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  * getPreloadText()
  30  *   Removes <noinclude> sections, and <includeonly> tags.
  31  *
  32  * Globals used:
  33  *    objects:   $wgLang, $wgContLang
  34  *
  35  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  36  *
  37  * settings:
  38  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  39  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  40  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  41  *  $wgMaxArticleSize*
  42  *
  43  *  * only within ParserOptions
  44  * </pre>
  45  *
  46  * @ingroup Parser
  47  */
  48 class Parser
  49 {
  50         /**
  51          * Update this version number when the ParserOutput format
  52          * changes in an incompatible way, so the parser cache
  53          * can automatically discard old data.
  54          */
  55         const VERSION = '1.6.4';
  56
  57         # Flags for Parser::setFunctionHook
  58         # Also available as global constants from Defines.php
  59         const SFH_NO_HASH = 1;
  60         const SFH_OBJECT_ARGS = 2;
  61
  62         # Constants needed for external link processing
  63         # Everything except bracket, space, or control characters
  64         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  65         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  66                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  67
  68         // State constants for the definition list colon extraction
  69         const COLON_STATE_TEXT = 0;
  70         const COLON_STATE_TAG = 1;
  71         const COLON_STATE_TAGSTART = 2;
  72         const COLON_STATE_CLOSETAG = 3;
  73         const COLON_STATE_TAGSLASH = 4;
  74         const COLON_STATE_COMMENT = 5;
  75         const COLON_STATE_COMMENTDASH = 6;
  76         const COLON_STATE_COMMENTDASHDASH = 7;
  77
  78         // Flags for preprocessToDom
  79         const PTD_FOR_INCLUSION = 1;
  80
  81         // Allowed values for $this->mOutputType
  82         // Parameter to startExternalParse().
  83         const OT_HTML = 1; // like parse()
  84         const OT_WIKI = 2; // like preSaveTransform()
  85         const OT_PREPROCESS = 3; // like preprocess()
  86         const OT_MSG = 3;
  87         const OT_PLAIN = 4; // like extractSections() - portions of the original are returned unchanged.
  88
  89         // Marker Suffix needs to be accessible staticly.
  90         const MARKER_SUFFIX = "-QINU\x7f";
  91
  92         /**#@+
  93          * @private
  94          */
  95         # Persistent:
  96         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  97                 $mSubstWords, $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex,
  98                 $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList,
  99                 $mVarCache, $mConf, $mFunctionTagHooks;
 100
 101
 102         # Cleared with clearState():
 103         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 104         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 105         var $mLinkHolders, $mLinkID;
 106         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 107         var $mTplExpandCache; // empty-frame expansion cache
 108         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 109         var $mExpensiveFunctionCount; // number of expensive parser function calls
 110
 111         # Temporary
 112         # These are variables reset at least once per parse regardless of $clearState
 113         var $mOptions,      // ParserOptions object
 114                 $mTitle,        // Title context, used for self-link rendering and similar things
 115                 $mOutputType,   // Output type, one of the OT_xxx constants
 116                 $ot,            // Shortcut alias, see setOutputType()
 117                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 118                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 119                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 120
 121         /**#@-*/
 122
 123         /**
 124          * Constructor
 125          *
 126          * @public
 127          */
 128         function __construct( $conf = array() ) {
 129                 $this->mConf = $conf;
 130                 $this->mTagHooks = array();
 131                 $this->mTransparentTagHooks = array();
 132                 $this->mFunctionHooks = array();
 133                 $this->mFunctionTagHooks = array();
 134                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 135                 $this->mDefaultStripList = $this->mStripList = array();
 136                 $this->mUrlProtocols = wfUrlProtocols();
 137                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 138                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 139                 $this->mVarCache = array();
 140                 if ( isset( $conf['preprocessorClass'] ) ) {
 141                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 142                 } elseif ( extension_loaded( 'domxml' ) ) {
 143                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 144                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 145                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 146                 } elseif ( extension_loaded( 'dom' ) ) {
 147                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 148                 } else {
 149                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 150                 }
 151                 $this->mMarkerIndex = 0;
 152                 $this->mFirstCall = true;
 153         }
 154
 155         /**
 156          * Reduce memory usage to reduce the impact of circular references
 157          */
 158         function __destruct() {
 159                 if ( isset( $this->mLinkHolders ) ) {
 160                         $this->mLinkHolders->__destruct();
 161                 }
 162                 foreach ( $this as $name => $value ) {
 163                         unset( $this->$name );
 164                 }
 165         }
 166
 167         /**
 168          * Do various kinds of initialisation on the first call of the parser
 169          */
 170         function firstCallInit() {
 171                 if ( !$this->mFirstCall ) {
 172                         return;
 173                 }
 174                 $this->mFirstCall = false;
 175
 176                 wfProfileIn( __METHOD__ );
 177
 178                 CoreParserFunctions::register( $this );
 179                 CoreTagHooks::register( $this );
 180                 $this->initialiseVariables();
 181
 182                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 183                 wfProfileOut( __METHOD__ );
 184         }
 185
 186         /**
 187          * Clear Parser state
 188          *
 189          * @private
 190          */
 191         function clearState() {
 192                 wfProfileIn( __METHOD__ );
 193                 if ( $this->mFirstCall ) {
 194                         $this->firstCallInit();
 195                 }
 196                 $this->mOutput = new ParserOutput;
 197                 $this->mAutonumber = 0;
 198                 $this->mLastSection = '';
 199                 $this->mDTopen = false;
 200                 $this->mIncludeCount = array();
 201                 $this->mStripState = new StripState;
 202                 $this->mArgStack = false;
 203                 $this->mInPre = false;
 204                 $this->mLinkHolders = new LinkHolderArray( $this );
 205                 $this->mLinkID = 0;
 206                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 207                 $this->mVarCache = array();
 208
 209                 /**
 210                  * Prefix for temporary replacement strings for the multipass parser.
 211                  * \x07 should never appear in input as it's disallowed in XML.
 212                  * Using it at the front also gives us a little extra robustness
 213                  * since it shouldn't match when butted up against identifier-like
 214                  * string constructs.
 215                  *
 216                  * Must not consist of all title characters, or else it will change
 217                  * the behaviour of <nowiki> in a link.
 218                  */
 219                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 220                 # Changed to \x7f to allow XML double-parsing -- TS
 221                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 222
 223
 224                 # Clear these on every parse, bug 4549
 225                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 226
 227                 $this->mShowToc = true;
 228                 $this->mForceTocPosition = false;
 229                 $this->mIncludeSizes = array(
 230                         'post-expand' => 0,
 231                         'arg' => 0,
 232                 );
 233                 $this->mPPNodeCount = 0;
 234                 $this->mDefaultSort = false;
 235                 $this->mHeadings = array();
 236                 $this->mDoubleUnderscores = array();
 237                 $this->mExpensiveFunctionCount = 0;
 238
 239                 # Fix cloning
 240                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 241                         $this->mPreprocessor = null;
 242                 }
 243
 244                 wfRunHooks( 'ParserClearState', array( &$this ) );
 245                 wfProfileOut( __METHOD__ );
 246         }
 247
 248         function setOutputType( $ot ) {
 249                 $this->mOutputType = $ot;
 250                 // Shortcut alias
 251                 $this->ot = array(
 252                         'html' => $ot == self::OT_HTML,
 253                         'wiki' => $ot == self::OT_WIKI,
 254                         'pre' => $ot == self::OT_PREPROCESS,
 255                         'plain' => $ot == self::OT_PLAIN,
 256                 );
 257         }
 258
 259         /**
 260          * Set the context title
 261          */
 262         function setTitle( $t ) {
 263                 if ( !$t || $t instanceof FakeTitle ) {
 264                         $t = Title::newFromText( 'NO TITLE' );
 265                 }
 266
 267                 if ( strval( $t->getFragment() ) !== '' ) {
 268                         # Strip the fragment to avoid various odd effects
 269                         $this->mTitle = clone $t;
 270                         $this->mTitle->setFragment( '' );
 271                 } else {
 272                         $this->mTitle = $t;
 273                 }
 274         }
 275
 276         /**
 277          * Accessor for mUniqPrefix.
 278          *
 279          * @public
 280          */
 281         function uniqPrefix() {
 282                 if( !isset( $this->mUniqPrefix ) ) {
 283                         // @todo Fixme: this is probably *horribly wrong*
 284                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 285                         // if this is called for a parser cache hit, the parser may not
 286                         // have ever been initialized in the first place.
 287                         // Not really sure what the heck is supposed to be going on here.
 288                         return '';
 289                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 290                 }
 291                 return $this->mUniqPrefix;
 292         }
 293
 294         /**
 295          * Convert wikitext to HTML
 296          * Do not call this function recursively.
 297          *
 298          * @param $text String: text we want to parse
 299          * @param $title A title object
 300          * @param $options ParserOptions
 301          * @param $linestart boolean
 302          * @param $clearState boolean
 303          * @param $revid Int: number to pass in {{REVISIONID}}
 304          * @return ParserOutput a ParserOutput
 305          */
 306         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 307                 /**
 308                  * First pass--just handle <nowiki> sections, pass the rest off
 309                  * to internalParse() which does all the real work.
 310                  */
 311
 312                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion;
 313                 $fname = __METHOD__.'-' . wfGetCaller();
 314                 wfProfileIn( __METHOD__ );
 315                 wfProfileIn( $fname );
 316
 317                 if ( $clearState ) {
 318                         $this->clearState();
 319                 }
 320
 321                 $this->mOptions = $options;
 322                 $this->setTitle( $title ); // Page title has to be set for the pre-processor
 323
 324                 $oldRevisionId = $this->mRevisionId;
 325                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 326                 if( $revid !== null ) {
 327                         $this->mRevisionId = $revid;
 328                         $this->mRevisionTimestamp = null;
 329                 }
 330                 $this->setOutputType( self::OT_HTML );
 331                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 332                 # No more strip!
 333                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 334                 $text = $this->internalParse( $text );
 335
 336                 $text = $this->mStripState->unstripGeneral( $text );
 337
 338                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 339                 $fixtags = array(
 340                         # french spaces, last one Guillemet-left
 341                         # only if there is something before the space
 342                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 343                         # french spaces, Guillemet-right
 344                         '/(\\302\\253) /' => '\\1&nbsp;',
 345                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 346                 );
 347                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 348
 349                 $text = $this->doBlockLevels( $text, $linestart );
 350
 351                 $this->replaceLinkHolders( $text );
 352
 353                 // The position of the convert() call should not be changed. it
 354                 // assumes that the links are all replaced and the only thing left
 355                 // is the <nowiki> mark.
 356                 if ( !( $wgDisableLangConversion
 357                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 358                                 || $this->mTitle->isTalkPage()
 359                                 || $this->mTitle->isConversionTable() ) ) {
 360                         $text = $wgContLang->convert( $text );
 361                 }
 362
 363                 // A title may have been set in a conversion rule.
 364                 // Note that if a user tries to set a title in a conversion
 365                 // rule but content conversion was not done, then the parser
 366                 // won't pick it up.  This is probably expected behavior.
 367                 if ( $wgContLang->getConvRuleTitle() ) {
 368                         $this->mOutput->setTitleText( $wgContLang->getConvRuleTitle() );
 369                 }
 370
 371                 $text = $this->mStripState->unstripNoWiki( $text );
 372
 373                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 374
 375 //!JF Move to its own function
 376
 377                 $uniq_prefix = $this->mUniqPrefix;
 378                 $matches = array();
 379                 $elements = array_keys( $this->mTransparentTagHooks );
 380                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 381
 382                 foreach( $matches as $marker => $data ) {
 383                         list( $element, $content, $params, $tag ) = $data;
 384                         $tagName = strtolower( $element );
 385                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 386                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 387                                         array( $content, $params, $this ) );
 388                         } else {
 389                                 $output = $tag;
 390                         }
 391                         $this->mStripState->general->setPair( $marker, $output );
 392                 }
 393                 $text = $this->mStripState->unstripGeneral( $text );
 394
 395                 $text = Sanitizer::normalizeCharReferences( $text );
 396
 397                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 398                         $text = MWTidy::tidy( $text );
 399                 } else {
 400                         # attempt to sanitize at least some nesting problems
 401                         # (bug #2702 and quite a few others)
 402                         $tidyregs = array(
 403                                 # ''Something [http://www.cool.com cool''] -->
 404                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 405                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 406                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 407                                 # fix up an anchor inside another anchor, only
 408                                 # at least for a single single nested link (bug 3695)
 409                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 410                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 411                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 412                                 # contains a div, so fix it up here; replace
 413                                 # div with escaped text
 414                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 415                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 416                                 # remove empty italic or bold tag pairs, some
 417                                 # introduced by rules above
 418                                 '/<([bi])><\/\\1>/' => '',
 419                         );
 420
 421                         $text = preg_replace(
 422                                 array_keys( $tidyregs ),
 423                                 array_values( $tidyregs ),
 424                                 $text );
 425                 }
 426                 global $wgExpensiveParserFunctionLimit;
 427                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 428                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 429                 }
 430
 431                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 432
 433                 # Information on include size limits, for the benefit of users who try to skirt them
 434                 if ( $this->mOptions->getEnableLimitReport() ) {
 435                         $max = $this->mOptions->getMaxIncludeSize();
 436                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 437                         $limitReport =
 438                                 "NewPP limit report\n" .
 439                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 440                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 441                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 442                                 $PFreport;
 443                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 444                         $text .= "\n<!-- \n$limitReport-->\n";
 445                 }
 446                 $this->mOutput->setText( $text );
 447
 448                 $this->mRevisionId = $oldRevisionId;
 449                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 450                 wfProfileOut( $fname );
 451                 wfProfileOut( __METHOD__ );
 452
 453                 return $this->mOutput;
 454         }
 455
 456         /**
 457          * Recursive parser entry point that can be called from an extension tag
 458          * hook.
 459          *
 460          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 461          *
 462          * @param $text String: text extension wants to have parsed
 463          * @param PPFrame $frame: The frame to use for expanding any template variables
 464          */
 465         function recursiveTagParse( $text, $frame=false ) {
 466                 wfProfileIn( __METHOD__ );
 467                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 468                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 469                 $text = $this->internalParse( $text, false, $frame );
 470                 wfProfileOut( __METHOD__ );
 471                 return $text;
 472         }
 473
 474         /**
 475          * Expand templates and variables in the text, producing valid, static wikitext.
 476          * Also removes comments.
 477          */
 478         function preprocess( $text, $title, $options, $revid = null ) {
 479                 wfProfileIn( __METHOD__ );
 480                 $this->clearState();
 481                 $this->setOutputType( self::OT_PREPROCESS );
 482                 $this->mOptions = $options;
 483                 $this->setTitle( $title );
 484                 if( $revid !== null ) {
 485                         $this->mRevisionId = $revid;
 486                 }
 487                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 488                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 489                 $text = $this->replaceVariables( $text );
 490                 $text = $this->mStripState->unstripBoth( $text );
 491                 wfProfileOut( __METHOD__ );
 492                 return $text;
 493         }
 494
 495         /**
 496          * Process the wikitext for the ?preload= feature. (bug 5210)
 497          *
 498          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 499          * comments, templates, arguments, tags hooks and parser functions are untouched.
 500          */
 501         public function getPreloadText( $text, $title, $options ) {
 502                 // Parser (re)initialisation
 503                 $this->clearState();
 504                 $this->setOutputType( self::OT_PLAIN );
 505                 $this->mOptions = $options;
 506                 $this->setTitle( $title );
 507
 508                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 509                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 510                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 511         }
 512
 513         /**
 514          * Get a random string
 515          *
 516          * @private
 517          * @static
 518          */
 519         function getRandomString() {
 520                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 521         }
 522
 523         function &getTitle() { return $this->mTitle; }
 524         function getOptions() { return $this->mOptions; }
 525         function getRevisionId() { return $this->mRevisionId; }
 526         function getOutput() { return $this->mOutput; }
 527         function nextLinkID() { return $this->mLinkID++; }
 528
 529         function getFunctionLang() {
 530                 global $wgLang, $wgContLang;
 531
 532                 $target = $this->mOptions->getTargetLanguage();
 533                 if ( $target !== null ) {
 534                         return $target;
 535                 } else {
 536                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 537                 }
 538         }
 539
 540         /**
 541          * Get a preprocessor object
 542          */
 543         function getPreprocessor() {
 544                 if ( !isset( $this->mPreprocessor ) ) {
 545                         $class = $this->mPreprocessorClass;
 546                         $this->mPreprocessor = new $class( $this );
 547                 }
 548                 return $this->mPreprocessor;
 549         }
 550
 551         /**
 552          * Replaces all occurrences of HTML-style comments and the given tags
 553          * in the text with a random marker and returns the next text. The output
 554          * parameter $matches will be an associative array filled with data in
 555          * the form:
 556          *   'UNIQ-xxxxx' => array(
 557          *     'element',
 558          *     'tag content',
 559          *     array( 'param' => 'x' ),
 560          *     '<element param="x">tag content</element>' ) )
 561          *
 562          * @param $elements list of element names. Comments are always extracted.
