includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * preprocess()
  24  *   removes HTML comments and expands templates
  25  * cleanSig()
  26  *   Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *   Extracts sections from an article for section editing
  29  *
  30  * Globals used:
  31  *    objects:   $wgLang, $wgContLang
  32  *
  33  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  34  *
  35  * settings:
  36  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  37  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  38  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  39  *  $wgMaxArticleSize*
  40  *
  41  *  * only within ParserOptions
  42  * </pre>
  43  *
  44  * @ingroup Parser
  45  */
  46 class Parser
  47 {
  48         /**
  49          * Update this version number when the ParserOutput format
  50          * changes in an incompatible way, so the parser cache
  51          * can automatically discard old data.
  52          */
  53         const VERSION = '1.6.4';
  54
  55         # Flags for Parser::setFunctionHook
  56         # Also available as global constants from Defines.php
  57         const SFH_NO_HASH = 1;
  58         const SFH_OBJECT_ARGS = 2;
  59
  60         # Constants needed for external link processing
  61         # Everything except bracket, space, or control characters
  62         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  63         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  64                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  65
  66         // State constants for the definition list colon extraction
  67         const COLON_STATE_TEXT = 0;
  68         const COLON_STATE_TAG = 1;
  69         const COLON_STATE_TAGSTART = 2;
  70         const COLON_STATE_CLOSETAG = 3;
  71         const COLON_STATE_TAGSLASH = 4;
  72         const COLON_STATE_COMMENT = 5;
  73         const COLON_STATE_COMMENTDASH = 6;
  74         const COLON_STATE_COMMENTDASHDASH = 7;
  75
  76         // Flags for preprocessToDom
  77         const PTD_FOR_INCLUSION = 1;
  78
  79         // Allowed values for $this->mOutputType
  80         // Parameter to startExternalParse().
  81         const OT_HTML = 1;
  82         const OT_WIKI = 2;
  83         const OT_PREPROCESS = 3;
  84         const OT_MSG = 3;
  85
  86         // Marker Suffix needs to be accessible staticly.
  87         const MARKER_SUFFIX = "-QINU\x7f";
  88
  89         /**#@+
  90          * @private
  91          */
  92         # Persistent:
  93         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  94                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex, $mPreprocessor,
  95                 $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList, $mVarCache, $mConf,
  96                 $mFunctionTagHooks;
  97
  98
  99         # Cleared with clearState():
 100         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 101         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 102         var $mLinkHolders, $mLinkID;
 103         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 104         var $mTplExpandCache; // empty-frame expansion cache
 105         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 106         var $mExpensiveFunctionCount; // number of expensive parser function calls
 107
 108         # Temporary
 109         # These are variables reset at least once per parse regardless of $clearState
 110         var $mOptions,      // ParserOptions object
 111                 $mTitle,        // Title context, used for self-link rendering and similar things
 112                 $mOutputType,   // Output type, one of the OT_xxx constants
 113                 $ot,            // Shortcut alias, see setOutputType()
 114                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 115                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 116                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 117
 118         /**#@-*/
 119
 120         /**
 121          * Constructor
 122          *
 123          * @public
 124          */
 125         function __construct( $conf = array() ) {
 126                 $this->mConf = $conf;
 127                 $this->mTagHooks = array();
 128                 $this->mTransparentTagHooks = array();
 129                 $this->mFunctionHooks = array();
 130                 $this->mFunctionTagHooks = array();
 131                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 132                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 133                 $this->mUrlProtocols = wfUrlProtocols();
 134                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 135                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 136                 $this->mVarCache = array();
 137                 if ( isset( $conf['preprocessorClass'] ) ) {
 138                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 139                 } elseif ( extension_loaded( 'domxml' ) ) {
 140                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 141                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 142                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 143                 } elseif ( extension_loaded( 'dom' ) ) {
 144                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 145                 } else {
 146                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 147                 }
 148                 $this->mMarkerIndex = 0;
 149                 $this->mFirstCall = true;
 150         }
 151
 152         /**
 153          * Reduce memory usage to reduce the impact of circular references
 154          */
 155         function __destruct() {
 156                 if ( isset( $this->mLinkHolders ) ) {
 157                         $this->mLinkHolders->__destruct();
 158                 }
 159                 foreach ( $this as $name => $value ) {
 160                         unset( $this->$name );
 161                 }
 162         }
 163
 164         /**
 165          * Do various kinds of initialisation on the first call of the parser
 166          */
 167         function firstCallInit() {
 168                 if ( !$this->mFirstCall ) {
 169                         return;
 170                 }
 171                 $this->mFirstCall = false;
 172
 173                 wfProfileIn( __METHOD__ );
 174
 175                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 176                 CoreParserFunctions::register( $this );
 177                 $this->initialiseVariables();
 178
 179                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 180                 wfProfileOut( __METHOD__ );
 181         }
 182
 183         /**
 184          * Clear Parser state
 185          *
 186          * @private
 187          */
 188         function clearState() {
 189                 wfProfileIn( __METHOD__ );
 190                 if ( $this->mFirstCall ) {
 191                         $this->firstCallInit();
 192                 }
 193                 $this->mOutput = new ParserOutput;
 194                 $this->mAutonumber = 0;
 195                 $this->mLastSection = '';
 196                 $this->mDTopen = false;
 197                 $this->mIncludeCount = array();
 198                 $this->mStripState = new StripState;
 199                 $this->mArgStack = false;
 200                 $this->mInPre = false;
 201                 $this->mLinkHolders = new LinkHolderArray( $this );
 202                 $this->mLinkID = 0;
 203                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 204
 205                 /**
 206                  * Prefix for temporary replacement strings for the multipass parser.
 207                  * \x07 should never appear in input as it's disallowed in XML.
 208                  * Using it at the front also gives us a little extra robustness
 209                  * since it shouldn't match when butted up against identifier-like
 210                  * string constructs.
 211                  *
 212                  * Must not consist of all title characters, or else it will change
 213                  * the behaviour of <nowiki> in a link.
 214                  */
 215                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 216                 # Changed to \x7f to allow XML double-parsing -- TS
 217                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 218
 219
 220                 # Clear these on every parse, bug 4549
 221                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 222
 223                 $this->mShowToc = true;
 224                 $this->mForceTocPosition = false;
 225                 $this->mIncludeSizes = array(
 226                         'post-expand' => 0,
 227                         'arg' => 0,
 228                 );
 229                 $this->mPPNodeCount = 0;
 230                 $this->mDefaultSort = false;
 231                 $this->mHeadings = array();
 232                 $this->mDoubleUnderscores = array();
 233                 $this->mExpensiveFunctionCount = 0;
 234
 235                 # Fix cloning
 236                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 237                         $this->mPreprocessor = null;
 238                 }
 239
 240                 wfRunHooks( 'ParserClearState', array( &$this ) );
 241                 wfProfileOut( __METHOD__ );
 242         }
 243
 244         function setOutputType( $ot ) {
 245                 $this->mOutputType = $ot;
 246                 // Shortcut alias
 247                 $this->ot = array(
 248                         'html' => $ot == self::OT_HTML,
 249                         'wiki' => $ot == self::OT_WIKI,
 250                         'pre' => $ot == self::OT_PREPROCESS,
 251                 );
 252         }
 253
 254         /**
 255          * Set the context title
 256          */
 257         function setTitle( $t ) {
 258                 if ( !$t || $t instanceof FakeTitle ) {
 259                         $t = Title::newFromText( 'NO TITLE' );
 260                 }
 261                 if ( strval( $t->getFragment() ) !== '' ) {
 262                         # Strip the fragment to avoid various odd effects
 263                         $this->mTitle = clone $t;
 264                         $this->mTitle->setFragment( '' );
 265                 } else {
 266                         $this->mTitle = $t;
 267                 }
 268         }
 269
 270         /**
 271          * Accessor for mUniqPrefix.
 272          *
 273          * @public
 274          */
 275         function uniqPrefix() {
 276                 if( !isset( $this->mUniqPrefix ) ) {
 277                         // @fixme this is probably *horribly wrong*
 278                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 279                         // if this is called for a parser cache hit, the parser may not
 280                         // have ever been initialized in the first place.
 281                         // Not really sure what the heck is supposed to be going on here.
 282                         return '';
 283                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 284                 }
 285                 return $this->mUniqPrefix;
 286         }
 287
 288         /**
 289          * Convert wikitext to HTML
 290          * Do not call this function recursively.
 291          *
 292          * @param $text String: text we want to parse
 293          * @param $title A title object
 294          * @param $options ParserOptions
 295          * @param $linestart boolean
 296          * @param $clearState boolean
 297          * @param $revid Int: number to pass in {{REVISIONID}}
 298          * @return ParserOutput a ParserOutput
 299          */
 300         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 301                 /**
 302                  * First pass--just handle <nowiki> sections, pass the rest off
 303                  * to internalParse() which does all the real work.
 304                  */
 305
 306                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 307                 $fname = __METHOD__.'-' . wfGetCaller();
 308                 wfProfileIn( __METHOD__ );
 309                 wfProfileIn( $fname );
 310
 311                 if ( $clearState ) {
 312                         $this->clearState();
 313                 }
 314
 315                 $this->mOptions = $options;
 316                 $this->setTitle( $title );
 317                 $oldRevisionId = $this->mRevisionId;
 318                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 319                 if( $revid !== null ) {
 320                         $this->mRevisionId = $revid;
 321                         $this->mRevisionTimestamp = null;
 322                 }
 323                 $this->setOutputType( self::OT_HTML );
 324                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 325                 # No more strip!
 326                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 327                 $text = $this->internalParse( $text );
 328                 $text = $this->mStripState->unstripGeneral( $text );
 329
 330                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 331                 $fixtags = array(
 332                         # french spaces, last one Guillemet-left
 333                         # only if there is something before the space
 334                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 335                         # french spaces, Guillemet-right
 336                         '/(\\302\\253) /' => '\\1&nbsp;',
 337                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 338                 );
 339                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 340
 341                 $text = $this->doBlockLevels( $text, $linestart );
 342
 343                 $this->replaceLinkHolders( $text );
 344
 345                 # the position of the parserConvert() call should not be changed. it
 346                 # assumes that the links are all replaced and the only thing left
 347                 # is the <nowiki> mark.
 348                 # Side-effects: this calls $this->mOutput->setTitleText()
 349                 $text = $wgContLang->parserConvert( $text, $this );
 350
 351                 $text = $this->mStripState->unstripNoWiki( $text );
 352
 353                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 354
 355 //!JF Move to its own function
 356
 357                 $uniq_prefix = $this->mUniqPrefix;
 358                 $matches = array();
 359                 $elements = array_keys( $this->mTransparentTagHooks );
 360                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 361
 362                 foreach( $matches as $marker => $data ) {
 363                         list( $element, $content, $params, $tag ) = $data;
 364                         $tagName = strtolower( $element );
 365                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 366                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 367                                         array( $content, $params, $this ) );
 368                         } else {
 369                                 $output = $tag;
 370                         }
 371                         $this->mStripState->general->setPair( $marker, $output );
 372                 }
 373                 $text = $this->mStripState->unstripGeneral( $text );
 374
 375                 $text = Sanitizer::normalizeCharReferences( $text );
 376
 377                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 378                         $text = MWTidy::tidy( $text );
 379                 } else {
 380                         # attempt to sanitize at least some nesting problems
 381                         # (bug #2702 and quite a few others)
 382                         $tidyregs = array(
 383                                 # ''Something [http://www.cool.com cool''] -->
 384                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 385                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 386                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 387                                 # fix up an anchor inside another anchor, only
 388                                 # at least for a single single nested link (bug 3695)
 389                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 390                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 391                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 392                                 # contains a div, so fix it up here; replace
 393                                 # div with escaped text
 394                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 395                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 396                                 # remove empty italic or bold tag pairs, some
 397                                 # introduced by rules above
 398                                 '/<([bi])><\/\\1>/' => '',
 399                         );
 400
 401                         $text = preg_replace(
 402                                 array_keys( $tidyregs ),
 403                                 array_values( $tidyregs ),
 404                                 $text );
 405                 }
 406                 global $wgExpensiveParserFunctionLimit;
 407                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 408                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 409                 }
 410
 411                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 412
 413                 # Information on include size limits, for the benefit of users who try to skirt them
 414                 if ( $this->mOptions->getEnableLimitReport() ) {
 415                         global $wgExpensiveParserFunctionLimit;
 416                         $max = $this->mOptions->getMaxIncludeSize();
 417                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 418                         $limitReport =
 419                                 "NewPP limit report\n" .
 420                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 421                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 422                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 423                                 $PFreport;
 424                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 425                         $text .= "\n<!-- \n$limitReport-->\n";
 426                 }
 427                 $this->mOutput->setText( $text );
 428                 $this->mRevisionId = $oldRevisionId;
 429                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 430                 wfProfileOut( $fname );
 431                 wfProfileOut( __METHOD__ );
 432
 433                 return $this->mOutput;
 434         }
 435
 436         /**
 437          * Recursive parser entry point that can be called from an extension tag
 438          * hook.
 439          */
 440         function recursiveTagParse( $text ) {
 441                 wfProfileIn( __METHOD__ );
 442                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 443                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 444                 $text = $this->internalParse( $text, false );
 445                 wfProfileOut( __METHOD__ );
 446                 return $text;
 447         }
 448
 449         /**
 450          * Expand templates and variables in the text, producing valid, static wikitext.
 451          * Also removes comments.
 452          */
 453         function preprocess( $text, $title, $options, $revid = null ) {
 454                 wfProfileIn( __METHOD__ );
 455                 $this->clearState();
 456                 $this->setOutputType( self::OT_PREPROCESS );
 457                 $this->mOptions = $options;
 458                 $this->setTitle( $title );
 459                 if( $revid !== null ) {
 460                         $this->mRevisionId = $revid;
 461                 }
 462                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 463                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 464                 $text = $this->replaceVariables( $text );
 465                 $text = $this->mStripState->unstripBoth( $text );
 466                 wfProfileOut( __METHOD__ );
 467                 return $text;
 468         }
 469
 470         /**
 471          * Get a random string
 472          *
 473          * @private
 474          * @static
 475          */
 476         function getRandomString() {
 477                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 478         }
 479
 480         function &getTitle() { return $this->mTitle; }
 481         function getOptions() { return $this->mOptions; }
 482         function getRevisionId() { return $this->mRevisionId; }
 483         function getOutput() { return $this->mOutput; }
 484         function nextLinkID() { return $this->mLinkID++; }
 485
 486         function getFunctionLang() {
 487                 global $wgLang, $wgContLang;
 488
 489                 $target = $this->mOptions->getTargetLanguage();
 490                 if ( $target !== null ) {
 491                         return $target;
 492                 } else {
 493                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 494                 }
 495         }
 496
 497         /**
 498          * Get a preprocessor object
 499          */
 500         function getPreprocessor() {
 501                 if ( !isset( $this->mPreprocessor ) ) {
 502                         $class = $this->mPreprocessorClass;
 503                         $this->mPreprocessor = new $class( $this );
 504                 }
 505                 return $this->mPreprocessor;
 506         }
 507
 508         /**
 509          * Replaces all occurrences of HTML-style comments and the given tags
 510          * in the text with a random marker and returns the next text. The output
 511          * parameter $matches will be an associative array filled with data in
 512          * the form:
 513          *   'UNIQ-xxxxx' => array(
 514          *     'element',
 515          *     'tag content',
 516          *     array( 'param' => 'x' ),
 517          *     '<element param="x">tag content</element>' ) )
 518          *
 519          * @param $elements list of element names. Comments are always extracted.
 520          * @param $text Source text string.
 521          * @param $uniq_prefix
 522          *
 523          * @public
 524          * @static
 525          */
 526         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 527                 static $n = 1;
 528                 $stripped = '';
 529                 $matches = array();
 530
 531                 $taglist = implode( '|', $elements );
 532                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 533
 534                 while ( '' != $text ) {
 535                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 536                         $stripped .= $p[0];
 537                         if( count( $p ) < 5 ) {
 538                                 break;
 539                         }
 540                         if( count( $p ) > 5 ) {
 541                                 // comment
 542                                 $element    = $p[4];
 543                                 $attributes = '';
 544                                 $close      = '';
 545                                 $inside     = $p[5];
 546                         } else {
 547                                 // tag
 548                                 $element    = $p[1];
 549                                 $attributes = $p[2];
 550                                 $close      = $p[3];
 551                                 $inside     = $p[4];
 552                         }
 553
 554                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 555                         $stripped .= $marker;
 556
 557                         if ( $close === '/>' ) {
 558                                 // Empty element tag, <tag />
 559                                 $content = null;
 560                                 $text = $inside;
 561                                 $tail = null;
 562                         } else {
 563                                 if( $element === '!--' ) {
 564                                         $end = '/(-->)/';
 565                                 } else {
 566                                         $end = "/(<\\/$element\\s*>)/i";
 567                                 }
 568                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 569                                 $content = $q[0];
 570                                 if( count( $q ) < 3 ) {
 571                                         # No end tag -- let it run out to the end of the text.
 572                                         $tail = '';
 573                                         $text = '';
 574                                 } else {
 575                                         $tail = $q[1];
 576                                         $text = $q[2];
 577                                 }
 578                         }
 579
 580                         $matches[$marker] = array( $element,
 581                                 $content,
 582                                 Sanitizer::decodeTagAttributes( $attributes ),
 583                                 "<$element$attributes$close$content$tail" );
 584                 }
 585                 return $stripped;
 586         }
 587
 588         /**
 589          * Get a list of strippable XML-like elements
 590          */
 591         function getStripList() {
 592                 global $wgRawHtml;
 593                 $elements = $this->mStripList;
 594                 if( $wgRawHtml ) {
 595                         $elements[] = 'html';
 596                 }
 597                 if( $this->mOptions->getUseTeX() ) {
 598                         $elements[] = 'math';
 599                 }
 600                 return $elements;
 601         }
 602
 603         /**
 604          * @deprecated use replaceVariables
 605          */
 606         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 607                 return $text;
 608         }
 609
 610         /**
 611          * Restores pre, math, and other extensions removed by strip()
 612          *
 613          * always call unstripNoWiki() after this one
 614          * @private
 615          * @deprecated use $this->mStripState->unstrip()
 616          */
 617         function unstrip( $text, $state ) {
 618                 return $state->unstripGeneral( $text );
 619         }
 620
 621         /**
 622          * Always call this after unstrip() to preserve the order
 623          *
 624          * @private
 625          * @deprecated use $this->mStripState->unstrip()
 626          */
 627         function unstripNoWiki( $text, $state ) {
 628                 return $state->unstripNoWiki( $text );
 629         }
 630
 631         /**
 632          * @deprecated use $this->mStripState->unstripBoth()
 633          */
 634         function unstripForHTML( $text ) {
 635                 return $this->mStripState->unstripBoth( $text );
 636         }
 637
 638         /**
 639          * Add an item to the strip state
 640          * Returns the unique tag which must be inserted into the stripped text
 641          * The tag will be replaced with the original text in unstrip()
 642          *
 643          * @private
 644          */
 645         function insertStripItem( $text ) {
 646                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 647                 $this->mMarkerIndex++;
 648                 $this->mStripState->general->setPair( $rnd, $text );
 649                 return $rnd;
 650         }
 651
 652         /**
 653          * Interface with html tidy
 654          * @deprecated Use MWTidy::tidy()
 655          */
 656         public static function tidy( $text ) {
 657                 wfDeprecated( __METHOD__ );
 658                 return MWTidy::tidy( $text );
 659         }
 660
 661         /**
 662          * parse the wiki syntax used to render tables
 663          *
 664          * @private
 665          */
 666         function doTableStuff ( $text ) {
 667                 wfProfileIn( __METHOD__ );
 668
 669                 $lines = StringUtils::explode( "\n", $text );
 670                 $out = '';
 671                 $td_history = array (); // Is currently a td tag open?
 672                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 673                 $tr_history = array (); // Is currently a tr tag open?
 674                 $tr_attributes = array (); // history of tr attributes
 675                 $has_opened_tr = array(); // Did this table open a <tr> element?
