {
1483                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1484                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1485                                         continue;
1486                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1487                                         # Image matches a whitelist entry
1488                                         $text = $sk->makeExternalImage( $url );
1489                                         break;
1490                                 }
1491                         }
1492                 }
1493                 return $text;
1494         }
1495
1496         /**
1497          * Process [[ ]] wikilinks
1498          * @return processed text
1499          *
1500          * @private
1501          */
1502         function replaceInternalLinks( $s ) {
1503                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1504                 return $s;
1505         }
1506
1507         /**
1508          * Process [[ ]] wikilinks (RIL)
1509          * @return LinkHolderArray
1510          *
1511          * @private
1512          */
1513         function replaceInternalLinks2( &$s ) {
1514                 global $wgContLang;
1515
1516                 wfProfileIn( __METHOD__ );
1517
1518                 wfProfileIn( __METHOD__.'-setup' );
1519                 static $tc = FALSE, $e1, $e1_img;
1520                 # the % is needed to support urlencoded titles as well
1521                 if ( !$tc ) {
1522                         $tc = Title::legalChars() . '#%';
1523                         # Match a link having the form [[namespace:link|alternate]]trail
1524                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1525                         # Match cases where there is no "]]", which might still be images
1526                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1527                 }
1528
1529                 $sk = $this->mOptions->getSkin();
1530                 $holders = new LinkHolderArray( $this );
1531
1532                 #split the entire text string on occurences of [[
1533                 $a = StringUtils::explode( '[[', ' ' . $s );
1534                 #get the first element (all text up to first [[), and remove the space we added
1535                 $s = $a->current();
1536                 $a->next();
1537                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1538                 $s = substr( $s, 1 );
1539
1540                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1541                 $e2 = null;
1542                 if ( $useLinkPrefixExtension ) {
1543                         # Match the end of a line for a word that's not followed by whitespace,
1544                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1545                         $e2 = wfMsgForContent( 'linkprefix' );
1546                 }
1547
1548                 if( is_null( $this->mTitle ) ) {
1549                         wfProfileOut( __METHOD__.'-setup' );
1550                         wfProfileOut( __METHOD__ );
1551                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1552                 }
1553                 $nottalk = !$this->mTitle->isTalkPage();
1554
1555                 if ( $useLinkPrefixExtension ) {
1556                         $m = array();
1557                         if ( preg_match( $e2, $s, $m ) ) {
1558                                 $first_prefix = $m[2];
1559                         } else {
1560                                 $first_prefix = false;
1561                         }
1562                 } else {
1563                         $prefix = '';
1564                 }
1565
1566                 if($wgContLang->hasVariants()) {
1567                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1568                 } else {
1569                         $selflink = array($this->mTitle->getPrefixedText());
1570                 }
1571                 $useSubpages = $this->areSubpagesAllowed();
1572                 wfProfileOut( __METHOD__.'-setup' );
1573
1574                 # Loop for each link
1575                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1576                         # Check for excessive memory usage
1577                         if ( $holders->isBig() ) {
1578                                 # Too big
1579                                 # Do the existence check, replace the link holders and clear the array
1580                                 $holders->replace( $s );
1581                                 $holders->clear();
1582                         }
1583
1584                         if ( $useLinkPrefixExtension ) {
1585                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1586                                 if ( preg_match( $e2, $s, $m ) ) {
1587                                         $prefix = $m[2];
1588                                         $s = $m[1];
1589                                 } else {
1590                                         $prefix='';
1591                                 }
1592                                 # first link
1593                                 if($first_prefix) {
1594                                         $prefix = $first_prefix;
1595                                         $first_prefix = false;
1596                                 }
1597                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1598                         }
1599
1600                         $might_be_img = false;
1601
1602                         wfProfileIn( __METHOD__."-e1" );
1603                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1604                                 $text = $m[2];
1605                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1606                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1607                                 # the real problem is with the $e1 regex
1608                                 # See bug 1300.
1609                                 #
1610                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1611                                 # and no image is in sight. See bug 2095.
1612                                 #
1613                                 if( $text !== '' &&
1614                                         substr( $m[3], 0, 1 ) === ']' &&
1615                                         strpos($text, '[') !== false
1616                                 )
1617                                 {
1618                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1619                                         $m[3] = substr( $m[3], 1 );
1620                                 }
1621                                 # fix up urlencoded title texts
1622                                 if( strpos( $m[1], '%' ) !== false ) {
1623                                         # Should anchors '#' also be rejected?
1624                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1625                                 }
1626                                 $trail = $m[3];
1627                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1628                                 $might_be_img = true;
1629                                 $text = $m[2];
1630                                 if ( strpos( $m[1], '%' ) !== false ) {
1631                                         $m[1] = urldecode($m[1]);
1632                                 }
1633                                 $trail = "";
1634                         } else { # Invalid form; output directly
1635                                 $s .= $prefix . '[[' . $line ;
1636                                 wfProfileOut( __METHOD__."-e1" );
1637                                 continue;
1638                         }
1639                         wfProfileOut( __METHOD__."-e1" );
1640                         wfProfileIn( __METHOD__."-misc" );
1641
1642                         # Don't allow internal links to pages containing
1643                         # PROTO: where PROTO is a valid URL protocol; these
1644                         # should be external links.
1645                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1646                                 $s .= $prefix . '[[' . $line ;
1647                                 wfProfileOut( __METHOD__."-misc" );
1648                                 continue;
1649                         }
1650
1651                         # Make subpage if necessary
1652                         if( $useSubpages ) {
1653                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1654                         } else {
1655                                 $link = $m[1];
1656                         }
1657
1658                         $noforce = (substr($m[1], 0, 1) !== ':');
1659                         if (!$noforce) {
1660                                 # Strip off leading ':'
1661                                 $link = substr($link, 1);
1662                         }
1663
1664                         wfProfileOut( __METHOD__."-misc" );
1665                         wfProfileIn( __METHOD__."-title" );
1666                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1667                         if( !$nt ) {
1668                                 $s .= $prefix . '[[' . $line;
1669                                 wfProfileOut( __METHOD__."-title" );
1670                                 continue;
1671                         }
1672
1673                         $ns = $nt->getNamespace();
1674                         $iw = $nt->getInterWiki();
1675                         wfProfileOut( __METHOD__."-title" );
1676
1677                         if ($might_be_img) { # if this is actually an invalid link
1678                                 wfProfileIn( __METHOD__."-might_be_img" );
1679                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1680                                         $found = false;
1681                                         while ( true ) {
1682                                                 #look at the next 'line' to see if we can close it there
1683                                                 $a->next();
1684                                                 $next_line = $a->current();
1685                                                 if ( $next_line === false || $next_line === null ) {
1686                                                         break;
1687                                                 }
1688                                                 $m = explode( ']]', $next_line, 3 );
1689                                                 if ( count( $m ) == 3 ) {
1690                                                         # the first ]] closes the inner link, the second the image
1691                                                         $found = true;
1692                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1693                                                         $trail = $m[2];
1694                                                         break;
1695                                                 } elseif ( count( $m ) == 2 ) {
1696                                                         #if there's exactly one ]] that's fine, we'll keep looking
1697                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1698                                                 } else {
1699                                                         #if $next_line is invalid too, we need look no further
1700                                                         $text .= '[[' . $next_line;
1701                                                         break;
1702                                                 }
1703                                         }
1704                                         if ( !$found ) {
1705                                                 # we couldn't find the end of this imageLink, so output it raw
1706                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1707                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1708                                                 $s .= "{$prefix}[[$link|$text";
1709                                                 # note: no $trail, because without an end, there *is* no trail
1710                                                 wfProfileOut( __METHOD__."-might_be_img" );
1711                                                 continue;
1712                                         }
1713                                 } else { #it's not an image, so output it raw
1714                                         $s .= "{$prefix}[[$link|$text";
1715                                         # note: no $trail, because without an end, there *is* no trail
1716                                         wfProfileOut( __METHOD__."-might_be_img" );
1717                                         continue;
1718                                 }
1719                                 wfProfileOut( __METHOD__."-might_be_img" );
1720                         }
1721
1722                         $wasblank = ( '' == $text );
1723                         if( $wasblank ) $text = $link;
1724
1725                         # Link not escaped by : , create the various objects
1726                         if( $noforce ) {
1727
1728                                 # Interwikis
1729                                 wfProfileIn( __METHOD__."-interwiki" );
1730                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1731                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1732                                         $s = rtrim($s . $prefix);
1733                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1734                                         wfProfileOut( __METHOD__."-interwiki" );
1735                                         continue;
1736                                 }
1737                                 wfProfileOut( __METHOD__."-interwiki" );
1738
1739                                 if ( $ns == NS_IMAGE ) {
1740                                         wfProfileIn( __METHOD__."-image" );
1741                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1742                                                 # recursively parse links inside the image caption
1743                                                 # actually, this will parse them in any other parameters, too,
1744                                                 # but it might be hard to fix that, and it doesn't matter ATM
1745                                                 $text = $this->replaceExternalLinks($text);
1746                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1747
1748                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1749                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1750                                         }
1751                                         $this->mOutput->addImage( $nt->getDBkey() );
1752                                         wfProfileOut( __METHOD__."-image" );
1753                                         continue;
1754
1755                                 }
1756
1757                                 if ( $ns == NS_CATEGORY ) {
1758                                         wfProfileIn( __METHOD__."-category" );
1759                                         $s = rtrim($s . "\n"); # bug 87
1760
1761                                         if ( $wasblank ) {
1762                                                 $sortkey = $this->getDefaultSort();
1763                                         } else {
1764                                                 $sortkey = $text;
1765                                         }
1766                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1767                                         $sortkey = str_replace( "\n", '', $sortkey );
1768                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1769                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1770
1771                                         /**
1772                                          * Strip the whitespace Category links produce, see bug 87
1773                                          * @todo We might want to use trim($tmp, "\n") here.
1774                                          */
1775                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1776
1777                                         wfProfileOut( __METHOD__."-category" );
1778                                         continue;
1779                                 }
1780                         }
1781
1782                         # Self-link checking
1783                         if( $nt->getFragment() === '' && $nt->getNamespace() != NS_SPECIAL ) {
1784                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1785                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1786                                         continue;
1787                                 }
1788                         }
1789
1790                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1791                         if( $ns == NS_MEDIA ) {
1792                                 # Give extensions a chance to select the file revision for us
1793                                 $skip = $time = false;
1794                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1795                                 if ( $skip ) {
1796                                         $link = $sk->link( $nt );
1797                                 } else {
1798                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1799                                 }
1800                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1801                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1802                                 $this->mOutput->addImage( $nt->getDBkey() );
1803                                 continue;
1804                         } elseif( $ns == NS_SPECIAL ) {
1805                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1806                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1807                                 } else {
1808                                         $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1809                                 }
1810                                 continue;
1811                         } elseif( $ns == NS_IMAGE ) {
1812                                 $img = wfFindFile( $nt );
1813                                 if( $img ) {
1814                                         // Force a blue link if the file exists; may be a remote
1815                                         // upload on the shared repository, and we want to see its
1816                                         // auto-generated page.
1817                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1818                                         $this->mOutput->addLink( $nt );
1819                                         continue;
1820                                 }
1821                         }
1822                         $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1823                 }
1824                 wfProfileOut( __METHOD__ );
1825                 return $holders;
1826         }
1827
1828         /**
1829          * Make a link placeholder. The text returned can be later resolved to a real link with
1830          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1831          * parsing of interwiki links, and secondly to allow all existence checks and
1832          * article length checks (for stub links) to be bundled into a single query.
1833          *
1834          * @deprecated
1835          */
1836         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1837                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1838         }
1839
1840         /**
1841          * Render a forced-blue link inline; protect against double expansion of
1842          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1843          * Since this little disaster has to split off the trail text to avoid
1844          * breaking URLs in the following text without breaking trails on the
1845          * wiki links, it's been made into a horrible function.
1846          *
1847          * @param Title $nt
1848          * @param string $text
1849          * @param string $query
1850          * @param string $trail
1851          * @param string $prefix
1852          * @return string HTML-wikitext mix oh yuck
1853          */
1854         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1855                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1856                 $sk = $this->mOptions->getSkin();
1857                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1858                 return $this->armorLinks( $link ) . $trail;
1859         }
1860
1861         /**
1862          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1863          * going to go through further parsing steps before inline URL expansion.
1864          *
1865          * Not needed quite as much as it used to be since free links are a bit
1866          * more sensible these days. But bracketed links are still an issue.
1867          *
1868          * @param string more-or-less HTML
1869          * @return string less-or-more HTML with NOPARSE bits
1870          */
1871         function armorLinks( $text ) {
1872                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1873                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1874         }
1875
1876         /**
1877          * Return true if subpage links should be expanded on this page.