 563          * @param $text Source text string.
 564          * @param $uniq_prefix
 565          *
 566          * @public
 567          * @static
 568          */
 569         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 570                 static $n = 1;
 571                 $stripped = '';
 572                 $matches = array();
 573
 574                 $taglist = implode( '|', $elements );
 575                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 576
 577                 while ( $text != '' ) {
 578                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 579                         $stripped .= $p[0];
 580                         if( count( $p ) < 5 ) {
 581                                 break;
 582                         }
 583                         if( count( $p ) > 5 ) {
 584                                 // comment
 585                                 $element    = $p[4];
 586                                 $attributes = '';
 587                                 $close      = '';
 588                                 $inside     = $p[5];
 589                         } else {
 590                                 // tag
 591                                 $element    = $p[1];
 592                                 $attributes = $p[2];
 593                                 $close      = $p[3];
 594                                 $inside     = $p[4];
 595                         }
 596
 597                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 598                         $stripped .= $marker;
 599
 600                         if ( $close === '/>' ) {
 601                                 // Empty element tag, <tag />
 602                                 $content = null;
 603                                 $text = $inside;
 604                                 $tail = null;
 605                         } else {
 606                                 if( $element === '!--' ) {
 607                                         $end = '/(-->)/';
 608                                 } else {
 609                                         $end = "/(<\\/$element\\s*>)/i";
 610                                 }
 611                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 612                                 $content = $q[0];
 613                                 if( count( $q ) < 3 ) {
 614                                         # No end tag -- let it run out to the end of the text.
 615                                         $tail = '';
 616                                         $text = '';
 617                                 } else {
 618                                         $tail = $q[1];
 619                                         $text = $q[2];
 620                                 }
 621                         }
 622
 623                         $matches[$marker] = array( $element,
 624                                 $content,
 625                                 Sanitizer::decodeTagAttributes( $attributes ),
 626                                 "<$element$attributes$close$content$tail" );
 627                 }
 628                 return $stripped;
 629         }
 630
 631         /**
 632          * Get a list of strippable XML-like elements
 633          */
 634         function getStripList() {
 635                 return $this->mStripList;
 636         }
 637
 638         /**
 639          * @deprecated use replaceVariables
 640          */
 641         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 642                 return $text;
 643         }
 644
 645         /**
 646          * Restores pre, math, and other extensions removed by strip()
 647          *
 648          * always call unstripNoWiki() after this one
 649          * @private
 650          * @deprecated use $this->mStripState->unstrip()
 651          */
 652         function unstrip( $text, $state ) {
 653                 return $state->unstripGeneral( $text );
 654         }
 655
 656         /**
 657          * Always call this after unstrip() to preserve the order
 658          *
 659          * @private
 660          * @deprecated use $this->mStripState->unstrip()
 661          */
 662         function unstripNoWiki( $text, $state ) {
 663                 return $state->unstripNoWiki( $text );
 664         }
 665
 666         /**
 667          * @deprecated use $this->mStripState->unstripBoth()
 668          */
 669         function unstripForHTML( $text ) {
 670                 return $this->mStripState->unstripBoth( $text );
 671         }
 672
 673         /**
 674          * Add an item to the strip state
 675          * Returns the unique tag which must be inserted into the stripped text
 676          * The tag will be replaced with the original text in unstrip()
 677          *
 678          * @private
 679          */
 680         function insertStripItem( $text ) {
 681                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 682                 $this->mMarkerIndex++;
 683                 $this->mStripState->general->setPair( $rnd, $text );
 684                 return $rnd;
 685         }
 686
 687         /**
 688          * Interface with html tidy
 689          * @deprecated Use MWTidy::tidy()
 690          */
 691         public static function tidy( $text ) {
 692                 wfDeprecated( __METHOD__ );
 693                 return MWTidy::tidy( $text );
 694         }
 695
 696         /**
 697          * parse the wiki syntax used to render tables
 698          *
 699          * @private
 700          */
 701         function doTableStuff ( $text ) {
 702                 wfProfileIn( __METHOD__ );
 703
 704                 $lines = StringUtils::explode( "\n", $text );
 705                 $out = '';
 706                 $td_history = array (); // Is currently a td tag open?
 707                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 708                 $tr_history = array (); // Is currently a tr tag open?
 709                 $tr_attributes = array (); // history of tr attributes
 710                 $has_opened_tr = array(); // Did this table open a <tr> element?
 711                 $indent_level = 0; // indent level of the table
 712
 713                 foreach ( $lines as $outLine ) {
 714                         $line = trim( $outLine );
 715
 716                         if( $line == '' ) { // empty line, go to next line
 717                                 $out .= $outLine."\n";
 718                                 continue;
 719                         }
 720                         $first_character = $line[0];
 721                         $matches = array();
 722
 723                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 724                                 // First check if we are starting a new table
 725                                 $indent_level = strlen( $matches[1] );
 726
 727                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 728                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 729
 730                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 731                                 array_push( $td_history , false );
 732                                 array_push( $last_tag_history , '' );
 733                                 array_push( $tr_history , false );
 734                                 array_push( $tr_attributes , '' );
 735                                 array_push( $has_opened_tr , false );
 736                         } else if ( count ( $td_history ) == 0 ) {
 737                                 // Don't do any of the following
 738                                 $out .= $outLine."\n";
 739                                 continue;
 740                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 741                                 // We are ending a table
 742                                 $line = '</table>' . substr ( $line , 2 );
 743                                 $last_tag = array_pop ( $last_tag_history );
 744
 745                                 if ( !array_pop ( $has_opened_tr ) ) {
 746                                         $line = "<tr><td></td></tr>{$line}";
 747                                 }
 748
 749                                 if ( array_pop ( $tr_history ) ) {
 750                                         $line = "</tr>{$line}";
 751                                 }
 752
 753                                 if ( array_pop ( $td_history ) ) {
 754                                         $line = "</{$last_tag}>{$line}";
 755                                 }
 756                                 array_pop ( $tr_attributes );
 757                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 758                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 759                                 // Now we have a table row
 760                                 $line = preg_replace( '#^\|-+#', '', $line );
 761
 762                                 // Whats after the tag is now only attributes
 763                                 $attributes = $this->mStripState->unstripBoth( $line );
 764                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 765                                 array_pop( $tr_attributes );
 766                                 array_push( $tr_attributes, $attributes );
 767
 768                                 $line = '';
 769                                 $last_tag = array_pop ( $last_tag_history );
 770                                 array_pop ( $has_opened_tr );
 771                                 array_push ( $has_opened_tr , true );
 772
 773                                 if ( array_pop ( $tr_history ) ) {
 774                                         $line = '</tr>';
 775                                 }
 776
 777                                 if ( array_pop ( $td_history ) ) {
 778                                         $line = "</{$last_tag}>{$line}";
 779                                 }
 780
 781                                 $outLine = $line;
 782                                 array_push ( $tr_history , false );
 783                                 array_push ( $td_history , false );
 784                                 array_push ( $last_tag_history , '' );
 785                         }
 786                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 787                                 // This might be cell elements, td, th or captions
 788                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 789                                         $first_character = '+';
 790                                         $line = substr ( $line , 1 );
 791                                 }
 792
 793                                 $line = substr ( $line , 1 );
 794
 795                                 if ( $first_character === '!' ) {
 796                                         $line = str_replace ( '!!' , '||' , $line );
 797                                 }
 798
 799                                 // Split up multiple cells on the same line.
 800                                 // FIXME : This can result in improper nesting of tags processed
 801                                 // by earlier parser steps, but should avoid splitting up eg
 802                                 // attribute values containing literal "||".
 803                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 804
 805                                 $outLine = '';
 806
 807                                 // Loop through each table cell
 808                                 foreach ( $cells as $cell )
 809                                 {
 810                                         $previous = '';
 811                                         if ( $first_character !== '+' )
 812                                         {
 813                                                 $tr_after = array_pop ( $tr_attributes );
 814                                                 if ( !array_pop ( $tr_history ) ) {
 815                                                         $previous = "<tr{$tr_after}>\n";
 816                                                 }
 817                                                 array_push ( $tr_history , true );
 818                                                 array_push ( $tr_attributes , '' );
 819                                                 array_pop ( $has_opened_tr );
 820                                                 array_push ( $has_opened_tr , true );
 821                                         }
 822
 823                                         $last_tag = array_pop ( $last_tag_history );
 824
 825                                         if ( array_pop ( $td_history ) ) {
 826                                                 $previous = "</{$last_tag}>{$previous}";
 827                                         }
 828
 829                                         if ( $first_character === '|' ) {
 830                                                 $last_tag = 'td';
 831                                         } else if ( $first_character === '!' ) {
 832                                                 $last_tag = 'th';
 833                                         } else if ( $first_character === '+' ) {
 834                                                 $last_tag = 'caption';
 835                                         } else {
 836                                                 $last_tag = '';
 837                                         }
 838
 839                                         array_push ( $last_tag_history , $last_tag );
 840
 841                                         // A cell could contain both parameters and data
 842                                         $cell_data = explode ( '|' , $cell , 2 );
 843
 844                                         // Bug 553: Note that a '|' inside an invalid link should not
 845                                         // be mistaken as delimiting cell parameters
 846                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 847                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 848                                         } else if ( count ( $cell_data ) == 1 )
 849                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 850                                         else {
 851                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 852                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 853                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 854                                         }
 855
 856                                         $outLine .= $cell;
 857                                         array_push ( $td_history , true );
 858                                 }
 859                         }
 860                         $out .= $outLine . "\n";
 861                 }
 862
 863                 // Closing open td, tr && table
 864                 while ( count ( $td_history ) > 0 )
 865                 {
 866                         if ( array_pop ( $td_history ) ) {
 867                                 $out .= "</td>\n";
 868                         }
 869                         if ( array_pop ( $tr_history ) ) {
 870                                 $out .= "</tr>\n";
 871                         }
 872                         if ( !array_pop ( $has_opened_tr ) ) {
 873                                 $out .= "<tr><td></td></tr>\n" ;
 874                         }
 875
 876                         $out .= "</table>\n";
 877                 }
 878
 879                 // Remove trailing line-ending (b/c)
 880                 if ( substr( $out, -1 ) === "\n" ) {
 881                         $out = substr( $out, 0, -1 );
 882                 }
 883
 884                 // special case: don't return empty table
 885                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 886                         $out = '';
 887                 }
 888
 889                 wfProfileOut( __METHOD__ );
 890
 891                 return $out;
 892         }
 893
 894         /**
 895          * Helper function for parse() that transforms wiki markup into
 896          * HTML. Only called for $mOutputType == self::OT_HTML.
 897          *
 898          * @private
 899          */
 900         function internalParse( $text, $isMain = true, $frame=false ) {
 901                 wfProfileIn( __METHOD__ );
 902
 903                 $origText = $text;
 904
 905                 # Hook to suspend the parser in this state
 906                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 907                         wfProfileOut( __METHOD__ );
 908                         return $text ;
 909                 }
 910
 911                 // if $frame is provided, then use $frame for replacing any variables
 912                 if ($frame) {
 913                         // use frame depth to infer how include/noinclude tags should be handled
 914                         // depth=0 means this is the top-level document; otherwise it's an included document
 915                         if( !$frame->depth )
 916                                 $flag = 0;
 917                         else
 918                                 $flag = Parser::PTD_FOR_INCLUSION;
 919                         $dom = $this->preprocessToDom( $text, $flag );
 920                         $text = $frame->expand( $dom );
 921                 }
 922                 // if $frame is not provided, then use old-style replaceVariables
 923                 else {
 924                         $text = $this->replaceVariables( $text );
 925                 }
 926
 927                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 928                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 929
 930                 // Tables need to come after variable replacement for things to work
 931                 // properly; putting them before other transformations should keep
 932                 // exciting things like link expansions from showing up in surprising
 933                 // places.
 934                 $text = $this->doTableStuff( $text );
 935
 936                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 937
 938                 $text = $this->doDoubleUnderscore( $text );
 939
 940                 $text = $this->doHeadings( $text );
 941                 if( $this->mOptions->getUseDynamicDates() ) {
 942                         $df = DateFormatter::getInstance();
 943                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 944                 }
 945                 $text = $this->doAllQuotes( $text );
 946                 $text = $this->replaceInternalLinks( $text );
 947                 $text = $this->replaceExternalLinks( $text );
 948
 949                 # replaceInternalLinks may sometimes leave behind
 950                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 951                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 952
 953                 $text = $this->doMagicLinks( $text );
 954                 $text = $this->formatHeadings( $text, $origText, $isMain );
 955
 956                 wfProfileOut( __METHOD__ );
 957                 return $text;
 958         }
 959
 960         /**
 961          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 962          * magic external links.
 963          *
 964          * DML
 965          * @private
 966          */
 967         function doMagicLinks( $text ) {
 968                 wfProfileIn( __METHOD__ );
 969                 $prots = $this->mUrlProtocols;
 970                 $urlChar = self::EXT_LINK_URL_CLASS;
 971                 $text = preg_replace_callback(
 972                         '!(?:                           # Start cases
 973                                 (<a.*?</a>) |               # m[1]: Skip link text
 974                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 975                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 976                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 977                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 978                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 979                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 980                                     [0-9Xx]                 # check digit
 981                                     \b)
 982                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 983                 wfProfileOut( __METHOD__ );
 984                 return $text;
 985         }
 986
 987         function magicLinkCallback( $m ) {
 988                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 989                         # Skip anchor
 990                         return $m[0];
 991                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 992                         # Skip HTML element
 993                         return $m[0];
 994                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 995                         # Free external link
 996                         return $this->makeFreeExternalLink( $m[0] );
 997                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 998                         # RFC or PMID
 999                         $CssClass = '';
1000                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1001                                 $keyword = 'RFC';
1002                                 $urlmsg = 'rfcurl';
1003                                 $CssClass = 'mw-magiclink-rfc';
1004                                 $id = $m[4];
1005                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1006                                 $keyword = 'PMID';
1007                                 $urlmsg = 'pubmedurl';
1008                                 $CssClass = 'mw-magiclink-pmid';
1009                                 $id = $m[4];
1010                         } else {
1011                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1012                                         substr($m[0], 0, 20 ) . '"' );
1013                         }
1014                         $url = wfMsg( $urlmsg, $id);
1015                         $sk = $this->mOptions->getSkin();
1016                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1017                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1018                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1019                         # ISBN
1020                         $isbn = $m[5];
1021                         $num = strtr( $isbn, array(
1022                                 '-' => '',
1023                                 ' ' => '',
1024                                 'x' => 'X',
1025                         ));
1026                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1027                         return'<a href="' .
1028                                 $titleObj->escapeLocalUrl() .
1029                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1030                 } else {
1031                         return $m[0];
1032                 }
1033         }
1034
1035         /**
1036          * Make a free external link, given a user-supplied URL
1037          * @return HTML
1038          * @private
1039          */
1040         function makeFreeExternalLink( $url ) {
1041                 global $wgContLang;
1042                 wfProfileIn( __METHOD__ );
1043
1044                 $sk = $this->mOptions->getSkin();
1045                 $trail = '';
1046
1047                 # The characters '<' and '>' (which were escaped by
1048                 # removeHTMLtags()) should not be included in
1049                 # URLs, per RFC 2396.