 676                 $indent_level = 0; // indent level of the table
 677
 678                 foreach ( $lines as $outLine ) {
 679                         $line = trim( $outLine );
 680
 681                         if( $line == '' ) { // empty line, go to next line
 682                                 $out .= $outLine."\n";
 683                                 continue;
 684                         }
 685                         $first_character = $line[0];
 686                         $matches = array();
 687
 688                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 689                                 // First check if we are starting a new table
 690                                 $indent_level = strlen( $matches[1] );
 691
 692                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 693                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 694
 695                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 696                                 array_push ( $td_history , false );
 697                                 array_push ( $last_tag_history , '' );
 698                                 array_push ( $tr_history , false );
 699                                 array_push ( $tr_attributes , '' );
 700                                 array_push ( $has_opened_tr , false );
 701                         } else if ( count ( $td_history ) == 0 ) {
 702                                 // Don't do any of the following
 703                                 $out .= $outLine."\n";
 704                                 continue;
 705                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 706                                 // We are ending a table
 707                                 $line = '</table>' . substr ( $line , 2 );
 708                                 $last_tag = array_pop ( $last_tag_history );
 709
 710                                 if ( !array_pop ( $has_opened_tr ) ) {
 711                                         $line = "<tr><td></td></tr>{$line}";
 712                                 }
 713
 714                                 if ( array_pop ( $tr_history ) ) {
 715                                         $line = "</tr>{$line}";
 716                                 }
 717
 718                                 if ( array_pop ( $td_history ) ) {
 719                                         $line = "</{$last_tag}>{$line}";
 720                                 }
 721                                 array_pop ( $tr_attributes );
 722                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 723                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 724                                 // Now we have a table row
 725                                 $line = preg_replace( '#^\|-+#', '', $line );
 726
 727                                 // Whats after the tag is now only attributes
 728                                 $attributes = $this->mStripState->unstripBoth( $line );
 729                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 730                                 array_pop ( $tr_attributes );
 731                                 array_push ( $tr_attributes , $attributes );
 732
 733                                 $line = '';
 734                                 $last_tag = array_pop ( $last_tag_history );
 735                                 array_pop ( $has_opened_tr );
 736                                 array_push ( $has_opened_tr , true );
 737
 738                                 if ( array_pop ( $tr_history ) ) {
 739                                         $line = '</tr>';
 740                                 }
 741
 742                                 if ( array_pop ( $td_history ) ) {
 743                                         $line = "</{$last_tag}>{$line}";
 744                                 }
 745
 746                                 $outLine = $line;
 747                                 array_push ( $tr_history , false );
 748                                 array_push ( $td_history , false );
 749                                 array_push ( $last_tag_history , '' );
 750                         }
 751                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 752                                 // This might be cell elements, td, th or captions
 753                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 754                                         $first_character = '+';
 755                                         $line = substr ( $line , 1 );
 756                                 }
 757
 758                                 $line = substr ( $line , 1 );
 759
 760                                 if ( $first_character === '!' ) {
 761                                         $line = str_replace ( '!!' , '||' , $line );
 762                                 }
 763
 764                                 // Split up multiple cells on the same line.
 765                                 // FIXME : This can result in improper nesting of tags processed
 766                                 // by earlier parser steps, but should avoid splitting up eg
 767                                 // attribute values containing literal "||".
 768                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 769
 770                                 $outLine = '';
 771
 772                                 // Loop through each table cell
 773                                 foreach ( $cells as $cell )
 774                                 {
 775                                         $previous = '';
 776                                         if ( $first_character !== '+' )
 777                                         {
 778                                                 $tr_after = array_pop ( $tr_attributes );
 779                                                 if ( !array_pop ( $tr_history ) ) {
 780                                                         $previous = "<tr{$tr_after}>\n";
 781                                                 }
 782                                                 array_push ( $tr_history , true );
 783                                                 array_push ( $tr_attributes , '' );
 784                                                 array_pop ( $has_opened_tr );
 785                                                 array_push ( $has_opened_tr , true );
 786                                         }
 787
 788                                         $last_tag = array_pop ( $last_tag_history );
 789
 790                                         if ( array_pop ( $td_history ) ) {
 791                                                 $previous = "</{$last_tag}>{$previous}";
 792                                         }
 793
 794                                         if ( $first_character === '|' ) {
 795                                                 $last_tag = 'td';
 796                                         } else if ( $first_character === '!' ) {
 797                                                 $last_tag = 'th';
 798                                         } else if ( $first_character === '+' ) {
 799                                                 $last_tag = 'caption';
 800                                         } else {
 801                                                 $last_tag = '';
 802                                         }
 803
 804                                         array_push ( $last_tag_history , $last_tag );
 805
 806                                         // A cell could contain both parameters and data
 807                                         $cell_data = explode ( '|' , $cell , 2 );
 808
 809                                         // Bug 553: Note that a '|' inside an invalid link should not
 810                                         // be mistaken as delimiting cell parameters
 811                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 812                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 813                                         } else if ( count ( $cell_data ) == 1 )
 814                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 815                                         else {
 816                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 817                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 818                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 819                                         }
 820
 821                                         $outLine .= $cell;
 822                                         array_push ( $td_history , true );
 823                                 }
 824                         }
 825                         $out .= $outLine . "\n";
 826                 }
 827
 828                 // Closing open td, tr && table
 829                 while ( count ( $td_history ) > 0 )
 830                 {
 831                         if ( array_pop ( $td_history ) ) {
 832                                 $out .= "</td>\n";
 833                         }
 834                         if ( array_pop ( $tr_history ) ) {
 835                                 $out .= "</tr>\n";
 836                         }
 837                         if ( !array_pop ( $has_opened_tr ) ) {
 838                                 $out .= "<tr><td></td></tr>\n" ;
 839                         }
 840
 841                         $out .= "</table>\n";
 842                 }
 843
 844                 // Remove trailing line-ending (b/c)
 845                 if ( substr( $out, -1 ) === "\n" ) {
 846                         $out = substr( $out, 0, -1 );
 847                 }
 848
 849                 // special case: don't return empty table
 850                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 851                         $out = '';
 852                 }
 853
 854                 wfProfileOut( __METHOD__ );
 855
 856                 return $out;
 857         }
 858
 859         /**
 860          * Helper function for parse() that transforms wiki markup into
 861          * HTML. Only called for $mOutputType == self::OT_HTML.
 862          *
 863          * @private
 864          */
 865         function internalParse( $text, $isMain = true ) {
 866                 wfProfileIn( __METHOD__ );
 867
 868                 $origText = $text;
 869
 870                 # Hook to suspend the parser in this state
 871                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 872                         wfProfileOut( __METHOD__ );
 873                         return $text ;
 874                 }
 875
 876                 $text = $this->replaceVariables( $text );
 877                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 878                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 879
 880                 // Tables need to come after variable replacement for things to work
 881                 // properly; putting them before other transformations should keep
 882                 // exciting things like link expansions from showing up in surprising
 883                 // places.
 884                 $text = $this->doTableStuff( $text );
 885
 886                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 887
 888                 $text = $this->doDoubleUnderscore( $text );
 889                 $text = $this->doHeadings( $text );
 890                 if( $this->mOptions->getUseDynamicDates() ) {
 891                         $df = DateFormatter::getInstance();
 892                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 893                 }
 894                 $text = $this->doAllQuotes( $text );
 895                 $text = $this->replaceInternalLinks( $text );
 896                 $text = $this->replaceExternalLinks( $text );
 897
 898                 # replaceInternalLinks may sometimes leave behind
 899                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 900                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 901
 902                 $text = $this->doMagicLinks( $text );
 903                 $text = $this->formatHeadings( $text, $origText, $isMain );
 904
 905                 wfProfileOut( __METHOD__ );
 906                 return $text;
 907         }
 908
 909         /**
 910          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 911          * magic external links.
 912          *
 913          * DML
 914          * @private
 915          */
 916         function doMagicLinks( $text ) {
 917                 wfProfileIn( __METHOD__ );
 918                 $prots = $this->mUrlProtocols;
 919                 $urlChar = self::EXT_LINK_URL_CLASS;
 920                 $text = preg_replace_callback(
 921                         '!(?:                           # Start cases
 922                                 (<a.*?</a>) |               # m[1]: Skip link text
 923                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 924                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 925                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 926                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 927                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 928                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 929                                     [0-9Xx]                 # check digit
 930                                     \b)
 931                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 932                 wfProfileOut( __METHOD__ );
 933                 return $text;
 934         }
 935
 936         function magicLinkCallback( $m ) {
 937                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 938                         # Skip anchor
 939                         return $m[0];
 940                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 941                         # Skip HTML element
 942                         return $m[0];
 943                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 944                         # Free external link
 945                         return $this->makeFreeExternalLink( $m[0] );
 946                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 947                         # RFC or PMID
 948                         $CssClass = '';
 949                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 950                                 $keyword = 'RFC';
 951                                 $urlmsg = 'rfcurl';
 952                                 $CssClass = 'mw-magiclink-rfc';
 953                                 $id = $m[4];
 954                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
 955                                 $keyword = 'PMID';
 956                                 $urlmsg = 'pubmedurl';
 957                                 $CssClass = 'mw-magiclink-pmid';
 958                                 $id = $m[4];
 959                         } else {
 960                                 throw new MWException( __METHOD__.': unrecognised match type "' .
 961                                         substr($m[0], 0, 20 ) . '"' );
 962                         }
 963                         $url = wfMsg( $urlmsg, $id);
 964                         $sk = $this->mOptions->getSkin();
 965                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
 966                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
 967                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
 968                         # ISBN
 969                         $isbn = $m[5];
 970                         $num = strtr( $isbn, array(
 971                                 '-' => '',
 972                                 ' ' => '',
 973                                 'x' => 'X',
 974                         ));
 975                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
 976                         return'<a href="' .
 977                                 $titleObj->escapeLocalUrl() .
 978                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
 979                 } else {
 980                         return $m[0];
 981                 }
 982         }
 983
 984         /**
 985          * Make a free external link, given a user-supplied URL
 986          * @return HTML
 987          * @private
 988          */
 989         function makeFreeExternalLink( $url ) {
 990                 global $wgContLang;
 991                 wfProfileIn( __METHOD__ );
 992
 993                 $sk = $this->mOptions->getSkin();
 994                 $trail = '';
 995
 996                 # The characters '<' and '>' (which were escaped by
 997                 # removeHTMLtags()) should not be included in
 998                 # URLs, per RFC 2396.
 999                 $m2 = array();
1000                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1001                         $trail = substr($url, $m2[0][1]) . $trail;
1002                         $url = substr($url, 0, $m2[0][1]);
1003                 }
1004
1005                 # Move trailing punctuation to $trail
1006                 $sep = ',;\.:!?';
1007                 # If there is no left bracket, then consider right brackets fair game too
1008                 if ( strpos( $url, '(' ) === false ) {
1009                         $sep .= ')';
1010                 }
1011
1012                 $numSepChars = strspn( strrev( $url ), $sep );
1013                 if ( $numSepChars ) {
1014                         $trail = substr( $url, -$numSepChars ) . $trail;
1015                         $url = substr( $url, 0, -$numSepChars );
1016                 }
1017
1018                 $url = Sanitizer::cleanUrl( $url );
1019
1020                 # Is this an external image?
1021                 $text = $this->maybeMakeExternalImage( $url );
1022                 if ( $text === false ) {
1023                         # Not an image, make a link
1024                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1025                                 $this->getExternalLinkAttribs( $url ) );
1026                         # Register it in the output object...
1027                         # Replace unnecessary URL escape codes with their equivalent characters
1028                         $pasteurized = self::replaceUnusualEscapes( $url );
1029                         $this->mOutput->addExternalLink( $pasteurized );
1030                 }
1031                 wfProfileOut( __METHOD__ );
1032                 return $text . $trail;
1033         }
1034
1035
1036         /**
1037          * Parse headers and return html
1038          *
1039          * @private
1040          */
1041         function doHeadings( $text ) {
1042                 wfProfileIn( __METHOD__ );
1043                 for ( $i = 6; $i >= 1; --$i ) {
1044                         $h = str_repeat( '=', $i );
1045                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1046                           "<h$i>\\1</h$i>", $text );
1047                 }
1048                 wfProfileOut( __METHOD__ );
1049                 return $text;
1050         }
1051
1052         /**
1053          * Replace single quotes with HTML markup
1054          * @private
1055          * @return string the altered text
1056          */
1057         function doAllQuotes( $text ) {
1058                 wfProfileIn( __METHOD__ );
1059                 $outtext = '';
1060                 $lines = StringUtils::explode( "\n", $text );
1061                 foreach ( $lines as $line ) {
1062                         $outtext .= $this->doQuotes( $line ) . "\n";
1063                 }
1064                 $outtext = substr($outtext, 0,-1);
1065                 wfProfileOut( __METHOD__ );
1066                 return $outtext;
1067         }
1068
1069         /**
1070          * Helper function for doAllQuotes()
1071          */
1072         public function doQuotes( $text ) {
1073                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1074                 if ( count( $arr ) == 1 )
1075                         return $text;
1076                 else
1077                 {
1078                         # First, do some preliminary work. This may shift some apostrophes from
1079                         # being mark-up to being text. It also counts the number of occurrences
1080                         # of bold and italics mark-ups.
1081                         $i = 0;
1082                         $numbold = 0;
1083                         $numitalics = 0;
1084                         foreach ( $arr as $r )
1085                         {
1086                                 if ( ( $i % 2 ) == 1 )
1087                                 {
1088                                         # If there are ever four apostrophes, assume the first is supposed to
1089                                         # be text, and the remaining three constitute mark-up for bold text.
1090                                         if ( strlen( $arr[$i] ) == 4 )
1091                                         {
1092                                                 $arr[$i-1] .= "'";
1093                                                 $arr[$i] = "'''";
1094                                         }
1095                                         # If there are more than 5 apostrophes in a row, assume they're all
1096                                         # text except for the last 5.
1097                                         else if ( strlen( $arr[$i] ) > 5 )
1098                                         {
1099                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1100                                                 $arr[$i] = "'''''";
1101                                         }
1102                                         # Count the number of occurrences of bold and italics mark-ups.
1103                                         # We are not counting sequences of five apostrophes.
1104                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1105                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1106                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1107                                 }
1108                                 $i++;
1109                         }
1110
1111                         # If there is an odd number of both bold and italics, it is likely
1112                         # that one of the bold ones was meant to be an apostrophe followed
1113                         # by italics. Which one we cannot know for certain, but it is more
1114                         # likely to be one that has a single-letter word before it.
1115                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1116                         {
1117                                 $i = 0;
1118                                 $firstsingleletterword = -1;
1119                                 $firstmultiletterword = -1;
1120                                 $firstspace = -1;
1121                                 foreach ( $arr as $r )
1122                                 {
1123                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1124                                         {
1125                                                 $x1 = substr ($arr[$i-1], -1);
1126                                                 $x2 = substr ($arr[$i-1], -2, 1);
1127                                                 if ($x1 === ' ') {
1128                                                         if ($firstspace == -1) $firstspace = $i;
1129                                                 } else if ($x2 === ' ') {
1130                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1131                                                 } else {
1132                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1133                                                 }
1134                                         }
1135                                         $i++;
1136                                 }
1137
1138                                 # If there is a single-letter word, use it!
1139                                 if ($firstsingleletterword > -1)
1140                                 {
1141                                         $arr [ $firstsingleletterword ] = "''";
1142                                         $arr [ $firstsingleletterword-1 ] .= "'";
1143                                 }
1144                                 # If not, but there's a multi-letter word, use that one.
1145                                 else if ($firstmultiletterword > -1)
1146                                 {
1147                                         $arr [ $firstmultiletterword ] = "''";
1148                                         $arr [ $firstmultiletterword-1 ] .= "'";
1149                                 }
1150                                 # ... otherwise use the first one that has neither.
1151                                 # (notice that it is possible for all three to be -1 if, for example,
1152                                 # there is only one pentuple-apostrophe in the line)
1153                                 else if ($firstspace > -1)
1154                                 {
1155                                         $arr [ $firstspace ] = "''";
1156                                         $arr [ $firstspace-1 ] .= "'";
1157                                 }
1158                         }
1159
1160                         # Now let's actually convert our apostrophic mush to HTML!
1161                         $output = '';
1162                         $buffer = '';
1163                         $state = '';
1164                         $i = 0;
1165                         foreach ($arr as $r)
1166                         {
1167                                 if (($i % 2) == 0)
1168                                 {
1169                                         if ($state === 'both')
1170                                                 $buffer .= $r;
1171                                         else
1172                                                 $output .= $r;
1173                                 }
1174                                 else
1175                                 {
1176                                         if (strlen ($r) == 2)
1177                                         {
1178                                                 if ($state === 'i')
1179                                                 { $output .= '</i>'; $state = ''; }
1180                                                 else if ($state === 'bi')
1181                                                 { $output .= '</i>'; $state = 'b'; }
1182                                                 else if ($state === 'ib')
1183                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1184                                                 else if ($state === 'both')
1185                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1186                                                 else # $state can be 'b' or ''
1187                                                 { $output .= '<i>'; $state .= 'i'; }
1188                                         }
1189                                         else if (strlen ($r) == 3)
1190                                         {
1191                                                 if ($state === 'b')
1192                                                 { $output .= '</b>'; $state = ''; }
1193                                                 else if ($state === 'bi')
1194                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1195                                                 else if ($state === 'ib')
1196                                                 { $output .= '</b>'; $state = 'i'; }
1197                                                 else if ($state === 'both')
1198                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1199                                                 else # $state can be 'i' or ''
1200                                                 { $output .= '<b>'; $state .= 'b'; }
1201                                         }
1202                                         else if (strlen ($r) == 5)
1203                                         {
1204                                                 if ($state === 'b')
1205                                                 { $output .= '</b><i>'; $state = 'i'; }
1206                                                 else if ($state === 'i')
1207                                                 { $output .= '</i><b>'; $state = 'b'; }
1208                                                 else if ($state === 'bi')
1209                                                 { $output .= '</i></b>'; $state = ''; }
1210                                                 else if ($state === 'ib')
1211                                                 { $output .= '</b></i>'; $state = ''; }
1212                                                 else if ($state === 'both')
1213                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1214                                                 else # ($state == '')
1215                                                 { $buffer = ''; $state = 'both'; }
1216                                         }
1217                                 }
1218                                 $i++;
1219                         }
1220                         # Now close all remaining tags.  Notice that the order is important.
1221                         if ($state === 'b' || $state === 'ib')
1222                                 $output .= '</b>';
1223                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1224                                 $output .= '</i>';
1225                         if ($state === 'bi')
1226                                 $output .= '</b>';
1227                         # There might be lonely ''''', so make sure we have a buffer
1228                         if ($state === 'both' && $buffer)
1229                                 $output .= '<b><i>'.$buffer.'</i></b>';
1230                         return $output;
1231                 }
1232         }
1233
1234         /**
1235          * Replace external links (REL)
1236          *
1237          * Note: this is all very hackish and the order of execution matters a lot.
1238          * Make sure to run maintenance/parserTests.php if you change this code.
1239          *
1240          * @private
1241          */
1242         function replaceExternalLinks( $text ) {
1243                 global $wgContLang;
1244                 wfProfileIn( __METHOD__ );
1245
1246                 $sk = $this->mOptions->getSkin();
1247
1248                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1249                 $s = array_shift( $bits );
1250
1251                 $i = 0;
1252                 while ( $i<count( $bits ) ) {
1253                         $url = $bits[$i++];
1254                         $protocol = $bits[$i++];
1255                         $text = $bits[$i++];
1256                         $trail = $bits[$i++];
1257
1258                         # The characters '<' and '>' (which were escaped by
1259                         # removeHTMLtags()) should not be included in
1260                         # URLs, per RFC 2396.
1261                         $m2 = array();
1262                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1263                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1264                                 $url = substr($url, 0, $m2[0][1]);
1265                         }
1266
1267                         # If the link text is an image URL, replace it with an <img> tag
1268                         # This happened by accident in the original parser, but some people used it extensively
1269                         $img = $this->maybeMakeExternalImage( $text );
1270                         if ( $img !== false ) {
1271                                 $text = $img;
1272                         }
1273
1274                         $dtrail = '';
1275
1276                         # Set linktype for CSS - if URL==text, link is essentially free
1277                         $linktype = ($text === $url) ? 'free' : 'text';
1278
1279                         # No link text, e.g. [http://domain.tld/some.link]
1280                         if ( $text == '' ) {
1281                                 # Autonumber if allowed. See bug #5918
1282                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1283                                         $langObj = $this->getFunctionLang();
1284                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1285                                         $linktype = 'autonumber';
1286                                 } else {
1287                                         # Otherwise just use the URL
1288                                         $text = htmlspecialchars( $url );
1289                                         $linktype = 'free';
1290                                 }
1291                         } else {
1292                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1293                                 # Check for trail
1294                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1295                         }
1296
1297                         $text = $wgContLang->markNoConversion($text);
1298
1299                         $url = Sanitizer::cleanUrl( $url );
1300
1301                         # Use the encoded URL
1302                         # This means that users can paste URLs directly into the text
1303                         # Funny characters like &ouml; aren't valid in URLs anyway
1304                         # This was changed in August 2004
1305                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1306                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1307
1308                         # Register link in the output object.