1878          * @return bool
1879          */
1880         function areSubpagesAllowed() {
1881                 # Some namespaces don't allow subpages
1882                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1883         }
1884
1885         /**
1886          * Handle link to subpage if necessary
1887          * @param string $target the source of the link
1888          * @param string &$text the link text, modified as necessary
1889          * @return string the full name of the link
1890          * @private
1891          */
1892         function maybeDoSubpageLink($target, &$text) {
1893                 # Valid link forms:
1894                 # Foobar -- normal
1895                 # :Foobar -- override special treatment of prefix (images, language links)
1896                 # /Foobar -- convert to CurrentPage/Foobar
1897                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1898                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1899                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1900
1901                 wfProfileIn( __METHOD__ );
1902                 $ret = $target; # default return value is no change
1903
1904                 # Some namespaces don't allow subpages,
1905                 # so only perform processing if subpages are allowed
1906                 if( $this->areSubpagesAllowed() ) {
1907                         $hash = strpos( $target, '#' );
1908                         if( $hash !== false ) {
1909                                 $suffix = substr( $target, $hash );
1910                                 $target = substr( $target, 0, $hash );
1911                         } else {
1912                                 $suffix = '';
1913                         }
1914                         # bug 7425
1915                         $target = trim( $target );
1916                         # Look at the first character
1917                         if( $target != '' && $target{0} === '/' ) {
1918                                 # / at end means we don't want the slash to be shown
1919                                 $m = array();
1920                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1921                                 if( $trailingSlashes ) {
1922                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1923                                 } else {
1924                                         $noslash = substr( $target, 1 );
1925                                 }
1926
1927                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1928                                 if( '' === $text ) {
1929                                         $text = $target . $suffix;
1930                                 } # this might be changed for ugliness reasons
1931                         } else {
1932                                 # check for .. subpage backlinks
1933                                 $dotdotcount = 0;
1934                                 $nodotdot = $target;
1935                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1936                                         ++$dotdotcount;
1937                                         $nodotdot = substr( $nodotdot, 3 );
1938                                 }
1939                                 if($dotdotcount > 0) {
1940                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1941                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1942                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1943                                                 # / at the end means don't show full path
1944                                                 if( substr( $nodotdot, -1, 1 ) === '/' ) {
1945                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1946                                                         if( '' === $text ) {
1947                                                                 $text = $nodotdot . $suffix;
1948                                                         }
1949                                                 }
1950                                                 $nodotdot = trim( $nodotdot );
1951                                                 if( $nodotdot != '' ) {
1952                                                         $ret .= '/' . $nodotdot;
1953                                                 }
1954                                                 $ret .= $suffix;
1955                                         }
1956                                 }
1957                         }
1958                 }
1959
1960                 wfProfileOut( __METHOD__ );
1961                 return $ret;
1962         }
1963
1964         /**#@+
1965          * Used by doBlockLevels()
1966          * @private
1967          */
1968         /* private */ function closeParagraph() {
1969                 $result = '';
1970                 if ( '' != $this->mLastSection ) {
1971                         $result = '</' . $this->mLastSection  . ">\n";
1972                 }
1973                 $this->mInPre = false;
1974                 $this->mLastSection = '';
1975                 return $result;
1976         }
1977         # getCommon() returns the length of the longest common substring
1978         # of both arguments, starting at the beginning of both.
1979         #
1980         /* private */ function getCommon( $st1, $st2 ) {
1981                 $fl = strlen( $st1 );
1982                 $shorter = strlen( $st2 );
1983                 if ( $fl < $shorter ) { $shorter = $fl; }
1984
1985                 for ( $i = 0; $i < $shorter; ++$i ) {
1986                         if ( $st1{$i} != $st2{$i} ) { break; }
1987                 }
1988                 return $i;
1989         }
1990         # These next three functions open, continue, and close the list
1991         # element appropriate to the prefix character passed into them.
1992         #
1993         /* private */ function openList( $char ) {
1994                 $result = $this->closeParagraph();
1995
1996                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1997                 else if ( '#' === $char ) { $result .= '<ol><li>'; }
1998                 else if ( ':' === $char ) { $result .= '<dl><dd>'; }
1999                 else if ( ';' === $char ) {
2000                         $result .= '<dl><dt>';
2001                         $this->mDTopen = true;
2002                 }
2003                 else { $result = '<!-- ERR 1 -->'; }
2004
2005                 return $result;
2006         }
2007
2008         /* private */ function nextItem( $char ) {
2009                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
2010                 else if ( ':' === $char || ';' === $char ) {
2011                         $close = '</dd>';
2012                         if ( $this->mDTopen ) { $close = '</dt>'; }
2013                         if ( ';' === $char ) {
2014                                 $this->mDTopen = true;
2015                                 return $close . '<dt>';
2016                         } else {
2017                                 $this->mDTopen = false;
2018                                 return $close . '<dd>';
2019                         }
2020                 }
2021                 return '<!-- ERR 2 -->';
2022         }
2023
2024         /* private */ function closeList( $char ) {
2025                 if ( '*' === $char ) { $text = '</li></ul>'; }
2026                 else if ( '#' === $char ) { $text = '</li></ol>'; }
2027                 else if ( ':' === $char ) {
2028                         if ( $this->mDTopen ) {
2029                                 $this->mDTopen = false;
2030                                 $text = '</dt></dl>';
2031                         } else {
2032                                 $text = '</dd></dl>';
2033                         }
2034                 }
2035                 else {  return '<!-- ERR 3 -->'; }
2036                 return $text."\n";
2037         }
2038         /**#@-*/
2039
2040         /**
2041          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2042          *
2043          * @private
2044          * @return string the lists rendered as HTML
2045          */
2046         function doBlockLevels( $text, $linestart ) {
2047                 wfProfileIn( __METHOD__ );
2048
2049                 # Parsing through the text line by line.  The main thing
2050                 # happening here is handling of block-level elements p, pre,
2051                 # and making lists from lines starting with * # : etc.
2052                 #
2053                 $textLines = StringUtils::explode( "\n", $text );
2054
2055                 $lastPrefix = $output = '';
2056                 $this->mDTopen = $inBlockElem = false;
2057                 $prefixLength = 0;
2058                 $paragraphStack = false;
2059
2060                 foreach ( $textLines as $oLine ) {
2061                         # Fix up $linestart
2062                         if ( !$linestart ) {
2063                                 $output .= $oLine;
2064                                 $linestart = true;
2065                                 continue;
2066                         }
2067
2068                         $lastPrefixLength = strlen( $lastPrefix );
2069                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2070                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2071                         if ( !$this->mInPre ) {
2072                                 # Multiple prefixes may abut each other for nested lists.
2073                                 $prefixLength = strspn( $oLine, '*#:;' );
2074                                 $prefix = substr( $oLine, 0, $prefixLength );
2075
2076                                 # eh?
2077                                 $prefix2 = str_replace( ';', ':', $prefix );
2078                                 $t = substr( $oLine, $prefixLength );
2079                                 $this->mInPre = (bool)$preOpenMatch;
2080                         } else {
2081                                 # Don't interpret any other prefixes in preformatted text
2082                                 $prefixLength = 0;
2083                                 $prefix = $prefix2 = '';
2084                                 $t = $oLine;
2085                         }
2086
2087                         # List generation
2088                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2089                                 # Same as the last item, so no need to deal with nesting or opening stuff
2090                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2091                                 $paragraphStack = false;
2092
2093                                 if ( substr( $prefix, -1 ) === ';') {
2094                                         # The one nasty exception: definition lists work like this:
2095                                         # ; title : definition text
2096                                         # So we check for : in the remainder text to split up the
2097                                         # title and definition, without b0rking links.
2098                                         $term = $t2 = '';
2099                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2100                                                 $t = $t2;
2101                                                 $output .= $term . $this->nextItem( ':' );
2102                                         }
2103                                 }
2104                         } elseif( $prefixLength || $lastPrefixLength ) {
2105                                 # Either open or close a level...
2106                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2107                                 $paragraphStack = false;
2108
2109                                 while( $commonPrefixLength < $lastPrefixLength ) {
2110                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2111                                         --$lastPrefixLength;
2112                                 }
2113                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2114                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2115                                 }
2116                                 while ( $prefixLength > $commonPrefixLength ) {
2117                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2118                                         $output .= $this->openList( $char );
2119
2120                                         if ( ';' === $char ) {
2121                                                 # FIXME: This is dupe of code above
2122                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2123                                                         $t = $t2;
2124                                                         $output .= $term . $this->nextItem( ':' );
2125                                                 }
2126                                         }
2127                                         ++$commonPrefixLength;
2128                                 }
2129                                 $lastPrefix = $prefix2;
2130                         }
2131                         if( 0 == $prefixLength ) {
2132                                 wfProfileIn( __METHOD__."-paragraph" );
2133                                 # No prefix (not in list)--go to paragraph mode
2134                                 // XXX: use a stack for nestable elements like span, table and div
2135                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2136                                 $closematch = preg_match(
2137                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2138                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2139                                 if ( $openmatch or $closematch ) {
2140                                         $paragraphStack = false;
2141                                         # TODO bug 5718: paragraph closed
2142                                         $output .= $this->closeParagraph();
2143                                         if ( $preOpenMatch and !$preCloseMatch ) {
2144                                                 $this->mInPre = true;
2145                                         }
2146                                         if ( $closematch ) {
2147                                                 $inBlockElem = false;
2148                                         } else {
2149                                                 $inBlockElem = true;
2150                                         }
2151                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2152                                         if ( ' ' == $t{0} and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2153                                                 // pre
2154                                                 if ($this->mLastSection !== 'pre') {
2155                                                         $paragraphStack = false;
2156                                                         $output .= $this->closeParagraph().'<pre>';
2157                                                         $this->mLastSection = 'pre';
2158                                                 }
2159                                                 $t = substr( $t, 1 );
2160                                         } else {
2161                                                 // paragraph
2162                                                 if ( '' == trim($t) ) {
2163                                                         if ( $paragraphStack ) {
2164                                                                 $output .= $paragraphStack.'<br />';
2165                                                                 $paragraphStack = false;
2166                                                                 $this->mLastSection = 'p';
2167                                                         } else {
2168                                                                 if ($this->mLastSection !== 'p' ) {
2169                                                                         $output .= $this->closeParagraph();
2170                                                                         $this->mLastSection = '';
2171                                                                         $paragraphStack = '<p>';
2172                                                                 } else {
2173                                                                         $paragraphStack = '</p><p>';
2174                                                                 }
2175                                                         }
2176                                                 } else {
2177                                                         if ( $paragraphStack ) {
2178                                                                 $output .= $paragraphStack;
2179                                                                 $paragraphStack = false;
2180                                                                 $this->mLastSection = 'p';
2181                                                         } else if ($this->mLastSection !== 'p') {
2182                                                                 $output .= $this->closeParagraph().'<p>';
2183                                                                 $this->mLastSection = 'p';
2184                                                         }
2185                                                 }
2186                                         }
2187                                 }
2188                                 wfProfileOut( __METHOD__."-paragraph" );
2189                         }
2190                         // somewhere above we forget to get out of pre block (bug 785)
2191                         if($preCloseMatch && $this->mInPre) {
2192                                 $this->mInPre = false;
2193                         }
2194                         if ($paragraphStack === false) {
2195                                 $output .= $t."\n";
2196                         }
2197                 }
2198                 while ( $prefixLength ) {
2199                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2200                         --$prefixLength;
2201                 }
2202                 if ( '' != $this->mLastSection ) {
2203                         $output .= '</' . $this->mLastSection . '>';
2204                         $this->mLastSection = '';
2205                 }
2206
2207                 wfProfileOut( __METHOD__ );
2208                 return $output;
2209         }
2210
2211         /**
2212          * Split up a string on ':', ignoring any occurences inside tags
2213          * to prevent illegal overlapping.
2214          * @param string $str the string to split
2215          * @param string &$before set to everything before the ':'
2216          * @param string &$after set to everything after the ':'
2217          * return string the position of the ':', or false if none found
2218          */
2219         function findColonNoLinks($str, &$before, &$after) {
2220                 wfProfileIn( __METHOD__ );
2221
2222                 $pos = strpos( $str, ':' );
2223                 if( $pos === false ) {
2224                         // Nothing to find!
2225                         wfProfileOut( __METHOD__ );
2226                         return false;
2227                 }
2228
2229                 $lt = strpos( $str, '<' );
2230                 if( $lt === false || $lt > $pos ) {
2231                         // Easy; no tag nesting to worry about
2232                         $before = substr( $str, 0, $pos );
2233                         $after = substr( $str, $pos+1 );
2234                         wfProfileOut( __METHOD__ );
2235                         return $pos;
2236                 }
2237
2238                 // Ugly state machine to walk through avoiding tags.
2239                 $state = self::COLON_STATE_TEXT;
2240                 $stack = 0;
2241                 $len = strlen( $str );
2242                 for( $i = 0; $i < $len; $i++ ) {
2243                         $c = $str{$i};
2244
2245                         switch( $state ) {
2246                         // (Using the number is a performance hack for common cases)
2247                         case 0: // self::COLON_STATE_TEXT:
2248                                 switch( $c ) {
2249                                 case "<":
2250                                         // Could be either a <start> tag or an </end> tag
2251                                         $state = self::COLON_STATE_TAGSTART;
2252                                         break;
2253                                 case ":":
2254                                         if( $stack == 0 ) {
2255                                                 // We found it!
2256                                                 $before = substr( $str, 0, $i );
2257                                                 $after = substr( $str, $i + 1 );
2258                                                 wfProfileOut( __METHOD__ );
2259                                                 return $i;
2260                                         }
2261                                         // Embedded in a tag; don't break it.