1050                 $m2 = array();
1051                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1052                         $trail = substr($url, $m2[0][1]) . $trail;
1053                         $url = substr($url, 0, $m2[0][1]);
1054                 }
1055
1056                 # Move trailing punctuation to $trail
1057                 $sep = ',;\.:!?';
1058                 # If there is no left bracket, then consider right brackets fair game too
1059                 if ( strpos( $url, '(' ) === false ) {
1060                         $sep .= ')';
1061                 }
1062
1063                 $numSepChars = strspn( strrev( $url ), $sep );
1064                 if ( $numSepChars ) {
1065                         $trail = substr( $url, -$numSepChars ) . $trail;
1066                         $url = substr( $url, 0, -$numSepChars );
1067                 }
1068
1069                 $url = Sanitizer::cleanUrl( $url );
1070
1071                 # Is this an external image?
1072                 $text = $this->maybeMakeExternalImage( $url );
1073                 if ( $text === false ) {
1074                         # Not an image, make a link
1075                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1076                                 $this->getExternalLinkAttribs( $url ) );
1077                         # Register it in the output object...
1078                         # Replace unnecessary URL escape codes with their equivalent characters
1079                         $pasteurized = self::replaceUnusualEscapes( $url );
1080                         $this->mOutput->addExternalLink( $pasteurized );
1081                 }
1082                 wfProfileOut( __METHOD__ );
1083                 return $text . $trail;
1084         }
1085
1086
1087         /**
1088          * Parse headers and return html
1089          *
1090          * @private
1091          */
1092         function doHeadings( $text ) {
1093                 wfProfileIn( __METHOD__ );
1094                 for ( $i = 6; $i >= 1; --$i ) {
1095                         $h = str_repeat( '=', $i );
1096                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1097                           "<h$i>\\1</h$i>", $text );
1098                 }
1099                 wfProfileOut( __METHOD__ );
1100                 return $text;
1101         }
1102
1103         /**
1104          * Replace single quotes with HTML markup
1105          * @private
1106          * @return string the altered text
1107          */
1108         function doAllQuotes( $text ) {
1109                 wfProfileIn( __METHOD__ );
1110                 $outtext = '';
1111                 $lines = StringUtils::explode( "\n", $text );
1112                 foreach ( $lines as $line ) {
1113                         $outtext .= $this->doQuotes( $line ) . "\n";
1114                 }
1115                 $outtext = substr($outtext, 0,-1);
1116                 wfProfileOut( __METHOD__ );
1117                 return $outtext;
1118         }
1119
1120         /**
1121          * Helper function for doAllQuotes()
1122          */
1123         public function doQuotes( $text ) {
1124                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1125                 if ( count( $arr ) == 1 )
1126                         return $text;
1127                 else
1128                 {
1129                         # First, do some preliminary work. This may shift some apostrophes from
1130                         # being mark-up to being text. It also counts the number of occurrences
1131                         # of bold and italics mark-ups.
1132                         $i = 0;
1133                         $numbold = 0;
1134                         $numitalics = 0;
1135                         foreach ( $arr as $r )
1136                         {
1137                                 if ( ( $i % 2 ) == 1 )
1138                                 {
1139                                         # If there are ever four apostrophes, assume the first is supposed to
1140                                         # be text, and the remaining three constitute mark-up for bold text.
1141                                         if ( strlen( $arr[$i] ) == 4 )
1142                                         {
1143                                                 $arr[$i-1] .= "'";
1144                                                 $arr[$i] = "'''";
1145                                         }
1146                                         # If there are more than 5 apostrophes in a row, assume they're all
1147                                         # text except for the last 5.
1148                                         else if ( strlen( $arr[$i] ) > 5 )
1149                                         {
1150                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1151                                                 $arr[$i] = "'''''";
1152                                         }
1153                                         # Count the number of occurrences of bold and italics mark-ups.
1154                                         # We are not counting sequences of five apostrophes.
1155                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1156                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1157                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1158                                 }
1159                                 $i++;
1160                         }
1161
1162                         # If there is an odd number of both bold and italics, it is likely
1163                         # that one of the bold ones was meant to be an apostrophe followed
1164                         # by italics. Which one we cannot know for certain, but it is more
1165                         # likely to be one that has a single-letter word before it.
1166                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1167                         {
1168                                 $i = 0;
1169                                 $firstsingleletterword = -1;
1170                                 $firstmultiletterword = -1;
1171                                 $firstspace = -1;
1172                                 foreach ( $arr as $r )
1173                                 {
1174                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1175                                         {
1176                                                 $x1 = substr ($arr[$i-1], -1);
1177                                                 $x2 = substr ($arr[$i-1], -2, 1);
1178                                                 if ($x1 === ' ') {
1179                                                         if ($firstspace == -1) $firstspace = $i;
1180                                                 } else if ($x2 === ' ') {
1181                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1182                                                 } else {
1183                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1184                                                 }
1185                                         }
1186                                         $i++;
1187                                 }
1188
1189                                 # If there is a single-letter word, use it!
1190                                 if ($firstsingleletterword > -1)
1191                                 {
1192                                         $arr [ $firstsingleletterword ] = "''";
1193                                         $arr [ $firstsingleletterword-1 ] .= "'";
1194                                 }
1195                                 # If not, but there's a multi-letter word, use that one.
1196                                 else if ($firstmultiletterword > -1)
1197                                 {
1198                                         $arr [ $firstmultiletterword ] = "''";
1199                                         $arr [ $firstmultiletterword-1 ] .= "'";
1200                                 }
1201                                 # ... otherwise use the first one that has neither.
1202                                 # (notice that it is possible for all three to be -1 if, for example,
1203                                 # there is only one pentuple-apostrophe in the line)
1204                                 else if ($firstspace > -1)
1205                                 {
1206                                         $arr [ $firstspace ] = "''";
1207                                         $arr [ $firstspace-1 ] .= "'";
1208                                 }
1209                         }
1210
1211                         # Now let's actually convert our apostrophic mush to HTML!
1212                         $output = '';
1213                         $buffer = '';
1214                         $state = '';
1215                         $i = 0;
1216                         foreach ($arr as $r)
1217                         {
1218                                 if (($i % 2) == 0)
1219                                 {
1220                                         if ($state === 'both')
1221                                                 $buffer .= $r;
1222                                         else
1223                                                 $output .= $r;
1224                                 }
1225                                 else
1226                                 {
1227                                         if (strlen ($r) == 2)
1228                                         {
1229                                                 if ($state === 'i')
1230                                                 { $output .= '</i>'; $state = ''; }
1231                                                 else if ($state === 'bi')
1232                                                 { $output .= '</i>'; $state = 'b'; }
1233                                                 else if ($state === 'ib')
1234                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1235                                                 else if ($state === 'both')
1236                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1237                                                 else # $state can be 'b' or ''
1238                                                 { $output .= '<i>'; $state .= 'i'; }
1239                                         }
1240                                         else if (strlen ($r) == 3)
1241                                         {
1242                                                 if ($state === 'b')
1243                                                 { $output .= '</b>'; $state = ''; }
1244                                                 else if ($state === 'bi')
1245                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1246                                                 else if ($state === 'ib')
1247                                                 { $output .= '</b>'; $state = 'i'; }
1248                                                 else if ($state === 'both')
1249                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1250                                                 else # $state can be 'i' or ''
1251                                                 { $output .= '<b>'; $state .= 'b'; }
1252                                         }
1253                                         else if (strlen ($r) == 5)
1254                                         {
1255                                                 if ($state === 'b')
1256                                                 { $output .= '</b><i>'; $state = 'i'; }
1257                                                 else if ($state === 'i')
1258                                                 { $output .= '</i><b>'; $state = 'b'; }
1259                                                 else if ($state === 'bi')
1260                                                 { $output .= '</i></b>'; $state = ''; }
1261                                                 else if ($state === 'ib')
1262                                                 { $output .= '</b></i>'; $state = ''; }
1263                                                 else if ($state === 'both')
1264                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1265                                                 else # ($state == '')
1266                                                 { $buffer = ''; $state = 'both'; }
1267                                         }
1268                                 }
1269                                 $i++;
1270                         }
1271                         # Now close all remaining tags.  Notice that the order is important.
1272                         if ($state === 'b' || $state === 'ib')
1273                                 $output .= '</b>';
1274                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1275                                 $output .= '</i>';
1276                         if ($state === 'bi')
1277                                 $output .= '</b>';
1278                         # There might be lonely ''''', so make sure we have a buffer
1279                         if ($state === 'both' && $buffer)
1280                                 $output .= '<b><i>'.$buffer.'</i></b>';
1281                         return $output;
1282                 }
1283         }
1284
1285         /**
1286          * Replace external links (REL)
1287          *
1288          * Note: this is all very hackish and the order of execution matters a lot.
1289          * Make sure to run maintenance/parserTests.php if you change this code.
1290          *
1291          * @private
1292          */
1293         function replaceExternalLinks( $text ) {
1294                 global $wgContLang;
1295                 wfProfileIn( __METHOD__ );
1296
1297                 $sk = $this->mOptions->getSkin();
1298
1299                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1300                 $s = array_shift( $bits );
1301
1302                 $i = 0;
1303                 while ( $i<count( $bits ) ) {
1304                         $url = $bits[$i++];
1305                         $protocol = $bits[$i++];
1306                         $text = $bits[$i++];
1307                         $trail = $bits[$i++];
1308
1309                         # The characters '<' and '>' (which were escaped by
1310                         # removeHTMLtags()) should not be included in
1311                         # URLs, per RFC 2396.
1312                         $m2 = array();
1313                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1314                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1315                                 $url = substr($url, 0, $m2[0][1]);
1316                         }
1317
1318                         # If the link text is an image URL, replace it with an <img> tag
1319                         # This happened by accident in the original parser, but some people used it extensively
1320                         $img = $this->maybeMakeExternalImage( $text );
1321                         if ( $img !== false ) {
1322                                 $text = $img;
1323                         }
1324
1325                         $dtrail = '';
1326
1327                         # Set linktype for CSS - if URL==text, link is essentially free
1328                         $linktype = ($text === $url) ? 'free' : 'text';
1329
1330                         # No link text, e.g. [http://domain.tld/some.link]
1331                         if ( $text == '' ) {
1332                                 # Autonumber if allowed. See bug #5918
1333                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1334                                         $langObj = $this->getFunctionLang();
1335                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1336                                         $linktype = 'autonumber';
1337                                 } else {
1338                                         # Otherwise just use the URL
1339                                         $text = htmlspecialchars( $url );
1340                                         $linktype = 'free';
1341                                 }
1342                         } else {
1343                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1344                                 # Check for trail
1345                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1346                         }
1347
1348                         $text = $wgContLang->markNoConversion($text);
1349
1350                         $url = Sanitizer::cleanUrl( $url );
1351
1352                         # Use the encoded URL
1353                         # This means that users can paste URLs directly into the text
1354                         # Funny characters like &ouml; aren't valid in URLs anyway
1355                         # This was changed in August 2004
1356                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1357                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1358
1359                         # Register link in the output object.
1360                         # Replace unnecessary URL escape codes with the referenced character
1361                         # This prevents spammers from hiding links from the filters
1362                         $pasteurized = self::replaceUnusualEscapes( $url );
1363                         $this->mOutput->addExternalLink( $pasteurized );
1364                 }
1365
1366                 wfProfileOut( __METHOD__ );
1367                 return $s;
1368         }
1369
1370         /**
1371          * Get an associative array of additional HTML attributes appropriate for a
1372          * particular external link.  This currently may include rel => nofollow
1373          * (depending on configuration, namespace, and the URL's domain) and/or a
1374          * target attribute (depending on configuration).
1375          *
1376          * @param string $url Optional URL, to extract the domain from for rel =>
1377          *   nofollow if appropriate
1378          * @return array Associative array of HTML attributes
1379          */
1380         function getExternalLinkAttribs( $url = false ) {
1381                 $attribs = array();
1382                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1383                 $ns = $this->mTitle->getNamespace();
1384                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1385                         $attribs['rel'] = 'nofollow';
1386
1387                         global $wgNoFollowDomainExceptions;
1388                         if ( $wgNoFollowDomainExceptions ) {
1389                                 $bits = wfParseUrl( $url );
1390                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1391                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1392                                                 if( substr( $bits['host'], -strlen( $domain ) )
1393                                                 == $domain ) {
1394                                                         unset( $attribs['rel'] );
1395                                                         break;
1396                                                 }
1397                                         }
1398                                 }
1399                         }
1400                 }
1401                 if ( $this->mOptions->getExternalLinkTarget() ) {
1402                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1403                 }
1404                 return $attribs;
1405         }
1406
1407
1408         /**
1409          * Replace unusual URL escape codes with their equivalent characters
1410          * @param string
1411          * @return string
1412          * @static
1413          * @todo  This can merge genuinely required bits in the path or query string,
1414          *        breaking legit URLs. A proper fix would treat the various parts of
1415          *        the URL differently; as a workaround, just use the output for
1416          *        statistical records, not for actual linking/output.
1417          */
1418         static function replaceUnusualEscapes( $url ) {
1419                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1420                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1421         }
1422
1423         /**
1424          * Callback function used in replaceUnusualEscapes().
1425          * Replaces unusual URL escape codes with their equivalent character
1426          * @static
1427          * @private
1428          */
1429         private static function replaceUnusualEscapesCallback( $matches ) {
1430                 $char = urldecode( $matches[0] );
1431                 $ord = ord( $char );
1432                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1433                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1434                         // No, shouldn't be escaped
1435                         return $char;
1436                 } else {
1437                         // Yes, leave it escaped
1438                         return $matches[0];
1439                 }
1440         }
1441
1442         /**
1443          * make an image if it's allowed, either through the global
1444          * option, through the exception, or through the on-wiki whitelist
1445          * @private
1446          */
1447         function maybeMakeExternalImage( $url ) {
1448                 $sk = $this->mOptions->getSkin();
1449                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1450                 $imagesexception = !empty($imagesfrom);
1451                 $text = false;
1452                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1453                 if( $imagesexception && is_array( $imagesfrom ) ) {
1454                         $imagematch = false;
1455                         foreach( $imagesfrom as $match ) {
1456                                 if( strpos( $url, $match ) === 0 ) {
1457                                         $imagematch = true;
1458                                         break;
1459                                 }
1460                         }
1461                 } elseif( $imagesexception ) {
1462                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1463                 } else {
1464                         $imagematch = false;
1465                 }
1466                 if ( $this->mOptions->getAllowExternalImages()
1467                      || ( $imagesexception && $imagematch ) ) {
1468                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1469                                 # Image found
1470                                 $text = $sk->makeExternalImage( $url );
1471                         }
1472                 }
1473                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1474                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1475                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1476                         foreach( $whitelist as $entry ) {
1477                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1478                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1479                                         continue;
1480                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1481                                         # Image matches a whitelist entry
1482                                         $text = $sk->makeExternalImage( $url );
1483                                         break;
1484                                 }
1485                         }
1486                 }
1487                 return $text;
1488         }
1489
1490         /**
1491          * Process [[ ]] wikilinks
1492          * @return processed text
1493          *
1494          * @private
1495          */
1496         function replaceInternalLinks( $s ) {
1497                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1498                 return $s;
1499         }
1500
1501         /**
1502          * Process [[ ]] wikilinks (RIL)
1503          * @return LinkHolderArray
1504          *
1505          * @private
1506          */
1507         function replaceInternalLinks2( &$s ) {
1508                 global $wgContLang;
1509
1510                 wfProfileIn( __METHOD__ );
1511
1512                 wfProfileIn( __METHOD__.'-setup' );
1513                 static $tc = FALSE, $e1, $e1_img;
1514                 # the % is needed to support urlencoded titles as well
1515                 if ( !$tc ) {
1516                         $tc = Title::legalChars() . '#%';
1517                         # Match a link having the form [[namespace:link|alternate]]trail
1518                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1519                         # Match cases where there is no "]]", which might still be images
1520                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1521                 }
1522
1523                 $sk = $this->mOptions->getSkin();
1524                 $holders = new LinkHolderArray( $this );
1525
1526                 #split the entire text string on occurences of [[
1527                 $a = StringUtils::explode( '[[', ' ' . $s );
1528                 #get the first element (all text up to first [[), and remove the space we added
1529                 $s = $a->current();
1530                 $a->next();
1531                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1532                 $s = substr( $s, 1 );
1533
1534                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1535                 $e2 = null;
1536                 if ( $useLinkPrefixExtension ) {
1537                         # Match the end of a line for a word that's not followed by whitespace,
1538                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1539                         $e2 = wfMsgForContent( 'linkprefix' );
1540                 }
1541
1542                 if( is_null( $this->mTitle ) ) {
1543                         wfProfileOut( __METHOD__.'-setup' );
1544                         wfProfileOut( __METHOD__ );
1545                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1546                 }
1547                 $nottalk = !$this->mTitle->isTalkPage();
1548
1549                 if ( $useLinkPrefixExtension ) {
1550                         $m = array();
1551                         if ( preg_match( $e2, $s, $m ) ) {
1552                                 $first_prefix = $m[2];
1553                         } else {
1554                                 $first_prefix = false;
1555                         }
1556                 } else {
1557                         $prefix = '';
1558                 }
1559
1560                 if($wgContLang->hasVariants()) {
1561                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1562                 } else {
1563                         $selflink = array($this->mTitle->getPrefixedText());
1564                 }
1565                 $useSubpages = $this->areSubpagesAllowed();
1566                 wfProfileOut( __METHOD__.'-setup' );
1567
1568                 # Loop for each link
1569                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1570                         # Check for excessive memory usage
1571                         if ( $holders->isBig() ) {
1572                                 # Too big
1573                                 # Do the existence check, replace the link holders and clear the array
1574                                 $holders->replace( $s );
1575                                 $holders->clear();
1576                         }
1577
1578                         if ( $useLinkPrefixExtension ) {
1579                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1580                                 if ( preg_match( $e2, $s, $m ) ) {
1581                                         $prefix = $m[2];
1582                                         $s = $m[1];
1583                                 } else {
1584                                         $prefix='';
1585                                 }
1586                                 # first link
1587                                 if($first_prefix) {
1588                                         $prefix = $first_prefix;
1589                                         $first_prefix = false;
1590                                 }
1591                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1592                         }
1593
1594                         $might_be_img = false;
1595
1596                         wfProfileIn( __METHOD__."-e1" );
1597                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1598                                 $text = $m[2];
1599                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1600                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1601                                 # the real problem is with the $e1 regex
1602                                 # See bug 1300.