1309                         # Replace unnecessary URL escape codes with the referenced character
1310                         # This prevents spammers from hiding links from the filters
1311                         $pasteurized = self::replaceUnusualEscapes( $url );
1312                         $this->mOutput->addExternalLink( $pasteurized );
1313                 }
1314
1315                 wfProfileOut( __METHOD__ );
1316                 return $s;
1317         }
1318
1319         /**
1320          * Get an associative array of additional HTML attributes appropriate for a
1321          * particular external link.  This currently may include rel => nofollow
1322          * (depending on configuration, namespace, and the URL's domain) and/or a
1323          * target attribute (depending on configuration).
1324          *
1325          * @param string $url Optional URL, to extract the domain from for rel =>
1326          *   nofollow if appropriate
1327          * @return array Associative array of HTML attributes
1328          */
1329         function getExternalLinkAttribs( $url = false ) {
1330                 $attribs = array();
1331                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1332                 $ns = $this->mTitle->getNamespace();
1333                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1334                         $attribs['rel'] = 'nofollow';
1335
1336                         global $wgNoFollowDomainExceptions;
1337                         if ( $wgNoFollowDomainExceptions ) {
1338                                 $bits = wfParseUrl( $url );
1339                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1340                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1341                                                 if( substr( $bits['host'], -strlen( $domain ) )
1342                                                 == $domain ) {
1343                                                         unset( $attribs['rel'] );
1344                                                         break;
1345                                                 }
1346                                         }
1347                                 }
1348                         }
1349                 }
1350                 if ( $this->mOptions->getExternalLinkTarget() ) {
1351                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1352                 }
1353                 return $attribs;
1354         }
1355
1356
1357         /**
1358          * Replace unusual URL escape codes with their equivalent characters
1359          * @param string
1360          * @return string
1361          * @static
1362          * @todo  This can merge genuinely required bits in the path or query string,
1363          *        breaking legit URLs. A proper fix would treat the various parts of
1364          *        the URL differently; as a workaround, just use the output for
1365          *        statistical records, not for actual linking/output.
1366          */
1367         static function replaceUnusualEscapes( $url ) {
1368                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1369                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1370         }
1371
1372         /**
1373          * Callback function used in replaceUnusualEscapes().
1374          * Replaces unusual URL escape codes with their equivalent character
1375          * @static
1376          * @private
1377          */
1378         private static function replaceUnusualEscapesCallback( $matches ) {
1379                 $char = urldecode( $matches[0] );
1380                 $ord = ord( $char );
1381                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1382                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1383                         // No, shouldn't be escaped
1384                         return $char;
1385                 } else {
1386                         // Yes, leave it escaped
1387                         return $matches[0];
1388                 }
1389         }
1390
1391         /**
1392          * make an image if it's allowed, either through the global
1393          * option, through the exception, or through the on-wiki whitelist
1394          * @private
1395          */
1396         function maybeMakeExternalImage( $url ) {
1397                 $sk = $this->mOptions->getSkin();
1398                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1399                 $imagesexception = !empty($imagesfrom);
1400                 $text = false;
1401                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1402                 if( $imagesexception && is_array( $imagesfrom ) ) {
1403                         $imagematch = false;
1404                         foreach( $imagesfrom as $match ) {
1405                                 if( strpos( $url, $match ) === 0 ) {
1406                                         $imagematch = true;
1407                                         break;
1408                                 }
1409                         }
1410                 } elseif( $imagesexception ) {
1411                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1412                 } else {
1413                         $imagematch = false;
1414                 }
1415                 if ( $this->mOptions->getAllowExternalImages()
1416                      || ( $imagesexception && $imagematch ) ) {
1417                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1418                                 # Image found
1419                                 $text = $sk->makeExternalImage( $url );
1420                         }
1421                 }
1422                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1423                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1424                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1425                         foreach( $whitelist as $entry ) {
1426                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1427                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1428                                         continue;
1429                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1430                                         # Image matches a whitelist entry
1431                                         $text = $sk->makeExternalImage( $url );
1432                                         break;
1433                                 }
1434                         }
1435                 }
1436                 return $text;
1437         }
1438
1439         /**
1440          * Process [[ ]] wikilinks
1441          * @return processed text
1442          *
1443          * @private
1444          */
1445         function replaceInternalLinks( $s ) {
1446                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1447                 return $s;
1448         }
1449
1450         /**
1451          * Process [[ ]] wikilinks (RIL)
1452          * @return LinkHolderArray
1453          *
1454          * @private
1455          */
1456         function replaceInternalLinks2( &$s ) {
1457                 global $wgContLang;
1458
1459                 wfProfileIn( __METHOD__ );
1460
1461                 wfProfileIn( __METHOD__.'-setup' );
1462                 static $tc = FALSE, $e1, $e1_img;
1463                 # the % is needed to support urlencoded titles as well
1464                 if ( !$tc ) {
1465                         $tc = Title::legalChars() . '#%';
1466                         # Match a link having the form [[namespace:link|alternate]]trail
1467                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1468                         # Match cases where there is no "]]", which might still be images
1469                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1470                 }
1471
1472                 $sk = $this->mOptions->getSkin();
1473                 $holders = new LinkHolderArray( $this );
1474
1475                 #split the entire text string on occurences of [[
1476                 $a = StringUtils::explode( '[[', ' ' . $s );
1477                 #get the first element (all text up to first [[), and remove the space we added
1478                 $s = $a->current();
1479                 $a->next();
1480                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1481                 $s = substr( $s, 1 );
1482
1483                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1484                 $e2 = null;
1485                 if ( $useLinkPrefixExtension ) {
1486                         # Match the end of a line for a word that's not followed by whitespace,
1487                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1488                         $e2 = wfMsgForContent( 'linkprefix' );
1489                 }
1490
1491                 if( is_null( $this->mTitle ) ) {
1492                         wfProfileOut( __METHOD__.'-setup' );
1493                         wfProfileOut( __METHOD__ );
1494                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1495                 }
1496                 $nottalk = !$this->mTitle->isTalkPage();
1497
1498                 if ( $useLinkPrefixExtension ) {
1499                         $m = array();
1500                         if ( preg_match( $e2, $s, $m ) ) {
1501                                 $first_prefix = $m[2];
1502                         } else {
1503                                 $first_prefix = false;
1504                         }
1505                 } else {
1506                         $prefix = '';
1507                 }
1508
1509                 if($wgContLang->hasVariants()) {
1510                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1511                 } else {
1512                         $selflink = array($this->mTitle->getPrefixedText());
1513                 }
1514                 $useSubpages = $this->areSubpagesAllowed();
1515                 wfProfileOut( __METHOD__.'-setup' );
1516
1517                 # Loop for each link
1518                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1519                         # Check for excessive memory usage
1520                         if ( $holders->isBig() ) {
1521                                 # Too big
1522                                 # Do the existence check, replace the link holders and clear the array
1523                                 $holders->replace( $s );
1524                                 $holders->clear();
1525                         }
1526
1527                         if ( $useLinkPrefixExtension ) {
1528                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1529                                 if ( preg_match( $e2, $s, $m ) ) {
1530                                         $prefix = $m[2];
1531                                         $s = $m[1];
1532                                 } else {
1533                                         $prefix='';
1534                                 }
1535                                 # first link
1536                                 if($first_prefix) {
1537                                         $prefix = $first_prefix;
1538                                         $first_prefix = false;
1539                                 }
1540                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1541                         }
1542
1543                         $might_be_img = false;
1544
1545                         wfProfileIn( __METHOD__."-e1" );
1546                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1547                                 $text = $m[2];
1548                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1549                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1550                                 # the real problem is with the $e1 regex
1551                                 # See bug 1300.
1552                                 #
1553                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1554                                 # and no image is in sight. See bug 2095.
1555                                 #
1556                                 if( $text !== '' &&
1557                                         substr( $m[3], 0, 1 ) === ']' &&
1558                                         strpos($text, '[') !== false
1559                                 )
1560                                 {
1561                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1562                                         $m[3] = substr( $m[3], 1 );
1563                                 }
1564                                 # fix up urlencoded title texts
1565                                 if( strpos( $m[1], '%' ) !== false ) {
1566                                         # Should anchors '#' also be rejected?
1567                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1568                                 }
1569                                 $trail = $m[3];
1570                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1571                                 $might_be_img = true;
1572                                 $text = $m[2];
1573                                 if ( strpos( $m[1], '%' ) !== false ) {
1574                                         $m[1] = urldecode($m[1]);
1575                                 }
1576                                 $trail = "";
1577                         } else { # Invalid form; output directly
1578                                 $s .= $prefix . '[[' . $line ;
1579                                 wfProfileOut( __METHOD__."-e1" );
1580                                 continue;
1581                         }
1582                         wfProfileOut( __METHOD__."-e1" );
1583                         wfProfileIn( __METHOD__."-misc" );
1584
1585                         # Don't allow internal links to pages containing
1586                         # PROTO: where PROTO is a valid URL protocol; these
1587                         # should be external links.
1588                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1589                                 $s .= $prefix . '[[' . $line ;
1590                                 wfProfileOut( __METHOD__."-misc" );
1591                                 continue;
1592                         }
1593
1594                         # Make subpage if necessary
1595                         if ( $useSubpages ) {
1596                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1597                         } else {
1598                                 $link = $m[1];
1599                         }
1600
1601                         $noforce = (substr( $m[1], 0, 1 ) !== ':');
1602                         if (!$noforce) {
1603                                 # Strip off leading ':'
1604                                 $link = substr( $link, 1 );
1605                         }
1606
1607                         wfProfileOut( __METHOD__."-misc" );
1608                         wfProfileIn( __METHOD__."-title" );
1609                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1610                         if ( $nt === NULL ) {
1611                                 $s .= $prefix . '[[' . $line;
1612                                 wfProfileOut( __METHOD__."-title" );
1613                                 continue;
1614                         }
1615
1616                         $ns = $nt->getNamespace();
1617                         $iw = $nt->getInterWiki();
1618                         wfProfileOut( __METHOD__."-title" );
1619
1620                         if ( $might_be_img ) { # if this is actually an invalid link
1621                                 wfProfileIn( __METHOD__."-might_be_img" );
1622                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1623                                         $found = false;
1624                                         while ( true ) {
1625                                                 #look at the next 'line' to see if we can close it there
1626                                                 $a->next();
1627                                                 $next_line = $a->current();
1628                                                 if ( $next_line === false || $next_line === null ) {
1629                                                         break;
1630                                                 }
1631                                                 $m = explode( ']]', $next_line, 3 );
1632                                                 if ( count( $m ) == 3 ) {
1633                                                         # the first ]] closes the inner link, the second the image
1634                                                         $found = true;
1635                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1636                                                         $trail = $m[2];
1637                                                         break;
1638                                                 } elseif ( count( $m ) == 2 ) {
1639                                                         #if there's exactly one ]] that's fine, we'll keep looking
1640                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1641                                                 } else {
1642                                                         #if $next_line is invalid too, we need look no further
1643                                                         $text .= '[[' . $next_line;
1644                                                         break;
1645                                                 }
1646                                         }
1647                                         if ( !$found ) {
1648                                                 # we couldn't find the end of this imageLink, so output it raw
1649                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1650                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1651                                                 $s .= "{$prefix}[[$link|$text";
1652                                                 # note: no $trail, because without an end, there *is* no trail
1653                                                 wfProfileOut( __METHOD__."-might_be_img" );
1654                                                 continue;
1655                                         }
1656                                 } else { #it's not an image, so output it raw
1657                                         $s .= "{$prefix}[[$link|$text";
1658                                         # note: no $trail, because without an end, there *is* no trail
1659                                         wfProfileOut( __METHOD__."-might_be_img" );
1660                                         continue;
1661                                 }
1662                                 wfProfileOut( __METHOD__."-might_be_img" );
1663                         }
1664
1665                         $wasblank = ( '' == $text );
1666                         if ( $wasblank ) $text = $link;
1667
1668                         # Link not escaped by : , create the various objects
1669                         if ( $noforce ) {
1670
1671                                 # Interwikis
1672                                 wfProfileIn( __METHOD__."-interwiki" );
1673                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1674                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1675                                         $s = rtrim($s . $prefix);
1676                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1677                                         wfProfileOut( __METHOD__."-interwiki" );
1678                                         continue;
1679                                 }
1680                                 wfProfileOut( __METHOD__."-interwiki" );
1681
1682                                 if ( $ns == NS_FILE ) {
1683                                         wfProfileIn( __METHOD__."-image" );
1684                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1685                                                 if ( $wasblank ) {
1686                                                         # if no parameters were passed, $text
1687                                                         # becomes something like "File:Foo.png",
1688                                                         # which we don't want to pass on to the
1689                                                         # image generator
1690                                                         $text = '';
1691                                                 } else {
1692                                                         # recursively parse links inside the image caption
1693                                                         # actually, this will parse them in any other parameters, too,
1694                                                         # but it might be hard to fix that, and it doesn't matter ATM
1695                                                         $text = $this->replaceExternalLinks($text);
1696                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1697                                                 }
1698                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1699                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1700                                         }
1701                                         $this->mOutput->addImage( $nt->getDBkey() );
1702                                         wfProfileOut( __METHOD__."-image" );
1703                                         continue;
1704
1705                                 }
1706
1707                                 if ( $ns == NS_CATEGORY ) {
1708                                         wfProfileIn( __METHOD__."-category" );
1709                                         $s = preg_replace( "/(\s*\n)+\s*$/D", '', $s ); # bug 87
1710
1711                                         if ( $wasblank ) {
1712                                                 $sortkey = $this->getDefaultSort();
1713                                         } else {
1714                                                 $sortkey = $text;
1715                                         }
1716                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1717                                         $sortkey = str_replace( "\n", '', $sortkey );
1718                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1719                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1720
1721                                         /**
1722                                          * Strip the whitespace Category links produce, see bug 87
1723                                          * @todo We might want to use trim($tmp, "\n") here.
1724                                          */
1725                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1726
1727                                         wfProfileOut( __METHOD__."-category" );
1728                                         continue;
1729                                 }
1730                         }
1731
1732                         # Self-link checking
1733                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1734                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1735                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1736                                         continue;
1737                                 }
1738                         }
1739
1740                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1741                         # FIXME: Should do batch file existence checks, see comment below
1742                         if( $ns == NS_MEDIA ) {
1743                                 wfProfileIn( __METHOD__."-media" );
1744                                 # Give extensions a chance to select the file revision for us
1745                                 $skip = $time = false;
1746                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1747                                 if ( $skip ) {
1748                                         $link = $sk->link( $nt );
1749                                 } else {
1750                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1751                                 }
1752                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1753                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1754                                 $this->mOutput->addImage( $nt->getDBkey() );
1755                                 wfProfileOut( __METHOD__."-media" );
1756                                 continue;
1757                         }
1758
1759                         wfProfileIn( __METHOD__."-always_known" );
1760                         # Some titles, such as valid special pages or files in foreign repos, should
1761                         # be shown as bluelinks even though they're not included in the page table
1762                         #
1763                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1764                         # batch file existence checks for NS_FILE and NS_MEDIA
1765                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1766                                 $this->mOutput->addLink( $nt );
1767                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1768                         } else {
1769                                 # Links will be added to the output link list after checking
1770                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1771                         }
1772                         wfProfileOut( __METHOD__."-always_known" );
1773                 }
1774                 wfProfileOut( __METHOD__ );
1775                 return $holders;
1776         }
1777
1778         /**
1779          * Make a link placeholder. The text returned can be later resolved to a real link with
1780          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1781          * parsing of interwiki links, and secondly to allow all existence checks and
1782          * article length checks (for stub links) to be bundled into a single query.
1783          *
1784          * @deprecated
1785          */
1786         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1787                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1788         }
1789
1790         /**
1791          * Render a forced-blue link inline; protect against double expansion of
1792          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1793          * Since this little disaster has to split off the trail text to avoid
1794          * breaking URLs in the following text without breaking trails on the
1795          * wiki links, it's been made into a horrible function.
1796          *
1797          * @param Title $nt
1798          * @param string $text
1799          * @param string $query
1800          * @param string $trail
1801          * @param string $prefix
1802          * @return string HTML-wikitext mix oh yuck
1803          */
1804         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1805                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1806                 $sk = $this->mOptions->getSkin();
1807                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1808                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1809                 return $this->armorLinks( $link ) . $trail;
1810         }
1811
1812         /**
1813          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1814          * going to go through further parsing steps before inline URL expansion.
1815          *
1816          * Not needed quite as much as it used to be since free links are a bit
1817          * more sensible these days. But bracketed links are still an issue.
1818          *
1819          * @param string more-or-less HTML
1820          * @return string less-or-more HTML with NOPARSE bits
1821          */
1822         function armorLinks( $text ) {
1823                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1824                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1825         }
1826
1827         /**
1828          * Return true if subpage links should be expanded on this page.
1829          * @return bool
1830          */
1831         function areSubpagesAllowed() {
1832                 # Some namespaces don't allow subpages
1833                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1834         }
1835
1836         /**
1837          * Handle link to subpage if necessary
1838          * @param string $target the source of the link
1839          * @param string &$text the link text, modified as necessary
1840          * @return string the full name of the link
1841          * @private
1842          */
1843         function maybeDoSubpageLink($target, &$text) {
1844                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1845         }
1846
1847         /**#@+
1848          * Used by doBlockLevels()
1849          * @private
1850          */
1851         /* private */ function closeParagraph() {
1852                 $result = '';
1853                 if ( '' != $this->mLastSection ) {
1854                         $result = '</' . $this->mLastSection  . ">\n";
1855                 }
1856                 $this->mInPre = false;
1857                 $this->mLastSection = '';
1858                 return $result;
1859         }
1860         # getCommon() returns the length of the longest common substring
1861         # of both arguments, starting at the beginning of both.
1862         #
1863         /* private */ function getCommon( $st1, $st2 ) {
1864                 $fl = strlen( $st1 );
1865                 $shorter = strlen( $st2 );
1866                 if ( $fl < $shorter ) { $shorter = $fl; }
1867
1868                 for ( $i = 0; $i < $shorter; ++$i ) {
1869                         if ( $st1{$i} != $st2{$i} ) { break; }
1870                 }
1871                 return $i;
1872         }
1873         # These next three functions open, continue, and close the list
1874         # element appropriate to the prefix character passed into them.
1875         #
1876         /* private */ function openList( $char ) {
1877                 $result = $this->closeParagraph();
1878
1879                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1880                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1881                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1882                 elseif ( ';' === $char ) {
1883                         $result .= '<dl><dt>';
1884                         $this->mDTopen = true;
1885                 } elseif ( '>' === $char ) { $result .= "<blockquote><p>"; }
1886                 else { $result = '<!-- ERR 1 -->'; }
1887
1888                 return $result;
1889         }
1890
1891         /* private */ function nextItem( $char ) {
1892                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1893                 elseif ( '>' === $char ) { return "</p><p>"; }
1894                 elseif ( ':' === $char || ';' === $char ) {
1895                         $close = '</dd>';
1896                         if ( $this->mDTopen ) { $close = '</dt>'; }
1897                         if ( ';' === $char ) {
1898                                 $this->mDTopen = true;
1899                                 return $close . '<dt>';
1900                         } else {
1901                                 $this->mDTopen = false;
1902                                 return $close . '<dd>';
1903                         }
1904                 }
1905                 return '<!-- ERR 2 -->';
1906         }
1907
1908         /* private */ function closeList( $char ) {
1909                 if ( '*' === $char ) { $text = '</li></ul>'; }
1910                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
1911                 elseif ( '>' === $char ) { $text = "</p></blockquote>"; }
1912                 elseif ( ':' === $char ) {
1913                         if ( $this->mDTopen ) {
1914                                 $this->mDTopen = false;
1915                                 $text = '</dt></dl>';
1916                         } else {
1917                                 $text = '</dd></dl>';
1918                         }
1919                 }
1920                 else {  return '<!-- ERR 3 -->'; }
1921                 return $text."\n";
1922         }
1923         /**#@-*/
1924
1925         /**
1926          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1927          *
1928          * @param $linestart bool whether or not this is at the start of a line.
1929          * @private
1930          * @return string the lists rendered as HTML
1931          */
1932         function doBlockLevels( $text, $linestart ) {
1933                 wfProfileIn( __METHOD__ );
1934
1935                 # Parsing through the text line by line.  The main thing
1936                 # happening here is handling of block-level elements p, pre,
1937                 # and making lists from lines starting with * # : etc.