2262                                         break;
2263                                 default:
2264                                         // Skip ahead looking for something interesting
2265                                         $colon = strpos( $str, ':', $i );
2266                                         if( $colon === false ) {
2267                                                 // Nothing else interesting
2268                                                 wfProfileOut( __METHOD__ );
2269                                                 return false;
2270                                         }
2271                                         $lt = strpos( $str, '<', $i );
2272                                         if( $stack === 0 ) {
2273                                                 if( $lt === false || $colon < $lt ) {
2274                                                         // We found it!
2275                                                         $before = substr( $str, 0, $colon );
2276                                                         $after = substr( $str, $colon + 1 );
2277                                                         wfProfileOut( __METHOD__ );
2278                                                         return $i;
2279                                                 }
2280                                         }
2281                                         if( $lt === false ) {
2282                                                 // Nothing else interesting to find; abort!
2283                                                 // We're nested, but there's no close tags left. Abort!
2284                                                 break 2;
2285                                         }
2286                                         // Skip ahead to next tag start
2287                                         $i = $lt;
2288                                         $state = self::COLON_STATE_TAGSTART;
2289                                 }
2290                                 break;
2291                         case 1: // self::COLON_STATE_TAG:
2292                                 // In a <tag>
2293                                 switch( $c ) {
2294                                 case ">":
2295                                         $stack++;
2296                                         $state = self::COLON_STATE_TEXT;
2297                                         break;
2298                                 case "/":
2299                                         // Slash may be followed by >?
2300                                         $state = self::COLON_STATE_TAGSLASH;
2301                                         break;
2302                                 default:
2303                                         // ignore
2304                                 }
2305                                 break;
2306                         case 2: // self::COLON_STATE_TAGSTART:
2307                                 switch( $c ) {
2308                                 case "/":
2309                                         $state = self::COLON_STATE_CLOSETAG;
2310                                         break;
2311                                 case "!":
2312                                         $state = self::COLON_STATE_COMMENT;
2313                                         break;
2314                                 case ">":
2315                                         // Illegal early close? This shouldn't happen D:
2316                                         $state = self::COLON_STATE_TEXT;
2317                                         break;
2318                                 default:
2319                                         $state = self::COLON_STATE_TAG;
2320                                 }
2321                                 break;
2322                         case 3: // self::COLON_STATE_CLOSETAG:
2323                                 // In a </tag>
2324                                 if( $c === ">" ) {
2325                                         $stack--;
2326                                         if( $stack < 0 ) {
2327                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2328                                                 wfProfileOut( __METHOD__ );
2329                                                 return false;
2330                                         }
2331                                         $state = self::COLON_STATE_TEXT;
2332                                 }
2333                                 break;
2334                         case self::COLON_STATE_TAGSLASH:
2335                                 if( $c === ">" ) {
2336                                         // Yes, a self-closed tag <blah/>
2337                                         $state = self::COLON_STATE_TEXT;
2338                                 } else {
2339                                         // Probably we're jumping the gun, and this is an attribute
2340                                         $state = self::COLON_STATE_TAG;
2341                                 }
2342                                 break;
2343                         case 5: // self::COLON_STATE_COMMENT:
2344                                 if( $c === "-" ) {
2345                                         $state = self::COLON_STATE_COMMENTDASH;
2346                                 }
2347                                 break;
2348                         case self::COLON_STATE_COMMENTDASH:
2349                                 if( $c === "-" ) {
2350                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2351                                 } else {
2352                                         $state = self::COLON_STATE_COMMENT;
2353                                 }
2354                                 break;
2355                         case self::COLON_STATE_COMMENTDASHDASH:
2356                                 if( $c === ">" ) {
2357                                         $state = self::COLON_STATE_TEXT;
2358                                 } else {
2359                                         $state = self::COLON_STATE_COMMENT;
2360                                 }
2361                                 break;
2362                         default:
2363                                 throw new MWException( "State machine error in " . __METHOD__ );
2364                         }
2365                 }
2366                 if( $stack > 0 ) {
2367                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2368                         return false;
2369                 }
2370                 wfProfileOut( __METHOD__ );
2371                 return false;
2372         }
2373
2374         /**
2375          * Return value of a magic variable (like PAGENAME)
2376          *
2377          * @private
2378          */
2379         function getVariableValue( $index ) {
2380                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2381
2382                 /**
2383                  * Some of these require message or data lookups and can be
2384                  * expensive to check many times.
2385                  */
2386                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2387                         if ( isset( $this->mVarCache[$index] ) ) {
2388                                 return $this->mVarCache[$index];
2389                         }
2390                 }
2391
2392                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2393                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2394
2395                 # Use the time zone
2396                 global $wgLocaltimezone;
2397                 if ( isset( $wgLocaltimezone ) ) {
2398                         $oldtz = getenv( 'TZ' );
2399                         putenv( 'TZ='.$wgLocaltimezone );
2400                 }
2401
2402                 wfSuppressWarnings(); // E_STRICT system time bitching
2403                 $localTimestamp = date( 'YmdHis', $ts );
2404                 $localMonth = date( 'm', $ts );
2405                 $localMonthName = date( 'n', $ts );
2406                 $localDay = date( 'j', $ts );
2407                 $localDay2 = date( 'd', $ts );
2408                 $localDayOfWeek = date( 'w', $ts );
2409                 $localWeek = date( 'W', $ts );
2410                 $localYear = date( 'Y', $ts );
2411                 $localHour = date( 'H', $ts );
2412                 if ( isset( $wgLocaltimezone ) ) {
2413                         putenv( 'TZ='.$oldtz );
2414                 }
2415                 wfRestoreWarnings();
2416
2417                 switch ( $index ) {
2418                         case 'currentmonth':
2419                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2420                         case 'currentmonthname':
2421                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2422                         case 'currentmonthnamegen':
2423                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2424                         case 'currentmonthabbrev':
2425                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2426                         case 'currentday':
2427                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2428                         case 'currentday2':
2429                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2430                         case 'localmonth':
2431                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2432                         case 'localmonthname':
2433                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2434                         case 'localmonthnamegen':
2435                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2436                         case 'localmonthabbrev':
2437                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2438                         case 'localday':
2439                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2440                         case 'localday2':
2441                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2442                         case 'pagename':
2443                                 return wfEscapeWikiText( $this->mTitle->getText() );
2444                         case 'pagenamee':
2445                                 return $this->mTitle->getPartialURL();
2446                         case 'fullpagename':
2447                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2448                         case 'fullpagenamee':
2449                                 return $this->mTitle->getPrefixedURL();
2450                         case 'subpagename':
2451                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2452                         case 'subpagenamee':
2453                                 return $this->mTitle->getSubpageUrlForm();
2454                         case 'basepagename':
2455                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2456                         case 'basepagenamee':
2457                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2458                         case 'talkpagename':
2459                                 if( $this->mTitle->canTalk() ) {
2460                                         $talkPage = $this->mTitle->getTalkPage();
2461                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2462                                 } else {
2463                                         return '';
2464                                 }
2465                         case 'talkpagenamee':
2466                                 if( $this->mTitle->canTalk() ) {
2467                                         $talkPage = $this->mTitle->getTalkPage();
2468                                         return $talkPage->getPrefixedUrl();
2469                                 } else {
2470                                         return '';
2471                                 }
2472                         case 'subjectpagename':
2473                                 $subjPage = $this->mTitle->getSubjectPage();
2474                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2475                         case 'subjectpagenamee':
2476                                 $subjPage = $this->mTitle->getSubjectPage();
2477                                 return $subjPage->getPrefixedUrl();
2478                         case 'revisionid':
2479                                 // Let the edit saving system know we should parse the page
2480                                 // *after* a revision ID has been assigned.
2481                                 $this->mOutput->setFlag( 'vary-revision' );
2482                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2483                                 return $this->mRevisionId;
2484                         case 'revisionday':
2485                                 // Let the edit saving system know we should parse the page
2486                                 // *after* a revision ID has been assigned. This is for null edits.
2487                                 $this->mOutput->setFlag( 'vary-revision' );
2488                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2489                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2490                         case 'revisionday2':
2491                                 // Let the edit saving system know we should parse the page
2492                                 // *after* a revision ID has been assigned. This is for null edits.
2493                                 $this->mOutput->setFlag( 'vary-revision' );
2494                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2495                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2496                         case 'revisionmonth':
2497                                 // Let the edit saving system know we should parse the page
2498                                 // *after* a revision ID has been assigned. This is for null edits.
2499                                 $this->mOutput->setFlag( 'vary-revision' );
2500                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2501                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2502                         case 'revisionyear':
2503                                 // Let the edit saving system know we should parse the page
2504                                 // *after* a revision ID has been assigned. This is for null edits.
2505                                 $this->mOutput->setFlag( 'vary-revision' );
2506                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2507                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2508                         case 'revisiontimestamp':
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__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2513                                 return $this->getRevisionTimestamp();
2514                         case 'namespace':
2515                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2516                         case 'namespacee':
2517                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2518                         case 'talkspace':
2519                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2520                         case 'talkspacee':
2521                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2522                         case 'subjectspace':
2523                                 return $this->mTitle->getSubjectNsText();
2524                         case 'subjectspacee':
2525                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2526                         case 'currentdayname':
2527                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2528                         case 'currentyear':
2529                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2530                         case 'currenttime':
2531                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2532                         case 'currenthour':
2533                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2534                         case 'currentweek':
2535                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2536                                 // int to remove the padding
2537                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2538                         case 'currentdow':
2539                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2540                         case 'localdayname':
2541                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2542                         case 'localyear':
2543                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2544                         case 'localtime':
2545                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2546                         case 'localhour':
2547                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2548                         case 'localweek':
2549                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2550                                 // int to remove the padding
2551                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2552                         case 'localdow':
2553                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2554                         case 'numberofarticles':
2555                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2556                         case 'numberoffiles':
2557                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2558                         case 'numberofusers':
2559                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2560                         case 'numberofpages':
2561                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2562                         case 'numberofadmins':
2563                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2564                         case 'numberofedits':
2565                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2566                         case 'currenttimestamp':
2567                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2568                         case 'localtimestamp':
2569                                 return $this->mVarCache[$index] = $localTimestamp;
2570                         case 'currentversion':
2571                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2572                         case 'sitename':
2573                                 return $wgSitename;
2574                         case 'server':
2575                                 return $wgServer;
2576                         case 'servername':
2577                                 return $wgServerName;
2578                         case 'scriptpath':
2579                                 return $wgScriptPath;
2580                         case 'directionmark':
2581                                 return $wgContLang->getDirMark();
2582                         case 'contentlanguage':
2583                                 global $wgContLanguageCode;
2584                                 return $wgContLanguageCode;
2585                         default:
2586                                 $ret = null;
2587                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2588                                         return $ret;
2589                                 else
2590                                         return null;
2591                 }
2592         }
2593
2594         /**
2595          * initialise the magic variables (like CURRENTMONTHNAME)
2596          *
2597          * @private
2598          */
2599         function initialiseVariables() {
2600                 wfProfileIn( __METHOD__ );
2601                 $variableIDs = MagicWord::getVariableIDs();
2602
2603                 $this->mVariables = new MagicWordArray( $variableIDs );
2604                 wfProfileOut( __METHOD__ );
2605         }
2606
2607         /**
2608          * Preprocess some wikitext and return the document tree.
2609          * This is the ghost of replace_variables().
2610          *
2611          * @param string $text The text to parse
2612          * @param integer flags Bitwise combination of:
2613          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2614          *                                     included. Default is to assume a direct page view.
2615          *
2616          * The generated DOM tree must depend only on the input text and the flags.
2617          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2618          *
2619          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2620          * change in the DOM tree for a given text, must be passed through the section identifier
2621          * in the section edit link and thus back to extractSections().
2622          *
2623          * The output of this function is currently only cached in process memory, but a persistent
2624          * cache may be implemented at a later date which takes further advantage of these strict
2625          * dependency requirements.
2626          *
2627          * @private
2628          */
2629         function preprocessToDom ( $text, $flags = 0 ) {
2630                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2631                 return $dom;
2632         }
2633
2634         /*
2635          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2636          */
2637         public static function splitWhitespace( $s ) {
2638                 $ltrimmed = ltrim( $s );
2639                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2640                 $trimmed = rtrim( $ltrimmed );
2641                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2642                 if ( $diff > 0 ) {
2643                         $w2 = substr( $ltrimmed, -$diff );
2644                 } else {
2645                         $w2 = '';
2646                 }
2647                 return array( $w1, $trimmed, $w2 );
2648         }
2649
2650         /**
2651          * Replace magic variables, templates, and template arguments
2652          * with the appropriate text. Templates are substituted recursively,
2653          * taking care to avoid infinite loops.
2654          *
2655          * Note that the substitution depends on value of $mOutputType:
2656          *  self::OT_WIKI: only {{subst:}} templates
2657          *  self::OT_PREPROCESS: templates but not extension tags
2658          *  self::OT_HTML: all templates and extension tags
2659          *
2660          * @param string $tex The text to transform
2661          * @param PPFrame $frame Object describing the arguments passed to the template.