1603                                 #
1604                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1605                                 # and no image is in sight. See bug 2095.
1606                                 #
1607                                 if( $text !== '' &&
1608                                         substr( $m[3], 0, 1 ) === ']' &&
1609                                         strpos($text, '[') !== false
1610                                 )
1611                                 {
1612                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1613                                         $m[3] = substr( $m[3], 1 );
1614                                 }
1615                                 # fix up urlencoded title texts
1616                                 if( strpos( $m[1], '%' ) !== false ) {
1617                                         # Should anchors '#' also be rejected?
1618                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1619                                 }
1620                                 $trail = $m[3];
1621                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1622                                 $might_be_img = true;
1623                                 $text = $m[2];
1624                                 if ( strpos( $m[1], '%' ) !== false ) {
1625                                         $m[1] = urldecode($m[1]);
1626                                 }
1627                                 $trail = "";
1628                         } else { # Invalid form; output directly
1629                                 $s .= $prefix . '[[' . $line ;
1630                                 wfProfileOut( __METHOD__."-e1" );
1631                                 continue;
1632                         }
1633                         wfProfileOut( __METHOD__."-e1" );
1634                         wfProfileIn( __METHOD__."-misc" );
1635
1636                         # Don't allow internal links to pages containing
1637                         # PROTO: where PROTO is a valid URL protocol; these
1638                         # should be external links.
1639                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1640                                 $s .= $prefix . '[[' . $line ;
1641                                 wfProfileOut( __METHOD__."-misc" );
1642                                 continue;
1643                         }
1644
1645                         # Make subpage if necessary
1646                         if ( $useSubpages ) {
1647                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1648                         } else {
1649                                 $link = $m[1];
1650                         }
1651
1652                         $noforce = (substr( $m[1], 0, 1 ) !== ':');
1653                         if (!$noforce) {
1654                                 # Strip off leading ':'
1655                                 $link = substr( $link, 1 );
1656                         }
1657
1658                         wfProfileOut( __METHOD__."-misc" );
1659                         wfProfileIn( __METHOD__."-title" );
1660                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1661                         if ( $nt === null ) {
1662                                 $s .= $prefix . '[[' . $line;
1663                                 wfProfileOut( __METHOD__."-title" );
1664                                 continue;
1665                         }
1666
1667                         $ns = $nt->getNamespace();
1668                         $iw = $nt->getInterWiki();
1669                         wfProfileOut( __METHOD__."-title" );
1670
1671                         if ( $might_be_img ) { # if this is actually an invalid link
1672                                 wfProfileIn( __METHOD__."-might_be_img" );
1673                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1674                                         $found = false;
1675                                         while ( true ) {
1676                                                 #look at the next 'line' to see if we can close it there
1677                                                 $a->next();
1678                                                 $next_line = $a->current();
1679                                                 if ( $next_line === false || $next_line === null ) {
1680                                                         break;
1681                                                 }
1682                                                 $m = explode( ']]', $next_line, 3 );
1683                                                 if ( count( $m ) == 3 ) {
1684                                                         # the first ]] closes the inner link, the second the image
1685                                                         $found = true;
1686                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1687                                                         $trail = $m[2];
1688                                                         break;
1689                                                 } elseif ( count( $m ) == 2 ) {
1690                                                         #if there's exactly one ]] that's fine, we'll keep looking
1691                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1692                                                 } else {
1693                                                         #if $next_line is invalid too, we need look no further
1694                                                         $text .= '[[' . $next_line;
1695                                                         break;
1696                                                 }
1697                                         }
1698                                         if ( !$found ) {
1699                                                 # we couldn't find the end of this imageLink, so output it raw
1700                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1701                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1702                                                 $s .= "{$prefix}[[$link|$text";
1703                                                 # note: no $trail, because without an end, there *is* no trail
1704                                                 wfProfileOut( __METHOD__."-might_be_img" );
1705                                                 continue;
1706                                         }
1707                                 } else { #it's not an image, so output it raw
1708                                         $s .= "{$prefix}[[$link|$text";
1709                                         # note: no $trail, because without an end, there *is* no trail
1710                                         wfProfileOut( __METHOD__."-might_be_img" );
1711                                         continue;
1712                                 }
1713                                 wfProfileOut( __METHOD__."-might_be_img" );
1714                         }
1715
1716                         $wasblank = ( $text  == '' );
1717                         if ( $wasblank ) $text = $link;
1718
1719                         # Link not escaped by : , create the various objects
1720                         if ( $noforce ) {
1721
1722                                 # Interwikis
1723                                 wfProfileIn( __METHOD__."-interwiki" );
1724                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1725                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1726                                         $s = rtrim($s . $prefix);
1727                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1728                                         wfProfileOut( __METHOD__."-interwiki" );
1729                                         continue;
1730                                 }
1731                                 wfProfileOut( __METHOD__."-interwiki" );
1732
1733                                 if ( $ns == NS_FILE ) {
1734                                         wfProfileIn( __METHOD__."-image" );
1735                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1736                                                 if ( $wasblank ) {
1737                                                         # if no parameters were passed, $text
1738                                                         # becomes something like "File:Foo.png",
1739                                                         # which we don't want to pass on to the
1740                                                         # image generator
1741                                                         $text = '';
1742                                                 } else {
1743                                                         # recursively parse links inside the image caption
1744                                                         # actually, this will parse them in any other parameters, too,
1745                                                         # but it might be hard to fix that, and it doesn't matter ATM
1746                                                         $text = $this->replaceExternalLinks($text);
1747                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1748                                                 }
1749                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1750                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1751                                         } else {
1752                                                 $s .= $prefix . $trail;
1753                                         }
1754                                         $this->mOutput->addImage( $nt->getDBkey() );
1755                                         wfProfileOut( __METHOD__."-image" );
1756                                         continue;
1757
1758                                 }
1759
1760                                 if ( $ns == NS_CATEGORY ) {
1761                                         wfProfileIn( __METHOD__."-category" );
1762                                         $s = rtrim($s . "\n"); # bug 87
1763
1764                                         if ( $wasblank ) {
1765                                                 $sortkey = $this->getDefaultSort();
1766                                         } else {
1767                                                 $sortkey = $text;
1768                                         }
1769                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1770                                         $sortkey = str_replace( "\n", '', $sortkey );
1771                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1772                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1773
1774                                         /**
1775                                          * Strip the whitespace Category links produce, see bug 87
1776                                          * @todo We might want to use trim($tmp, "\n") here.
1777                                          */
1778                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1779
1780                                         wfProfileOut( __METHOD__."-category" );
1781                                         continue;
1782                                 }
1783                         }
1784
1785                         # Self-link checking
1786                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1787                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1788                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1789                                         continue;
1790                                 }
1791                         }
1792
1793                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1794                         # FIXME: Should do batch file existence checks, see comment below
1795                         if( $ns == NS_MEDIA ) {
1796                                 wfProfileIn( __METHOD__."-media" );
1797                                 # Give extensions a chance to select the file revision for us
1798                                 $skip = $time = false;
1799                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1800                                 if ( $skip ) {
1801                                         $link = $sk->link( $nt );
1802                                 } else {
1803                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1804                                 }
1805                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1806                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1807                                 $this->mOutput->addImage( $nt->getDBkey() );
1808                                 wfProfileOut( __METHOD__."-media" );
1809                                 continue;
1810                         }
1811
1812                         wfProfileIn( __METHOD__."-always_known" );
1813                         # Some titles, such as valid special pages or files in foreign repos, should
1814                         # be shown as bluelinks even though they're not included in the page table
1815                         #
1816                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1817                         # batch file existence checks for NS_FILE and NS_MEDIA
1818                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1819                                 $this->mOutput->addLink( $nt );
1820                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1821                         } else {
1822                                 # Links will be added to the output link list after checking
1823                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1824                         }
1825                         wfProfileOut( __METHOD__."-always_known" );
1826                 }
1827                 wfProfileOut( __METHOD__ );
1828                 return $holders;
1829         }
1830
1831         /**
1832          * Make a link placeholder. The text returned can be later resolved to a real link with
1833          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1834          * parsing of interwiki links, and secondly to allow all existence checks and
1835          * article length checks (for stub links) to be bundled into a single query.
1836          *
1837          * @deprecated
1838          */
1839         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1840                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1841         }
1842
1843         /**
1844          * Render a forced-blue link inline; protect against double expansion of
1845          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1846          * Since this little disaster has to split off the trail text to avoid
1847          * breaking URLs in the following text without breaking trails on the
1848          * wiki links, it's been made into a horrible function.
1849          *
1850          * @param Title $nt
1851          * @param string $text
1852          * @param string $query
1853          * @param string $trail
1854          * @param string $prefix
1855          * @return string HTML-wikitext mix oh yuck
1856          */
1857         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1858                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1859                 $sk = $this->mOptions->getSkin();
1860                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1861                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1862                 return $this->armorLinks( $link ) . $trail;
1863         }
1864
1865         /**
1866          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1867          * going to go through further parsing steps before inline URL expansion.
1868          *
1869          * Not needed quite as much as it used to be since free links are a bit
1870          * more sensible these days. But bracketed links are still an issue.
1871          *
1872          * @param string more-or-less HTML
1873          * @return string less-or-more HTML with NOPARSE bits
1874          */
1875         function armorLinks( $text ) {
1876                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1877                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1878         }
1879
1880         /**
1881          * Return true if subpage links should be expanded on this page.
1882          * @return bool
1883          */
1884         function areSubpagesAllowed() {
1885                 # Some namespaces don't allow subpages
1886                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1887         }
1888
1889         /**
1890          * Handle link to subpage if necessary
1891          * @param string $target the source of the link
1892          * @param string &$text the link text, modified as necessary
1893          * @return string the full name of the link
1894          * @private
1895          */
1896         function maybeDoSubpageLink($target, &$text) {
1897                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1898         }
1899
1900         /**#@+
1901          * Used by doBlockLevels()
1902          * @private
1903          */
1904         /* private */ function closeParagraph() {
1905                 $result = '';
1906                 if ( $this->mLastSection != '' ) {
1907                         $result = '</' . $this->mLastSection  . ">\n";
1908                 }
1909                 $this->mInPre = false;
1910                 $this->mLastSection = '';
1911                 return $result;
1912         }
1913         # getCommon() returns the length of the longest common substring
1914         # of both arguments, starting at the beginning of both.
1915         #
1916         /* private */ function getCommon( $st1, $st2 ) {
1917                 $fl = strlen( $st1 );
1918                 $shorter = strlen( $st2 );
1919                 if ( $fl < $shorter ) { $shorter = $fl; }
1920
1921                 for ( $i = 0; $i < $shorter; ++$i ) {
1922                         if ( $st1{$i} != $st2{$i} ) { break; }
1923                 }
1924                 return $i;
1925         }
1926         # These next three functions open, continue, and close the list
1927         # element appropriate to the prefix character passed into them.
1928         #
1929         /* private */ function openList( $char ) {
1930                 $result = $this->closeParagraph();
1931
1932                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1933                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1934                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1935                 elseif ( ';' === $char ) {
1936                         $result .= '<dl><dt>';
1937                         $this->mDTopen = true;
1938                 }
1939                 else { $result = '<!-- ERR 1 -->'; }
1940
1941                 return $result;
1942         }
1943
1944         /* private */ function nextItem( $char ) {
1945                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1946                 elseif ( ':' === $char || ';' === $char ) {
1947                         $close = '</dd>';
1948                         if ( $this->mDTopen ) { $close = '</dt>'; }
1949                         if ( ';' === $char ) {
1950                                 $this->mDTopen = true;
1951                                 return $close . '<dt>';
1952                         } else {
1953                                 $this->mDTopen = false;
1954                                 return $close . '<dd>';
1955                         }
1956                 }
1957                 return '<!-- ERR 2 -->';
1958         }
1959
1960         /* private */ function closeList( $char ) {
1961                 if ( '*' === $char ) { $text = '</li></ul>'; }
1962                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
1963                 elseif ( ':' === $char ) {
1964                         if ( $this->mDTopen ) {
1965                                 $this->mDTopen = false;
1966                                 $text = '</dt></dl>';
1967                         } else {
1968                                 $text = '</dd></dl>';
1969                         }
1970                 }
1971                 else {  return '<!-- ERR 3 -->'; }
1972                 return $text."\n";
1973         }
1974         /**#@-*/
1975
1976         /**
1977          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1978          *
1979          * @param $linestart bool whether or not this is at the start of a line.
1980          * @private
1981          * @return string the lists rendered as HTML
1982          */
1983         function doBlockLevels( $text, $linestart ) {
1984                 wfProfileIn( __METHOD__ );
1985
1986                 # Parsing through the text line by line.  The main thing
1987                 # happening here is handling of block-level elements p, pre,
1988                 # and making lists from lines starting with * # : etc.
1989                 #
1990                 $textLines = StringUtils::explode( "\n", $text );
1991
1992                 $lastPrefix = $output = '';
1993                 $this->mDTopen = $inBlockElem = false;
1994                 $prefixLength = 0;
1995                 $paragraphStack = false;
1996
1997                 foreach ( $textLines as $oLine ) {
1998                         # Fix up $linestart
1999                         if ( !$linestart ) {
2000                                 $output .= $oLine;
2001                                 $linestart = true;
2002                                 continue;
2003                         }
2004                         // * = ul
2005                         // # = ol
2006                         // ; = dt
2007                         // : = dd
2008
2009                         $lastPrefixLength = strlen( $lastPrefix );
2010                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2011                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2012                         // If not in a <pre> element, scan for and figure out what prefixes are there.
2013                         if ( !$this->mInPre ) {
2014                                 # Multiple prefixes may abut each other for nested lists.
2015                                 $prefixLength = strspn( $oLine, '*#:;' );
2016                                 $prefix = substr( $oLine, 0, $prefixLength );
2017
2018                                 # eh?