1938                 #
1939                 $textLines = StringUtils::explode( "\n", $text );
1940
1941                 $lastPrefix = $output = '';
1942                 $this->mDTopen = $inBlockElem = false;
1943                 $prefixLength = 0;
1944                 $paragraphStack = false;
1945
1946                 foreach ( $textLines as $oLine ) {
1947                         # Fix up $linestart
1948                         if ( !$linestart ) {
1949                                 $output .= $oLine;
1950                                 $linestart = true;
1951                                 continue;
1952                         }
1953                         // * = ul
1954                         // # = ol
1955                         // ; = dt
1956                         // : = dd
1957                         // > = blockquote
1958
1959                         $lastPrefixLength = strlen( $lastPrefix );
1960                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1961                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1962
1963                         // Need to decode &gt; --> > for blockquote syntax. Re-encode later.
1964                         // To avoid collision with real >s, we temporarily convert them to &gt;
1965                         // This is a weird choice of armouring, but it's totally resistant to any
1966                         //  collision.
1967                         $orig = $oLine;
1968                         $oLine = strtr( $oLine, array( '&gt;' => '>', '>' => '&gt;' ) );
1969
1970                         // If not in a <pre> element, scan for and figure out what prefixes are there.
1971                         if ( !$this->mInPre ) {
1972                                 # Multiple prefixes may abut each other for nested lists.
1973                                 $prefixLength = strspn( $oLine, '*#:;>' );
1974                                 $prefix = substr( $oLine, 0, $prefixLength );
1975
1976                                 # eh?
1977                                 // ; and : are both from definition-lists, so they're equivalent
1978                                 //  for the purposes of determining whether or not we need to open/close
1979                                 //  elements.
1980                                 $prefix2 = str_replace( ';', ':', $prefix );
1981                                 $t = substr( $oLine, $prefixLength );
1982                                 $this->mInPre = (bool)$preOpenMatch;
1983                         } else {
1984                                 # Don't interpret any other prefixes in preformatted text
1985                                 $prefixLength = 0;
1986                                 $prefix = $prefix2 = '';
1987                                 $t = $oLine;
1988                         }
1989
1990                         // Re-encode >s now
1991                         $t = strtr( $t, array( '&gt;' => '>', '>' => '&gt;' ) );
1992
1993                         # List generation
1994                         if( $prefixLength && $lastPrefix === $prefix2 ) {
1995                                 # Same as the last item, so no need to deal with nesting or opening stuff
1996                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
1997                                 $paragraphStack = false;
1998
1999                                 if ( substr( $prefix, -1 ) === ';') {
2000                                         # The one nasty exception: definition lists work like this:
2001                                         # ; title : definition text
2002                                         # So we check for : in the remainder text to split up the
2003                                         # title and definition, without b0rking links.
2004                                         $term = $t2 = '';
2005                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2006                                                 $t = $t2;
2007                                                 $output .= $term . $this->nextItem( ':' );
2008                                         }
2009                                 }
2010                         } elseif( $prefixLength || $lastPrefixLength ) {
2011                                 // We need to open or close prefixes, or both.
2012
2013                                 # Either open or close a level...
2014                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2015                                 $paragraphStack = false;
2016
2017                                 // Close all the prefixes which aren't shared.
2018                                 while( $commonPrefixLength < $lastPrefixLength ) {
2019                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2020                                         --$lastPrefixLength;
2021                                 }
2022
2023                                 // Continue the current prefix if appropriate.
2024                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2025                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2026                                 }
2027
2028                                 // Open prefixes where appropriate.
2029                                 while ( $prefixLength > $commonPrefixLength ) {
2030                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2031                                         $output .= $this->openList( $char );
2032
2033                                         if ( ';' === $char ) {
2034                                                 # FIXME: This is dupe of code above
2035                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2036                                                         $t = $t2;
2037                                                         $output .= $term . $this->nextItem( ':' );
2038                                                 }
2039                                         }
2040                                         ++$commonPrefixLength;
2041                                 }
2042                                 $lastPrefix = $prefix2;
2043                         }
2044
2045                         // If we have no prefixes, go to paragraph mode.
2046                         if( 0 == $prefixLength ) {
2047                                 wfProfileIn( __METHOD__."-paragraph" );
2048                                 # No prefix (not in list)--go to paragraph mode
2049                                 // XXX: use a stack for nestable elements like span, table and div
2050                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2051                                 $closematch = preg_match(
2052                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2053                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2054                                 if ( $openmatch or $closematch ) {
2055                                         $paragraphStack = false;
2056                                         # TODO bug 5718: paragraph closed
2057                                         $output .= $this->closeParagraph();
2058                                         if ( $preOpenMatch and !$preCloseMatch ) {
2059                                                 $this->mInPre = true;
2060                                         }
2061                                         if ( $closematch ) {
2062                                                 $inBlockElem = false;
2063                                         } else {
2064                                                 $inBlockElem = true;
2065                                         }
2066                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2067                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2068                                                 // pre
2069                                                 if ($this->mLastSection !== 'pre') {
2070                                                         $paragraphStack = false;
2071                                                         $output .= $this->closeParagraph().'<pre>';
2072                                                         $this->mLastSection = 'pre';
2073                                                 }
2074                                                 $t = substr( $t, 1 );
2075                                         } else {
2076                                                 // paragraph
2077                                                 if ( '' == trim($t) ) {
2078                                                         if ( $paragraphStack ) {
2079                                                                 $output .= $paragraphStack.'<br />';
2080                                                                 $paragraphStack = false;
2081                                                                 $this->mLastSection = 'p';
2082                                                         } else {
2083                                                                 if ($this->mLastSection !== 'p' ) {
2084                                                                         $output .= $this->closeParagraph();
2085                                                                         $this->mLastSection = '';
2086                                                                         $paragraphStack = '<p>';
2087                                                                 } else {
2088                                                                         $paragraphStack = '</p><p>';
2089                                                                 }
2090                                                         }
2091                                                 } else {
2092                                                         if ( $paragraphStack ) {
2093                                                                 $output .= $paragraphStack;
2094                                                                 $paragraphStack = false;
2095                                                                 $this->mLastSection = 'p';
2096                                                         } else if ($this->mLastSection !== 'p') {
2097                                                                 $output .= $this->closeParagraph().'<p>';
2098                                                                 $this->mLastSection = 'p';
2099                                                         }
2100                                                 }
2101                                         }
2102                                 }
2103                                 wfProfileOut( __METHOD__."-paragraph" );
2104                         }
2105                         // somewhere above we forget to get out of pre block (bug 785)
2106                         if($preCloseMatch && $this->mInPre) {
2107                                 $this->mInPre = false;
2108                         }
2109                         if ($paragraphStack === false) {
2110                                 $output .= $t."\n";
2111                         }
2112                 }
2113                 while ( $prefixLength ) {
2114                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2115                         --$prefixLength;
2116                 }
2117                 if ( '' != $this->mLastSection ) {
2118                         $output .= '</' . $this->mLastSection . '>';
2119                         $this->mLastSection = '';
2120                 }
2121
2122                 wfProfileOut( __METHOD__ );
2123                 return $output;
2124         }
2125
2126         /**
2127          * Split up a string on ':', ignoring any occurences inside tags
2128          * to prevent illegal overlapping.
2129          * @param string $str the string to split
2130          * @param string &$before set to everything before the ':'
2131          * @param string &$after set to everything after the ':'
2132          * return string the position of the ':', or false if none found
2133          */
2134         function findColonNoLinks($str, &$before, &$after) {
2135                 wfProfileIn( __METHOD__ );
2136
2137                 $pos = strpos( $str, ':' );
2138                 if( $pos === false ) {
2139                         // Nothing to find!
2140                         wfProfileOut( __METHOD__ );
2141                         return false;
2142                 }
2143
2144                 $lt = strpos( $str, '<' );
2145                 if( $lt === false || $lt > $pos ) {
2146                         // Easy; no tag nesting to worry about
2147                         $before = substr( $str, 0, $pos );
2148                         $after = substr( $str, $pos+1 );
2149                         wfProfileOut( __METHOD__ );
2150                         return $pos;
2151                 }
2152
2153                 // Ugly state machine to walk through avoiding tags.
2154                 $state = self::COLON_STATE_TEXT;
2155                 $stack = 0;
2156                 $len = strlen( $str );
2157                 for( $i = 0; $i < $len; $i++ ) {
2158                         $c = $str{$i};
2159
2160                         switch( $state ) {
2161                         // (Using the number is a performance hack for common cases)
2162                         case 0: // self::COLON_STATE_TEXT:
2163                                 switch( $c ) {
2164                                 case "<":
2165                                         // Could be either a <start> tag or an </end> tag
2166                                         $state = self::COLON_STATE_TAGSTART;
2167                                         break;
2168                                 case ":":
2169                                         if( $stack == 0 ) {
2170                                                 // We found it!
2171                                                 $before = substr( $str, 0, $i );
2172                                                 $after = substr( $str, $i + 1 );
2173                                                 wfProfileOut( __METHOD__ );
2174                                                 return $i;
2175                                         }
2176                                         // Embedded in a tag; don't break it.
2177                                         break;
2178                                 default:
2179                                         // Skip ahead looking for something interesting
2180                                         $colon = strpos( $str, ':', $i );
2181                                         if( $colon === false ) {
2182                                                 // Nothing else interesting
2183                                                 wfProfileOut( __METHOD__ );
2184                                                 return false;
2185                                         }
2186                                         $lt = strpos( $str, '<', $i );
2187                                         if( $stack === 0 ) {
2188                                                 if( $lt === false || $colon < $lt ) {
2189                                                         // We found it!
2190                                                         $before = substr( $str, 0, $colon );
2191                                                         $after = substr( $str, $colon + 1 );
2192                                                         wfProfileOut( __METHOD__ );
2193                                                         return $i;
2194                                                 }
2195                                         }
2196                                         if( $lt === false ) {
2197                                                 // Nothing else interesting to find; abort!
2198                                                 // We're nested, but there's no close tags left. Abort!
2199                                                 break 2;
2200                                         }
2201                                         // Skip ahead to next tag start
2202                                         $i = $lt;
2203                                         $state = self::COLON_STATE_TAGSTART;
2204                                 }
2205                                 break;
2206                         case 1: // self::COLON_STATE_TAG:
2207                                 // In a <tag>
2208                                 switch( $c ) {
2209                                 case ">":
2210                                         $stack++;
2211                                         $state = self::COLON_STATE_TEXT;
2212                                         break;
2213                                 case "/":
2214                                         // Slash may be followed by >?
2215                                         $state = self::COLON_STATE_TAGSLASH;
2216                                         break;
2217                                 default:
2218                                         // ignore
2219                                 }
2220                                 break;
2221                         case 2: // self::COLON_STATE_TAGSTART:
2222                                 switch( $c ) {
2223                                 case "/":
2224                                         $state = self::COLON_STATE_CLOSETAG;
2225                                         break;
2226                                 case "!":
2227                                         $state = self::COLON_STATE_COMMENT;
2228                                         break;
2229                                 case ">":
2230                                         // Illegal early close? This shouldn't happen D:
2231                                         $state = self::COLON_STATE_TEXT;
2232                                         break;
2233                                 default:
2234                                         $state = self::COLON_STATE_TAG;
2235                                 }
2236                                 break;
2237                         case 3: // self::COLON_STATE_CLOSETAG:
2238                                 // In a </tag>
2239                                 if( $c === ">" ) {
2240                                         $stack--;
2241                                         if( $stack < 0 ) {
2242                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2243                                                 wfProfileOut( __METHOD__ );
2244                                                 return false;
2245                                         }
2246                                         $state = self::COLON_STATE_TEXT;
2247                                 }
2248                                 break;
2249                         case self::COLON_STATE_TAGSLASH:
2250                                 if( $c === ">" ) {
2251                                         // Yes, a self-closed tag <blah/>
2252                                         $state = self::COLON_STATE_TEXT;
2253                                 } else {
2254                                         // Probably we're jumping the gun, and this is an attribute
2255                                         $state = self::COLON_STATE_TAG;
2256                                 }
2257                                 break;
2258                         case 5: // self::COLON_STATE_COMMENT:
2259                                 if( $c === "-" ) {
2260                                         $state = self::COLON_STATE_COMMENTDASH;
2261                                 }
2262                                 break;
2263                         case self::COLON_STATE_COMMENTDASH:
2264                                 if( $c === "-" ) {
2265                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2266                                 } else {
2267                                         $state = self::COLON_STATE_COMMENT;
2268                                 }
2269                                 break;
2270                         case self::COLON_STATE_COMMENTDASHDASH:
2271                                 if( $c === ">" ) {
2272                                         $state = self::COLON_STATE_TEXT;
2273                                 } else {
2274                                         $state = self::COLON_STATE_COMMENT;
2275                                 }
2276                                 break;
2277                         default:
2278                                 throw new MWException( "State machine error in " . __METHOD__ );
2279                         }
2280                 }
2281                 if( $stack > 0 ) {
2282                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2283                         return false;
2284                 }
2285                 wfProfileOut( __METHOD__ );
2286                 return false;
2287         }
2288
2289         /**
2290          * Return value of a magic variable (like PAGENAME)
2291          *
2292          * @private
2293          */
2294         function getVariableValue( $index ) {
2295                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2296
2297                 /**
2298                  * Some of these require message or data lookups and can be
2299                  * expensive to check many times.
2300                  */
2301                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2302                         if ( isset( $this->mVarCache[$index] ) ) {
2303                                 return $this->mVarCache[$index];
2304                         }
2305                 }
2306
2307                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2308                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2309
2310                 # Use the time zone
2311                 global $wgLocaltimezone;
2312                 if ( isset( $wgLocaltimezone ) ) {
2313                         $oldtz = getenv( 'TZ' );
2314                         putenv( 'TZ='.$wgLocaltimezone );
2315                 }
2316
2317                 wfSuppressWarnings(); // E_STRICT system time bitching
2318                 $localTimestamp = date( 'YmdHis', $ts );
2319                 $localMonth = date( 'm', $ts );
2320                 $localMonth1 = date( 'n', $ts );
2321                 $localMonthName = date( 'n', $ts );
2322                 $localDay = date( 'j', $ts );
2323                 $localDay2 = date( 'd', $ts );
2324                 $localDayOfWeek = date( 'w', $ts );
2325                 $localWeek = date( 'W', $ts );
2326                 $localYear = date( 'Y', $ts );
2327                 $localHour = date( 'H', $ts );
2328                 if ( isset( $wgLocaltimezone ) ) {
2329                         putenv( 'TZ='.$oldtz );
2330                 }
2331                 wfRestoreWarnings();
2332
2333                 switch ( $index ) {
2334                         case 'currentmonth':
2335                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2336                         case 'currentmonth1':
2337                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2338                         case 'currentmonthname':
2339                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2340                         case 'currentmonthnamegen':
2341                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2342                         case 'currentmonthabbrev':
2343                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2344                         case 'currentday':
2345                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2346                         case 'currentday2':
2347                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2348                         case 'localmonth':
2349                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2350                         case 'localmonth1':
2351                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth1 );
2352                         case 'localmonthname':
2353                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2354                         case 'localmonthnamegen':
2355                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2356                         case 'localmonthabbrev':
2357                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2358                         case 'localday':
2359                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2360                         case 'localday2':
2361                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2362                         case 'pagename':
2363                                 return wfEscapeWikiText( $this->mTitle->getText() );
2364                         case 'pagenamee':
2365                                 return $this->mTitle->getPartialURL();
2366                         case 'fullpagename':
2367                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2368                         case 'fullpagenamee':
2369                                 return $this->mTitle->getPrefixedURL();
2370                         case 'subpagename':
2371                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2372                         case 'subpagenamee':
2373                                 return $this->mTitle->getSubpageUrlForm();
2374                         case 'basepagename':
2375                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2376                         case 'basepagenamee':
2377                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2378                         case 'talkpagename':
2379                                 if( $this->mTitle->canTalk() ) {
2380                                         $talkPage = $this->mTitle->getTalkPage();
2381                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2382                                 } else {
2383                                         return '';
2384                                 }
2385                         case 'talkpagenamee':
2386                                 if( $this->mTitle->canTalk() ) {
2387                                         $talkPage = $this->mTitle->getTalkPage();
2388                                         return $talkPage->getPrefixedUrl();
2389                                 } else {
2390                                         return '';
2391                                 }
2392                         case 'subjectpagename':
2393                                 $subjPage = $this->mTitle->getSubjectPage();
2394                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2395                         case 'subjectpagenamee':
2396                                 $subjPage = $this->mTitle->getSubjectPage();
2397                                 return $subjPage->getPrefixedUrl();
2398                         case 'revisionid':
2399                                 // Let the edit saving system know we should parse the page
2400                                 // *after* a revision ID has been assigned.
2401                                 $this->mOutput->setFlag( 'vary-revision' );
2402                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2403                                 return $this->mRevisionId;
2404                         case 'revisionday':
2405                                 // Let the edit saving system know we should parse the page
2406                                 // *after* a revision ID has been assigned. This is for null edits.
2407                                 $this->mOutput->setFlag( 'vary-revision' );
2408                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2409                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2410                         case 'revisionday2':
2411                                 // Let the edit saving system know we should parse the page
2412                                 // *after* a revision ID has been assigned. This is for null edits.
2413                                 $this->mOutput->setFlag( 'vary-revision' );
2414                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2415                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2416                         case 'revisionmonth':
2417                                 // Let the edit saving system know we should parse the page
2418                                 // *after* a revision ID has been assigned. This is for null edits.
2419                                 $this->mOutput->setFlag( 'vary-revision' );
2420                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2421                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2422                         case 'revisionyear':
2423                                 // Let the edit saving system know we should parse the page
2424                                 // *after* a revision ID has been assigned. This is for null edits.
2425                                 $this->mOutput->setFlag( 'vary-revision' );
2426                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2427                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2428                         case 'revisiontimestamp':
2429                                 // Let the edit saving system know we should parse the page
2430                                 // *after* a revision ID has been assigned. This is for null edits.
2431                                 $this->mOutput->setFlag( 'vary-revision' );
2432                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2433                                 return $this->getRevisionTimestamp();
2434                         case 'revisionuser':
2435                                 // Let the edit saving system know we should parse the page
2436                                 // *after* a revision ID has been assigned. This is for null edits.
2437                                 $this->mOutput->setFlag( 'vary-revision' );
2438                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2439                                 return $this->getRevisionUser();
2440                         case 'namespace':
2441                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2442                         case 'namespacee':
2443                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2444                         case 'talkspace':
2445                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2446                         case 'talkspacee':
2447                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2448                         case 'subjectspace':
2449                                 return $this->mTitle->getSubjectNsText();
2450                         case 'subjectspacee':
2451                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2452                         case 'currentdayname':
2453                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2454                         case 'currentyear':
2455                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2456                         case 'currenttime':
2457                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2458                         case 'currenthour':
2459                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2460                         case 'currentweek':
2461                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2462                                 // int to remove the padding
2463                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2464                         case 'currentdow':
2465                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2466                         case 'localdayname':
2467                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2468                         case 'localyear':
2469                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2470                         case 'localtime':
2471                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2472                         case 'localhour':
2473                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2474                         case 'localweek':
2475                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2476                                 // int to remove the padding
2477                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2478                         case 'localdow':
2479                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2480                         case 'numberofarticles':
2481                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2482                         case 'numberoffiles':
2483                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2484                         case 'numberofusers':
2485                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2486                         case 'numberofactiveusers':
2487                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::activeUsers() );
2488                         case 'numberofpages':
2489                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2490                         case 'numberofadmins':
2491                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2492                         case 'numberofedits':
2493                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2494                         case 'numberofviews':
2495                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::views() );
2496                         case 'currenttimestamp':
2497                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2498                         case 'localtimestamp':
2499                                 return $this->mVarCache[$index] = $localTimestamp;
2500                         case 'currentversion':
2501                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2502                         case 'sitename':
2503                                 return $wgSitename;
2504                         case 'server':
2505                                 return $wgServer;
2506                         case 'servername':
2507                                 return $wgServerName;
2508                         case 'scriptpath':
2509                                 return $wgScriptPath;
2510                         case 'directionmark':
2511                                 return $wgContLang->getDirMark();
2512                         case 'contentlanguage':
2513                                 global $wgContLanguageCode;
2514                                 return $wgContLanguageCode;
2515                         default:
2516                                 $ret = null;
2517                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2518                                         return $ret;
2519                                 else
2520                                         return null;
2521                 }
2522         }
2523
2524         /**
2525          * initialise the magic variables (like CURRENTMONTHNAME)
2526          *
2527          * @private
2528          */
2529         function initialiseVariables() {
2530                 wfProfileIn( __METHOD__ );
2531                 $variableIDs = MagicWord::getVariableIDs();
2532
2533                 $this->mVariables = new MagicWordArray( $variableIDs );
2534                 wfProfileOut( __METHOD__ );
2535         }
2536
2537         /**
2538          * Preprocess some wikitext and return the document tree.