2662          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2663          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2664          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2665          * @private
2666          */
2667         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2668                 # Prevent too big inclusions
2669                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2670                         return $text;
2671                 }
2672
2673                 wfProfileIn( __METHOD__ );
2674
2675                 if ( $frame === false ) {
2676                         $frame = $this->getPreprocessor()->newFrame();
2677                 } elseif ( !( $frame instanceof PPFrame ) ) {
2678                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2679                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2680                 }
2681
2682                 $dom = $this->preprocessToDom( $text );
2683                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2684                 $text = $frame->expand( $dom, $flags );
2685
2686                 wfProfileOut( __METHOD__ );
2687                 return $text;
2688         }
2689
2690         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2691         static function createAssocArgs( $args ) {
2692                 $assocArgs = array();
2693                 $index = 1;
2694                 foreach( $args as $arg ) {
2695                         $eqpos = strpos( $arg, '=' );
2696                         if ( $eqpos === false ) {
2697                                 $assocArgs[$index++] = $arg;
2698                         } else {
2699                                 $name = trim( substr( $arg, 0, $eqpos ) );
2700                                 $value = trim( substr( $arg, $eqpos+1 ) );
2701                                 if ( $value === false ) {
2702                                         $value = '';
2703                                 }
2704                                 if ( $name !== false ) {
2705                                         $assocArgs[$name] = $value;
2706                                 }
2707                         }
2708                 }
2709
2710                 return $assocArgs;
2711         }
2712
2713         /**
2714          * Warn the user when a parser limitation is reached
2715          * Will warn at most once the user per limitation type
2716          *
2717          * @param string $limitationType, should be one of:
2718          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2719          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2720          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2721          * @params int $current, $max When an explicit limit has been
2722          *       exceeded, provide the values (optional)
2723          */
2724         function limitationWarn( $limitationType, $current=null, $max=null) {
2725                 $msgName = $limitationType . '-warning';
2726                 //does no harm if $current and $max are present but are unnecessary for the message
2727                 $warning = wfMsgExt( $msgName, array( 'parsemag', 'escape' ), $current, $max );
2728                 $this->mOutput->addWarning( $warning );
2729                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2730                 if ( $cat ) {
2731                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2732                 }
2733         }
2734
2735         /**
2736          * Return the text of a template, after recursively
2737          * replacing any variables or templates within the template.
2738          *
2739          * @param array $piece The parts of the template
2740          *  $piece['title']: the title, i.e. the part before the |
2741          *  $piece['parts']: the parameter array
2742          *  $piece['lineStart']: whether the brace was at the start of a line
2743          * @param PPFrame The current frame, contains template arguments
2744          * @return string the text of the template
2745          * @private
2746          */
2747         function braceSubstitution( $piece, $frame ) {
2748                 global $wgContLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2749                 wfProfileIn( __METHOD__ );
2750                 wfProfileIn( __METHOD__.'-setup' );
2751
2752                 # Flags
2753                 $found = false;             # $text has been filled
2754                 $nowiki = false;            # wiki markup in $text should be escaped
2755                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2756                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2757                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2758                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2759
2760                 # Title object, where $text came from
2761                 $title = NULL;
2762
2763                 # $part1 is the bit before the first |, and must contain only title characters.
2764                 # Various prefixes will be stripped from it later.
2765                 $titleWithSpaces = $frame->expand( $piece['title'] );
2766                 $part1 = trim( $titleWithSpaces );
2767                 $titleText = false;
2768
2769                 # Original title text preserved for various purposes
2770                 $originalTitle = $part1;
2771
2772                 # $args is a list of argument nodes, starting from index 0, not including $part1
2773                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2774                 wfProfileOut( __METHOD__.'-setup' );
2775
2776                 # SUBST
2777                 wfProfileIn( __METHOD__.'-modifiers' );
2778                 if ( !$found ) {
2779                         $mwSubst = MagicWord::get( 'subst' );
2780                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2781                                 # One of two possibilities is true:
2782                                 # 1) Found SUBST but not in the PST phase
2783                                 # 2) Didn't find SUBST and in the PST phase
2784                                 # In either case, return without further processing
2785                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2786                                 $isLocalObj = true;
2787                                 $found = true;
2788                         }
2789                 }
2790
2791                 # Variables
2792                 if ( !$found && $args->getLength() == 0 ) {
2793                         $id = $this->mVariables->matchStartToEnd( $part1 );
2794                         if ( $id !== false ) {
2795                                 $text = $this->getVariableValue( $id );
2796                                 if (MagicWord::getCacheTTL($id)>-1)
2797                                         $this->mOutput->mContainsOldMagic = true;
2798                                 $found = true;
2799                         }
2800                 }
2801
2802                 # MSG, MSGNW and RAW
2803                 if ( !$found ) {
2804                         # Check for MSGNW:
2805                         $mwMsgnw = MagicWord::get( 'msgnw' );
2806                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2807                                 $nowiki = true;
2808                         } else {
2809                                 # Remove obsolete MSG:
2810                                 $mwMsg = MagicWord::get( 'msg' );
2811                                 $mwMsg->matchStartAndRemove( $part1 );
2812                         }
2813
2814                         # Check for RAW:
2815                         $mwRaw = MagicWord::get( 'raw' );
2816                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2817                                 $forceRawInterwiki = true;
2818                         }
2819                 }
2820                 wfProfileOut( __METHOD__.'-modifiers' );
2821
2822                 # Parser functions
2823                 if ( !$found ) {
2824                         wfProfileIn( __METHOD__ . '-pfunc' );
2825
2826                         $colonPos = strpos( $part1, ':' );
2827                         if ( $colonPos !== false ) {
2828                                 # Case sensitive functions
2829                                 $function = substr( $part1, 0, $colonPos );
2830                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2831                                         $function = $this->mFunctionSynonyms[1][$function];
2832                                 } else {
2833                                         # Case insensitive functions
2834                                         $function = strtolower( $function );
2835                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2836                                                 $function = $this->mFunctionSynonyms[0][$function];
2837                                         } else {
2838                                                 $function = false;
2839                                         }
2840                                 }
2841                                 if ( $function ) {
2842                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2843                                         $initialArgs = array( &$this );
2844                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2845                                         if ( $flags & SFH_OBJECT_ARGS ) {
2846                                                 # Add a frame parameter, and pass the arguments as an array
2847                                                 $allArgs = $initialArgs;
2848                                                 $allArgs[] = $frame;
2849                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2850                                                         $funcArgs[] = $args->item( $i );
2851                                                 }
2852                                                 $allArgs[] = $funcArgs;
2853                                         } else {
2854                                                 # Convert arguments to plain text
2855                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2856                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2857                                                 }
2858                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2859                                         }
2860
2861                                         # Workaround for PHP bug 35229 and similar
2862                                         if ( !is_callable( $callback ) ) {
2863                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2864                                         }
2865                                         $result = call_user_func_array( $callback, $allArgs );
2866                                         $found = true;
2867                                         $noparse = true;
2868                                         $preprocessFlags = 0;
2869
2870                                         if ( is_array( $result ) ) {
2871                                                 if ( isset( $result[0] ) ) {
2872                                                         $text = $result[0];
2873                                                         unset( $result[0] );
2874                                                 }
2875
2876                                                 // Extract flags into the local scope
2877                                                 // This allows callers to set flags such as nowiki, found, etc.
2878                                                 extract( $result );
2879                                         } else {
2880                                                 $text = $result;
2881                                         }
2882                                         if ( !$noparse ) {
2883                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2884                                                 $isChildObj = true;
2885                                         }
2886                                 }
2887                         }
2888                         wfProfileOut( __METHOD__ . '-pfunc' );
2889                 }
2890
2891                 # Finish mangling title and then check for loops.
2892                 # Set $title to a Title object and $titleText to the PDBK
2893                 if ( !$found ) {
2894                         $ns = NS_TEMPLATE;
2895                         # Split the title into page and subpage
2896                         $subpage = '';
2897                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2898                         if ($subpage !== '') {
2899                                 $ns = $this->mTitle->getNamespace();
2900                         }
2901                         $title = Title::newFromText( $part1, $ns );
2902                         if ( $title ) {
2903                                 $titleText = $title->getPrefixedText();
2904                                 # Check for language variants if the template is not found
2905                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2906                                         $wgContLang->findVariantLink( $part1, $title, true );
2907                                 }
2908                                 # Do infinite loop check
2909                                 if ( !$frame->loopCheck( $title ) ) {
2910                                         $found = true;
2911                                         $text = "<span class=\"error\">Template loop detected: [[$titleText]]</span>";
2912                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2913                                 }
2914                                 # Do recursion depth check
2915                                 $limit = $this->mOptions->getMaxTemplateDepth();
2916                                 if ( $frame->depth >= $limit ) {
2917                                         $found = true;
2918                                         $text = "<span class=\"error\">Template recursion depth limit exceeded ($limit)</span>";
2919                                 }
2920                         }
2921                 }
2922
2923                 # Load from database
2924                 if ( !$found && $title ) {
2925                         wfProfileIn( __METHOD__ . '-loadtpl' );
2926                         if ( !$title->isExternal() ) {
2927                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2928                                         $text = SpecialPage::capturePath( $title );
2929                                         if ( is_string( $text ) ) {
2930                                                 $found = true;
2931                                                 $isHTML = true;
2932                                                 $this->disableCache();
2933                                         }
2934                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2935                                         $found = false; //access denied
2936                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2937                                 } else {
2938                                         list( $text, $title ) = $this->getTemplateDom( $title );
2939                                         if ( $text !== false ) {
2940                                                 $found = true;
2941                                                 $isChildObj = true;
2942                                         }
2943                                 }
2944
2945                                 # If the title is valid but undisplayable, make a link to it
2946                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2947                                         $text = "[[:$titleText]]";
2948                                         $found = true;
2949                                 }
2950                         } elseif ( $title->isTrans() ) {
2951                                 // Interwiki transclusion
2952                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2953                                         $text = $this->interwikiTransclude( $title, 'render' );
2954                                         $isHTML = true;
2955                                 } else {
2956                                         $text = $this->interwikiTransclude( $title, 'raw' );
2957                                         // Preprocess it like a template
2958                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2959                                         $isChildObj = true;
2960                                 }
2961                                 $found = true;
2962                         }
2963                         wfProfileOut( __METHOD__ . '-loadtpl' );
2964                 }
2965
2966                 # If we haven't found text to substitute by now, we're done
2967                 # Recover the source wikitext and return it
2968                 if ( !$found ) {
2969                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2970                         wfProfileOut( __METHOD__ );
2971                         return array( 'object' => $text );
2972                 }
2973
2974                 # Expand DOM-style return values in a child frame
2975                 if ( $isChildObj ) {
2976                         # Clean up argument array
2977                         $newFrame = $frame->newChild( $args, $title );
2978
2979                         if ( $nowiki ) {
2980                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2981                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2982                                 # Expansion is eligible for the empty-frame cache
2983                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2984                                         $text = $this->mTplExpandCache[$titleText];
2985                                 } else {
2986                                         $text = $newFrame->expand( $text );
2987                                         $this->mTplExpandCache[$titleText] = $text;
2988                                 }
2989                         } else {
2990                                 # Uncached expansion
2991                                 $text = $newFrame->expand( $text );
2992                         }
2993                 }
2994                 if ( $isLocalObj && $nowiki ) {
2995                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2996                         $isLocalObj = false;
2997                 }
2998
2999                 # Replace raw HTML by a placeholder
3000                 # Add a blank line preceding, to prevent it from mucking up
3001                 # immediately preceding headings
3002                 if ( $isHTML ) {
3003                         $text = "\n\n" . $this->insertStripItem( $text );
3004                 }
3005                 # Escape nowiki-style return values
3006                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3007                         $text = wfEscapeWikiText( $text );
3008                 }
3009                 # Bug 529: if the template begins with a table or block-level
3010                 # element, it should be treated as beginning a new line.
3011                 # This behaviour is somewhat controversial.
3012                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3013                         $text = "\n" . $text;
3014                 }
3015
3016                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3017                         # Error, oversize inclusion
3018                         $text = "[[$originalTitle]]" .
3019                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3020                         $this->limitationWarn( 'post-expand-template-inclusion' );
3021                 }
3022
3023                 if ( $isLocalObj ) {
3024                         $ret = array( 'object' => $text );
3025                 } else {
3026                         $ret = array( 'text' => $text );
3027                 }
3028
3029                 wfProfileOut( __METHOD__ );
3030                 return $ret;
3031         }
3032
3033         /**
3034          * Get the semi-parsed DOM representation of a template with a given title,
3035          * and its redirect destination title. Cached.
3036          */
3037         function getTemplateDom( $title ) {
3038                 $cacheTitle = $title;
3039                 $titleText = $title->getPrefixedDBkey();
3040
3041                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3042                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3043                         $title = Title::makeTitle( $ns, $dbk );
3044                         $titleText = $title->getPrefixedDBkey();
3045                 }
3046                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3047                         return array( $this->mTplDomCache[$titleText], $title );
3048                 }
3049
3050                 // Cache miss, go to the database
3051                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3052
3053                 if ( $text === false ) {
3054                         $this->mTplDomCache[$titleText] = false;
3055                         return array( false, $title );
3056                 }
3057
3058                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3059                 $this->mTplDomCache[ $titleText ] = $dom;
3060
3061                 if (! $title->equals($cacheTitle)) {
3062                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3063                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3064                 }
3065
3066                 return array( $dom, $title );
3067         }
3068
3069         /**
3070          * Fetch the unparsed text of a template and register a reference to it.