2019                                 // ; and : are both from definition-lists, so they're equivalent
2020                                 //  for the purposes of determining whether or not we need to open/close
2021                                 //  elements.
2022                                 $prefix2 = str_replace( ';', ':', $prefix );
2023                                 $t = substr( $oLine, $prefixLength );
2024                                 $this->mInPre = (bool)$preOpenMatch;
2025                         } else {
2026                                 # Don't interpret any other prefixes in preformatted text
2027                                 $prefixLength = 0;
2028                                 $prefix = $prefix2 = '';
2029                                 $t = $oLine;
2030                         }
2031
2032                         # List generation
2033                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2034                                 # Same as the last item, so no need to deal with nesting or opening stuff
2035                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2036                                 $paragraphStack = false;
2037
2038                                 if ( substr( $prefix, -1 ) === ';') {
2039                                         # The one nasty exception: definition lists work like this:
2040                                         # ; title : definition text
2041                                         # So we check for : in the remainder text to split up the
2042                                         # title and definition, without b0rking links.
2043                                         $term = $t2 = '';
2044                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2045                                                 $t = $t2;
2046                                                 $output .= $term . $this->nextItem( ':' );
2047                                         }
2048                                 }
2049                         } elseif( $prefixLength || $lastPrefixLength ) {
2050                                 // We need to open or close prefixes, or both.
2051
2052                                 # Either open or close a level...
2053                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2054                                 $paragraphStack = false;
2055
2056                                 // Close all the prefixes which aren't shared.
2057                                 while( $commonPrefixLength < $lastPrefixLength ) {
2058                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2059                                         --$lastPrefixLength;
2060                                 }
2061
2062                                 // Continue the current prefix if appropriate.
2063                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2064                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2065                                 }
2066
2067                                 // Open prefixes where appropriate.
2068                                 while ( $prefixLength > $commonPrefixLength ) {
2069                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2070                                         $output .= $this->openList( $char );
2071
2072                                         if ( ';' === $char ) {
2073                                                 # FIXME: This is dupe of code above
2074                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2075                                                         $t = $t2;
2076                                                         $output .= $term . $this->nextItem( ':' );
2077                                                 }
2078                                         }
2079                                         ++$commonPrefixLength;
2080                                 }
2081                                 $lastPrefix = $prefix2;
2082                         }
2083
2084                         // If we have no prefixes, go to paragraph mode.
2085                         if( 0 == $prefixLength ) {
2086                                 wfProfileIn( __METHOD__."-paragraph" );
2087                                 # No prefix (not in list)--go to paragraph mode
2088                                 // XXX: use a stack for nestable elements like span, table and div
2089                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2090                                 $closematch = preg_match(
2091                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2092                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2093                                 if ( $openmatch or $closematch ) {
2094                                         $paragraphStack = false;
2095                                         # TODO bug 5718: paragraph closed
2096                                         $output .= $this->closeParagraph();
2097                                         if ( $preOpenMatch and !$preCloseMatch ) {
2098                                                 $this->mInPre = true;
2099                                         }
2100                                         if ( $closematch ) {
2101                                                 $inBlockElem = false;
2102                                         } else {
2103                                                 $inBlockElem = true;
2104                                         }
2105                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2106                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2107                                                 // pre
2108                                                 if ($this->mLastSection !== 'pre') {
2109                                                         $paragraphStack = false;
2110                                                         $output .= $this->closeParagraph().'<pre>';
2111                                                         $this->mLastSection = 'pre';
2112                                                 }
2113                                                 $t = substr( $t, 1 );
2114                                         } else {
2115                                                 // paragraph
2116                                                 if ( trim($t) == '' ) {
2117                                                         if ( $paragraphStack ) {
2118                                                                 $output .= $paragraphStack.'<br />';
2119                                                                 $paragraphStack = false;
2120                                                                 $this->mLastSection = 'p';
2121                                                         } else {
2122                                                                 if ($this->mLastSection !== 'p' ) {
2123                                                                         $output .= $this->closeParagraph();
2124                                                                         $this->mLastSection = '';
2125                                                                         $paragraphStack = '<p>';
2126                                                                 } else {
2127                                                                         $paragraphStack = '</p><p>';
2128                                                                 }
2129                                                         }
2130                                                 } else {
2131                                                         if ( $paragraphStack ) {
2132                                                                 $output .= $paragraphStack;
2133                                                                 $paragraphStack = false;
2134                                                                 $this->mLastSection = 'p';
2135                                                         } else if ($this->mLastSection !== 'p') {
2136                                                                 $output .= $this->closeParagraph().'<p>';
2137                                                                 $this->mLastSection = 'p';
2138                                                         }
2139                                                 }
2140                                         }
2141                                 }
2142                                 wfProfileOut( __METHOD__."-paragraph" );
2143                         }
2144                         // somewhere above we forget to get out of pre block (bug 785)
2145                         if($preCloseMatch && $this->mInPre) {
2146                                 $this->mInPre = false;
2147                         }
2148                         if ($paragraphStack === false) {
2149                                 $output .= $t."\n";
2150                         }
2151                 }
2152                 while ( $prefixLength ) {
2153                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2154                         --$prefixLength;
2155                 }
2156                 if ( $this->mLastSection != '' ) {
2157                         $output .= '</' . $this->mLastSection . '>';
2158                         $this->mLastSection = '';
2159                 }
2160
2161                 wfProfileOut( __METHOD__ );
2162                 return $output;
2163         }
2164
2165         /**
2166          * Split up a string on ':', ignoring any occurences inside tags
2167          * to prevent illegal overlapping.
2168          * @param string $str the string to split
2169          * @param string &$before set to everything before the ':'
2170          * @param string &$after set to everything after the ':'
2171          * return string the position of the ':', or false if none found
2172          */
2173         function findColonNoLinks($str, &$before, &$after) {
2174                 wfProfileIn( __METHOD__ );
2175
2176                 $pos = strpos( $str, ':' );
2177                 if( $pos === false ) {
2178                         // Nothing to find!
2179                         wfProfileOut( __METHOD__ );
2180                         return false;
2181                 }
2182
2183                 $lt = strpos( $str, '<' );
2184                 if( $lt === false || $lt > $pos ) {
2185                         // Easy; no tag nesting to worry about
2186                         $before = substr( $str, 0, $pos );
2187                         $after = substr( $str, $pos+1 );
2188                         wfProfileOut( __METHOD__ );
2189                         return $pos;
2190                 }
2191
2192                 // Ugly state machine to walk through avoiding tags.
2193                 $state = self::COLON_STATE_TEXT;
2194                 $stack = 0;
2195                 $len = strlen( $str );
2196                 for( $i = 0; $i < $len; $i++ ) {
2197                         $c = $str{$i};
2198
2199                         switch( $state ) {
2200                         // (Using the number is a performance hack for common cases)
2201                         case 0: // self::COLON_STATE_TEXT:
2202                                 switch( $c ) {
2203                                 case "<":
2204                                         // Could be either a <start> tag or an </end> tag
2205                                         $state = self::COLON_STATE_TAGSTART;
2206                                         break;
2207                                 case ":":
2208                                         if( $stack == 0 ) {
2209                                                 // We found it!
2210                                                 $before = substr( $str, 0, $i );
2211                                                 $after = substr( $str, $i + 1 );
2212                                                 wfProfileOut( __METHOD__ );
2213                                                 return $i;
2214                                         }
2215                                         // Embedded in a tag; don't break it.
2216                                         break;
2217                                 default:
2218                                         // Skip ahead looking for something interesting
2219                                         $colon = strpos( $str, ':', $i );
2220                                         if( $colon === false ) {
2221                                                 // Nothing else interesting
2222                                                 wfProfileOut( __METHOD__ );
2223                                                 return false;
2224                                         }
2225                                         $lt = strpos( $str, '<', $i );
2226                                         if( $stack === 0 ) {
2227                                                 if( $lt === false || $colon < $lt ) {
2228                                                         // We found it!
2229                                                         $before = substr( $str, 0, $colon );
2230                                                         $after = substr( $str, $colon + 1 );
2231                                                         wfProfileOut( __METHOD__ );
2232                                                         return $i;
2233                                                 }
2234                                         }
2235                                         if( $lt === false ) {
2236                                                 // Nothing else interesting to find; abort!
2237                                                 // We're nested, but there's no close tags left. Abort!
2238                                                 break 2;
2239                                         }
2240                                         // Skip ahead to next tag start
2241                                         $i = $lt;
2242                                         $state = self::COLON_STATE_TAGSTART;
2243                                 }
2244                                 break;
2245                         case 1: // self::COLON_STATE_TAG:
2246                                 // In a <tag>
2247                                 switch( $c ) {
2248                                 case ">":
2249                                         $stack++;
2250                                         $state = self::COLON_STATE_TEXT;
2251                                         break;
2252                                 case "/":
2253                                         // Slash may be followed by >?
2254                                         $state = self::COLON_STATE_TAGSLASH;
2255                                         break;
2256                                 default:
2257                                         // ignore
2258                                 }
2259                                 break;
2260                         case 2: // self::COLON_STATE_TAGSTART:
2261                                 switch( $c ) {
2262                                 case "/":
2263                                         $state = self::COLON_STATE_CLOSETAG;
2264                                         break;
2265                                 case "!":
2266                                         $state = self::COLON_STATE_COMMENT;
2267                                         break;
2268                                 case ">":
2269                                         // Illegal early close? This shouldn't happen D:
2270                                         $state = self::COLON_STATE_TEXT;
2271                                         break;
2272                                 default:
2273                                         $state = self::COLON_STATE_TAG;
2274                                 }
2275                                 break;
2276                         case 3: // self::COLON_STATE_CLOSETAG:
2277                                 // In a </tag>
2278                                 if( $c === ">" ) {
2279                                         $stack--;
2280                                         if( $stack < 0 ) {
2281                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2282                                                 wfProfileOut( __METHOD__ );
2283                                                 return false;
2284                                         }
2285                                         $state = self::COLON_STATE_TEXT;
2286                                 }
2287                                 break;
2288                         case self::COLON_STATE_TAGSLASH:
2289                                 if( $c === ">" ) {
2290                                         // Yes, a self-closed tag <blah/>
2291                                         $state = self::COLON_STATE_TEXT;
2292                                 } else {
2293                                         // Probably we're jumping the gun, and this is an attribute
2294                                         $state = self::COLON_STATE_TAG;
2295                                 }
2296                                 break;
2297                         case 5: // self::COLON_STATE_COMMENT:
2298                                 if( $c === "-" ) {
2299                                         $state = self::COLON_STATE_COMMENTDASH;
2300                                 }
2301                                 break;
2302                         case self::COLON_STATE_COMMENTDASH:
2303                                 if( $c === "-" ) {
2304                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2305                                 } else {
2306                                         $state = self::COLON_STATE_COMMENT;
2307                                 }
2308                                 break;
2309                         case self::COLON_STATE_COMMENTDASHDASH:
2310                                 if( $c === ">" ) {
2311                                         $state = self::COLON_STATE_TEXT;
2312                                 } else {
2313                                         $state = self::COLON_STATE_COMMENT;
2314                                 }
2315                                 break;
2316                         default:
2317                                 throw new MWException( "State machine error in " . __METHOD__ );
2318                         }
2319                 }
2320                 if( $stack > 0 ) {
2321                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2322                         return false;
2323                 }
2324                 wfProfileOut( __METHOD__ );
2325                 return false;
2326         }
2327
2328         /**
2329          * Return value of a magic variable (like PAGENAME)
2330          *
2331          * @private
2332          */
2333         function getVariableValue( $index, $frame=false ) {
2334                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2335                 global $wgScriptPath, $wgStylePath;
2336
2337                 /**
2338                  * Some of these require message or data lookups and can be
2339                  * expensive to check many times.
2340                  */
2341                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2342                         if ( isset( $this->mVarCache[$index] ) ) {
2343                                 return $this->mVarCache[$index];
2344                         }
2345                 }
2346
2347                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2348                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2349
2350                 # Use the time zone
2351                 global $wgLocaltimezone;
2352                 if ( isset( $wgLocaltimezone ) ) {
2353                         $oldtz = date_default_timezone_get();
2354                         date_default_timezone_set( $wgLocaltimezone );
2355                 }
2356
2357                 $localTimestamp = date( 'YmdHis', $ts );
2358                 $localMonth = date( 'm', $ts );
2359                 $localMonth1 = date( 'n', $ts );
2360                 $localMonthName = date( 'n', $ts );
2361                 $localDay = date( 'j', $ts );
2362                 $localDay2 = date( 'd', $ts );
2363                 $localDayOfWeek = date( 'w', $ts );
2364                 $localWeek = date( 'W', $ts );
2365                 $localYear = date( 'Y', $ts );
2366                 $localHour = date( 'H', $ts );
2367                 if ( isset( $wgLocaltimezone ) ) {
2368                         date_default_timezone_set( $oldtz );
2369                 }
2370
2371                 switch ( $index ) {
2372                         case 'currentmonth':
2373                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2374                                 break;
2375                         case 'currentmonth1':
2376                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2377                                 break;
2378                         case 'currentmonthname':
2379                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2380                                 break;
2381                         case 'currentmonthnamegen':
2382                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2383                                 break;
2384                         case 'currentmonthabbrev':
2385                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2386                                 break;
2387                         case 'currentday':
2388                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2389                                 break;
2390                         case 'currentday2':
2391                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2392                                 break;
2393                         case 'localmonth':
2394                                 $value = $wgContLang->formatNum( $localMonth );
2395                                 break;
2396                         case 'localmonth1':
2397                                 $value = $wgContLang->formatNum( $localMonth1 );
2398                                 break;
2399                         case 'localmonthname':
2400                                 $value = $wgContLang->getMonthName( $localMonthName );
2401                                 break;
2402                         case 'localmonthnamegen':
2403                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2404                                 break;
2405                         case 'localmonthabbrev':
2406                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2407                                 break;
2408                         case 'localday':
2409                                 $value = $wgContLang->formatNum( $localDay );
2410                                 break;
2411                         case 'localday2':
2412                                 $value = $wgContLang->formatNum( $localDay2 );
2413                                 break;
2414                         case 'pagename':
2415                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2416                                 break;
2417                         case 'pagenamee':
2418                                 $value = $this->mTitle->getPartialURL();
2419                                 break;
2420                         case 'fullpagename':
2421                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2422                                 break;
2423                         case 'fullpagenamee':
2424                                 $value = $this->mTitle->getPrefixedURL();
2425                                 break;
2426                         case 'subpagename':
2427                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2428                                 break;
2429                         case 'subpagenamee':
2430                                 $value = $this->mTitle->getSubpageUrlForm();
2431                                 break;
2432                         case 'basepagename':
2433                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2434                                 break;
2435                         case 'basepagenamee':
2436                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2437                                 break;
2438                         case 'talkpagename':
2439                                 if( $this->mTitle->canTalk() ) {
2440                                         $talkPage = $this->mTitle->getTalkPage();
2441                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2442                                 } else {
2443                                         $value = '';
2444                                 }
2445                                 break;
2446                         case 'talkpagenamee':
2447                                 if( $this->mTitle->canTalk() ) {
2448                                         $talkPage = $this->mTitle->getTalkPage();
2449                                         $value = $talkPage->getPrefixedUrl();
2450                                 } else {
2451                                         $value = '';
2452                                 }
2453                                 break;
2454                         case 'subjectpagename':
2455                                 $subjPage = $this->mTitle->getSubjectPage();
2456                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2457                                 break;
2458                         case 'subjectpagenamee':
2459                                 $subjPage = $this->mTitle->getSubjectPage();
2460                                 $value = $subjPage->getPrefixedUrl();
2461                                 break;
2462                         case 'revisionid':
2463                                 // Let the edit saving system know we should parse the page
2464                                 // *after* a revision ID has been assigned.
2465                                 $this->mOutput->setFlag( 'vary-revision' );
2466                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2467                                 $value = $this->mRevisionId;
2468                                 break;
2469                         case 'revisionday':
2470                                 // Let the edit saving system know we should parse the page
2471                                 // *after* a revision ID has been assigned. This is for null edits.