2539          * This is the ghost of replace_variables().
2540          *
2541          * @param string $text The text to parse
2542          * @param integer flags Bitwise combination of:
2543          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2544          *                                     included. Default is to assume a direct page view.
2545          *
2546          * The generated DOM tree must depend only on the input text and the flags.
2547          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2548          *
2549          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2550          * change in the DOM tree for a given text, must be passed through the section identifier
2551          * in the section edit link and thus back to extractSections().
2552          *
2553          * The output of this function is currently only cached in process memory, but a persistent
2554          * cache may be implemented at a later date which takes further advantage of these strict
2555          * dependency requirements.
2556          *
2557          * @private
2558          */
2559         function preprocessToDom ( $text, $flags = 0 ) {
2560                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2561                 return $dom;
2562         }
2563
2564         /*
2565          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2566          */
2567         public static function splitWhitespace( $s ) {
2568                 $ltrimmed = ltrim( $s );
2569                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2570                 $trimmed = rtrim( $ltrimmed );
2571                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2572                 if ( $diff > 0 ) {
2573                         $w2 = substr( $ltrimmed, -$diff );
2574                 } else {
2575                         $w2 = '';
2576                 }
2577                 return array( $w1, $trimmed, $w2 );
2578         }
2579
2580         /**
2581          * Replace magic variables, templates, and template arguments
2582          * with the appropriate text. Templates are substituted recursively,
2583          * taking care to avoid infinite loops.
2584          *
2585          * Note that the substitution depends on value of $mOutputType:
2586          *  self::OT_WIKI: only {{subst:}} templates
2587          *  self::OT_PREPROCESS: templates but not extension tags
2588          *  self::OT_HTML: all templates and extension tags
2589          *
2590          * @param string $tex The text to transform
2591          * @param PPFrame $frame Object describing the arguments passed to the template.
2592          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2593          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2594          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2595          * @private
2596          */
2597         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2598                 # Is there any text? Also, Prevent too big inclusions!
2599                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2600                         return $text;
2601                 }
2602                 wfProfileIn( __METHOD__ );
2603
2604                 if ( $frame === false ) {
2605                         $frame = $this->getPreprocessor()->newFrame();
2606                 } elseif ( !( $frame instanceof PPFrame ) ) {
2607                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2608                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2609                 }
2610
2611                 $dom = $this->preprocessToDom( $text );
2612                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2613                 $text = $frame->expand( $dom, $flags );
2614
2615                 wfProfileOut( __METHOD__ );
2616                 return $text;
2617         }
2618
2619         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2620         static function createAssocArgs( $args ) {
2621                 $assocArgs = array();
2622                 $index = 1;
2623                 foreach( $args as $arg ) {
2624                         $eqpos = strpos( $arg, '=' );
2625                         if ( $eqpos === false ) {
2626                                 $assocArgs[$index++] = $arg;
2627                         } else {
2628                                 $name = trim( substr( $arg, 0, $eqpos ) );
2629                                 $value = trim( substr( $arg, $eqpos+1 ) );
2630                                 if ( $value === false ) {
2631                                         $value = '';
2632                                 }
2633                                 if ( $name !== false ) {
2634                                         $assocArgs[$name] = $value;
2635                                 }
2636                         }
2637                 }
2638
2639                 return $assocArgs;
2640         }
2641
2642         /**
2643          * Warn the user when a parser limitation is reached
2644          * Will warn at most once the user per limitation type
2645          *
2646          * @param string $limitationType, should be one of:
2647          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2648          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2649          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2650          * @params int $current, $max When an explicit limit has been
2651          *       exceeded, provide the values (optional)
2652          */
2653         function limitationWarn( $limitationType, $current=null, $max=null) {
2654                 $msgName = $limitationType . '-warning';
2655                 //does no harm if $current and $max are present but are unnecessary for the message
2656                 $warning = wfMsgExt( $msgName, array( 'parsemag', 'escape' ), $current, $max );
2657                 $this->mOutput->addWarning( $warning );
2658                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2659                 if ( $cat ) {
2660                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2661                 }
2662         }
2663
2664         /**
2665          * Return the text of a template, after recursively
2666          * replacing any variables or templates within the template.
2667          *
2668          * @param array $piece The parts of the template
2669          *  $piece['title']: the title, i.e. the part before the |
2670          *  $piece['parts']: the parameter array
2671          *  $piece['lineStart']: whether the brace was at the start of a line
2672          * @param PPFrame The current frame, contains template arguments
2673          * @return string the text of the template
2674          * @private
2675          */
2676         function braceSubstitution( $piece, $frame ) {
2677                 global $wgContLang, $wgNonincludableNamespaces;
2678                 wfProfileIn( __METHOD__ );
2679                 wfProfileIn( __METHOD__.'-setup' );
2680
2681                 # Flags
2682                 $found = false;             # $text has been filled
2683                 $nowiki = false;            # wiki markup in $text should be escaped
2684                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2685                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2686                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2687                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2688
2689                 # Title object, where $text came from
2690                 $title = NULL;
2691
2692                 # $part1 is the bit before the first |, and must contain only title characters.
2693                 # Various prefixes will be stripped from it later.
2694                 $titleWithSpaces = $frame->expand( $piece['title'] );
2695                 $part1 = trim( $titleWithSpaces );
2696                 $titleText = false;
2697
2698                 # Original title text preserved for various purposes
2699                 $originalTitle = $part1;
2700
2701                 # $args is a list of argument nodes, starting from index 0, not including $part1
2702                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2703                 wfProfileOut( __METHOD__.'-setup' );
2704
2705                 # SUBST
2706                 wfProfileIn( __METHOD__.'-modifiers' );
2707                 if ( !$found ) {
2708                         $mwSubst = MagicWord::get( 'subst' );
2709                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2710                                 # One of two possibilities is true:
2711                                 # 1) Found SUBST but not in the PST phase
2712                                 # 2) Didn't find SUBST and in the PST phase
2713                                 # In either case, return without further processing
2714                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2715                                 $isLocalObj = true;
2716                                 $found = true;
2717                         }
2718                 }
2719
2720                 # Variables
2721                 if ( !$found && $args->getLength() == 0 ) {
2722                         $id = $this->mVariables->matchStartToEnd( $part1 );
2723                         if ( $id !== false ) {
2724                                 $text = $this->getVariableValue( $id );
2725                                 if (MagicWord::getCacheTTL($id)>-1)
2726                                         $this->mOutput->mContainsOldMagic = true;
2727                                 $found = true;
2728                         }
2729                 }
2730
2731                 # MSG, MSGNW and RAW
2732                 if ( !$found ) {
2733                         # Check for MSGNW:
2734                         $mwMsgnw = MagicWord::get( 'msgnw' );
2735                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2736                                 $nowiki = true;
2737                         } else {
2738                                 # Remove obsolete MSG:
2739                                 $mwMsg = MagicWord::get( 'msg' );
2740                                 $mwMsg->matchStartAndRemove( $part1 );
2741                         }
2742
2743                         # Check for RAW:
2744                         $mwRaw = MagicWord::get( 'raw' );
2745                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2746                                 $forceRawInterwiki = true;
2747                         }
2748                 }
2749                 wfProfileOut( __METHOD__.'-modifiers' );
2750
2751                 # Parser functions
2752                 if ( !$found ) {
2753                         wfProfileIn( __METHOD__ . '-pfunc' );
2754
2755                         $colonPos = strpos( $part1, ':' );
2756                         if ( $colonPos !== false ) {
2757                                 # Case sensitive functions
2758                                 $function = substr( $part1, 0, $colonPos );
2759                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2760                                         $function = $this->mFunctionSynonyms[1][$function];
2761                                 } else {
2762                                         # Case insensitive functions
2763                                         $function = strtolower( $function );
2764                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2765                                                 $function = $this->mFunctionSynonyms[0][$function];
2766                                         } else {
2767                                                 $function = false;
2768                                         }
2769                                 }
2770                                 if ( $function ) {
2771                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2772                                         $initialArgs = array( &$this );
2773                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2774                                         if ( $flags & SFH_OBJECT_ARGS ) {
2775                                                 # Add a frame parameter, and pass the arguments as an array
2776                                                 $allArgs = $initialArgs;
2777                                                 $allArgs[] = $frame;
2778                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2779                                                         $funcArgs[] = $args->item( $i );
2780                                                 }
2781                                                 $allArgs[] = $funcArgs;
2782                                         } else {
2783                                                 # Convert arguments to plain text
2784                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2785                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2786                                                 }
2787                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2788                                         }
2789
2790                                         # Workaround for PHP bug 35229 and similar
2791                                         if ( !is_callable( $callback ) ) {
2792                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2793                                                 wfProfileOut( __METHOD__ );
2794                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2795                                         }
2796                                         $result = call_user_func_array( $callback, $allArgs );
2797                                         $found = true;
2798                                         $noparse = true;
2799                                         $preprocessFlags = 0;
2800
2801                                         if ( is_array( $result ) ) {
2802                                                 if ( isset( $result[0] ) ) {
2803                                                         $text = $result[0];
2804                                                         unset( $result[0] );
2805                                                 }
2806
2807                                                 // Extract flags into the local scope
2808                                                 // This allows callers to set flags such as nowiki, found, etc.
2809                                                 extract( $result );
2810                                         } else {
2811                                                 $text = $result;
2812                                         }
2813                                         if ( !$noparse ) {
2814                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2815                                                 $isChildObj = true;
2816                                         }
2817                                 }
2818                         }
2819                         wfProfileOut( __METHOD__ . '-pfunc' );
2820                 }
2821
2822                 # Finish mangling title and then check for loops.
2823                 # Set $title to a Title object and $titleText to the PDBK
2824                 if ( !$found ) {
2825                         $ns = NS_TEMPLATE;
2826                         # Split the title into page and subpage
2827                         $subpage = '';
2828                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2829                         if ($subpage !== '') {
2830                                 $ns = $this->mTitle->getNamespace();
2831                         }
2832                         $title = Title::newFromText( $part1, $ns );
2833                         if ( $title ) {
2834                                 $titleText = $title->getPrefixedText();
2835                                 # Check for language variants if the template is not found
2836                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2837                                         $wgContLang->findVariantLink( $part1, $title, true );
2838                                 }
2839                                 # Do recursion depth check
2840                                 $limit = $this->mOptions->getMaxTemplateDepth();
2841                                 if ( $frame->depth >= $limit ) {
2842                                         $found = true;
2843                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2844                                 }
2845                         }
2846                 }
2847
2848                 # Load from database
2849                 if ( !$found && $title ) {
2850                         wfProfileIn( __METHOD__ . '-loadtpl' );
2851                         if ( !$title->isExternal() ) {
2852                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2853                                         $text = SpecialPage::capturePath( $title );
2854                                         if ( is_string( $text ) ) {
2855                                                 $found = true;
2856                                                 $isHTML = true;
2857                                                 $this->disableCache();
2858                                         }
2859                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2860                                         $found = false; //access denied
2861                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2862                                 } else {
2863                                         list( $text, $title ) = $this->getTemplateDom( $title );
2864                                         if ( $text !== false ) {
2865                                                 $found = true;
2866                                                 $isChildObj = true;
2867                                         }
2868                                 }
2869
2870                                 # If the title is valid but undisplayable, make a link to it
2871                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2872                                         $text = "[[:$titleText]]";
2873                                         $found = true;
2874                                 }
2875                         } elseif ( $title->isTrans() ) {
2876                                 // Interwiki transclusion
2877                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2878                                         $text = $this->interwikiTransclude( $title, 'render' );
2879                                         $isHTML = true;
2880                                 } else {
2881                                         $text = $this->interwikiTransclude( $title, 'raw' );
2882                                         // Preprocess it like a template
2883                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2884                                         $isChildObj = true;
2885                                 }
2886                                 $found = true;
2887                         }
2888
2889                         # Do infinite loop check
2890                         # This has to be done after redirect resolution to avoid infinite loops via redirects
2891                         if ( !$frame->loopCheck( $title ) ) {
2892                                 $found = true;
2893                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2894                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2895                         }
2896                         wfProfileOut( __METHOD__ . '-loadtpl' );
2897                 }
2898
2899                 # If we haven't found text to substitute by now, we're done
2900                 # Recover the source wikitext and return it
2901                 if ( !$found ) {
2902                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2903                         wfProfileOut( __METHOD__ );
2904                         return array( 'object' => $text );
2905                 }
2906
2907                 # Expand DOM-style return values in a child frame
2908                 if ( $isChildObj ) {
2909                         # Clean up argument array
2910                         $newFrame = $frame->newChild( $args, $title );
2911
2912                         if ( $nowiki ) {
2913                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2914                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2915                                 # Expansion is eligible for the empty-frame cache
2916                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2917                                         $text = $this->mTplExpandCache[$titleText];
2918                                 } else {
2919                                         $text = $newFrame->expand( $text );
2920                                         $this->mTplExpandCache[$titleText] = $text;
2921                                 }
2922                         } else {
2923                                 # Uncached expansion
2924                                 $text = $newFrame->expand( $text );
2925                         }
2926                 }
2927                 if ( $isLocalObj && $nowiki ) {
2928                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2929                         $isLocalObj = false;
2930                 }
2931
2932                 # Replace raw HTML by a placeholder
2933                 # Add a blank line preceding, to prevent it from mucking up
2934                 # immediately preceding headings
2935                 if ( $isHTML ) {
2936                         $text = "\n\n" . $this->insertStripItem( $text );
2937                 }
2938                 # Escape nowiki-style return values
2939                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2940                         $text = wfEscapeWikiText( $text );
2941                 }
2942                 # Bug 529: if the template begins with a table or block-level
2943                 # element, it should be treated as beginning a new line.
2944                 # This behaviour is somewhat controversial.
2945                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
2946                         $text = "\n" . $text;
2947                 }
2948
2949                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
2950                         # Error, oversize inclusion
2951                         $text = "[[$originalTitle]]" .
2952                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
2953                         $this->limitationWarn( 'post-expand-template-inclusion' );
2954                 }
2955
2956                 if ( $isLocalObj ) {
2957                         $ret = array( 'object' => $text );
2958                 } else {
2959                         $ret = array( 'text' => $text );
2960                 }
2961
2962                 wfProfileOut( __METHOD__ );
2963                 return $ret;
2964         }
2965
2966         /**
2967          * Get the semi-parsed DOM representation of a template with a given title,
2968          * and its redirect destination title. Cached.
2969          */
2970         function getTemplateDom( $title ) {
2971                 $cacheTitle = $title;
2972                 $titleText = $title->getPrefixedDBkey();
2973
2974                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
2975                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
2976                         $title = Title::makeTitle( $ns, $dbk );
2977                         $titleText = $title->getPrefixedDBkey();
2978                 }
2979                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
2980                         return array( $this->mTplDomCache[$titleText], $title );
2981                 }
2982
2983                 // Cache miss, go to the database
2984                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
2985
2986                 if ( $text === false ) {
2987                         $this->mTplDomCache[$titleText] = false;
2988                         return array( false, $title );
2989                 }
2990
2991                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2992                 $this->mTplDomCache[ $titleText ] = $dom;
2993
2994                 if (! $title->equals($cacheTitle)) {
2995                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
2996                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
2997                 }
2998
2999                 return array( $dom, $title );
3000         }
3001
3002         /**
3003          * Fetch the unparsed text of a template and register a reference to it.
3004          */
3005         function fetchTemplateAndTitle( $title ) {
3006                 $templateCb = $this->mOptions->getTemplateCallback();
3007                 $stuff = call_user_func( $templateCb, $title, $this );
3008                 $text = $stuff['text'];
3009                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3010                 if ( isset( $stuff['deps'] ) ) {
3011                         foreach ( $stuff['deps'] as $dep ) {
3012                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3013                         }
3014                 }
3015                 return array($text,$finalTitle);
3016         }
3017
3018         function fetchTemplate( $title ) {
3019                 $rv = $this->fetchTemplateAndTitle($title);
3020                 return $rv[0];
3021         }
3022
3023         /**
3024          * Static function to get a template
3025          * Can be overridden via ParserOptions::setTemplateCallback().
3026          */
3027         static function statelessFetchTemplate( $title, $parser=false ) {
3028                 $text = $skip = false;
3029                 $finalTitle = $title;
3030                 $deps = array();
3031
3032                 // Loop to fetch the article, with up to 1 redirect
3033                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3034                         # Give extensions a chance to select the revision instead
3035                         $id = false; // Assume current
3036                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3037
3038                         if( $skip ) {
3039                                 $text = false;
3040                                 $deps[] = array(
3041                                         'title' => $title,
3042                                         'page_id' => $title->getArticleID(),
3043                                         'rev_id' => null );
3044                                 break;
3045                         }
3046                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3047                         $rev_id = $rev ? $rev->getId() : 0;
3048                         // If there is no current revision, there is no page
3049                         if( $id === false && !$rev ) {
3050                                 $linkCache = LinkCache::singleton();
3051                                 $linkCache->addBadLinkObj( $title );
3052                         }
3053
3054                         $deps[] = array(
3055                                 'title' => $title,
3056                                 'page_id' => $title->getArticleID(),
3057                                 'rev_id' => $rev_id );
3058
3059                         if( $rev ) {
3060                                 $text = $rev->getText();
3061                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3062                                 global $wgContLang;
3063                                 $message = $wgContLang->lcfirst( $title->getText() );
3064                                 $text = wfMsgForContentNoTrans( $message );
3065                                 if( wfEmptyMsg( $message, $text ) ) {
3066                                         $text = false;
3067                                         break;
3068                                 }
3069                         } else {
3070                                 break;
3071                         }
3072                         if ( $text === false ) {
3073                                 break;
3074                         }
3075                         // Redirect?
3076                         $finalTitle = $title;
3077                         $title = Title::newFromRedirect( $text );
3078                 }
3079                 return array(
3080                         'text' => $text,
3081                         'finalTitle' => $finalTitle,
3082                         'deps' => $deps );
3083         }
3084
3085         /**
3086          * Transclude an interwiki link.
3087          */
3088         function interwikiTransclude( $title, $action ) {
3089                 global $wgEnableScaryTranscluding;
3090
3091                 if (!$wgEnableScaryTranscluding)
3092                         return wfMsg('scarytranscludedisabled');
3093
3094                 $url = $title->getFullUrl( "action=$action" );
3095
3096                 if (strlen($url) > 255)
3097                         return wfMsg('scarytranscludetoolong');
3098                 return $this->fetchScaryTemplateMaybeFromCache($url);
3099         }
3100
3101         function fetchScaryTemplateMaybeFromCache($url) {
3102                 global $wgTranscludeCacheExpiry;
3103                 $dbr = wfGetDB(DB_SLAVE);
3104                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3105                                 array('tc_url' => $url));
3106                 if ($obj) {
3107                         $time = $obj->tc_time;
3108                         $text = $obj->tc_contents;
3109                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3110                                 return $text;
3111                         }
3112                 }
3113
3114                 $text = Http::get($url);
3115                 if (!$text)
3116                         return wfMsg('scarytranscludefailed', $url);
3117
3118                 $dbw = wfGetDB(DB_MASTER);
3119                 $dbw->replace('transcache', array('tc_url'), array(
3120                         'tc_url' => $url,
3121                         'tc_time' => time(),
3122                         'tc_contents' => $text));
3123                 return $text;
3124         }
3125
3126
3127         /**
3128          * Triple brace replacement -- used for template arguments
3129          * @private
3130          */
3131         function argSubstitution( $piece, $frame ) {
3132                 wfProfileIn( __METHOD__ );
3133
3134                 $error = false;
3135                 $parts = $piece['parts'];
3136                 $nameWithSpaces = $frame->expand( $piece['title'] );
3137                 $argName = trim( $nameWithSpaces );
3138                 $object = false;
3139                 $text = $frame->getArgument( $argName );
3140                 if (  $text === false && $parts->getLength() > 0
3141                   && (
3142                     $this->ot['html']
3143                     || $this->ot['pre']
3144                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3145                   )
3146                 ) {
3147                         # No match in frame, use the supplied default
3148                         $object = $parts->item( 0 )->getChildren();
3149                 }
3150                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3151                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3152                         $this->limitationWarn( 'post-expand-template-argument' );
3153                 }
3154
3155                 if ( $text === false && $object === false ) {
3156                         # No match anywhere
3157                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3158                 }
3159                 if ( $error !== false ) {
3160                         $text .= $error;
3161                 }
3162                 if ( $object !== false ) {
3163                         $ret = array( 'object' => $object );
3164                 } else {
3165                         $ret = array( 'text' => $text );
3166                 }
3167
3168                 wfProfileOut( __METHOD__ );
3169                 return $ret;
3170         }
3171
3172         /**
3173          * Return the text to be used for a given extension tag.