3071          */
3072         function fetchTemplateAndTitle( $title ) {
3073                 $templateCb = $this->mOptions->getTemplateCallback();
3074                 $stuff = call_user_func( $templateCb, $title, $this );
3075                 $text = $stuff['text'];
3076                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3077                 if ( isset( $stuff['deps'] ) ) {
3078                         foreach ( $stuff['deps'] as $dep ) {
3079                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3080                         }
3081                 }
3082                 return array($text,$finalTitle);
3083         }
3084
3085         function fetchTemplate( $title ) {
3086                 $rv = $this->fetchTemplateAndTitle($title);
3087                 return $rv[0];
3088         }
3089
3090         /**
3091          * Static function to get a template
3092          * Can be overridden via ParserOptions::setTemplateCallback().
3093          */
3094         static function statelessFetchTemplate( $title, $parser=false ) {
3095                 $text = $skip = false;
3096                 $finalTitle = $title;
3097                 $deps = array();
3098
3099                 // Loop to fetch the article, with up to 1 redirect
3100                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3101                         # Give extensions a chance to select the revision instead
3102                         $id = false; // Assume current
3103                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3104
3105                         if( $skip ) {
3106                                 $text = false;
3107                                 $deps[] = array(
3108                                         'title' => $title,
3109                                         'page_id' => $title->getArticleID(),
3110                                         'rev_id' => null );
3111                                 break;
3112                         }
3113                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3114                         $rev_id = $rev ? $rev->getId() : 0;
3115                         // If there is no current revision, there is no page
3116                         if( $id === false && !$rev ) {
3117                                 $linkCache = LinkCache::singleton();
3118                                 $linkCache->addBadLinkObj( $title );
3119                         }
3120
3121                         $deps[] = array(
3122                                 'title' => $title,
3123                                 'page_id' => $title->getArticleID(),
3124                                 'rev_id' => $rev_id );
3125
3126                         if( $rev ) {
3127                                 $text = $rev->getText();
3128                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3129                                 global $wgContLang;
3130                                 $message = $wgContLang->lcfirst( $title->getText() );
3131                                 $text = wfMsgForContentNoTrans( $message );
3132                                 if( wfEmptyMsg( $message, $text ) ) {
3133                                         $text = false;
3134                                         break;
3135                                 }
3136                         } else {
3137                                 break;
3138                         }
3139                         if ( $text === false ) {
3140                                 break;
3141                         }
3142                         // Redirect?
3143                         $finalTitle = $title;
3144                         $title = Title::newFromRedirect( $text );
3145                 }
3146                 return array(
3147                         'text' => $text,
3148                         'finalTitle' => $finalTitle,
3149                         'deps' => $deps );
3150         }
3151
3152         /**
3153          * Transclude an interwiki link.
3154          */
3155         function interwikiTransclude( $title, $action ) {
3156                 global $wgEnableScaryTranscluding;
3157
3158                 if (!$wgEnableScaryTranscluding)
3159                         return wfMsg('scarytranscludedisabled');
3160
3161                 $url = $title->getFullUrl( "action=$action" );
3162
3163                 if (strlen($url) > 255)
3164                         return wfMsg('scarytranscludetoolong');
3165                 return $this->fetchScaryTemplateMaybeFromCache($url);
3166         }
3167
3168         function fetchScaryTemplateMaybeFromCache($url) {
3169                 global $wgTranscludeCacheExpiry;
3170                 $dbr = wfGetDB(DB_SLAVE);
3171                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3172                                 array('tc_url' => $url));
3173                 if ($obj) {
3174                         $time = $obj->tc_time;
3175                         $text = $obj->tc_contents;
3176                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3177                                 return $text;
3178                         }
3179                 }
3180
3181                 $text = Http::get($url);
3182                 if (!$text)
3183                         return wfMsg('scarytranscludefailed', $url);
3184
3185                 $dbw = wfGetDB(DB_MASTER);
3186                 $dbw->replace('transcache', array('tc_url'), array(
3187                         'tc_url' => $url,
3188                         'tc_time' => time(),
3189                         'tc_contents' => $text));
3190                 return $text;
3191         }
3192
3193
3194         /**
3195          * Triple brace replacement -- used for template arguments
3196          * @private
3197          */
3198         function argSubstitution( $piece, $frame ) {
3199                 wfProfileIn( __METHOD__ );
3200
3201                 $error = false;
3202                 $parts = $piece['parts'];
3203                 $nameWithSpaces = $frame->expand( $piece['title'] );
3204                 $argName = trim( $nameWithSpaces );
3205                 $object = false;
3206                 $text = $frame->getArgument( $argName );
3207                 if (  $text === false && $parts->getLength() > 0
3208                   && (
3209                     $this->ot['html']
3210                     || $this->ot['pre']
3211                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3212                   )
3213                 ) {
3214                         # No match in frame, use the supplied default
3215                         $object = $parts->item( 0 )->getChildren();
3216                 }
3217                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3218                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3219                         $this->limitationWarn( 'post-expand-template-argument' );
3220                 }
3221
3222                 if ( $text === false && $object === false ) {
3223                         # No match anywhere
3224                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3225                 }
3226                 if ( $error !== false ) {
3227                         $text .= $error;
3228                 }
3229                 if ( $object !== false ) {
3230                         $ret = array( 'object' => $object );
3231                 } else {
3232                         $ret = array( 'text' => $text );
3233                 }
3234
3235                 wfProfileOut( __METHOD__ );
3236                 return $ret;
3237         }
3238
3239         /**
3240          * Return the text to be used for a given extension tag.
3241          * This is the ghost of strip().
3242          *
3243          * @param array $params Associative array of parameters:
3244          *     name       PPNode for the tag name
3245          *     attr       PPNode for unparsed text where tag attributes are thought to be
3246          *     attributes Optional associative array of parsed attributes
3247          *     inner      Contents of extension element
3248          *     noClose    Original text did not have a close tag
3249          * @param PPFrame $frame
3250          */
3251         function extensionSubstitution( $params, $frame ) {
3252                 global $wgRawHtml, $wgContLang;
3253
3254                 $name = $frame->expand( $params['name'] );
3255                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3256                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3257
3258                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3259
3260                 if ( $this->ot['html'] ) {
3261                         $name = strtolower( $name );
3262
3263                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3264                         if ( isset( $params['attributes'] ) ) {
3265                                 $attributes = $attributes + $params['attributes'];
3266                         }
3267                         switch ( $name ) {
3268                                 case 'html':
3269                                         if( $wgRawHtml ) {
3270                                                 $output = $content;
3271                                                 break;
3272                                         } else {
3273                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3274                                         }
3275                                 case 'nowiki':
3276                                         $output = Xml::escapeTagsOnly( $content );
3277                                         break;
3278                                 case 'math':
3279                                         $output = $wgContLang->armourMath(
3280                                                 MathRenderer::renderMath( $content, $attributes ) );
3281                                         break;
3282                                 case 'gallery':
3283                                         $output = $this->renderImageGallery( $content, $attributes );
3284                                         break;
3285                                 default:
3286                                         if( isset( $this->mTagHooks[$name] ) ) {
3287                                                 # Workaround for PHP bug 35229 and similar
3288                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3289                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3290                                                 }
3291                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3292                                                         array( $content, $attributes, $this ) );
3293                                         } else {
3294                                                 $output = '<span class="error">Invalid tag extension name: ' .
3295                                                         htmlspecialchars( $name ) . '</span>';
3296                                         }
3297                         }
3298                 } else {
3299                         if ( is_null( $attrText ) ) {
3300                                 $attrText = '';
3301                         }
3302                         if ( isset( $params['attributes'] ) ) {
3303                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3304                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3305                                                 htmlspecialchars( $attrValue ) . '"';
3306                                 }
3307                         }
3308                         if ( $content === null ) {
3309                                 $output = "<$name$attrText/>";
3310                         } else {
3311                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3312                                 $output = "<$name$attrText>$content$close";
3313                         }
3314                 }
3315
3316                 if ( $name === 'html' || $name === 'nowiki' ) {
3317                         $this->mStripState->nowiki->setPair( $marker, $output );
3318                 } else {
3319                         $this->mStripState->general->setPair( $marker, $output );
3320                 }
3321                 return $marker;
3322         }
3323
3324         /**
3325          * Increment an include size counter
3326          *
3327          * @param string $type The type of expansion
3328          * @param integer $size The size of the text
3329          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3330          */
3331         function incrementIncludeSize( $type, $size ) {
3332                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3333                         return false;
3334                 } else {
3335                         $this->mIncludeSizes[$type] += $size;
3336                         return true;
3337                 }
3338         }
3339
3340         /**
3341          * Increment the expensive function count
3342          *
3343          * @return boolean False if the limit has been exceeded
3344          */
3345         function incrementExpensiveFunctionCount() {
3346                 global $wgExpensiveParserFunctionLimit;
3347                 $this->mExpensiveFunctionCount++;
3348                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3349                         return true;
3350                 }
3351                 return false;
3352         }
3353
3354         /**
3355          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3356          * Fills $this->mDoubleUnderscores, returns the modified text
3357          */
3358         function doDoubleUnderscore( $text ) {
3359                 // The position of __TOC__ needs to be recorded
3360                 $mw = MagicWord::get( 'toc' );
3361                 if( $mw->match( $text ) ) {
3362                         $this->mShowToc = true;
3363                         $this->mForceTocPosition = true;
3364
3365                         // Set a placeholder. At the end we'll fill it in with the TOC.
3366                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3367
3368                         // Only keep the first one.
3369                         $text = $mw->replace( '', $text );
3370                 }
3371
3372                 // Now match and remove the rest of them
3373                 $mwa = MagicWord::getDoubleUnderscoreArray();
3374                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3375
3376                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3377                         $this->mOutput->mNoGallery = true;
3378                 }
3379                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3380                         $this->mShowToc = false;
3381                 }
3382                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3383                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3384
3385                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3386                         if ( $containerCategory ) {
3387                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3388                         } else {
3389                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3390                         }
3391                 }
3392                 # (bug 8068) Allow control over whether robots index a page.
3393                 #
3394                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3395                 # is not desirable, the last one on the page should win.
3396                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3397                         $this->mOutput->setIndexPolicy( 'noindex' );
3398                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3399                         $this->mOutput->setIndexPolicy( 'index' );
3400                 }
3401
3402                 return $text;
3403         }
3404
3405         /**
3406          * This function accomplishes several tasks:
3407          * 1) Auto-number headings if that option is enabled
3408          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3409          * 3) Add a Table of contents on the top for users who have enabled the option
3410          * 4) Auto-anchor headings
3411          *
3412          * It loops through all headlines, collects the necessary data, then splits up the
3413          * string and re-inserts the newly formatted headlines.
3414          *
3415          * @param string $text
3416          * @param boolean $isMain
3417          * @private
3418          */
3419         function formatHeadings( $text, $isMain=true ) {
3420                 global $wgMaxTocLevel, $wgContLang;
3421
3422                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3423                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3424                         $showEditLink = 0;
3425                 } else {
3426                         $showEditLink = $this->mOptions->getEditSection();
3427                 }
3428
3429                 # Inhibit editsection links if requested in the page
3430                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3431                         $showEditLink = 0;
3432                 }
3433
3434                 # Get all headlines for numbering them and adding funky stuff like [edit]
3435                 # links - this is for later, but we need the number of headlines right now
3436                 $matches = array();
3437                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3438
3439                 # if there are fewer than 4 headlines in the article, do not show TOC
3440                 # unless it's been explicitly enabled.