2472                                 $this->mOutput->setFlag( 'vary-revision' );
2473                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2474                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2475                                 break;
2476                         case 'revisionday2':
2477                                 // Let the edit saving system know we should parse the page
2478                                 // *after* a revision ID has been assigned. This is for null edits.
2479                                 $this->mOutput->setFlag( 'vary-revision' );
2480                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2481                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2482                                 break;
2483                         case 'revisionmonth':
2484                                 // Let the edit saving system know we should parse the page
2485                                 // *after* a revision ID has been assigned. This is for null edits.
2486                                 $this->mOutput->setFlag( 'vary-revision' );
2487                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2488                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2489                                 break;
2490                         case 'revisionyear':
2491                                 // Let the edit saving system know we should parse the page
2492                                 // *after* a revision ID has been assigned. This is for null edits.
2493                                 $this->mOutput->setFlag( 'vary-revision' );
2494                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2495                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2496                                 break;
2497                         case 'revisiontimestamp':
2498                                 // Let the edit saving system know we should parse the page
2499                                 // *after* a revision ID has been assigned. This is for null edits.
2500                                 $this->mOutput->setFlag( 'vary-revision' );
2501                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2502                                 $value = $this->getRevisionTimestamp();
2503                                 break;
2504                         case 'revisionuser':
2505                                 // Let the edit saving system know we should parse the page
2506                                 // *after* a revision ID has been assigned. This is for null edits.
2507                                 $this->mOutput->setFlag( 'vary-revision' );
2508                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2509                                 $value = $this->getRevisionUser();
2510                                 break;
2511                         case 'namespace':
2512                                 $value = str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2513                                 break;
2514                         case 'namespacee':
2515                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2516                                 break;
2517                         case 'talkspace':
2518                                 $value = $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2519                                 break;
2520                         case 'talkspacee':
2521                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2522                                 break;
2523                         case 'subjectspace':
2524                                 $value = $this->mTitle->getSubjectNsText();
2525                                 break;
2526                         case 'subjectspacee':
2527                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2528                                 break;
2529                         case 'currentdayname':
2530                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2531                                 break;
2532                         case 'currentyear':
2533                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2534                                 break;
2535                         case 'currenttime':
2536                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2537                                 break;
2538                         case 'currenthour':
2539                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2540                                 break;
2541                         case 'currentweek':
2542                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2543                                 // int to remove the padding
2544                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2545                                 break;
2546                         case 'currentdow':
2547                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2548                                 break;
2549                         case 'localdayname':
2550                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2551                                 break;
2552                         case 'localyear':
2553                                 $value = $wgContLang->formatNum( $localYear, true );
2554                                 break;
2555                         case 'localtime':
2556                                 $value = $wgContLang->time( $localTimestamp, false, false );
2557                                 break;
2558                         case 'localhour':
2559                                 $value = $wgContLang->formatNum( $localHour, true );
2560                                 break;
2561                         case 'localweek':
2562                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2563                                 // int to remove the padding
2564                                 $value = $wgContLang->formatNum( (int)$localWeek );
2565                                 break;
2566                         case 'localdow':
2567                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2568                                 break;
2569                         case 'numberofarticles':
2570                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2571                                 break;
2572                         case 'numberoffiles':
2573                                 $value = $wgContLang->formatNum( SiteStats::images() );
2574                                 break;
2575                         case 'numberofusers':
2576                                 $value = $wgContLang->formatNum( SiteStats::users() );
2577                                 break;
2578                         case 'numberofactiveusers':
2579                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2580                                 break;
2581                         case 'numberofpages':
2582                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2583                                 break;
2584                         case 'numberofadmins':
2585                                 $value = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2586                                 break;
2587                         case 'numberofedits':
2588                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2589                                 break;
2590                         case 'numberofviews':
2591                                 $value = $wgContLang->formatNum( SiteStats::views() );
2592                                 break;
2593                         case 'currenttimestamp':
2594                                 $value = wfTimestamp( TS_MW, $ts );
2595                                 break;
2596                         case 'localtimestamp':
2597                                 $value = $localTimestamp;
2598                                 break;
2599                         case 'currentversion':
2600                                 $value = SpecialVersion::getVersion();
2601                                 break;
2602                         case 'sitename':
2603                                 return $wgSitename;
2604                         case 'server':
2605                                 return $wgServer;
2606                         case 'servername':
2607                                 return $wgServerName;
2608                         case 'scriptpath':
2609                                 return $wgScriptPath;
2610                         case 'stylepath':
2611                                 return $wgStylePath;
2612                         case 'directionmark':
2613                                 return $wgContLang->getDirMark();
2614                         case 'contentlanguage':
2615                                 global $wgContLanguageCode;
2616                                 return $wgContLanguageCode;
2617                         default:
2618                                 $ret = null;
2619                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) )
2620                                         return $ret;
2621                                 else
2622                                         return null;
2623                 }
2624
2625                 if ( $index )
2626                         $this->mVarCache[$index] = $value;
2627
2628                 return $value;
2629         }
2630
2631         /**
2632          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2633          *
2634          * @private
2635          */
2636         function initialiseVariables() {
2637                 wfProfileIn( __METHOD__ );
2638                 $variableIDs = MagicWord::getVariableIDs();
2639                 $substIDs = MagicWord::getSubstIDs();
2640
2641                 $this->mVariables = new MagicWordArray( $variableIDs );
2642                 $this->mSubstWords = new MagicWordArray( $substIDs );
2643                 wfProfileOut( __METHOD__ );
2644         }
2645
2646         /**
2647          * Preprocess some wikitext and return the document tree.
2648          * This is the ghost of replace_variables().
2649          *
2650          * @param string $text The text to parse
2651          * @param integer flags Bitwise combination of:
2652          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2653          *                                     included. Default is to assume a direct page view.
2654          *
2655          * The generated DOM tree must depend only on the input text and the flags.
2656          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2657          *
2658          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2659          * change in the DOM tree for a given text, must be passed through the section identifier
2660          * in the section edit link and thus back to extractSections().
2661          *
2662          * The output of this function is currently only cached in process memory, but a persistent
2663          * cache may be implemented at a later date which takes further advantage of these strict
2664          * dependency requirements.
2665          *
2666          * @private
2667          */
2668         function preprocessToDom ( $text, $flags = 0 ) {
2669                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2670                 return $dom;
2671         }
2672
2673         /*
2674          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2675          */
2676         public static function splitWhitespace( $s ) {
2677                 $ltrimmed = ltrim( $s );
2678                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2679                 $trimmed = rtrim( $ltrimmed );
2680                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2681                 if ( $diff > 0 ) {
2682                         $w2 = substr( $ltrimmed, -$diff );
2683                 } else {
2684                         $w2 = '';
2685                 }
2686                 return array( $w1, $trimmed, $w2 );
2687         }
2688
2689         /**
2690          * Replace magic variables, templates, and template arguments
2691          * with the appropriate text. Templates are substituted recursively,
2692          * taking care to avoid infinite loops.
2693          *
2694          * Note that the substitution depends on value of $mOutputType:
2695          *  self::OT_WIKI: only {{subst:}} templates
2696          *  self::OT_PREPROCESS: templates but not extension tags
2697          *  self::OT_HTML: all templates and extension tags
2698          *
2699          * @param string $tex The text to transform
2700          * @param PPFrame $frame Object describing the arguments passed to the template.
2701          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2702          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2703          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2704          * @private
2705          */
2706         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2707                 # Is there any text? Also, Prevent too big inclusions!
2708                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2709                         return $text;
2710                 }
2711                 wfProfileIn( __METHOD__ );
2712
2713                 if ( $frame === false ) {
2714                         $frame = $this->getPreprocessor()->newFrame();
2715                 } elseif ( !( $frame instanceof PPFrame ) ) {
2716                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2717                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2718                 }
2719
2720                 $dom = $this->preprocessToDom( $text );
2721                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2722                 $text = $frame->expand( $dom, $flags );
2723
2724                 wfProfileOut( __METHOD__ );
2725                 return $text;
2726         }
2727
2728         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2729         static function createAssocArgs( $args ) {
2730                 $assocArgs = array();
2731                 $index = 1;
2732                 foreach( $args as $arg ) {
2733                         $eqpos = strpos( $arg, '=' );
2734                         if ( $eqpos === false ) {
2735                                 $assocArgs[$index++] = $arg;
2736                         } else {
2737                                 $name = trim( substr( $arg, 0, $eqpos ) );
2738                                 $value = trim( substr( $arg, $eqpos+1 ) );
2739                                 if ( $value === false ) {
2740                                         $value = '';
2741                                 }
2742                                 if ( $name !== false ) {
2743                                         $assocArgs[$name] = $value;
2744                                 }
2745                         }
2746                 }
2747
2748                 return $assocArgs;
2749         }
2750
2751         /**
2752          * Warn the user when a parser limitation is reached
2753          * Will warn at most once the user per limitation type
2754          *
2755          * @param string $limitationType, should be one of:
2756          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2757          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2758          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2759          * @params int $current, $max When an explicit limit has been
2760          *       exceeded, provide the values (optional)
2761          */
2762         function limitationWarn( $limitationType, $current=null, $max=null) {
2763                 //does no harm if $current and $max are present but are unnecessary for the message
2764                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2765                 $this->mOutput->addWarning( $warning );
2766                 $this->addTrackingCategory( "$limitationType-category" );
2767         }
2768
2769         /**
2770          * Return the text of a template, after recursively
2771          * replacing any variables or templates within the template.
2772          *
2773          * @param array $piece The parts of the template
2774          *  $piece['title']: the title, i.e. the part before the |
2775          *  $piece['parts']: the parameter array
2776          *  $piece['lineStart']: whether the brace was at the start of a line
2777          * @param PPFrame The current frame, contains template arguments
2778          * @return string the text of the template
2779          * @private
2780          */
2781         function braceSubstitution( $piece, $frame ) {
2782                 global $wgContLang, $wgNonincludableNamespaces;
2783                 wfProfileIn( __METHOD__ );
2784                 wfProfileIn( __METHOD__.'-setup' );
2785
2786                 # Flags
2787                 $found = false;             # $text has been filled
2788                 $nowiki = false;            # wiki markup in $text should be escaped
2789                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2790                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2791                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2792                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2793
2794                 # Title object, where $text came from
2795                 $title = null;
2796
2797                 # $part1 is the bit before the first |, and must contain only title characters.
2798                 # Various prefixes will be stripped from it later.
2799                 $titleWithSpaces = $frame->expand( $piece['title'] );
2800                 $part1 = trim( $titleWithSpaces );
2801                 $titleText = false;
2802
2803                 # Original title text preserved for various purposes
2804                 $originalTitle = $part1;
2805
2806                 # $args is a list of argument nodes, starting from index 0, not including $part1
2807                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2808                 wfProfileOut( __METHOD__.'-setup' );
2809
2810                 # SUBST
2811                 wfProfileIn( __METHOD__.'-modifiers' );
2812                 if ( !$found ) {
2813
2814                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2815
2816                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2817                         # Decide whether to expand template or keep wikitext as-is.
2818                         if ( $this->ot['wiki'] ) {
2819                                 if ( $substMatch === false ) {
2820                                         $literal = true;  # literal when in PST with no prefix
2821                                 } else {
2822                                         $literal = false; # expand when in PST with subst: or safesubst:
2823                                 }
2824                         } else {
2825                                 if ( $substMatch == 'subst' ) {
2826                                         $literal = true;  # literal when not in PST with plain subst:
2827                                 } else {
2828                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2829                                 }
2830                         }
2831                         if ( $literal ) {
2832                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2833                                 $isLocalObj = true;
2834                                 $found = true;
2835                         }
2836                 }
2837
2838                 # Variables
2839                 if ( !$found && $args->getLength() == 0 ) {
2840                         $id = $this->mVariables->matchStartToEnd( $part1 );
2841                         if ( $id !== false ) {
2842                                 $text = $this->getVariableValue( $id, $frame );
2843                                 if (MagicWord::getCacheTTL($id)>-1)
2844                                         $this->mOutput->mContainsOldMagic = true;
2845                                 $found = true;
2846                         }
2847                 }
2848
2849                 # MSG, MSGNW and RAW
2850                 if ( !$found ) {
2851                         # Check for MSGNW:
2852                         $mwMsgnw = MagicWord::get( 'msgnw' );
2853                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2854                                 $nowiki = true;
2855                         } else {
2856                                 # Remove obsolete MSG:
2857                                 $mwMsg = MagicWord::get( 'msg' );
2858                                 $mwMsg->matchStartAndRemove( $part1 );
2859                         }
2860
2861                         # Check for RAW:
2862                         $mwRaw = MagicWord::get( 'raw' );
2863                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2864                                 $forceRawInterwiki = true;
2865                         }
2866                 }
2867                 wfProfileOut( __METHOD__.'-modifiers' );
2868
2869                 # Parser functions
2870                 if ( !$found ) {
2871                         wfProfileIn( __METHOD__ . '-pfunc' );
2872
2873                         $colonPos = strpos( $part1, ':' );
2874                         if ( $colonPos !== false ) {
2875                                 # Case sensitive functions
2876                                 $function = substr( $part1, 0, $colonPos );
2877                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2878                                         $function = $this->mFunctionSynonyms[1][$function];
2879                                 } else {
2880                                         # Case insensitive functions
2881                                         $function = $wgContLang->lc( $function );
2882                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2883                                                 $function = $this->mFunctionSynonyms[0][$function];
2884                                         } else {
2885                                                 $function = false;
2886                                         }
2887                                 }
2888                                 if ( $function ) {
2889                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2890                                         $initialArgs = array( &$this );
2891                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2892                                         if ( $flags & SFH_OBJECT_ARGS ) {
2893                                                 # Add a frame parameter, and pass the arguments as an array
2894                                                 $allArgs = $initialArgs;
2895                                                 $allArgs[] = $frame;
2896                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2897                                                         $funcArgs[] = $args->item( $i );
2898                                                 }
2899                                                 $allArgs[] = $funcArgs;
2900                                         } else {
2901                                                 # Convert arguments to plain text
2902                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2903                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2904                                                 }
2905                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2906                                         }
2907
2908                                         # Workaround for PHP bug 35229 and similar
2909                                         if ( !is_callable( $callback ) ) {
2910                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2911                                                 wfProfileOut( __METHOD__ );
2912                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2913                                         }
2914                                         $result = call_user_func_array( $callback, $allArgs );
2915                                         $found = true;
2916                                         $noparse = true;
2917                                         $preprocessFlags = 0;
2918
2919                                         if ( is_array( $result ) ) {
2920                                                 if ( isset( $result[0] ) ) {
2921                                                         $text = $result[0];
2922                                                         unset( $result[0] );
2923                                                 }
2924
2925                                                 // Extract flags into the local scope
2926                                                 // This allows callers to set flags such as nowiki, found, etc.
2927                                                 extract( $result );
2928                                         } else {
2929                                                 $text = $result;
2930                                         }
2931                                         if ( !$noparse ) {
2932                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2933                                                 $isChildObj = true;
2934                                         }
2935                                 }
2936                         }
2937                         wfProfileOut( __METHOD__ . '-pfunc' );
2938                 }
2939
2940                 # Finish mangling title and then check for loops.