3174          * This is the ghost of strip().
3175          *
3176          * @param array $params Associative array of parameters:
3177          *     name       PPNode for the tag name
3178          *     attr       PPNode for unparsed text where tag attributes are thought to be
3179          *     attributes Optional associative array of parsed attributes
3180          *     inner      Contents of extension element
3181          *     noClose    Original text did not have a close tag
3182          * @param PPFrame $frame
3183          */
3184         function extensionSubstitution( $params, $frame ) {
3185                 global $wgRawHtml, $wgContLang;
3186
3187                 $name = $frame->expand( $params['name'] );
3188                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3189                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3190
3191                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3192
3193                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3194                         ( $this->ot['html'] || $this->ot['pre'] );
3195                 if ( $this->ot['html'] || $isFunctionTag ) {
3196                         $name = strtolower( $name );
3197                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3198                         if ( isset( $params['attributes'] ) ) {
3199                                 $attributes = $attributes + $params['attributes'];
3200                         }
3201                         switch ( $name ) {
3202                                 case 'html':
3203                                         if( $wgRawHtml ) {
3204                                                 $output = $content;
3205                                                 break;
3206                                         } else {
3207                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3208                                         }
3209                                 case 'nowiki':
3210                                         $content = strtr($content, array('-{' => '-&#123;', '}-' => '&#125;-'));
3211                                         $output = Xml::escapeTagsOnly( $content );
3212                                         break;
3213                                 case 'math':
3214                                         $output = $wgContLang->armourMath(
3215                                                 MathRenderer::renderMath( $content, $attributes ) );
3216                                         break;
3217                                 case 'gallery':
3218                                         $output = $this->renderImageGallery( $content, $attributes );
3219                                         break;
3220                                 default:
3221                                         if( isset( $this->mTagHooks[$name] ) ) {
3222                                                 # Workaround for PHP bug 35229 and similar
3223                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3224                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3225                                                 }
3226                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3227                                                         array( $content, $attributes, $this ) );
3228                                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3229                                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3230                                                 if( !is_callable( $callback ) )
3231                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3232
3233                                                 $output = call_user_func_array( $callback,
3234                                                         array( &$this, $frame, $content, $attributes ) );
3235                                         } else {
3236                                                 $output = '<span class="error">Invalid tag extension name: ' .
3237                                                         htmlspecialchars( $name ) . '</span>';
3238                                         }
3239                         }
3240                 } else {
3241                         if ( is_null( $attrText ) ) {
3242                                 $attrText = '';
3243                         }
3244                         if ( isset( $params['attributes'] ) ) {
3245                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3246                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3247                                                 htmlspecialchars( $attrValue ) . '"';
3248                                 }
3249                         }
3250                         if ( $content === null ) {
3251                                 $output = "<$name$attrText/>";
3252                         } else {
3253                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3254                                 $output = "<$name$attrText>$content$close";
3255                         }
3256                 }
3257
3258                 if( $isFunctionTag ) {
3259                         return $output;
3260                 } elseif ( $name === 'html' || $name === 'nowiki' ) {
3261                         $this->mStripState->nowiki->setPair( $marker, $output );
3262                 } else {
3263                         $this->mStripState->general->setPair( $marker, $output );
3264                 }
3265                 return $marker;
3266         }
3267
3268         /**
3269          * Increment an include size counter
3270          *
3271          * @param string $type The type of expansion
3272          * @param integer $size The size of the text
3273          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3274          */
3275         function incrementIncludeSize( $type, $size ) {
3276                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3277                         return false;
3278                 } else {
3279                         $this->mIncludeSizes[$type] += $size;
3280                         return true;
3281                 }
3282         }
3283
3284         /**
3285          * Increment the expensive function count
3286          *
3287          * @return boolean False if the limit has been exceeded
3288          */
3289         function incrementExpensiveFunctionCount() {
3290                 global $wgExpensiveParserFunctionLimit;
3291                 $this->mExpensiveFunctionCount++;
3292                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3293                         return true;
3294                 }
3295                 return false;
3296         }
3297
3298         /**
3299          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3300          * Fills $this->mDoubleUnderscores, returns the modified text
3301          */
3302         function doDoubleUnderscore( $text ) {
3303                 wfProfileIn( __METHOD__ );
3304                 // The position of __TOC__ needs to be recorded
3305                 $mw = MagicWord::get( 'toc' );
3306                 if( $mw->match( $text ) ) {
3307                         $this->mShowToc = true;
3308                         $this->mForceTocPosition = true;
3309
3310                         // Set a placeholder. At the end we'll fill it in with the TOC.
3311                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3312
3313                         // Only keep the first one.
3314                         $text = $mw->replace( '', $text );
3315                 }
3316
3317                 // Now match and remove the rest of them
3318                 $mwa = MagicWord::getDoubleUnderscoreArray();
3319                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3320
3321                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3322                         $this->mOutput->mNoGallery = true;
3323                 }
3324                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3325                         $this->mShowToc = false;
3326                 }
3327                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3328                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3329
3330                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3331                         if ( $containerCategory ) {
3332                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3333                         } else {
3334                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3335                         }
3336                 }
3337                 # (bug 8068) Allow control over whether robots index a page.
3338                 #
3339                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3340                 # is not desirable, the last one on the page should win.
3341                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3342                         $this->mOutput->setIndexPolicy( 'noindex' );
3343                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3344                         $this->mOutput->setIndexPolicy( 'index' );
3345                 }
3346                 wfProfileOut( __METHOD__ );
3347                 return $text;
3348         }
3349
3350         /**
3351          * This function accomplishes several tasks:
3352          * 1) Auto-number headings if that option is enabled
3353          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3354          * 3) Add a Table of contents on the top for users who have enabled the option
3355          * 4) Auto-anchor headings
3356          *
3357          * It loops through all headlines, collects the necessary data, then splits up the
3358          * string and re-inserts the newly formatted headlines.
3359          *
3360          * @param string $text
3361          * @param string $origText Original, untouched wikitext
3362          * @param boolean $isMain
3363          * @private
3364          */
3365         function formatHeadings( $text, $origText, $isMain=true ) {
3366                 global $wgMaxTocLevel, $wgContLang, $wgEnforceHtmlIds;
3367
3368                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3369                 $showEditLink = $this->mOptions->getEditSection();
3370
3371                 // Do not call quickUserCan unless necessary
3372                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3373                         $showEditLink = 0;
3374                 }
3375
3376                 # Inhibit editsection links if requested in the page
3377                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3378                         $showEditLink = 0;
3379                 }
3380
3381                 # Get all headlines for numbering them and adding funky stuff like [edit]
3382                 # links - this is for later, but we need the number of headlines right now
3383                 $matches = array();
3384                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3385
3386                 # if there are fewer than 4 headlines in the article, do not show TOC
3387                 # unless it's been explicitly enabled.
3388                 $enoughToc = $this->mShowToc &&
3389                         (($numMatches >= 4) || $this->mForceTocPosition);
3390
3391                 # Allow user to stipulate that a page should have a "new section"
3392                 # link added via __NEWSECTIONLINK__
3393                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3394                         $this->mOutput->setNewSection( true );
3395                 }
3396
3397                 # Allow user to remove the "new section"
3398                 # link via __NONEWSECTIONLINK__
3399                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3400                         $this->mOutput->hideNewSection( true );
3401                 }
3402
3403                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3404                 # override above conditions and always show TOC above first header
3405                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3406                         $this->mShowToc = true;
3407                         $enoughToc = true;
3408                 }
3409
3410                 # We need this to perform operations on the HTML
3411                 $sk = $this->mOptions->getSkin();
3412
3413                 # headline counter
3414                 $headlineCount = 0;
3415                 $numVisible = 0;
3416
3417                 # Ugh .. the TOC should have neat indentation levels which can be
3418                 # passed to the skin functions. These are determined here
3419                 $toc = '';
3420                 $full = '';
3421                 $head = array();
3422                 $sublevelCount = array();
3423                 $levelCount = array();
3424                 $toclevel = 0;
3425                 $level = 0;
3426                 $prevlevel = 0;
3427                 $toclevel = 0;
3428                 $prevtoclevel = 0;
3429                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3430                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3431                 $oldType = $this->mOutputType;
3432                 $this->setOutputType( self::OT_WIKI );
3433                 $frame = $this->getPreprocessor()->newFrame();
3434                 $root = $this->preprocessToDom( $origText );
3435                 $node = $root->getFirstChild();
3436                 $byteOffset = 0;
3437                 $tocraw = array();
3438
3439                 foreach( $matches[3] as $headline ) {
3440                         $isTemplate = false;
3441                         $titleText = false;
3442                         $sectionIndex = false;
3443                         $numbering = '';
3444                         $markerMatches = array();
3445                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3446                                 $serial = $markerMatches[1];
3447                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3448                                 $isTemplate = ($titleText != $baseTitleText);
3449                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3450                         }
3451
3452                         if( $toclevel ) {
3453                                 $prevlevel = $level;
3454                                 $prevtoclevel = $toclevel;
3455                         }
3456                         $level = $matches[1][$headlineCount];
3457
3458                         if ( $level > $prevlevel ) {
3459                                 # Increase TOC level
3460                                 $toclevel++;
3461                                 $sublevelCount[$toclevel] = 0;
3462                                 if( $toclevel<$wgMaxTocLevel ) {
3463                                         $prevtoclevel = $toclevel;
3464                                         $toc .= $sk->tocIndent();
3465                                         $numVisible++;
3466                                 }
3467                         }
3468                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3469                                 # Decrease TOC level, find level to jump to
3470
3471                                 for ($i = $toclevel; $i > 0; $i--) {
3472                                         if ( $levelCount[$i] == $level ) {
3473                                                 # Found last matching level
3474                                                 $toclevel = $i;
3475                                                 break;
3476                                         }
3477                                         elseif ( $levelCount[$i] < $level ) {
3478                                                 # Found first matching level below current level
3479                                                 $toclevel = $i + 1;
3480                                                 break;
3481                                         }
3482                                 }
3483                                 if( $i == 0 ) $toclevel = 1;
3484                                 if( $toclevel<$wgMaxTocLevel ) {
3485                                         if($prevtoclevel < $wgMaxTocLevel) {
3486                                                 # Unindent only if the previous toc level was shown :p
3487                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3488                                                 $prevtoclevel = $toclevel;
3489                                         } else {
3490                                                 $toc .= $sk->tocLineEnd();
3491                                         }
3492                                 }
3493                         }
3494                         else {
3495                                 # No change in level, end TOC line
3496                                 if( $toclevel<$wgMaxTocLevel ) {
3497                                         $toc .= $sk->tocLineEnd();
3498                                 }
3499                         }
3500
3501                         $levelCount[$toclevel] = $level;
3502
3503                         # count number of headlines for each level
3504                         @$sublevelCount[$toclevel]++;
3505                         $dot = 0;
3506                         for( $i = 1; $i <= $toclevel; $i++ ) {
3507                                 if( !empty( $sublevelCount[$i] ) ) {
3508                                         if( $dot ) {
3509                                                 $numbering .= '.';
3510                                         }
3511                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3512                                         $dot = 1;
3513                                 }
3514                         }
3515
3516                         # The safe header is a version of the header text safe to use for links
3517                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3518                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3519
3520                         # Remove link placeholders by the link text.
3521                         #     <!--LINK number-->
3522                         # turns into
3523                         #     link text with suffix
3524                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3525
3526                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3527                         $tocline = preg_replace(
3528                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3529                                 array( '',                          '<$1>'),
3530                                 $safeHeadline
3531                         );
3532                         $tocline = trim( $tocline );
3533
3534                         # For the anchor, strip out HTML-y stuff period
3535                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3536                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3537                         $safeHeadline = trim( $safeHeadline );
3538
3539                         # Save headline for section edit hint before it's escaped
3540                         $headlineHint = $safeHeadline;
3541
3542                         if ( $wgEnforceHtmlIds ) {
3543                                 $legacyHeadline = false;
3544                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3545                                         'noninitial' );
3546                         } else {
3547                                 # For reverse compatibility, provide an id that's
3548                                 # HTML4-compatible, like we used to.
3549                                 #
3550                                 # It may be worth noting, academically, that it's possible for
3551                                 # the legacy anchor to conflict with a non-legacy headline
3552                                 # anchor on the page.  In this case likely the "correct" thing
3553                                 # would be to either drop the legacy anchors or make sure
3554                                 # they're numbered first.  However, this would require people
3555                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3556                                 # manually, so let's not bother worrying about it.
3557                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3558                                         'noninitial' );
3559                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline, 'xml' );
3560
3561                                 if ( $legacyHeadline == $safeHeadline ) {
3562                                         # No reason to have both (in fact, we can't)
3563                                         $legacyHeadline = false;
3564                                 } elseif ( $legacyHeadline != Sanitizer::escapeId(
3565                                 $legacyHeadline, 'xml' ) ) {
3566                                         # The legacy id is invalid XML.  We used to allow this, but
3567                                         # there's no reason to do so anymore.  Backward
3568                                         # compatibility will fail slightly in this case, but it's
3569                                         # no big deal.
3570                                         $legacyHeadline = false;
3571                                 }
3572                         }
3573
3574                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3575                         # Does this apply to Unicode characters?  Because we aren't
3576                         # handling those here.
3577                         $arrayKey = strtolower( $safeHeadline );
3578                         if ( $legacyHeadline === false ) {
3579                                 $legacyArrayKey = false;
3580                         } else {
3581                                 $legacyArrayKey = strtolower( $legacyHeadline );
3582                         }
3583
3584                         # count how many in assoc. array so we can track dupes in anchors
3585                         if ( isset( $refers[$arrayKey] ) ) {
3586                                 $refers[$arrayKey]++;
3587                         } else {
3588                                 $refers[$arrayKey] = 1;
3589                         }
3590                         if ( isset( $refers[$legacyArrayKey] ) ) {
3591                                 $refers[$legacyArrayKey]++;
3592                         } else {
3593                                 $refers[$legacyArrayKey] = 1;
3594                         }
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                         $legacyAnchor = $legacyHeadline;
3605                         if ( $refers[$arrayKey] > 1 ) {
3606                                 $anchor .= '_' . $refers[$arrayKey];
3607                         }
3608                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3609                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3610                         }
3611                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3612                                 $toc .= $sk->tocLine($anchor, $tocline,
3613                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3614                         }
3615
3616                         # Add the section to the section tree
3617                         # Find the DOM node for this header
3618                         while ( $node && !$isTemplate ) {
3619                                 if ( $node->getName() === 'h' ) {
3620                                         $bits = $node->splitHeading();
3621                                         if ( $bits['i'] == $sectionIndex )
3622                                                 break;
3623                                 }
3624                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3625                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3626                                 $node = $node->getNextSibling();
3627                         }
3628                         $tocraw[] = array(
3629                                 'toclevel' => $toclevel,
3630                                 'level' => $level,
3631                                 'line' => $tocline,
3632                                 'number' => $numbering,
3633                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3634                                 'fromtitle' => $titleText,
3635                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3636                                 'anchor' => $anchor,
3637                         );
3638
3639                         # give headline the correct <h#> tag
3640                         if( $showEditLink && $sectionIndex !== false ) {
3641                                 if( $isTemplate ) {
3642                                         # Put a T flag in the section identifier, to indicate to extractSections()
3643                                         # that sections inside <includeonly> should be counted.
3644                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3645                                 } else {
3646                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3647                                 }
3648                         } else {
3649                                 $editlink = '';
3650                         }
3651                         $head[$headlineCount] = $sk->makeHeadline( $level,
3652                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3653                                 $editlink, $legacyAnchor );
3654
3655                         $headlineCount++;
3656                 }
3657
3658                 $this->setOutputType( $oldType );
3659
3660                 # Never ever show TOC if no headers
3661                 if( $numVisible < 1 ) {
3662                         $enoughToc = false;
3663                 }
3664
3665                 if( $enoughToc ) {
3666                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3667                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3668                         }
3669                         $toc = $sk->tocList( $toc );
3670                         $this->mOutput->setTOCHTML( $toc );
3671                 }
3672
3673                 if ( $isMain ) {
3674                         $this->mOutput->setSections( $tocraw );
3675                 }
3676
3677                 # split up and insert constructed headlines
3678
3679                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3680                 $i = 0;
3681
3682                 foreach( $blocks as $block ) {
3683                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3684                                 # This is the [edit] link that appears for the top block of text when
3685                                 # section editing is enabled
3686
3687                                 # Disabled because it broke block formatting
3688                                 # For example, a bullet point in the top line
3689                                 # $full .= $sk->editSectionLink(0);
3690                         }
3691                         $full .= $block;
3692                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3693                                 # Top anchor now in skin
3694                                 $full = $full.$toc;
3695                         }
3696
3697                         if( !empty( $head[$i] ) ) {
3698                                 $full .= $head[$i];
3699                         }
3700                         $i++;
3701                 }
3702                 if( $this->mForceTocPosition ) {
3703                         return str_replace( '<!--MWTOC-->', $toc, $full );
3704                 } else {
3705                         return $full;
3706                 }
3707         }
3708
3709         /**
3710          * Merge $tree2 into $tree1 by replacing the section with index
3711          * $section in $tree1 and its descendants with the sections in $tree2.
3712          * Note that in the returned section tree, only the 'index' and
3713          * 'byteoffset' fields are guaranteed to be correct.
3714          * @param $tree1 array Section tree from ParserOutput::getSectons()
3715         *  @param $tree2 array Section tree
3716          * @param $section int Section index
3717          * @param $title Title Title both section trees come from
3718          * @param $len2 int Length of the original wikitext for $tree2
3719          * @return array Merged section tree
3720          */
3721         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3722                 global $wgContLang;
3723                 $newTree = array();
3724                 $targetLevel = false;
3725                 $merged = false;
3726                 $lastLevel = 1;
3727                 $nextIndex = 1;
3728                 $numbering = array( 0 );
3729                 $titletext = $title->getPrefixedDBkey();
3730                 foreach ( $tree1 as $s ) {
3731                         if ( $targetLevel !== false ) {
3732                                 if ( $s['level'] <= $targetLevel )
3733                                         // We've skipped enough
3734                                         $targetLevel = false;
3735                                 else
3736                                         continue;
3737                         }
3738                         if ( $s['index'] != $section ||
3739                                         $s['fromtitle'] != $titletext ) {
3740                                 self::incrementNumbering( $numbering,
3741                                         $s['toclevel'], $lastLevel );
3742
3743                                 // Rewrite index, byteoffset and number
3744                                 if ( $s['fromtitle'] == $titletext ) {
3745                                         $s['index'] = $nextIndex++;
3746                                         if ( $merged )
3747                                                 $s['byteoffset'] += $len2;
3748                                 }
3749                                 $s['number']  = implode( '.', array_map(
3750                                         array( $wgContLang, 'formatnum' ),
3751                                         $numbering ) );
3752                                 $lastLevel = $s['toclevel'];
3753                                 $newTree[] = $s;
3754                         } else {
3755                                 // We're at $section
3756                                 // Insert sections from $tree2 here
3757                                 foreach ( $tree2 as $s2 ) {
3758                                         // Rewrite the fields in $s2
3759                                         // before inserting it
3760                                         $s2['toclevel'] += $s['toclevel'] - 1;
3761                                         $s2['level'] += $s['level'] - 1;
3762                                         $s2['index'] = $nextIndex++;
3763                                         $s2['byteoffset'] += $s['byteoffset'];
3764
3765                                         self::incrementNumbering( $numbering,
3766                                                 $s2['toclevel'], $lastLevel );
3767                                         $s2['number']  = implode( '.', array_map(
3768                                                 array( $wgContLang, 'formatnum' ),
3769                                                 $numbering ) );
3770                                         $lastLevel = $s2['toclevel'];
3771                                         $newTree[] = $s2;
3772                                 }
3773                                 // Skip all descendants of $section in $tree1
3774                                 $targetLevel = $s['level'];
3775                                 $merged = true;
3776                         }
3777                 }
3778                 return $newTree;
3779         }
3780
3781         /**
3782          * Increment a section number. Helper function for mergeSectionTrees()
3783          * @param $number array Array representing a section number
3784          * @param $level int Current TOC level (depth)
3785          * @param $lastLevel int Level of previous TOC entry
3786          */
3787         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3788                 if ( $level > $lastLevel )
3789                         $number[$level - 1] = 1;
3790                 else if ( $level < $lastLevel ) {
3791                         foreach ( $number as $key => $unused )
3792                                 if ( $key >= $level )
3793                                         unset( $number[$key] );
3794                         $number[$level - 1]++;
3795                 } else
3796                         $number[$level - 1]++;
3797         }
3798
3799         /**
3800          * Transform wiki markup when saving a page by doing \r\n -> \n
3801          * conversion, substitting signatures, {{subst:}} templates, etc.