3441                 $enoughToc = $this->mShowToc &&
3442                         (($numMatches >= 4) || $this->mForceTocPosition);
3443
3444                 # Allow user to stipulate that a page should have a "new section"
3445                 # link added via __NEWSECTIONLINK__
3446                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3447                         $this->mOutput->setNewSection( true );
3448                 }
3449
3450                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3451                 # override above conditions and always show TOC above first header
3452                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3453                         $this->mShowToc = true;
3454                         $enoughToc = true;
3455                 }
3456
3457                 # We need this to perform operations on the HTML
3458                 $sk = $this->mOptions->getSkin();
3459
3460                 # headline counter
3461                 $headlineCount = 0;
3462                 $numVisible = 0;
3463
3464                 # Ugh .. the TOC should have neat indentation levels which can be
3465                 # passed to the skin functions. These are determined here
3466                 $toc = '';
3467                 $full = '';
3468                 $head = array();
3469                 $sublevelCount = array();
3470                 $levelCount = array();
3471                 $toclevel = 0;
3472                 $level = 0;
3473                 $prevlevel = 0;
3474                 $toclevel = 0;
3475                 $prevtoclevel = 0;
3476                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3477                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3478                 $tocraw = array();
3479
3480                 foreach( $matches[3] as $headline ) {
3481                         $isTemplate = false;
3482                         $titleText = false;
3483                         $sectionIndex = false;
3484                         $numbering = '';
3485                         $markerMatches = array();
3486                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3487                                 $serial = $markerMatches[1];
3488                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3489                                 $isTemplate = ($titleText != $baseTitleText);
3490                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3491                         }
3492
3493                         if( $toclevel ) {
3494                                 $prevlevel = $level;
3495                                 $prevtoclevel = $toclevel;
3496                         }
3497                         $level = $matches[1][$headlineCount];
3498
3499                         if( $doNumberHeadings || $enoughToc ) {
3500
3501                                 if ( $level > $prevlevel ) {
3502                                         # Increase TOC level
3503                                         $toclevel++;
3504                                         $sublevelCount[$toclevel] = 0;
3505                                         if( $toclevel<$wgMaxTocLevel ) {
3506                                                 $prevtoclevel = $toclevel;
3507                                                 $toc .= $sk->tocIndent();
3508                                                 $numVisible++;
3509                                         }
3510                                 }
3511                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3512                                         # Decrease TOC level, find level to jump to
3513
3514                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3515                                                 # Can only go down to level 1
3516                                                 $toclevel = 1;
3517                                         } else {
3518                                                 for ($i = $toclevel; $i > 0; $i--) {
3519                                                         if ( $levelCount[$i] == $level ) {
3520                                                                 # Found last matching level
3521                                                                 $toclevel = $i;
3522                                                                 break;
3523                                                         }
3524                                                         elseif ( $levelCount[$i] < $level ) {
3525                                                                 # Found first matching level below current level
3526                                                                 $toclevel = $i + 1;
3527                                                                 break;
3528                                                         }
3529                                                 }
3530                                         }
3531                                         if( $toclevel<$wgMaxTocLevel ) {
3532                                                 if($prevtoclevel < $wgMaxTocLevel) {
3533                                                         # Unindent only if the previous toc level was shown :p
3534                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3535                                                         $prevtoclevel = $toclevel;
3536                                                 } else {
3537                                                         $toc .= $sk->tocLineEnd();
3538                                                 }
3539                                         }
3540                                 }
3541                                 else {
3542                                         # No change in level, end TOC line
3543                                         if( $toclevel<$wgMaxTocLevel ) {
3544                                                 $toc .= $sk->tocLineEnd();
3545                                         }
3546                                 }
3547
3548                                 $levelCount[$toclevel] = $level;
3549
3550                                 # count number of headlines for each level
3551                                 @$sublevelCount[$toclevel]++;
3552                                 $dot = 0;
3553                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3554                                         if( !empty( $sublevelCount[$i] ) ) {
3555                                                 if( $dot ) {
3556                                                         $numbering .= '.';
3557                                                 }
3558                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3559                                                 $dot = 1;
3560                                         }
3561                                 }
3562                         }
3563
3564                         # The safe header is a version of the header text safe to use for links
3565                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3566                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3567
3568                         # Remove link placeholders by the link text.
3569                         #     <!--LINK number-->
3570                         # turns into
3571                         #     link text with suffix
3572                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3573
3574                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3575                         $tocline = preg_replace(
3576                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3577                                 array( '',                          '<$1>'),
3578                                 $safeHeadline
3579                         );
3580                         $tocline = trim( $tocline );
3581
3582                         # For the anchor, strip out HTML-y stuff period
3583                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3584                         $safeHeadline = trim( $safeHeadline );
3585
3586                         # Save headline for section edit hint before it's escaped
3587                         $headlineHint = $safeHeadline;
3588                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3589                         # HTML names must be case-insensitively unique (bug 10721)
3590                         $arrayKey = strtolower( $safeHeadline );
3591
3592                         # count how many in assoc. array so we can track dupes in anchors
3593                         isset( $refers[$arrayKey] ) ? $refers[$arrayKey]++ : $refers[$arrayKey] = 1;
3594                         $refcount[$headlineCount] = $refers[$arrayKey];
3595
3596                         # Don't number the heading if it is the only one (looks silly)
3597                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3598                                 # the two are different if the line contains a link
3599                                 $headline=$numbering . ' ' . $headline;
3600                         }
3601
3602                         # Create the anchor for linking from the TOC to the section
3603                         $anchor = $safeHeadline;
3604                         if($refcount[$headlineCount] > 1 ) {
3605                                 $anchor .= '_' . $refcount[$headlineCount];
3606                         }
3607                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3608                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3609                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3610                         }
3611                         # give headline the correct <h#> tag
3612                         if( $showEditLink && $sectionIndex !== false ) {
3613                                 if( $isTemplate ) {
3614                                         # Put a T flag in the section identifier, to indicate to extractSections()
3615                                         # that sections inside <includeonly> should be counted.
3616                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3617                                 } else {
3618                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3619                                 }
3620                         } else {
3621                                 $editlink = '';
3622                         }
3623                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3624
3625                         $headlineCount++;
3626                 }
3627
3628                 $this->mOutput->setSections( $tocraw );
3629
3630                 # Never ever show TOC if no headers
3631                 if( $numVisible < 1 ) {
3632                         $enoughToc = false;
3633                 }
3634
3635                 if( $enoughToc ) {
3636                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3637                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3638                         }
3639                         $toc = $sk->tocList( $toc );
3640                 }
3641
3642                 # split up and insert constructed headlines
3643
3644                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3645                 $i = 0;
3646
3647                 foreach( $blocks as $block ) {
3648                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3649                                 # This is the [edit] link that appears for the top block of text when
3650                                 # section editing is enabled
3651
3652                                 # Disabled because it broke block formatting
3653                                 # For example, a bullet point in the top line
3654                                 # $full .= $sk->editSectionLink(0);
3655                         }
3656                         $full .= $block;
3657                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3658                                 # Top anchor now in skin
3659                                 $full = $full.$toc;
3660                         }
3661
3662                         if( !empty( $head[$i] ) ) {
3663                                 $full .= $head[$i];
3664                         }
3665                         $i++;
3666                 }
3667                 if( $this->mForceTocPosition ) {
3668                         return str_replace( '<!--MWTOC-->', $toc, $full );
3669                 } else {
3670                         return $full;
3671                 }
3672         }
3673
3674         /**
3675          * Transform wiki markup when saving a page by doing \r\n -> \n
3676          * conversion, substitting signatures, {{subst:}} templates, etc.
3677          *
3678          * @param string $text the text to transform
3679          * @param Title &$title the Title object for the current article
3680          * @param User &$user the User object describing the current user
3681          * @param ParserOptions $options parsing options
3682          * @param bool $clearState whether to clear the parser state first
3683          * @return string the altered wiki markup
3684          * @public
3685          */
3686         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3687                 $this->mOptions = $options;
3688                 $this->setTitle( $title );
3689                 $this->setOutputType( self::OT_WIKI );
3690
3691                 if ( $clearState ) {
3692                         $this->clearState();
3693                 }
3694
3695                 $pairs = array(
3696                         "\r\n" => "\n",
3697                 );
3698                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3699                 $text = $this->pstPass2( $text, $user );
3700                 $text = $this->mStripState->unstripBoth( $text );
3701                 return $text;
3702         }
3703
3704         /**
3705          * Pre-save transform helper function
3706          * @private
3707          */
3708         function pstPass2( $text, $user ) {
3709                 global $wgContLang, $wgLocaltimezone;
3710
3711                 /* Note: This is the timestamp saved as hardcoded wikitext to
3712                  * the database, we use $wgContLang here in order to give
3713                  * everyone the same signature and use the default one rather
3714                  * than the one selected in each user's preferences.
3715                  *
3716                  * (see also bug 12815)
3717                  */
3718                 $ts = $this->mOptions->getTimestamp();
3719                 $tz = wfMsgForContent( 'timezone-utc' );
3720                 if ( isset( $wgLocaltimezone ) ) {
3721                         $unixts = wfTimestamp( TS_UNIX, $ts );
3722                         $oldtz = getenv( 'TZ' );
3723                         putenv( 'TZ='.$wgLocaltimezone );
3724                         $ts = date( 'YmdHis', $unixts );
3725                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3726                         putenv( 'TZ='.$oldtz );
3727                 }
3728
3729                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3730
3731                 # Variable replacement
3732                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3733                 $text = $this->replaceVariables( $text );
3734
3735                 # Signatures
3736                 $sigText = $this->getUserSig( $user );
3737                 $text = strtr( $text, array(
3738                         '~~~~~' => $d,
3739                         '~~~~' => "$sigText $d",
3740                         '~~~' => $sigText
3741                 ) );
3742
3743                 # Context links: [[|name]] and [[name (context)|]]
3744                 #
3745                 global $wgLegalTitleChars;
3746                 $tc = "[$wgLegalTitleChars]";
3747                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3748
3749                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3750                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3751                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3752                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3753
3754                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3755                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3756                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3757                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3758
3759                 $t = $this->mTitle->getText();
3760                 $m = array();
3761                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3762                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3763                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3764                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3765                 } else {
3766                         # if there's no context, don't bother duplicating the title
3767                         $text = preg_replace( $p2, '[[\\1]]', $text );
3768                 }
3769
3770                 # Trim trailing whitespace
3771                 $text = rtrim( $text );
3772
3773                 return $text;
3774         }
3775
3776         /**
3777          * Fetch the user's signature text, if any, and normalize to
3778          * validated, ready-to-insert wikitext.
3779          *
3780          * @param User $user
3781          * @return string
3782          * @private
3783          */
3784         function getUserSig( &$user ) {
3785                 global $wgMaxSigChars;
3786
3787                 $username = $user->getName();
3788                 $nickname = $user->getOption( 'nickname' );
3789                 $nickname = $nickname === '' ? $username : $nickname;
3790
3791                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3792                         $nickname = $username;
3793                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3794                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3795                         # Sig. might contain markup; validate this
3796                         if( $this->validateSig( $nickname ) !== false ) {
3797                                 # Validated; clean up (if needed) and return it
3798                                 return $this->cleanSig( $nickname, true );
3799                         } else {
3800                                 # Failed to validate; fall back to the default
3801                                 $nickname = $username;
3802                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3803                         }
3804                 }
3805
3806                 // Make sure nickname doesnt get a sig in a sig
3807                 $nickname = $this->cleanSigInSig( $nickname );
3808
3809                 # If we're still here, make it a link to the user page
3810                 $userText = wfEscapeWikiText( $username );
3811                 $nickText = wfEscapeWikiText( $nickname );
3812                 if ( $user->isAnon() )  {
3813                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3814                 } else {
3815                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3816                 }
3817         }
3818
3819         /**
3820          * Check that the user's signature contains no bad XML
3821          *
3822          * @param string $text
3823          * @return mixed An expanded string, or false if invalid.
3824          */
3825         function validateSig( $text ) {
3826                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3827         }
3828
3829         /**
3830          * Clean up signature text
3831          *
3832          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3833          * 2) Substitute all transclusions
3834          *
3835          * @param string $text
3836          * @param $parsing Whether we're cleaning (preferences save) or parsing
3837          * @return string Signature text
3838          */
3839         function cleanSig( $text, $parsing = false ) {
3840                 if ( !$parsing ) {
3841                         global $wgTitle;
3842                         $this->clearState();
3843                         $this->setTitle( $wgTitle );
3844                         $this->mOptions = new ParserOptions;
3845                         $this->setOutputType = self::OT_PREPROCESS;
3846                 }
3847
3848                 # Option to disable this feature
3849                 if ( !$this->mOptions->getCleanSignatures() ) {
3850                         return $text;
3851                 }
3852
3853                 # FIXME: regex doesn't respect extension tags or nowiki
3854                 #  => Move this logic to braceSubstitution()
3855                 $substWord = MagicWord::get( 'subst' );
3856                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3857                 $substText = '{{' . $substWord->getSynonym( 0 );
3858
3859                 $text = preg_replace( $substRegex, $substText, $text );
3860                 $text = $this->cleanSigInSig( $text );
3861                 $dom = $this->preprocessToDom( $text );
3862                 $frame = $this->getPreprocessor()->newFrame();
3863                 $text = $frame->expand( $dom );
3864
3865                 if ( !$parsing ) {
3866                         $text = $this->mStripState->unstripBoth( $text );
3867                 }
3868
3869                 return $text;
3870         }
3871
3872         /**
3873          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3874          * @param string $text
3875          * @return string Signature text with /~{3,5}/ removed
3876          */
3877         function cleanSigInSig( $text ) {
3878                 $text = preg_replace( '/~{3,5}/', '', $text );
3879                 return $text;
3880         }
3881
3882         /**
3883          * Set up some variables which are usually set up in parse()
3884          * so that an external function can call some class members with confidence
3885          * @public
3886          */
3887         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3888                 $this->setTitle( $title );
3889                 $this->mOptions = $options;
3890                 $this->setOutputType( $outputType );
3891                 if ( $clearState ) {
3892                         $this->clearState();
3893                 }
3894         }
3895
3896         /**
3897          * Wrapper for preprocess()
3898          *
3899          * @param string $text the text to preprocess
3900          * @param ParserOptions $options  options
3901          * @return string
3902          * @public
3903          */
3904         function transformMsg( $text, $options ) {
3905                 global $wgTitle;
3906                 static $executing = false;
3907
3908                 # Guard against infinite recursion
3909                 if ( $executing ) {
3910                         return $text;
3911                 }
3912                 $executing = true;
3913
3914                 wfProfileIn(__METHOD__);
3915                 $text = $this->preprocess( $text, $wgTitle, $options );
3916
3917                 $executing = false;
3918                 wfProfileOut(__METHOD__);
3919                 return $text;
3920         }
3921
3922         /**
3923          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3924          * The callback should have the following form:
3925          *    function myParserHook( $text, $params, &$parser ) { ... }
3926          *
3927          * Transform and return $text. Use $parser for any required context, e.g. use
3928          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3929          *
3930          * @public
3931          *
3932          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3933          * @param mixed $callback The callback function (and object) to use for the tag
3934          *
3935          * @return The old value of the mTagHooks array associated with the hook
3936          */
3937         function setHook( $tag, $callback ) {
3938                 $tag = strtolower( $tag );
3939                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3940                 $this->mTagHooks[$tag] = $callback;
3941                 if( !in_array( $tag, $this->mStripList ) ) {
3942                         $this->mStripList[] = $tag;
3943                 }
3944
3945                 return $oldVal;
3946         }
3947
3948         function setTransparentTagHook( $tag, $callback ) {
3949                 $tag = strtolower( $tag );
3950                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3951                 $this->mTransparentTagHooks[$tag] = $callback;
3952
3953                 return $oldVal;
3954         }
3955
3956         /**
3957          * Remove all tag hooks
3958          */
3959         function clearTagHooks() {
3960                 $this->mTagHooks = array();
3961                 $this->mStripList = $this->mDefaultStripList;
3962         }
3963
3964         /**
3965          * Create a function, e.g. {{sum:1|2|3}}
3966          * The callback function should have the form:
3967          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3968          *
3969          * Or with SFH_OBJECT_ARGS:
3970          *    function myParserFunction( $parser, $frame, $args ) { ... }
3971          *
3972          * The callback may either return the text result of the function, or an array with the text
3973          * in element 0, and a number of flags in the other elements. The names of the flags are
3974          * specified in the keys. Valid flags are:
3975          *   found                     The text returned is valid, stop processing the template. This
3976          *                             is on by default.