2941                 # Set $title to a Title object and $titleText to the PDBK
2942                 if ( !$found ) {
2943                         $ns = NS_TEMPLATE;
2944                         # Split the title into page and subpage
2945                         $subpage = '';
2946                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2947                         if ($subpage !== '') {
2948                                 $ns = $this->mTitle->getNamespace();
2949                         }
2950                         $title = Title::newFromText( $part1, $ns );
2951                         if ( $title ) {
2952                                 $titleText = $title->getPrefixedText();
2953                                 # Check for language variants if the template is not found
2954                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2955                                         $wgContLang->findVariantLink( $part1, $title, true );
2956                                 }
2957                                 # Do recursion depth check
2958                                 $limit = $this->mOptions->getMaxTemplateDepth();
2959                                 if ( $frame->depth >= $limit ) {
2960                                         $found = true;
2961                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2962                                 }
2963                         }
2964                 }
2965
2966                 # Load from database
2967                 if ( !$found && $title ) {
2968                         wfProfileIn( __METHOD__ . '-loadtpl' );
2969                         if ( !$title->isExternal() ) {
2970                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2971                                         $text = SpecialPage::capturePath( $title );
2972                                         if ( is_string( $text ) ) {
2973                                                 $found = true;
2974                                                 $isHTML = true;
2975                                                 $this->disableCache();
2976                                         }
2977                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2978                                         $found = false; //access denied
2979                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2980                                 } else {
2981                                         list( $text, $title ) = $this->getTemplateDom( $title );
2982                                         if ( $text !== false ) {
2983                                                 $found = true;
2984                                                 $isChildObj = true;
2985                                         }
2986                                 }
2987
2988                                 # If the title is valid but undisplayable, make a link to it
2989                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2990                                         $text = "[[:$titleText]]";
2991                                         $found = true;
2992                                 }
2993                         } elseif ( $title->isTrans() ) {
2994                                 // Interwiki transclusion
2995                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2996                                         $text = $this->interwikiTransclude( $title, 'render' );
2997                                         $isHTML = true;
2998                                 } else {
2999                                         $text = $this->interwikiTransclude( $title, 'raw' );
3000                                         // Preprocess it like a template
3001                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3002                                         $isChildObj = true;
3003                                 }
3004                                 $found = true;
3005                         }
3006
3007                         # Do infinite loop check
3008                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3009                         if ( !$frame->loopCheck( $title ) ) {
3010                                 $found = true;
3011                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3012                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3013                         }
3014                         wfProfileOut( __METHOD__ . '-loadtpl' );
3015                 }
3016
3017                 # If we haven't found text to substitute by now, we're done
3018                 # Recover the source wikitext and return it
3019                 if ( !$found ) {
3020                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3021                         wfProfileOut( __METHOD__ );
3022                         return array( 'object' => $text );
3023                 }
3024
3025                 # Expand DOM-style return values in a child frame
3026                 if ( $isChildObj ) {
3027                         # Clean up argument array
3028                         $newFrame = $frame->newChild( $args, $title );
3029
3030                         if ( $nowiki ) {
3031                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3032                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3033                                 # Expansion is eligible for the empty-frame cache
3034                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3035                                         $text = $this->mTplExpandCache[$titleText];
3036                                 } else {
3037                                         $text = $newFrame->expand( $text );
3038                                         $this->mTplExpandCache[$titleText] = $text;
3039                                 }
3040                         } else {
3041                                 # Uncached expansion
3042                                 $text = $newFrame->expand( $text );
3043                         }
3044                 }
3045                 if ( $isLocalObj && $nowiki ) {
3046                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3047                         $isLocalObj = false;
3048                 }
3049
3050                 # Replace raw HTML by a placeholder
3051                 # Add a blank line preceding, to prevent it from mucking up
3052                 # immediately preceding headings
3053                 if ( $isHTML ) {
3054                         $text = "\n\n" . $this->insertStripItem( $text );
3055                 }
3056                 # Escape nowiki-style return values
3057                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3058                         $text = wfEscapeWikiText( $text );
3059                 }
3060                 # Bug 529: if the template begins with a table or block-level
3061                 # element, it should be treated as beginning a new line.
3062                 # This behaviour is somewhat controversial.
3063                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3064                         $text = "\n" . $text;
3065                 }
3066
3067                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3068                         # Error, oversize inclusion
3069                         $text = "[[$originalTitle]]" .
3070                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3071                         $this->limitationWarn( 'post-expand-template-inclusion' );
3072                 }
3073
3074                 if ( $isLocalObj ) {
3075                         $ret = array( 'object' => $text );
3076                 } else {
3077                         $ret = array( 'text' => $text );
3078                 }
3079
3080                 wfProfileOut( __METHOD__ );
3081                 return $ret;
3082         }
3083
3084         /**
3085          * Get the semi-parsed DOM representation of a template with a given title,
3086          * and its redirect destination title. Cached.
3087          */
3088         function getTemplateDom( $title ) {
3089                 $cacheTitle = $title;
3090                 $titleText = $title->getPrefixedDBkey();
3091
3092                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3093                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3094                         $title = Title::makeTitle( $ns, $dbk );
3095                         $titleText = $title->getPrefixedDBkey();
3096                 }
3097                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3098                         return array( $this->mTplDomCache[$titleText], $title );
3099                 }
3100
3101                 // Cache miss, go to the database
3102                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3103
3104                 if ( $text === false ) {
3105                         $this->mTplDomCache[$titleText] = false;
3106                         return array( false, $title );
3107                 }
3108
3109                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3110                 $this->mTplDomCache[ $titleText ] = $dom;
3111
3112                 if (! $title->equals($cacheTitle)) {
3113                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3114                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3115                 }
3116
3117                 return array( $dom, $title );
3118         }
3119
3120         /**
3121          * Fetch the unparsed text of a template and register a reference to it.
3122          */
3123         function fetchTemplateAndTitle( $title ) {
3124                 $templateCb = $this->mOptions->getTemplateCallback();
3125                 $stuff = call_user_func( $templateCb, $title, $this );
3126                 $text = $stuff['text'];
3127                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3128                 if ( isset( $stuff['deps'] ) ) {
3129                         foreach ( $stuff['deps'] as $dep ) {
3130                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3131                         }
3132                 }
3133                 return array($text,$finalTitle);
3134         }
3135
3136         function fetchTemplate( $title ) {
3137                 $rv = $this->fetchTemplateAndTitle($title);
3138                 return $rv[0];
3139         }
3140
3141         /**
3142          * Static function to get a template
3143          * Can be overridden via ParserOptions::setTemplateCallback().
3144          */
3145         static function statelessFetchTemplate( $title, $parser=false ) {
3146                 $text = $skip = false;
3147                 $finalTitle = $title;
3148                 $deps = array();
3149
3150                 // Loop to fetch the article, with up to 1 redirect
3151                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3152                         # Give extensions a chance to select the revision instead
3153                         $id = false; // Assume current
3154                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3155
3156                         if( $skip ) {
3157                                 $text = false;
3158                                 $deps[] = array(
3159                                         'title' => $title,
3160                                         'page_id' => $title->getArticleID(),
3161                                         'rev_id' => null );
3162                                 break;
3163                         }
3164                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3165                         $rev_id = $rev ? $rev->getId() : 0;
3166                         // If there is no current revision, there is no page
3167                         if( $id === false && !$rev ) {
3168                                 $linkCache = LinkCache::singleton();
3169                                 $linkCache->addBadLinkObj( $title );
3170                         }
3171
3172                         $deps[] = array(
3173                                 'title' => $title,
3174                                 'page_id' => $title->getArticleID(),
3175                                 'rev_id' => $rev_id );
3176
3177                         if( $rev ) {
3178                                 $text = $rev->getText();
3179                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3180                                 global $wgContLang;
3181                                 $message = $wgContLang->lcfirst( $title->getText() );
3182                                 $text = wfMsgForContentNoTrans( $message );
3183                                 if( wfEmptyMsg( $message, $text ) ) {
3184                                         $text = false;
3185                                         break;
3186                                 }
3187                         } else {
3188                                 break;
3189                         }
3190                         if ( $text === false ) {
3191                                 break;
3192                         }
3193                         // Redirect?
3194                         $finalTitle = $title;
3195                         $title = Title::newFromRedirect( $text );
3196                 }
3197                 return array(
3198                         'text' => $text,
3199                         'finalTitle' => $finalTitle,
3200                         'deps' => $deps );
3201         }
3202
3203         /**
3204          * Transclude an interwiki link.
3205          */
3206         function interwikiTransclude( $title, $action ) {
3207                 global $wgEnableScaryTranscluding;
3208
3209                 if (!$wgEnableScaryTranscluding)
3210                         return wfMsg('scarytranscludedisabled');
3211
3212                 $url = $title->getFullUrl( "action=$action" );
3213
3214                 if (strlen($url) > 255)
3215                         return wfMsg('scarytranscludetoolong');
3216                 return $this->fetchScaryTemplateMaybeFromCache($url);
3217         }
3218
3219         function fetchScaryTemplateMaybeFromCache($url) {
3220                 global $wgTranscludeCacheExpiry;
3221                 $dbr = wfGetDB(DB_SLAVE);
3222                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3223                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3224                                 array('tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3225                 if ($obj) {
3226                         return $obj->tc_contents;
3227                 }
3228
3229                 $text = Http::get($url);
3230                 if (!$text)
3231                         return wfMsg('scarytranscludefailed', $url);
3232
3233                 $dbw = wfGetDB(DB_MASTER);
3234                 $dbw->replace('transcache', array('tc_url'), array(
3235                         'tc_url' => $url,
3236                         'tc_time' => $dbw->timestamp( time() ),
3237                         'tc_contents' => $text));
3238                 return $text;
3239         }
3240
3241
3242         /**
3243          * Triple brace replacement -- used for template arguments
3244          * @private
3245          */
3246         function argSubstitution( $piece, $frame ) {
3247                 wfProfileIn( __METHOD__ );
3248
3249                 $error = false;
3250                 $parts = $piece['parts'];
3251                 $nameWithSpaces = $frame->expand( $piece['title'] );
3252                 $argName = trim( $nameWithSpaces );
3253                 $object = false;
3254                 $text = $frame->getArgument( $argName );
3255                 if (  $text === false && $parts->getLength() > 0
3256                   && (
3257                     $this->ot['html']
3258                     || $this->ot['pre']
3259                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3260                   )
3261                 ) {
3262                         # No match in frame, use the supplied default
3263                         $object = $parts->item( 0 )->getChildren();
3264                 }
3265                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3266                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3267                         $this->limitationWarn( 'post-expand-template-argument' );
3268                 }
3269
3270                 if ( $text === false && $object === false ) {
3271                         # No match anywhere
3272                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3273                 }
3274                 if ( $error !== false ) {
3275                         $text .= $error;
3276                 }
3277                 if ( $object !== false ) {
3278                         $ret = array( 'object' => $object );
3279                 } else {
3280                         $ret = array( 'text' => $text );
3281                 }
3282
3283                 wfProfileOut( __METHOD__ );
3284                 return $ret;
3285         }
3286
3287         /**
3288          * Return the text to be used for a given extension tag.
3289          * This is the ghost of strip().
3290          *
3291          * @param array $params Associative array of parameters:
3292          *     name       PPNode for the tag name
3293          *     attr       PPNode for unparsed text where tag attributes are thought to be
3294          *     attributes Optional associative array of parsed attributes
3295          *     inner      Contents of extension element
3296          *     noClose    Original text did not have a close tag
3297          * @param PPFrame $frame
3298          */
3299         function extensionSubstitution( $params, $frame ) {
3300                 global $wgRawHtml, $wgContLang;
3301
3302                 $name = $frame->expand( $params['name'] );
3303                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3304                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3305                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3306
3307                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3308                         ( $this->ot['html'] || $this->ot['pre'] );
3309                 if ( $isFunctionTag ) {
3310                         $markerType = 'none';
3311                 } else {
3312                         $markerType = 'general';
3313                 }
3314                 if ( $this->ot['html'] || $isFunctionTag ) {
3315                         $name = strtolower( $name );
3316                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3317                         if ( isset( $params['attributes'] ) ) {
3318                                 $attributes = $attributes + $params['attributes'];
3319                         }
3320
3321                         if( isset( $this->mTagHooks[$name] ) ) {
3322                                 # Workaround for PHP bug 35229 and similar
3323                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3324                                         throw new MWException( "Tag hook for $name is not callable\n" );
3325                                 }
3326                                 $output = call_user_func_array( $this->mTagHooks[$name],
3327                                         array( $content, $attributes, $this, $frame ) );
3328                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3329                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3330                                 if( !is_callable( $callback ) )
3331                                         throw new MWException( "Tag hook for $name is not callable\n" );
3332
3333                                 $output = call_user_func_array( $callback,
3334                                         array( &$this, $frame, $content, $attributes ) );
3335                         } else {
3336                                 $output = '<span class="error">Invalid tag extension name: ' .
3337                                         htmlspecialchars( $name ) . '</span>';
3338                         }
3339
3340                         if ( is_array( $output ) ) {
3341                                 // Extract flags to local scope (to override $markerType)
3342                                 $flags = $output;
3343                                 $output = $flags[0];
3344                                 unset( $flags[0] );
3345                                 extract( $flags );
3346                         }
3347                 } else {
3348                         if ( is_null( $attrText ) ) {
3349                                 $attrText = '';
3350                         }
3351                         if ( isset( $params['attributes'] ) ) {
3352                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3353                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3354                                                 htmlspecialchars( $attrValue ) . '"';
3355                                 }
3356                         }
3357                         if ( $content === null ) {
3358                                 $output = "<$name$attrText/>";
3359                         } else {
3360                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3361                                 $output = "<$name$attrText>$content$close";
3362                         }
3363                 }
3364
3365                 if( $markerType === 'none' ) {
3366                         return $output;
3367                 } elseif ( $markerType === 'nowiki' ) {
3368                         $this->mStripState->nowiki->setPair( $marker, $output );
3369                 } elseif ( $markerType === 'general' ) {
3370                         $this->mStripState->general->setPair( $marker, $output );
3371                 } else {
3372                         throw new MWException( __METHOD__.': invalid marker type' );
3373                 }
3374                 return $marker;
3375         }
3376
3377         /**
3378          * Increment an include size counter
3379          *
3380          * @param string $type The type of expansion
3381          * @param integer $size The size of the text
3382          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3383          */
3384         function incrementIncludeSize( $type, $size ) {
3385                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3386                         return false;
3387                 } else {
3388                         $this->mIncludeSizes[$type] += $size;
3389                         return true;
3390                 }
3391         }
3392
3393         /**
3394          * Increment the expensive function count
3395          *
3396          * @return boolean False if the limit has been exceeded
3397          */
3398         function incrementExpensiveFunctionCount() {
3399                 global $wgExpensiveParserFunctionLimit;
3400                 $this->mExpensiveFunctionCount++;
3401                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3402                         return true;
3403                 }
3404                 return false;
3405         }
3406
3407         /**
3408          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3409          * Fills $this->mDoubleUnderscores, returns the modified text
3410          */
3411         function doDoubleUnderscore( $text ) {
3412                 wfProfileIn( __METHOD__ );
3413
3414                 // The position of __TOC__ needs to be recorded
3415                 $mw = MagicWord::get( 'toc' );
3416                 if( $mw->match( $text ) ) {
3417                         $this->mShowToc = true;
3418                         $this->mForceTocPosition = true;
3419
3420                         // Set a placeholder. At the end we'll fill it in with the TOC.
3421                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3422
3423                         // Only keep the first one.
3424                         $text = $mw->replace( '', $text );
3425                 }
3426
3427                 // Now match and remove the rest of them
3428                 $mwa = MagicWord::getDoubleUnderscoreArray();
3429                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3430
3431                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3432                         $this->mOutput->mNoGallery = true;
3433                 }
3434                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3435                         $this->mShowToc = false;
3436                 }
3437                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3438                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3439                         $this->addTrackingCategory( 'hidden-category-category' );
3440                 }
3441                 # (bug 8068) Allow control over whether robots index a page.
3442                 #
3443                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3444                 # is not desirable, the last one on the page should win.
3445                 if( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3446                         $this->mOutput->setIndexPolicy( 'noindex' );
3447                         $this->addTrackingCategory( 'noindex-category' );
3448                 }
3449                 if( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ){
3450                         $this->mOutput->setIndexPolicy( 'index' );
3451                         $this->addTrackingCategory( 'index-category' );
3452                 }
3453
3454                 wfProfileOut( __METHOD__ );
3455                 return $text;
3456         }
3457
3458         /**
3459          * Add a tracking category, getting the title from a system message,
3460          * or print a debug message if the title is invalid.
3461          * @param $msg String message key
3462          * @return Bool whether the addition was successful
3463          */
3464         protected function addTrackingCategory( $msg ){
3465                 $cat = wfMsgForContent( $msg );
3466
3467                 # Allow tracking categories to be disabled by setting them to "-"
3468                 if( $cat === '-' ) return false;
3469
3470                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3471                 if ( $containerCategory ) {
3472                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3473                         return true;
3474                 } else {
3475                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3476                         return false;
3477                 }
3478         }
3479
3480         /**
3481          * This function accomplishes several tasks:
3482          * 1) Auto-number headings if that option is enabled
3483          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3484          * 3) Add a Table of contents on the top for users who have enabled the option
3485          * 4) Auto-anchor headings
3486          *
3487          * It loops through all headlines, collects the necessary data, then splits up the
3488          * string and re-inserts the newly formatted headlines.