3802          *
3803          * @param string $text the text to transform
3804          * @param Title &$title the Title object for the current article
3805          * @param User $user the User object describing the current user
3806          * @param ParserOptions $options parsing options
3807          * @param bool $clearState whether to clear the parser state first
3808          * @return string the altered wiki markup
3809          * @public
3810          */
3811         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3812                 $this->mOptions = $options;
3813                 $this->setTitle( $title );
3814                 $this->setOutputType( self::OT_WIKI );
3815
3816                 if ( $clearState ) {
3817                         $this->clearState();
3818                 }
3819
3820                 $pairs = array(
3821                         "\r\n" => "\n",
3822                 );
3823                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3824                 $text = $this->pstPass2( $text, $user );
3825                 $text = $this->mStripState->unstripBoth( $text );
3826                 return $text;
3827         }
3828
3829         /**
3830          * Pre-save transform helper function
3831          * @private
3832          */
3833         function pstPass2( $text, $user ) {
3834                 global $wgContLang, $wgLocaltimezone;
3835
3836                 /* Note: This is the timestamp saved as hardcoded wikitext to
3837                  * the database, we use $wgContLang here in order to give
3838                  * everyone the same signature and use the default one rather
3839                  * than the one selected in each user's preferences.
3840                  *
3841                  * (see also bug 12815)
3842                  */
3843                 $ts = $this->mOptions->getTimestamp();
3844                 $tz = wfMsgForContent( 'timezone-utc' );
3845                 if ( isset( $wgLocaltimezone ) ) {
3846                         $unixts = wfTimestamp( TS_UNIX, $ts );
3847                         $oldtz = getenv( 'TZ' );
3848                         putenv( 'TZ='.$wgLocaltimezone );
3849                         $ts = date( 'YmdHis', $unixts );
3850                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3851
3852                         /* Allow translation of timezones trough wiki. date() can return
3853                          * whatever crap the system uses, localised or not, so we cannot
3854                          * ship premade translations.
3855                          */
3856                         $key = 'timezone-' . strtolower( trim( $tz ) );
3857                         $value = wfMsgForContent( $key );
3858                         if ( !wfEmptyMsg( $key, $value ) ) $tz = $value;
3859
3860                         putenv( 'TZ='.$oldtz );
3861                 }
3862
3863                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3864
3865                 # Variable replacement
3866                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3867                 $text = $this->replaceVariables( $text );
3868
3869                 # Signatures
3870                 $sigText = $this->getUserSig( $user );
3871                 $text = strtr( $text, array(
3872                         '~~~~~' => $d,
3873                         '~~~~' => "$sigText $d",
3874                         '~~~' => $sigText
3875                 ) );
3876
3877                 # Context links: [[|name]] and [[name (context)|]]
3878                 #
3879                 global $wgLegalTitleChars;
3880                 $tc = "[$wgLegalTitleChars]";
3881                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3882
3883                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3884                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3885                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3886                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3887
3888                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3889                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3890                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3891                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3892
3893                 $t = $this->mTitle->getText();
3894                 $m = array();
3895                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3896                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3897                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3898                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3899                 } else {
3900                         # if there's no context, don't bother duplicating the title
3901                         $text = preg_replace( $p2, '[[\\1]]', $text );
3902                 }
3903
3904                 # Trim trailing whitespace
3905                 $text = rtrim( $text );
3906
3907                 return $text;
3908         }
3909
3910         /**
3911          * Fetch the user's signature text, if any, and normalize to
3912          * validated, ready-to-insert wikitext.
3913          *
3914          * @param User $user
3915          * @return string
3916          * @private
3917          */
3918         function getUserSig( &$user ) {
3919                 global $wgMaxSigChars;
3920
3921                 $username = $user->getName();
3922                 $nickname = $user->getOption( 'nickname' );
3923                 $nickname = $nickname === null ? $username : $nickname;
3924
3925                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3926                         $nickname = $username;
3927                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3928                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3929                         # Sig. might contain markup; validate this
3930                         if( $this->validateSig( $nickname ) !== false ) {
3931                                 # Validated; clean up (if needed) and return it
3932                                 return $this->cleanSig( $nickname, true );
3933                         } else {
3934                                 # Failed to validate; fall back to the default
3935                                 $nickname = $username;
3936                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3937                         }
3938                 }
3939
3940                 // Make sure nickname doesnt get a sig in a sig
3941                 $nickname = $this->cleanSigInSig( $nickname );
3942
3943                 # If we're still here, make it a link to the user page
3944                 $userText = wfEscapeWikiText( $username );
3945                 $nickText = wfEscapeWikiText( $nickname );
3946                 if ( $user->isAnon() )  {
3947                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3948                 } else {
3949                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3950                 }
3951         }
3952
3953         /**
3954          * Check that the user's signature contains no bad XML
3955          *
3956          * @param string $text
3957          * @return mixed An expanded string, or false if invalid.
3958          */
3959         function validateSig( $text ) {
3960                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3961         }
3962
3963         /**
3964          * Clean up signature text
3965          *
3966          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3967          * 2) Substitute all transclusions
3968          *
3969          * @param string $text
3970          * @param $parsing Whether we're cleaning (preferences save) or parsing
3971          * @return string Signature text
3972          */
3973         function cleanSig( $text, $parsing = false ) {
3974                 if ( !$parsing ) {
3975                         global $wgTitle;
3976                         $this->clearState();
3977                         $this->setTitle( $wgTitle );
3978                         $this->mOptions = new ParserOptions;
3979                         $this->setOutputType = self::OT_PREPROCESS;
3980                 }
3981
3982                 # Option to disable this feature
3983                 if ( !$this->mOptions->getCleanSignatures() ) {
3984                         return $text;
3985                 }
3986
3987                 # FIXME: regex doesn't respect extension tags or nowiki
3988                 #  => Move this logic to braceSubstitution()
3989                 $substWord = MagicWord::get( 'subst' );
3990                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3991                 $substText = '{{' . $substWord->getSynonym( 0 );
3992
3993                 $text = preg_replace( $substRegex, $substText, $text );
3994                 $text = $this->cleanSigInSig( $text );
3995                 $dom = $this->preprocessToDom( $text );
3996                 $frame = $this->getPreprocessor()->newFrame();
3997                 $text = $frame->expand( $dom );
3998
3999                 if ( !$parsing ) {
4000                         $text = $this->mStripState->unstripBoth( $text );
4001                 }
4002
4003                 return $text;
4004         }
4005
4006         /**
4007          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4008          * @param string $text
4009          * @return string Signature text with /~{3,5}/ removed
4010          */
4011         function cleanSigInSig( $text ) {
4012                 $text = preg_replace( '/~{3,5}/', '', $text );
4013                 return $text;
4014         }
4015
4016         /**
4017          * Set up some variables which are usually set up in parse()
4018          * so that an external function can call some class members with confidence
4019          * @public
4020          */
4021         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4022                 $this->setTitle( $title );
4023                 $this->mOptions = $options;
4024                 $this->setOutputType( $outputType );
4025                 if ( $clearState ) {
4026                         $this->clearState();
4027                 }
4028         }
4029
4030         /**
4031          * Wrapper for preprocess()
4032          *
4033          * @param string $text the text to preprocess
4034          * @param ParserOptions $options  options
4035          * @return string
4036          * @public
4037          */
4038         function transformMsg( $text, $options ) {
4039                 global $wgTitle;
4040                 static $executing = false;
4041
4042                 # Guard against infinite recursion
4043                 if ( $executing ) {
4044                         return $text;
4045                 }
4046                 $executing = true;
4047
4048                 wfProfileIn(__METHOD__);
4049                 $text = $this->preprocess( $text, $wgTitle, $options );
4050
4051                 $executing = false;
4052                 wfProfileOut(__METHOD__);
4053                 return $text;
4054         }
4055
4056         /**
4057          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4058          * The callback should have the following form:
4059          *    function myParserHook( $text, $params, &$parser ) { ... }
4060          *
4061          * Transform and return $text. Use $parser for any required context, e.g. use
4062          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4063          *
4064          * @public
4065          *
4066          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4067          * @param mixed $callback The callback function (and object) to use for the tag
4068          *
4069          * @return The old value of the mTagHooks array associated with the hook
4070          */
4071         function setHook( $tag, $callback ) {
4072                 $tag = strtolower( $tag );
4073                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4074                 $this->mTagHooks[$tag] = $callback;
4075                 if( !in_array( $tag, $this->mStripList ) ) {
4076                         $this->mStripList[] = $tag;
4077                 }
4078
4079                 return $oldVal;
4080         }
4081
4082         function setTransparentTagHook( $tag, $callback ) {
4083                 $tag = strtolower( $tag );
4084                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4085                 $this->mTransparentTagHooks[$tag] = $callback;
4086
4087                 return $oldVal;
4088         }
4089
4090         /**
4091          * Remove all tag hooks
4092          */
4093         function clearTagHooks() {
4094                 $this->mTagHooks = array();
4095                 $this->mStripList = $this->mDefaultStripList;
4096         }
4097
4098         /**
4099          * Create a function, e.g. {{sum:1|2|3}}
4100          * The callback function should have the form:
4101          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4102          *
4103          * Or with SFH_OBJECT_ARGS:
4104          *    function myParserFunction( $parser, $frame, $args ) { ... }
4105          *
4106          * The callback may either return the text result of the function, or an array with the text
4107          * in element 0, and a number of flags in the other elements. The names of the flags are
4108          * specified in the keys. Valid flags are:
4109          *   found                     The text returned is valid, stop processing the template. This
4110          *                             is on by default.
4111          *   nowiki                    Wiki markup in the return value should be escaped
4112          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4113          *
4114          * @public
4115          *
4116          * @param string $id The magic word ID
4117          * @param mixed $callback The callback function (and object) to use
4118          * @param integer $flags a combination of the following flags:
4119          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4120          *
4121          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4122          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4123          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4124          *     the arguments, and to control the way they are expanded.
4125          *
4126          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4127          *     arguments, for instance:
4128          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4129          *
4130          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4131          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4132          *     working if/when this is changed.
4133          *
4134          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4135          *     expansion.
4136          *
4137          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4138          *     about the methods available in PPFrame and PPNode.
4139          *
4140          * @return The old callback function for this name, if any
4141          */
4142         function setFunctionHook( $id, $callback, $flags = 0 ) {
4143                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4144                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4145
4146                 # Add to function cache
4147                 $mw = MagicWord::get( $id );
4148                 if( !$mw )
4149                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4150
4151                 $synonyms = $mw->getSynonyms();
4152                 $sensitive = intval( $mw->isCaseSensitive() );
4153
4154                 foreach ( $synonyms as $syn ) {
4155                         # Case
4156                         if ( !$sensitive ) {
4157                                 $syn = strtolower( $syn );
4158                         }
4159                         # Add leading hash
4160                         if ( !( $flags & SFH_NO_HASH ) ) {
4161                                 $syn = '#' . $syn;
4162                         }
4163                         # Remove trailing colon
4164                         if ( substr( $syn, -1, 1 ) === ':' ) {
4165                                 $syn = substr( $syn, 0, -1 );
4166                         }
4167                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4168                 }
4169                 return $oldVal;
4170         }
4171
4172         /**
4173          * Get all registered function hook identifiers
4174          *
4175          * @return array
4176          */
4177         function getFunctionHooks() {
4178                 return array_keys( $this->mFunctionHooks );
4179         }
4180
4181         /**
4182          * Create a tag function, e.g. <test>some stuff</test>.
4183          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4184          * Unlike parser functions, their content is not preprocessed.
4185          */
4186         function setFunctionTagHook( $tag, $callback, $flags ) {
4187                 $tag = strtolower( $tag );
4188                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4189                         $this->mFunctionTagHooks[$tag] : null;
4190                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4191
4192                 if( !in_array( $tag, $this->mStripList ) ) {
4193                         $this->mStripList[] = $tag;
4194                 }
4195
4196                 return $old;
4197         }
4198
4199         /**
4200          * FIXME: update documentation. makeLinkObj() is deprecated.
4201          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4202          * Placeholders created in Skin::makeLinkObj()
4203          * Returns an array of link CSS classes, indexed by PDBK.
4204          */
4205         function replaceLinkHolders( &$text, $options = 0 ) {
4206                 return $this->mLinkHolders->replace( $text );
4207         }
4208
4209         /**
4210          * Replace <!--LINK--> link placeholders with plain text of links
4211          * (not HTML-formatted).
4212          * @param string $text
4213          * @return string
4214          */
4215         function replaceLinkHoldersText( $text ) {
4216                 return $this->mLinkHolders->replaceText( $text );
4217         }
4218
4219         /**
4220          * Tag hook handler for 'pre'.
4221          */
4222         function renderPreTag( $text, $attribs ) {
4223                 // Backwards-compatibility hack
4224                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4225
4226                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4227                 return Xml::openElement( 'pre', $attribs ) .
4228                         Xml::escapeTagsOnly( $content ) .
4229                         '</pre>';
4230         }
4231
4232         /**
4233          * Renders an image gallery from a text with one line per image.
4234          * text labels may be given by using |-style alternative text. E.g.
4235          *   Image:one.jpg|The number "1"
4236          *   Image:tree.jpg|A tree
4237          * given as text will return the HTML of a gallery with two images,
4238          * labeled 'The number "1"' and
4239          * 'A tree'.
4240          */
4241         function renderImageGallery( $text, $params ) {
4242                 $ig = new ImageGallery();
4243                 $ig->setContextTitle( $this->mTitle );
4244                 $ig->setShowBytes( false );
4245                 $ig->setShowFilename( false );
4246                 $ig->setParser( $this );
4247                 $ig->setHideBadImages();
4248                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4249                 $ig->useSkin( $this->mOptions->getSkin() );
4250                 $ig->mRevisionId = $this->mRevisionId;
4251
4252                 if( isset( $params['caption'] ) ) {
4253                         $caption = $params['caption'];
4254                         $caption = htmlspecialchars( $caption );
4255                         $caption = $this->replaceInternalLinks( $caption );
4256                         $ig->setCaptionHtml( $caption );
4257                 }
4258                 if( isset( $params['perrow'] ) ) {
4259                         $ig->setPerRow( $params['perrow'] );
4260                 }
4261                 if( isset( $params['widths'] ) ) {
4262                         $ig->setWidths( $params['widths'] );
4263                 }
4264                 if( isset( $params['heights'] ) ) {
4265                         $ig->setHeights( $params['heights'] );
4266                 }
4267
4268                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4269
4270                 $lines = StringUtils::explode( "\n", $text );
4271                 foreach ( $lines as $line ) {
4272                         # match lines like these:
4273                         # Image:someimage.jpg|This is some image
4274                         $matches = array();
4275                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4276                         # Skip empty lines
4277                         if ( count( $matches ) == 0 ) {
4278                                 continue;
4279                         }
4280
4281                         if ( strpos( $matches[0], '%' ) !== false )
4282                                 $matches[1] = urldecode( $matches[1] );
4283                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4284                         $nt =& $tp;
4285                         if( is_null( $nt ) ) {
4286                                 # Bogus title. Ignore these so we don't bomb out later.
4287                                 continue;
4288                         }
4289                         if ( isset( $matches[3] ) ) {
4290                                 $label = $matches[3];
4291                         } else {
4292                                 $label = '';
4293                         }
4294
4295                         $html = $this->recursiveTagParse( trim( $label ) );
4296
4297                         $ig->add( $nt, $html );
4298
4299                         # Only add real images (bug #5586)
4300                         if ( $nt->getNamespace() == NS_FILE ) {
4301                                 $this->mOutput->addImage( $nt->getDBkey() );
4302                         }
4303                 }
4304                 return $ig->toHTML();
4305         }
4306
4307         function getImageParams( $handler ) {
4308                 if ( $handler ) {
4309                         $handlerClass = get_class( $handler );
4310                 } else {
4311                         $handlerClass = '';
4312                 }
4313                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4314                         // Initialise static lists
4315                         static $internalParamNames = array(
4316                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4317                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4318                                         'bottom', 'text-bottom' ),
4319                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4320                                         'upright', 'border', 'link', 'alt' ),
4321                         );
4322                         static $internalParamMap;
4323                         if ( !$internalParamMap ) {
4324                                 $internalParamMap = array();
4325                                 foreach ( $internalParamNames as $type => $names ) {
4326                                         foreach ( $names as $name ) {
4327                                                 $magicName = str_replace( '-', '_', "img_$name" );
4328                                                 $internalParamMap[$magicName] = array( $type, $name );
4329                                         }
4330                                 }
4331                         }
4332
4333                         // Add handler params
4334                         $paramMap = $internalParamMap;
4335                         if ( $handler ) {
4336                                 $handlerParamMap = $handler->getParamMap();
4337                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4338                                         $paramMap[$magic] = array( 'handler', $paramName );
4339                                 }
4340                         }
4341                         $this->mImageParams[$handlerClass] = $paramMap;
4342                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4343                 }
4344                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4345         }
4346
4347         /**
4348          * Parse image options text and use it to make an image
4349          * @param Title $title
4350          * @param string $options
4351          * @param LinkHolderArray $holders
4352          */
4353         function makeImage( $title, $options, $holders = false ) {
4354                 # Check if the options text is of the form "options|alt text"
4355                 # Options are:
4356                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4357                 #  * left       no resizing, just left align. label is used for alt= only
4358                 #  * right      same, but right aligned
4359                 #  * none       same, but not aligned
4360                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4361                 #  * center     center the image
4362                 #  * frame      Keep original image size, no magnify-button.
4363                 #  * framed     Same as "frame"
4364                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4365                 #  * upright    reduce width for upright images, rounded to full __0 px
4366                 #  * border     draw a 1px border around the image
4367                 #  * alt        Text for HTML alt attribute (defaults to empty)
4368                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4369                 # vertical-align values (no % or length right now):
4370                 #  * baseline
4371                 #  * sub
4372                 #  * super
4373                 #  * top
4374                 #  * text-top
4375                 #  * middle
4376                 #  * bottom
4377                 #  * text-bottom
4378
4379                 $parts = StringUtils::explode( "|", $options );
4380                 $sk = $this->mOptions->getSkin();
4381
4382                 # Give extensions a chance to select the file revision for us
4383                 $skip = $time = $descQuery = false;
4384                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4385
4386                 if ( $skip ) {
4387                         return $sk->link( $title );
4388                 }
4389
4390                 # Get the file
4391                 $imagename = $title->getDBkey();
4392                 $file = wfFindFile( $title, array( 'time' => $time ) );
4393                 # Get parameter map
4394                 $handler = $file ? $file->getHandler() : false;
4395
4396                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4397
4398                 # Process the input parameters
4399                 $caption = '';
4400                 $params = array( 'frame' => array(), 'handler' => array(),
4401                         'horizAlign' => array(), 'vertAlign' => array() );
4402                 foreach( $parts as $part ) {
4403                         $part = trim( $part );
4404                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4405                         $validated = false;
4406                         if( isset( $paramMap[$magicName] ) ) {
4407                                 list( $type, $paramName ) = $paramMap[$magicName];
4408
4409                                 // Special case; width and height come in one variable together
4410                                 if( $type === 'handler' && $paramName === 'width' ) {
4411                                         $m = array();
4412                                         # (bug 13500) In both cases (width/height and width only),
4413                                         # permit trailing "px" for backward compatibility.
4414                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4415                                                 $width = intval( $m[1] );
4416                                                 $height = intval( $m[2] );
4417                                                 if ( $handler->validateParam( 'width', $width ) ) {
4418                                                         $params[$type]['width'] = $width;
4419                                                         $validated = true;
4420                                                 }
4421                                                 if ( $handler->validateParam( 'height', $height ) ) {
4422                                                         $params[$type]['height'] = $height;
4423                                                         $validated = true;
4424                                                 }
4425                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4426                                                 $width = intval( $value );
4427                                                 if ( $handler->validateParam( 'width', $width ) ) {
4428                                                         $params[$type]['width'] = $width;
4429                                                         $validated = true;
4430                                                 }
4431                                         } // else no validation -- bug 13436
4432                                 } else {
4433                                         if ( $type === 'handler' ) {
4434                                                 # Validate handler parameter
4435                                                 $validated = $handler->validateParam( $paramName, $value );
4436                                         } else {
4437                                                 # Validate internal parameters
4438                                                 switch( $paramName ) {
4439                                                 case 'manualthumb':
4440                                                 case 'alt':
4441                                                         // @fixme - possibly check validity here for
4442                                                         // manualthumb? downstream behavior seems odd with
4443                                                         // missing manual thumbs.