3977          *   nowiki                    Wiki markup in the return value should be escaped
3978          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3979          *
3980          * @public
3981          *
3982          * @param string $id The magic word ID
3983          * @param mixed $callback The callback function (and object) to use
3984          * @param integer $flags a combination of the following flags:
3985          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3986          *
3987          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
3988          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
3989          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
3990          *     the arguments, and to control the way they are expanded.
3991          *
3992          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
3993          *     arguments, for instance:
3994          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
3995          *
3996          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
3997          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
3998          *     working if/when this is changed.
3999          *
4000          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4001          *     expansion.
4002          *
4003          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4004          *     about the methods available in PPFrame and PPNode.
4005          *
4006          * @return The old callback function for this name, if any
4007          */
4008         function setFunctionHook( $id, $callback, $flags = 0 ) {
4009                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4010                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4011
4012                 # Add to function cache
4013                 $mw = MagicWord::get( $id );
4014                 if( !$mw )
4015                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4016
4017                 $synonyms = $mw->getSynonyms();
4018                 $sensitive = intval( $mw->isCaseSensitive() );
4019
4020                 foreach ( $synonyms as $syn ) {
4021                         # Case
4022                         if ( !$sensitive ) {
4023                                 $syn = strtolower( $syn );
4024                         }
4025                         # Add leading hash
4026                         if ( !( $flags & SFH_NO_HASH ) ) {
4027                                 $syn = '#' . $syn;
4028                         }
4029                         # Remove trailing colon
4030                         if ( substr( $syn, -1, 1 ) === ':' ) {
4031                                 $syn = substr( $syn, 0, -1 );
4032                         }
4033                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4034                 }
4035                 return $oldVal;
4036         }
4037
4038         /**
4039          * Get all registered function hook identifiers
4040          *
4041          * @return array
4042          */
4043         function getFunctionHooks() {
4044                 return array_keys( $this->mFunctionHooks );
4045         }
4046
4047         /**
4048          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4049          * Placeholders created in Skin::makeLinkObj()
4050          * Returns an array of link CSS classes, indexed by PDBK.
4051          */
4052         function replaceLinkHolders( &$text, $options = 0 ) {
4053                 return $this->mLinkHolders->replace( $text );
4054         }
4055
4056         /**
4057          * Replace <!--LINK--> link placeholders with plain text of links
4058          * (not HTML-formatted).
4059          * @param string $text
4060          * @return string
4061          */
4062         function replaceLinkHoldersText( $text ) {
4063                 return $this->mLinkHolders->replaceText( $text );
4064         }
4065
4066         /**
4067          * Tag hook handler for 'pre'.
4068          */
4069         function renderPreTag( $text, $attribs ) {
4070                 // Backwards-compatibility hack
4071                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4072
4073                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4074                 return wfOpenElement( 'pre', $attribs ) .
4075                         Xml::escapeTagsOnly( $content ) .
4076                         '</pre>';
4077         }
4078
4079         /**
4080          * Renders an image gallery from a text with one line per image.
4081          * text labels may be given by using |-style alternative text. E.g.
4082          *   Image:one.jpg|The number "1"
4083          *   Image:tree.jpg|A tree
4084          * given as text will return the HTML of a gallery with two images,
4085          * labeled 'The number "1"' and
4086          * 'A tree'.
4087          */
4088         function renderImageGallery( $text, $params ) {
4089                 $ig = new ImageGallery();
4090                 $ig->setContextTitle( $this->mTitle );
4091                 $ig->setShowBytes( false );
4092                 $ig->setShowFilename( false );
4093                 $ig->setParser( $this );
4094                 $ig->setHideBadImages();
4095                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4096                 $ig->useSkin( $this->mOptions->getSkin() );
4097                 $ig->mRevisionId = $this->mRevisionId;
4098
4099                 if( isset( $params['caption'] ) ) {
4100                         $caption = $params['caption'];
4101                         $caption = htmlspecialchars( $caption );
4102                         $caption = $this->replaceInternalLinks( $caption );
4103                         $ig->setCaptionHtml( $caption );
4104                 }
4105                 if( isset( $params['perrow'] ) ) {
4106                         $ig->setPerRow( $params['perrow'] );
4107                 }
4108                 if( isset( $params['widths'] ) ) {
4109                         $ig->setWidths( $params['widths'] );
4110                 }
4111                 if( isset( $params['heights'] ) ) {
4112                         $ig->setHeights( $params['heights'] );
4113                 }
4114
4115                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4116
4117                 $lines = StringUtils::explode( "\n", $text );
4118                 foreach ( $lines as $line ) {
4119                         # match lines like these:
4120                         # Image:someimage.jpg|This is some image
4121                         $matches = array();
4122                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4123                         # Skip empty lines
4124                         if ( count( $matches ) == 0 ) {
4125                                 continue;
4126                         }
4127
4128                         if ( strpos( $matches[0], '%' ) !== false )
4129                                 $matches[1] = urldecode( $matches[1] );
4130                         $tp = Title::newFromText( $matches[1]/*, NS_IMAGE*/ );
4131                         $nt =& $tp;
4132                         if( is_null( $nt ) ) {
4133                                 # Bogus title. Ignore these so we don't bomb out later.
4134                                 continue;
4135                         }
4136                         if ( isset( $matches[3] ) ) {
4137                                 $label = $matches[3];
4138                         } else {
4139                                 $label = '';
4140                         }
4141
4142                         $html = $this->recursiveTagParse( trim( $label ) );
4143
4144                         $ig->add( $nt, $html );
4145
4146                         # Only add real images (bug #5586)
4147                         if ( $nt->getNamespace() == NS_IMAGE ) {
4148                                 $this->mOutput->addImage( $nt->getDBkey() );
4149                         }
4150                 }
4151                 return $ig->toHTML();
4152         }
4153
4154         function getImageParams( $handler ) {
4155                 if ( $handler ) {
4156                         $handlerClass = get_class( $handler );
4157                 } else {
4158                         $handlerClass = '';
4159                 }
4160                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4161                         // Initialise static lists
4162                         static $internalParamNames = array(
4163                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4164                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4165                                         'bottom', 'text-bottom' ),
4166                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4167                                         'upright', 'border', 'alt' ),
4168                         );
4169                         static $internalParamMap;
4170                         if ( !$internalParamMap ) {
4171                                 $internalParamMap = array();
4172                                 foreach ( $internalParamNames as $type => $names ) {
4173                                         foreach ( $names as $name ) {
4174                                                 $magicName = str_replace( '-', '_', "img_$name" );
4175                                                 $internalParamMap[$magicName] = array( $type, $name );
4176                                         }
4177                                 }
4178                         }
4179
4180                         // Add handler params
4181                         $paramMap = $internalParamMap;
4182                         if ( $handler ) {
4183                                 $handlerParamMap = $handler->getParamMap();
4184                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4185                                         $paramMap[$magic] = array( 'handler', $paramName );
4186                                 }
4187                         }
4188                         $this->mImageParams[$handlerClass] = $paramMap;
4189                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4190                 }
4191                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4192         }
4193
4194         /**
4195          * Parse image options text and use it to make an image
4196          * @param Title $title
4197          * @param string $options
4198          * @param LinkHolderArray $holders
4199          */
4200         function makeImage( $title, $options, $holders = false ) {
4201                 # Check if the options text is of the form "options|alt text"
4202                 # Options are:
4203                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4204                 #  * left       no resizing, just left align. label is used for alt= only
4205                 #  * right      same, but right aligned
4206                 #  * none       same, but not aligned
4207                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4208                 #  * center     center the image
4209                 #  * framed     Keep original image size, no magnify-button.
4210                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4211                 #  * upright    reduce width for upright images, rounded to full __0 px
4212                 #  * border     draw a 1px border around the image
4213                 #  * alt        Text for HTML alt attribute (defaults to empty)
4214                 # vertical-align values (no % or length right now):
4215                 #  * baseline
4216                 #  * sub
4217                 #  * super
4218                 #  * top
4219                 #  * text-top
4220                 #  * middle
4221                 #  * bottom
4222                 #  * text-bottom
4223
4224                 $parts = StringUtils::explode( "|", $options );
4225                 $sk = $this->mOptions->getSkin();
4226
4227                 # Give extensions a chance to select the file revision for us
4228                 $skip = $time = $descQuery = false;
4229                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4230
4231                 if ( $skip ) {
4232                         return $sk->link( $title );
4233                 }
4234
4235                 # Get parameter map
4236                 $file = wfFindFile( $title, $time );
4237                 $handler = $file ? $file->getHandler() : false;
4238
4239                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4240
4241                 # Process the input parameters
4242                 $caption = '';
4243                 $params = array( 'frame' => array(), 'handler' => array(),
4244                         'horizAlign' => array(), 'vertAlign' => array() );
4245                 foreach( $parts as $part ) {
4246                         $part = trim( $part );
4247                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4248                         $validated = false;
4249                         if( isset( $paramMap[$magicName] ) ) {
4250                                 list( $type, $paramName ) = $paramMap[$magicName];
4251
4252                                 // Special case; width and height come in one variable together
4253                                 if( $type === 'handler' && $paramName === 'width' ) {
4254                                         $m = array();
4255                                         # (bug 13500) In both cases (width/height and width only),
4256                                         # permit trailing "px" for backward compatibility.
4257                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4258                                                 $width = intval( $m[1] );
4259                                                 $height = intval( $m[2] );
4260                                                 if ( $handler->validateParam( 'width', $width ) ) {
4261                                                         $params[$type]['width'] = $width;
4262                                                         $validated = true;
4263                                                 }
4264                                                 if ( $handler->validateParam( 'height', $height ) ) {
4265                                                         $params[$type]['height'] = $height;
4266                                                         $validated = true;
4267                                                 }
4268                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4269                                                 $width = intval( $value );
4270                                                 if ( $handler->validateParam( 'width', $width ) ) {
4271                                                         $params[$type]['width'] = $width;
4272                                                         $validated = true;
4273                                                 }
4274                                         } // else no validation -- bug 13436
4275                                 } else {
4276                                         if ( $type === 'handler' ) {
4277                                                 # Validate handler parameter
4278                                                 $validated = $handler->validateParam( $paramName, $value );
4279                                         } else {
4280                                                 # Validate internal parameters
4281                                                 switch( $paramName ) {
4282                                                 case 'manualthumb':
4283                                                 case 'alt':
4284                                                         /// @fixme - possibly check validity here?
4285                                                         /// downstream behavior seems odd with missing manual thumbs.
4286                                                         $validated = true;
4287                                                         break;
4288                                                 default:
4289                                                         // Most other things appear to be empty or numeric...
4290                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4291                                                 }
4292                                         }
4293
4294                                         if ( $validated ) {
4295                                                 $params[$type][$paramName] = $value;
4296                                         }
4297                                 }
4298                         }
4299                         if ( !$validated ) {
4300                                 $caption = $part;
4301                         }
4302                 }
4303
4304                 # Process alignment parameters
4305                 if ( $params['horizAlign'] ) {
4306                         $params['frame']['align'] = key( $params['horizAlign'] );
4307                 }
4308                 if ( $params['vertAlign'] ) {
4309                         $params['frame']['valign'] = key( $params['vertAlign'] );
4310                 }
4311
4312                 $params['frame']['caption'] = $caption;
4313
4314                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4315
4316                 # Linker does the rest
4317                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4318
4319                 # Give the handler a chance to modify the parser object
4320                 if ( $handler ) {
4321                         $handler->parserTransformHook( $this, $file );
4322                 }
4323
4324                 return $ret;
4325         }
4326
4327         /**
4328          * Set a flag in the output object indicating that the content is dynamic and
4329          * shouldn't be cached.
4330          */
4331         function disableCache() {
4332                 wfDebug( "Parser output marked as uncacheable.\n" );
4333                 $this->mOutput->mCacheTime = -1;
4334         }
4335
4336         /**#@+
4337          * Callback from the Sanitizer for expanding items found in HTML attribute
4338          * values, so they can be safely tested and escaped.
4339          * @param string $text
4340          * @param PPFrame $frame
4341          * @return string
4342          * @private
4343          */
4344         function attributeStripCallback( &$text, $frame = false ) {
4345                 $text = $this->replaceVariables( $text, $frame );
4346                 $text = $this->mStripState->unstripBoth( $text );
4347                 return $text;
4348         }
4349
4350         /**#@-*/
4351
4352         /**#@+
4353          * Accessor/mutator
4354          */
4355         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4356         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4357         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4358         /**#@-*/
4359
4360         /**#@+
4361          * Accessor
4362          */
4363         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4364         /**#@-*/
4365
4366
4367         /**
4368          * Break wikitext input into sections, and either pull or replace
4369          * some particular section's text.
4370          *
4371          * External callers should use the getSection and replaceSection methods.
4372          *
4373          * @param string $text Page wikitext
4374          * @param string $section A section identifier string of the form:
4375          *   <flag1> - <flag2> - ... - <section number>
4376          *
4377          * Currently the only recognised flag is "T", which means the target section number
4378          * was derived during a template inclusion parse, in other words this is a template
4379          * section edit link. If no flags are given, it was an ordinary section edit link.
4380          * This flag is required to avoid a section numbering mismatch when a section is
4381          * enclosed by <includeonly> (bug 6563).
4382          *
4383          * The section number 0 pulls the text before the first heading; other numbers will
4384          * pull the given section along with its lower-level subsections. If the section is
4385          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4386          *
4387          * @param string $mode One of "get" or "replace"
4388          * @param string $newText Replacement text for section data.
4389          * @return string for "get", the extracted section text.
4390          *                for "replace", the whole page with the section replaced.
4391          */
4392         private function extractSections( $text, $section, $mode, $newText='' ) {
4393                 global $wgTitle;
4394                 $this->clearState();
4395                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4396                 $this->mOptions = new ParserOptions;
4397                 $this->setOutputType( self::OT_WIKI );
4398                 $outText = '';
4399                 $frame = $this->getPreprocessor()->newFrame();
4400
4401                 // Process section extraction flags
4402                 $flags = 0;
4403                 $sectionParts = explode( '-', $section );
4404                 $sectionIndex = array_pop( $sectionParts );
4405                 foreach ( $sectionParts as $part ) {
4406                         if ( $part === 'T' ) {
4407                                 $flags |= self::PTD_FOR_INCLUSION;
4408                         }
4409                 }
4410                 // Preprocess the text
4411                 $root = $this->preprocessToDom( $text, $flags );
4412
4413                 // <h> nodes indicate section breaks
4414                 // They can only occur at the top level, so we can find them by iterating the root's children
4415                 $node = $root->getFirstChild();
4416
4417                 // Find the target section
4418                 if ( $sectionIndex == 0 ) {
4419                         // Section zero doesn't nest, level=big
4420                         $targetLevel = 1000;
4421                 } else {
4422             while ( $node ) {
4423                 if ( $node->getName() === 'h' ) {
4424                     $bits = $node->splitHeading();
4425                                         if ( $bits['i'] == $sectionIndex ) {
4426                                         $targetLevel = $bits['level'];
4427                                                 break;
4428                                         }
4429                                 }
4430                                 if ( $mode === 'replace' ) {
4431                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4432                                 }
4433                                 $node = $node->getNextSibling();
4434                         }
4435                 }
4436
4437                 if ( !$node ) {
4438                         // Not found
4439                         if ( $mode === 'get' ) {
4440                                 return $newText;
4441                         } else {
4442                                 return $text;
4443                         }
4444                 }
4445
4446                 // Find the end of the section, including nested sections
4447                 do {
4448                         if ( $node->getName() === 'h' ) {
4449                                 $bits = $node->splitHeading();
4450                                 $curLevel = $bits['level'];
4451                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4452                                         break;
4453                                 }
4454                         }
4455                         if ( $mode === 'get' ) {
4456                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4457                         }
4458                         $node = $node->getNextSibling();
4459                 } while ( $node );
4460
4461                 // Write out the remainder (in replace mode only)
4462                 if ( $mode === 'replace' ) {
4463                         // Output the replacement text
4464                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4465                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4466                         $outText .= $newText . "\n\n";
4467                         while ( $node ) {
4468                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4469                                 $node = $node->getNextSibling();
4470                         }
4471                 }
4472
4473                 if ( is_string( $outText ) ) {
4474                         // Re-insert stripped tags
4475                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4476                 }
4477
4478                 return $outText;
4479         }
4480
4481         /**
4482          * This function returns the text of a section, specified by a number ($section).
4483          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4484          * the first section before any such heading (section 0).
4485          *
4486          * If a section contains subsections, these are also returned.
4487          *
4488          * @param string $text text to look in
4489          * @param string $section section identifier
4490          * @param string $deftext default to return if section is not found
4491          * @return string text of the requested section
4492          */
4493         public function getSection( $text, $section, $deftext='' ) {
4494                 return $this->extractSections( $text, $section, "get", $deftext );
4495         }
4496
4497         public function replaceSection( $oldtext, $section, $text ) {
4498                 return $this->extractSections( $oldtext, $section, "replace", $text );
4499         }
4500
4501         /**
4502          * Get the timestamp associated with the current revision, adjusted for
4503          * the default server-local timestamp
4504          */
4505         function getRevisionTimestamp() {
4506                 if ( is_null( $this->mRevisionTimestamp ) ) {
4507                         wfProfileIn( __METHOD__ );
4508                         global $wgContLang;
4509                         $dbr = wfGetDB( DB_SLAVE );
4510                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4511                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4512
4513                         // Normalize timestamp to internal MW format for timezone processing.
4514                         // This has the added side-effect of replacing a null value with
4515                         // the current time, which gives us more sensible behavior for
4516                         // previews.
4517                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4518
4519                         // The cryptic '' timezone parameter tells to use the site-default
4520                         // timezone offset instead of the user settings.
4521                         //
4522                         // Since this value will be saved into the parser cache, served
4523                         // to other users, and potentially even used inside links and such,
4524                         // it needs to be consistent for all visitors.
4525                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4526
4527                         wfProfileOut( __METHOD__ );
4528                 }
4529                 return $this->mRevisionTimestamp;
4530         }
4531
4532         /**
4533          * Mutator for $mDefaultSort
4534          *
4535          * @param $sort New value
4536          */
4537         public function setDefaultSort( $sort ) {
4538                 $this->mDefaultSort = $sort;
4539         }
4540
4541         /**
4542          * Accessor for $mDefaultSort
4543          * Will use the title/prefixed title if none is set
4544          *
4545          * @return string
4546          */
4547         public function getDefaultSort() {
4548                 global $wgCategoryPrefixedDefaultSortkey;
4549                 if( $this->mDefaultSort !== false ) {
4550                         return $this->mDefaultSort;
4551                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4552                         !$wgCategoryPrefixedDefaultSortkey) {
4553                         return $this->mTitle->getText();
4554                 } else {
4555                         return $this->mTitle->getPrefixedText();
4556                 }
4557         }
4558
4559         /**
4560          * Try to guess the section anchor name based on a wikitext fragment
4561          * presumably extracted from a heading, for example "Header" from
4562          * "== Header ==".
4563          */
4564         public function guessSectionNameFromWikiText( $text ) {
4565                 # Strip out wikitext links(they break the anchor)
4566                 $text = $this->stripSectionName( $text );
4567                 $headline = Sanitizer::decodeCharReferences( $text );
4568                 # strip out HTML
4569                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4570                 $headline = trim( $headline );
4571                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4572                 $replacearray = array(
4573                         '%3A' => ':',
4574                         '%' => '.'
4575                 );
4576                 return str_replace(
4577                         array_keys( $replacearray ),
4578                         array_values( $replacearray ),
4579                         $sectionanchor );
4580         }
4581
4582         /**
4583          * Strips a text string of wikitext for use in a section anchor
4584          *
4585          * Accepts a text string and then removes all wikitext from the
4586          * string and leaves only the resultant text (i.e. the result of
4587          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4588          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4589          * to create valid section anchors by mimicing the output of the
4590          * parser when headings are parsed.
4591          *
4592          * @param $text string Text string to be stripped of wikitext
4593          * for use in a Section anchor
4594          * @return Filtered text string
4595          */
4596         public function stripSectionName( $text ) {
4597                 # Strip internal link markup
4598                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4599                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4600
4601                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4602                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4603                 # on how many empty links there are on the page - need to figure that out.
4604                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4605
4606                 # Parse wikitext quotes (italics & bold)
4607                 $text = $this->doQuotes($text);
4608
4609                 # Strip HTML tags
4610                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4611                 return $text;
4612         }
4613
4614         function srvus( $text ) {
4615                 return $this->testSrvus( $text, $this->mOutputType );
4616         }
4617
4618         /**
4619          * strip/replaceVariables/unstrip for preprocessor regression testing
4620          */
4621         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4622                 $this->clearState();
4623                 if ( ! ( $title instanceof Title ) ) {
4624                         $title = Title::newFromText( $title );
4625                 }
4626                 $this->mTitle = $title;
4627                 $this->mOptions = $options;
4628                 $this->setOutputType( $outputType );
4629                 $text = $this->replaceVariables( $text );
4630                 $text = $this->mStripState->unstripBoth( $text );
4631                 $text = Sanitizer::removeHTMLtags( $text );
4632                 return $text;
4633         }
4634
4635         function testPst( $text, $title, $options ) {
4636                 global $wgUser;
4637                 if ( ! ( $title instanceof Title ) ) {
4638                         $title = Title::newFromText( $title );
4639                 }
4640                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4641         }
4642
4643         function testPreprocess( $text, $title, $options ) {
4644                 if ( ! ( $title instanceof Title ) ) {
4645                         $title = Title::newFromText( $title );
4646                 }
4647                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4648         }
4649
4650         function markerSkipCallback( $s, $callback ) {
4651                 $i = 0;
4652                 $out = '';
4653                 while ( $i < strlen( $s ) ) {
4654                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4655                         if ( $markerStart === false ) {
4656                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4657                                 break;
4658                         } else {
4659                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4660                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4661                                 if ( $markerEnd === false ) {
4662                                         $out .= substr( $s, $markerStart );
4663                                         break;
4664                                 } else {
4665                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4666                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4667                                         $i = $markerEnd;
4668                                 }
4669                         }
4670                 }
4671                 return $out;
4672         }
4673 }
4674
4675 /**
4676  * @todo document, briefly.
4677  * @ingroup Parser
4678  */
4679 class StripState {
4680         var $general, $nowiki;
4681
4682         function __construct() {
4683                 $this->general = new ReplacementArray;
4684                 $this->nowiki = new ReplacementArray;
4685         }
4686
4687         function unstripGeneral( $text ) {
4688                 wfProfileIn( __METHOD__ );
4689                 do {
4690                         $oldText = $text;
4691                         $text = $this->general->replace( $text );
4692                 } while ( $text !== $oldText );
4693                 wfProfileOut( __METHOD__ );
4694                 return $text;
4695         }
4696
4697         function unstripNoWiki( $text ) {
4698                 wfProfileIn( __METHOD__ );
4699                 do {
4700                         $oldText = $text;
4701                         $text = $this->nowiki->replace( $text );
4702                 } while ( $text !== $oldText );
4703                 wfProfileOut( __METHOD__ );
4704                 return $text;
4705         }
4706
4707         function unstripBoth( $text ) {
4708                 wfProfileIn( __METHOD__ );
4709                 do {
4710                         $oldText = $text;
4711                         $text = $this->general->replace( $text );
4712                         $text = $this->nowiki->replace( $text );
4713                 } while ( $text !== $oldText );
4714                 wfProfileOut( __METHOD__ );
4715                 return $text;
4716         }
4717 }
4718
4719 /**
4720  * @todo document, briefly.
4721  * @ingroup Parser
4722  */
4723 class OnlyIncludeReplacer {
4724         var $output = '';
4725
4726         function replace( $matches ) {
4727                 if ( substr( $matches[1], -1 ) === "\n" ) {
4728                         $this->output .= substr( $matches[1], 0, -1 );
4729                 } else {
4730                         $this->output .= $matches[1];
4731                 }
4732         }
4733 }