4444                                                         $validated = true;
4445                                                         $value = $this->stripAltText( $value, $holders );
4446                                                         break;
4447                                                 case 'link':
4448                                                         $chars = self::EXT_LINK_URL_CLASS;
4449                                                         $prots = $this->mUrlProtocols;
4450                                                         if ( $value === '' ) {
4451                                                                 $paramName = 'no-link';
4452                                                                 $value = true;
4453                                                                 $validated = true;
4454                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4455                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4456                                                                         $paramName = 'link-url';
4457                                                                         $this->mOutput->addExternalLink( $value );
4458                                                                         $validated = true;
4459                                                                 }
4460                                                         } else {
4461                                                                 $linkTitle = Title::newFromText( $value );
4462                                                                 if ( $linkTitle ) {
4463                                                                         $paramName = 'link-title';
4464                                                                         $value = $linkTitle;
4465                                                                         $this->mOutput->addLink( $linkTitle );
4466                                                                         $validated = true;
4467                                                                 }
4468                                                         }
4469                                                         break;
4470                                                 default:
4471                                                         // Most other things appear to be empty or numeric...
4472                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4473                                                 }
4474                                         }
4475
4476                                         if ( $validated ) {
4477                                                 $params[$type][$paramName] = $value;
4478                                         }
4479                                 }
4480                         }
4481                         if ( !$validated ) {
4482                                 $caption = $part;
4483                         }
4484                 }
4485
4486                 # Process alignment parameters
4487                 if ( $params['horizAlign'] ) {
4488                         $params['frame']['align'] = key( $params['horizAlign'] );
4489                 }
4490                 if ( $params['vertAlign'] ) {
4491                         $params['frame']['valign'] = key( $params['vertAlign'] );
4492                 }
4493
4494                 $params['frame']['caption'] = $caption;
4495
4496                 # Will the image be presented in a frame, with the caption below?
4497                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4498                                  isset( $params['frame']['framed'] ) ||
4499                                  isset( $params['frame']['thumbnail'] ) ||
4500                                  isset( $params['frame']['manualthumb'] );
4501
4502                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4503                 # came to also set the caption, ordinary text after the image -- which
4504                 # makes no sense, because that just repeats the text multiple times in
4505                 # screen readers.  It *also* came to set the title attribute.
4506                 #
4507                 # Now that we have an alt attribute, we should not set the alt text to
4508                 # equal the caption: that's worse than useless, it just repeats the
4509                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4510                 # use the unnamed parameter for alt text as well, just for the time be-
4511                 # ing, if the unnamed param is set and the alt param is not.
4512                 #
4513                 # For the future, we need to figure out if we want to tweak this more,
4514                 # e.g., introducing a title= parameter for the title; ignoring the un-
4515                 # named parameter entirely for images without a caption; adding an ex-
4516                 # plicit caption= parameter and preserving the old magic unnamed para-
4517                 # meter for BC; ...
4518                 if ( $imageIsFramed ) { # Framed image
4519                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4520                                 # No caption or alt text, add the filename as the alt text so
4521                                 # that screen readers at least get some description of the image
4522                                 $params['frame']['alt'] = $title->getText();
4523                         }
4524                         # Do not set $params['frame']['title'] because tooltips don't make sense
4525                         # for framed images
4526                 } else { # Inline image
4527                         if ( !isset( $params['frame']['alt'] ) ) {
4528                                 # No alt text, use the "caption" for the alt text
4529                                 if ( $caption !== '') {
4530                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4531                                 } else {
4532                                         # No caption, fall back to using the filename for the
4533                                         # alt text
4534                                         $params['frame']['alt'] = $title->getText();
4535                                 }
4536                         }
4537                         # Use the "caption" for the tooltip text
4538                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4539                 }
4540
4541                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4542
4543                 # Linker does the rest
4544                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4545
4546                 # Give the handler a chance to modify the parser object
4547                 if ( $handler ) {
4548                         $handler->parserTransformHook( $this, $file );
4549                 }
4550
4551                 return $ret;
4552         }
4553
4554         protected function stripAltText( $caption, $holders ) {
4555                 # Strip bad stuff out of the title (tooltip).  We can't just use
4556                 # replaceLinkHoldersText() here, because if this function is called
4557                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4558                 if ( $holders ) {
4559                         $tooltip = $holders->replaceText( $caption );
4560                 } else {
4561                         $tooltip = $this->replaceLinkHoldersText( $caption );
4562                 }
4563
4564                 # make sure there are no placeholders in thumbnail attributes
4565                 # that are later expanded to html- so expand them now and
4566                 # remove the tags
4567                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4568                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4569
4570                 return $tooltip;
4571         }
4572
4573         /**
4574          * Set a flag in the output object indicating that the content is dynamic and
4575          * shouldn't be cached.
4576          */
4577         function disableCache() {
4578                 wfDebug( "Parser output marked as uncacheable.\n" );
4579                 $this->mOutput->mCacheTime = -1;
4580         }
4581
4582         /**#@+
4583          * Callback from the Sanitizer for expanding items found in HTML attribute
4584          * values, so they can be safely tested and escaped.
4585          * @param string $text
4586          * @param PPFrame $frame
4587          * @return string
4588          * @private
4589          */
4590         function attributeStripCallback( &$text, $frame = false ) {
4591                 $text = $this->replaceVariables( $text, $frame );
4592                 $text = $this->mStripState->unstripBoth( $text );
4593                 return $text;
4594         }
4595
4596         /**#@-*/
4597
4598         /**#@+
4599          * Accessor/mutator
4600          */
4601         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4602         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4603         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4604         /**#@-*/
4605
4606         /**#@+
4607          * Accessor
4608          */
4609         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4610         /**#@-*/
4611
4612
4613         /**
4614          * Break wikitext input into sections, and either pull or replace
4615          * some particular section's text.
4616          *
4617          * External callers should use the getSection and replaceSection methods.
4618          *
4619          * @param string $text Page wikitext
4620          * @param string $section A section identifier string of the form:
4621          *   <flag1> - <flag2> - ... - <section number>
4622          *
4623          * Currently the only recognised flag is "T", which means the target section number
4624          * was derived during a template inclusion parse, in other words this is a template
4625          * section edit link. If no flags are given, it was an ordinary section edit link.
4626          * This flag is required to avoid a section numbering mismatch when a section is
4627          * enclosed by <includeonly> (bug 6563).
4628          *
4629          * The section number 0 pulls the text before the first heading; other numbers will
4630          * pull the given section along with its lower-level subsections. If the section is
4631          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4632          *
4633          * @param string $mode One of "get" or "replace"
4634          * @param string $newText Replacement text for section data.
4635          * @return string for "get", the extracted section text.
4636          *                for "replace", the whole page with the section replaced.
4637          */
4638         private function extractSections( $text, $section, $mode, $newText='' ) {
4639                 global $wgTitle;
4640                 $this->clearState();
4641                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4642                 $this->mOptions = new ParserOptions;
4643                 $this->setOutputType( self::OT_WIKI );
4644                 $outText = '';
4645                 $frame = $this->getPreprocessor()->newFrame();
4646
4647                 // Process section extraction flags
4648                 $flags = 0;
4649                 $sectionParts = explode( '-', $section );
4650                 $sectionIndex = array_pop( $sectionParts );
4651                 foreach ( $sectionParts as $part ) {
4652                         if ( $part === 'T' ) {
4653                                 $flags |= self::PTD_FOR_INCLUSION;
4654                         }
4655                 }
4656                 // Preprocess the text
4657                 $root = $this->preprocessToDom( $text, $flags );
4658
4659                 // <h> nodes indicate section breaks
4660                 // They can only occur at the top level, so we can find them by iterating the root's children
4661                 $node = $root->getFirstChild();
4662
4663                 // Find the target section
4664                 if ( $sectionIndex == 0 ) {
4665                         // Section zero doesn't nest, level=big
4666                         $targetLevel = 1000;
4667                 } else {
4668             while ( $node ) {
4669                 if ( $node->getName() === 'h' ) {
4670                     $bits = $node->splitHeading();
4671                                         if ( $bits['i'] == $sectionIndex ) {
4672                                         $targetLevel = $bits['level'];
4673                                                 break;
4674                                         }
4675                                 }
4676                                 if ( $mode === 'replace' ) {
4677                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4678                                 }
4679                                 $node = $node->getNextSibling();
4680                         }
4681                 }
4682
4683                 if ( !$node ) {
4684                         // Not found
4685                         if ( $mode === 'get' ) {
4686                                 return $newText;
4687                         } else {
4688                                 return $text;
4689                         }
4690                 }
4691
4692                 // Find the end of the section, including nested sections
4693                 do {
4694                         if ( $node->getName() === 'h' ) {
4695                                 $bits = $node->splitHeading();
4696                                 $curLevel = $bits['level'];
4697                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4698                                         break;
4699                                 }
4700                         }
4701                         if ( $mode === 'get' ) {
4702                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4703                         }
4704                         $node = $node->getNextSibling();
4705                 } while ( $node );
4706
4707                 // Write out the remainder (in replace mode only)
4708                 if ( $mode === 'replace' ) {
4709                         // Output the replacement text
4710                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4711                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4712                         // Only add trailing whitespace if there is newText
4713                         if($newText != "") {
4714                                 $outText .= $newText . "\n\n";
4715                         }
4716
4717                         while ( $node ) {
4718                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4719                                 $node = $node->getNextSibling();
4720                         }
4721                 }
4722
4723                 if ( is_string( $outText ) ) {
4724                         // Re-insert stripped tags
4725                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4726                 }
4727
4728                 return $outText;
4729         }
4730
4731         /**
4732          * This function returns the text of a section, specified by a number ($section).
4733          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4734          * the first section before any such heading (section 0).
4735          *
4736          * If a section contains subsections, these are also returned.
4737          *
4738          * @param string $text text to look in
4739          * @param string $section section identifier
4740          * @param string $deftext default to return if section is not found
4741          * @return string text of the requested section
4742          */
4743         public function getSection( $text, $section, $deftext='' ) {
4744                 return $this->extractSections( $text, $section, "get", $deftext );
4745         }
4746
4747         public function replaceSection( $oldtext, $section, $text ) {
4748                 return $this->extractSections( $oldtext, $section, "replace", $text );
4749         }
4750
4751         /**
4752          * Get the timestamp associated with the current revision, adjusted for
4753          * the default server-local timestamp
4754          */
4755         function getRevisionTimestamp() {
4756                 if ( is_null( $this->mRevisionTimestamp ) ) {
4757                         wfProfileIn( __METHOD__ );
4758                         global $wgContLang;
4759                         $dbr = wfGetDB( DB_SLAVE );
4760                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4761                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4762
4763                         // Normalize timestamp to internal MW format for timezone processing.
4764                         // This has the added side-effect of replacing a null value with
4765                         // the current time, which gives us more sensible behavior for
4766                         // previews.
4767                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4768
4769                         // The cryptic '' timezone parameter tells to use the site-default
4770                         // timezone offset instead of the user settings.
4771                         //
4772                         // Since this value will be saved into the parser cache, served
4773                         // to other users, and potentially even used inside links and such,
4774                         // it needs to be consistent for all visitors.
4775                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4776
4777                         wfProfileOut( __METHOD__ );
4778                 }
4779                 return $this->mRevisionTimestamp;
4780         }
4781
4782         /**
4783          * Get the name of the user that edited the last revision
4784          */
4785         function getRevisionUser() {
4786                 // if this template is subst: the revision id will be blank,
4787                 // so just use the current user's name
4788                 if( $this->mRevisionId ) {
4789                         $revision = Revision::newFromId( $this->mRevisionId );
4790                         $revuser = $revision->getUserText();
4791                 } else {
4792                         global $wgUser;
4793                         $revuser = $wgUser->getName();
4794                 }
4795                 return $revuser;
4796         }
4797
4798         /**
4799          * Mutator for $mDefaultSort
4800          *
4801          * @param $sort New value
4802          */
4803         public function setDefaultSort( $sort ) {
4804                 $this->mDefaultSort = $sort;
4805         }
4806
4807         /**
4808          * Accessor for $mDefaultSort
4809          * Will use the title/prefixed title if none is set
4810          *
4811          * @return string
4812          */
4813         public function getDefaultSort() {
4814                 global $wgCategoryPrefixedDefaultSortkey;
4815                 if( $this->mDefaultSort !== false ) {
4816                         return $this->mDefaultSort;
4817                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4818                         !$wgCategoryPrefixedDefaultSortkey) {
4819                         return $this->mTitle->getText();
4820                 } else {
4821                         return $this->mTitle->getPrefixedText();
4822                 }
4823         }
4824
4825         /**
4826          * Accessor for $mDefaultSort
4827          * Unlike getDefaultSort(), will return false if none is set
4828          *
4829          * @return string or false
4830          */
4831         public function getCustomDefaultSort() {
4832                 return $this->mDefaultSort;
4833         }
4834
4835         /**
4836          * Try to guess the section anchor name based on a wikitext fragment
4837          * presumably extracted from a heading, for example "Header" from
4838          * "== Header ==".
4839          */
4840         public function guessSectionNameFromWikiText( $text ) {
4841                 # Strip out wikitext links(they break the anchor)
4842                 $text = $this->stripSectionName( $text );
4843                 $headline = Sanitizer::decodeCharReferences( $text );
4844                 # strip out HTML
4845                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4846                 $headline = trim( $headline );
4847                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4848                 $replacearray = array(
4849                         '%3A' => ':',
4850                         '%' => '.'
4851                 );
4852                 return str_replace(
4853                         array_keys( $replacearray ),
4854                         array_values( $replacearray ),
4855                         $sectionanchor );
4856         }
4857
4858         /**
4859          * Strips a text string of wikitext for use in a section anchor
4860          *
4861          * Accepts a text string and then removes all wikitext from the
4862          * string and leaves only the resultant text (i.e. the result of
4863          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4864          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4865          * to create valid section anchors by mimicing the output of the
4866          * parser when headings are parsed.
4867          *
4868          * @param $text string Text string to be stripped of wikitext
4869          * for use in a Section anchor
4870          * @return Filtered text string
4871          */
4872         public function stripSectionName( $text ) {
4873                 # Strip internal link markup
4874                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4875                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4876
4877                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4878                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4879                 # on how many empty links there are on the page - need to figure that out.
4880                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4881
4882                 # Parse wikitext quotes (italics & bold)
4883                 $text = $this->doQuotes($text);
4884
4885                 # Strip HTML tags
4886                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4887                 return $text;
4888         }
4889
4890         function srvus( $text ) {
4891                 return $this->testSrvus( $text, $this->mOutputType );
4892         }
4893
4894         /**
4895          * strip/replaceVariables/unstrip for preprocessor regression testing
4896          */
4897         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4898                 $this->clearState();
4899                 if ( ! ( $title instanceof Title ) ) {
4900                         $title = Title::newFromText( $title );
4901                 }
4902                 $this->mTitle = $title;
4903                 $this->mOptions = $options;
4904                 $this->setOutputType( $outputType );
4905                 $text = $this->replaceVariables( $text );
4906                 $text = $this->mStripState->unstripBoth( $text );
4907                 $text = Sanitizer::removeHTMLtags( $text );
4908                 return $text;
4909         }
4910
4911         function testPst( $text, $title, $options ) {
4912                 global $wgUser;
4913                 if ( ! ( $title instanceof Title ) ) {
4914                         $title = Title::newFromText( $title );
4915                 }
4916                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4917         }
4918
4919         function testPreprocess( $text, $title, $options ) {
4920                 if ( ! ( $title instanceof Title ) ) {
4921                         $title = Title::newFromText( $title );
4922                 }
4923                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4924         }
4925
4926         function markerSkipCallback( $s, $callback ) {
4927                 $i = 0;
4928                 $out = '';
4929                 while ( $i < strlen( $s ) ) {
4930                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4931                         if ( $markerStart === false ) {
4932                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4933                                 break;
4934                         } else {
4935                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4936                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4937                                 if ( $markerEnd === false ) {
4938                                         $out .= substr( $s, $markerStart );
4939                                         break;
4940                                 } else {
4941                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4942                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4943                                         $i = $markerEnd;
4944                                 }
4945                         }
4946                 }
4947                 return $out;
4948         }
4949
4950         function serialiseHalfParsedText( $text ) {
4951                 $data = array();
4952                 $data['text'] = $text;
4953
4954                 // First, find all strip markers, and store their
4955                 //  data in an array.
4956                 $stripState = new StripState;
4957                 $pos = 0;
4958                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
4959                         $end_pos += strlen( self::MARKER_SUFFIX );
4960                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
4961
4962                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
4963                                 $replaceArray = $stripState->general;
4964                                 $stripText = $this->mStripState->general->data[$marker];
4965                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
4966                                 $replaceArray = $stripState->nowiki;
4967                                 $stripText = $this->mStripState->nowiki->data[$marker];
4968                         } else {
4969                                 throw new MWException( "Hanging strip marker: '$marker'." );
4970                         }
4971
4972                         $replaceArray->setPair( $marker, $stripText );
4973                         $pos = $end_pos;
4974                 }
4975                 $data['stripstate'] = $stripState;
4976
4977                 // Now, find all of our links, and store THEIR
4978                 //  data in an array! :)
4979                 $links = array( 'internal' => array(), 'interwiki' => array() );
4980                 $pos = 0;
4981
4982                 // Internal links
4983                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
4984                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
4985
4986                         $ns = trim($ns);
4987                         if (empty( $links['internal'][$ns] )) {
4988                                 $links['internal'][$ns] = array();
4989                         }
4990
4991                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
4992                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
4993                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
4994                 }
4995
4996                 $pos = 0;
4997
4998                 // Interwiki links
4999                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5000                         $data = substr( $text, $start_pos );
5001                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5002                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5003                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5004                 }
5005
5006                 $data['linkholder'] = $links;
5007
5008                 return $data;
5009         }
5010
5011         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5012                 if (!$intPrefix)
5013                         $intPrefix = $this->getRandomString();
5014
5015                 // First, extract the strip state.
5016                 $stripState = $data['stripstate'];
5017                 $this->mStripState->general->merge( $stripState->general );
5018                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5019
5020                 // Now, extract the text, and renumber links
5021                 $text = $data['text'];
5022                 $links = $data['linkholder'];
5023
5024                 // Internal...
5025                 foreach( $links['internal'] as $ns => $nsLinks ) {
5026                         foreach( $nsLinks as $key => $entry ) {
5027                                 $newKey = $intPrefix . '-' . $key;
5028                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5029
5030                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5031                         }
5032                 }
5033
5034                 // Interwiki...
5035                 foreach( $links['interwiki'] as $key => $entry ) {
5036                         $newKey = "$intPrefix-$key";
5037                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5038
5039                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5040                 }
5041
5042                 // Should be good to go.
5043                 return $text;
5044         }
5045 }
5046
5047 /**
5048  * @todo document, briefly.
5049  * @ingroup Parser
5050  */
5051 class StripState {
5052         var $general, $nowiki;
5053
5054         function __construct() {
5055                 $this->general = new ReplacementArray;
5056                 $this->nowiki = new ReplacementArray;
5057         }
5058
5059         function unstripGeneral( $text ) {
5060                 wfProfileIn( __METHOD__ );
5061                 do {
5062                         $oldText = $text;
5063                         $text = $this->general->replace( $text );
5064                 } while ( $text !== $oldText );
5065                 wfProfileOut( __METHOD__ );
5066                 return $text;
5067         }
5068
5069         function unstripNoWiki( $text ) {
5070                 wfProfileIn( __METHOD__ );
5071                 do {
5072                         $oldText = $text;
5073                         $text = $this->nowiki->replace( $text );
5074                 } while ( $text !== $oldText );
5075                 wfProfileOut( __METHOD__ );
5076                 return $text;
5077         }
5078
5079         function unstripBoth( $text ) {
5080                 wfProfileIn( __METHOD__ );
5081                 do {
5082                         $oldText = $text;
5083                         $text = $this->general->replace( $text );
5084                         $text = $this->nowiki->replace( $text );
5085                 } while ( $text !== $oldText );
5086                 wfProfileOut( __METHOD__ );
5087                 return $text;
5088         }
5089 }
5090
5091 /**
5092  * @todo document, briefly.
5093  * @ingroup Parser
5094  */
5095 class OnlyIncludeReplacer {
5096         var $output = '';
5097
5098         function replace( $matches ) {
5099                 if ( substr( $matches[1], -1 ) === "\n" ) {
5100                         $this->output .= substr( $matches[1], 0, -1 );
5101                 } else {
5102                         $this->output .= $matches[1];
5103                 }
5104         }
5105 }