3489          *
3490          * @param string $text
3491          * @param string $origText Original, untouched wikitext
3492          * @param boolean $isMain
3493          * @private
3494          */
3495         function formatHeadings( $text, $origText, $isMain=true ) {
3496                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3497
3498                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3499                 $showEditLink = $this->mOptions->getEditSection();
3500
3501                 // Do not call quickUserCan unless necessary
3502                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3503                         $showEditLink = 0;
3504                 }
3505
3506                 # Inhibit editsection links if requested in the page
3507                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3508                         $showEditLink = 0;
3509                 }
3510
3511                 # Get all headlines for numbering them and adding funky stuff like [edit]
3512                 # links - this is for later, but we need the number of headlines right now
3513                 $matches = array();
3514                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3515
3516                 # if there are fewer than 4 headlines in the article, do not show TOC
3517                 # unless it's been explicitly enabled.
3518                 $enoughToc = $this->mShowToc &&
3519                         (($numMatches >= 4) || $this->mForceTocPosition);
3520
3521                 # Allow user to stipulate that a page should have a "new section"
3522                 # link added via __NEWSECTIONLINK__
3523                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3524                         $this->mOutput->setNewSection( true );
3525                 }
3526
3527                 # Allow user to remove the "new section"
3528                 # link via __NONEWSECTIONLINK__
3529                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3530                         $this->mOutput->hideNewSection( true );
3531                 }
3532
3533                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3534                 # override above conditions and always show TOC above first header
3535                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3536                         $this->mShowToc = true;
3537                         $enoughToc = true;
3538                 }
3539
3540                 # We need this to perform operations on the HTML
3541                 $sk = $this->mOptions->getSkin();
3542
3543                 # headline counter
3544                 $headlineCount = 0;
3545                 $numVisible = 0;
3546
3547                 # Ugh .. the TOC should have neat indentation levels which can be
3548                 # passed to the skin functions. These are determined here
3549                 $toc = '';
3550                 $full = '';
3551                 $head = array();
3552                 $sublevelCount = array();
3553                 $levelCount = array();
3554                 $toclevel = 0;
3555                 $level = 0;
3556                 $prevlevel = 0;
3557                 $toclevel = 0;
3558                 $prevtoclevel = 0;
3559                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3560                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3561                 $oldType = $this->mOutputType;
3562                 $this->setOutputType( self::OT_WIKI );
3563                 $frame = $this->getPreprocessor()->newFrame();
3564                 $root = $this->preprocessToDom( $origText );
3565                 $node = $root->getFirstChild();
3566                 $byteOffset = 0;
3567                 $tocraw = array();
3568
3569                 foreach( $matches[3] as $headline ) {
3570                         $isTemplate = false;
3571                         $titleText = false;
3572                         $sectionIndex = false;
3573                         $numbering = '';
3574                         $markerMatches = array();
3575                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3576                                 $serial = $markerMatches[1];
3577                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3578                                 $isTemplate = ($titleText != $baseTitleText);
3579                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3580                         }
3581
3582                         if( $toclevel ) {
3583                                 $prevlevel = $level;
3584                                 $prevtoclevel = $toclevel;
3585                         }
3586                         $level = $matches[1][$headlineCount];
3587
3588                         if ( $level > $prevlevel ) {
3589                                 # Increase TOC level
3590                                 $toclevel++;
3591                                 $sublevelCount[$toclevel] = 0;
3592                                 if( $toclevel<$wgMaxTocLevel ) {
3593                                         $prevtoclevel = $toclevel;
3594                                         $toc .= $sk->tocIndent();
3595                                         $numVisible++;
3596                                 }
3597                         }
3598                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3599                                 # Decrease TOC level, find level to jump to
3600
3601                                 for ($i = $toclevel; $i > 0; $i--) {
3602                                         if ( $levelCount[$i] == $level ) {
3603                                                 # Found last matching level
3604                                                 $toclevel = $i;
3605                                                 break;
3606                                         }
3607                                         elseif ( $levelCount[$i] < $level ) {
3608                                                 # Found first matching level below current level
3609                                                 $toclevel = $i + 1;
3610                                                 break;
3611                                         }
3612                                 }
3613                                 if( $i == 0 ) $toclevel = 1;
3614                                 if( $toclevel<$wgMaxTocLevel ) {
3615                                         if($prevtoclevel < $wgMaxTocLevel) {
3616                                                 # Unindent only if the previous toc level was shown :p
3617                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3618                                                 $prevtoclevel = $toclevel;
3619                                         } else {
3620                                                 $toc .= $sk->tocLineEnd();
3621                                         }
3622                                 }
3623                         }
3624                         else {
3625                                 # No change in level, end TOC line
3626                                 if( $toclevel<$wgMaxTocLevel ) {
3627                                         $toc .= $sk->tocLineEnd();
3628                                 }
3629                         }
3630
3631                         $levelCount[$toclevel] = $level;
3632
3633                         # count number of headlines for each level
3634                         @$sublevelCount[$toclevel]++;
3635                         $dot = 0;
3636                         for( $i = 1; $i <= $toclevel; $i++ ) {
3637                                 if( !empty( $sublevelCount[$i] ) ) {
3638                                         if( $dot ) {
3639                                                 $numbering .= '.';
3640                                         }
3641                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3642                                         $dot = 1;
3643                                 }
3644                         }
3645
3646                         # The safe header is a version of the header text safe to use for links
3647                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3648                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3649
3650                         # Remove link placeholders by the link text.
3651                         #     <!--LINK number-->
3652                         # turns into
3653                         #     link text with suffix
3654                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3655
3656                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3657                         $tocline = preg_replace(
3658                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3659                                 array( '',                          '<$1>'),
3660                                 $safeHeadline
3661                         );
3662                         $tocline = trim( $tocline );
3663
3664                         # For the anchor, strip out HTML-y stuff period
3665                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3666                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3667                         $safeHeadline = trim( $safeHeadline );
3668
3669                         # Save headline for section edit hint before it's escaped
3670                         $headlineHint = $safeHeadline;
3671
3672                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3673                                 # For reverse compatibility, provide an id that's
3674                                 # HTML4-compatible, like we used to.
3675                                 #
3676                                 # It may be worth noting, academically, that it's possible for
3677                                 # the legacy anchor to conflict with a non-legacy headline
3678                                 # anchor on the page.  In this case likely the "correct" thing
3679                                 # would be to either drop the legacy anchors or make sure
3680                                 # they're numbered first.  However, this would require people
3681                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3682                                 # manually, so let's not bother worrying about it.
3683                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3684                                         array( 'noninitial', 'legacy' ) );
3685                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3686
3687                                 if ( $legacyHeadline == $safeHeadline ) {
3688                                         # No reason to have both (in fact, we can't)
3689                                         $legacyHeadline = false;
3690                                 }
3691                         } else {
3692                                 $legacyHeadline = false;
3693                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3694                                         'noninitial' );
3695                         }
3696
3697                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3698                         # Does this apply to Unicode characters?  Because we aren't
3699                         # handling those here.
3700                         $arrayKey = strtolower( $safeHeadline );
3701                         if ( $legacyHeadline === false ) {
3702                                 $legacyArrayKey = false;
3703                         } else {
3704                                 $legacyArrayKey = strtolower( $legacyHeadline );
3705                         }
3706
3707                         # count how many in assoc. array so we can track dupes in anchors
3708                         if ( isset( $refers[$arrayKey] ) ) {
3709                                 $refers[$arrayKey]++;
3710                         } else {
3711                                 $refers[$arrayKey] = 1;
3712                         }
3713                         if ( isset( $refers[$legacyArrayKey] ) ) {
3714                                 $refers[$legacyArrayKey]++;
3715                         } else {
3716                                 $refers[$legacyArrayKey] = 1;
3717                         }
3718
3719                         # Don't number the heading if it is the only one (looks silly)
3720                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3721                                 # the two are different if the line contains a link
3722                                 $headline = $numbering . ' ' . $headline;
3723                         }
3724
3725                         # Create the anchor for linking from the TOC to the section
3726                         $anchor = $safeHeadline;
3727                         $legacyAnchor = $legacyHeadline;
3728                         if ( $refers[$arrayKey] > 1 ) {
3729                                 $anchor .= '_' . $refers[$arrayKey];
3730                         }
3731                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3732                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3733                         }
3734                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3735                                 $toc .= $sk->tocLine($anchor, $tocline,
3736                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3737                         }
3738
3739                         # Add the section to the section tree
3740                         # Find the DOM node for this header
3741                         while ( $node && !$isTemplate ) {
3742                                 if ( $node->getName() === 'h' ) {
3743                                         $bits = $node->splitHeading();
3744                                         if ( $bits['i'] == $sectionIndex )
3745                                                 break;
3746                                 }
3747                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3748                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3749                                 $node = $node->getNextSibling();
3750                         }
3751                         $tocraw[] = array(
3752                                 'toclevel' => $toclevel,
3753                                 'level' => $level,
3754                                 'line' => $tocline,
3755                                 'number' => $numbering,
3756                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3757                                 'fromtitle' => $titleText,
3758                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3759                                 'anchor' => $anchor,
3760                         );
3761
3762                         # give headline the correct <h#> tag
3763                         if( $showEditLink && $sectionIndex !== false ) {
3764                                 if( $isTemplate ) {
3765                                         # Put a T flag in the section identifier, to indicate to extractSections()
3766                                         # that sections inside <includeonly> should be counted.
3767                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3768                                 } else {
3769                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3770                                 }
3771                         } else {
3772                                 $editlink = '';
3773                         }
3774                         $head[$headlineCount] = $sk->makeHeadline( $level,
3775                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3776                                 $editlink, $legacyAnchor );
3777
3778                         $headlineCount++;
3779                 }
3780
3781                 $this->setOutputType( $oldType );
3782
3783                 # Never ever show TOC if no headers
3784                 if( $numVisible < 1 ) {
3785                         $enoughToc = false;
3786                 }
3787
3788                 if( $enoughToc ) {
3789                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3790                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3791                         }
3792                         $toc = $sk->tocList( $toc );
3793                         $this->mOutput->setTOCHTML( $toc );
3794                 }
3795
3796                 if ( $isMain ) {
3797                         $this->mOutput->setSections( $tocraw );
3798                 }
3799
3800                 # split up and insert constructed headlines
3801
3802                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3803                 $i = 0;
3804
3805                 foreach( $blocks as $block ) {
3806                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3807                                 # This is the [edit] link that appears for the top block of text when
3808                                 # section editing is enabled
3809
3810                                 # Disabled because it broke block formatting
3811                                 # For example, a bullet point in the top line
3812                                 # $full .= $sk->editSectionLink(0);
3813                         }
3814                         $full .= $block;
3815                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3816                                 # Top anchor now in skin
3817                                 $full = $full.$toc;
3818                         }
3819
3820                         if( !empty( $head[$i] ) ) {
3821                                 $full .= $head[$i];
3822                         }
3823                         $i++;
3824                 }
3825                 if( $this->mForceTocPosition ) {
3826                         return str_replace( '<!--MWTOC-->', $toc, $full );
3827                 } else {
3828                         return $full;
3829                 }
3830         }
3831
3832         /**
3833          * Merge $tree2 into $tree1 by replacing the section with index
3834          * $section in $tree1 and its descendants with the sections in $tree2.
3835          * Note that in the returned section tree, only the 'index' and
3836          * 'byteoffset' fields are guaranteed to be correct.
3837          * @param $tree1 array Section tree from ParserOutput::getSectons()
3838         *  @param $tree2 array Section tree
3839          * @param $section int Section index
3840          * @param $title Title Title both section trees come from
3841          * @param $len2 int Length of the original wikitext for $tree2
3842          * @return array Merged section tree
3843          */
3844         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3845                 global $wgContLang;
3846                 $newTree = array();
3847                 $targetLevel = false;
3848                 $merged = false;
3849                 $lastLevel = 1;
3850                 $nextIndex = 1;
3851                 $numbering = array( 0 );
3852                 $titletext = $title->getPrefixedDBkey();
3853                 foreach ( $tree1 as $s ) {
3854                         if ( $targetLevel !== false ) {
3855                                 if ( $s['level'] <= $targetLevel )
3856                                         // We've skipped enough
3857                                         $targetLevel = false;
3858                                 else
3859                                         continue;
3860                         }
3861                         if ( $s['index'] != $section ||
3862                                         $s['fromtitle'] != $titletext ) {
3863                                 self::incrementNumbering( $numbering,
3864                                         $s['toclevel'], $lastLevel );
3865
3866                                 // Rewrite index, byteoffset and number
3867                                 if ( $s['fromtitle'] == $titletext ) {
3868                                         $s['index'] = $nextIndex++;
3869                                         if ( $merged )
3870                                                 $s['byteoffset'] += $len2;
3871                                 }
3872                                 $s['number']  = implode( '.', array_map(
3873                                         array( $wgContLang, 'formatnum' ),
3874                                         $numbering ) );
3875                                 $lastLevel = $s['toclevel'];
3876                                 $newTree[] = $s;
3877                         } else {
3878                                 // We're at $section
3879                                 // Insert sections from $tree2 here
3880                                 foreach ( $tree2 as $s2 ) {
3881                                         // Rewrite the fields in $s2
3882                                         // before inserting it
3883                                         $s2['toclevel'] += $s['toclevel'] - 1;
3884                                         $s2['level'] += $s['level'] - 1;
3885                                         $s2['index'] = $nextIndex++;
3886                                         $s2['byteoffset'] += $s['byteoffset'];
3887
3888                                         self::incrementNumbering( $numbering,
3889                                                 $s2['toclevel'], $lastLevel );
3890                                         $s2['number']  = implode( '.', array_map(
3891                                                 array( $wgContLang, 'formatnum' ),
3892                                                 $numbering ) );
3893                                         $lastLevel = $s2['toclevel'];
3894                                         $newTree[] = $s2;
3895                                 }
3896                                 // Skip all descendants of $section in $tree1
3897                                 $targetLevel = $s['level'];
3898                                 $merged = true;
3899                         }
3900                 }
3901                 return $newTree;
3902         }
3903
3904         /**
3905          * Increment a section number. Helper function for mergeSectionTrees()
3906          * @param $number array Array representing a section number
3907          * @param $level int Current TOC level (depth)
3908          * @param $lastLevel int Level of previous TOC entry
3909          */
3910         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3911                 if ( $level > $lastLevel )
3912                         $number[$level - 1] = 1;
3913                 else if ( $level < $lastLevel ) {
3914                         foreach ( $number as $key => $unused )
3915                                 if ( $key >= $level )
3916                                         unset( $number[$key] );
3917                         $number[$level - 1]++;
3918                 } else
3919                         $number[$level - 1]++;
3920         }
3921
3922         /**
3923          * Transform wiki markup when saving a page by doing \r\n -> \n
3924          * conversion, substitting signatures, {{subst:}} templates, etc.
3925          *
3926          * @param string $text the text to transform
3927          * @param Title &$title the Title object for the current article
3928          * @param User $user the User object describing the current user
3929          * @param ParserOptions $options parsing options
3930          * @param bool $clearState whether to clear the parser state first
3931          * @return string the altered wiki markup
3932          * @public
3933          */
3934         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3935                 $this->mOptions = $options;
3936                 $this->setTitle( $title );
3937                 $this->setOutputType( self::OT_WIKI );
3938
3939                 if ( $clearState ) {
3940                         $this->clearState();
3941                 }
3942
3943                 $pairs = array(
3944                         "\r\n" => "\n",
3945                 );
3946                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3947                 $text = $this->pstPass2( $text, $user );
3948                 $text = $this->mStripState->unstripBoth( $text );
3949                 return $text;
3950         }
3951
3952         /**
3953          * Pre-save transform helper function
3954          * @private
3955          */
3956         function pstPass2( $text, $user ) {
3957                 global $wgContLang, $wgLocaltimezone;
3958
3959                 /* Note: This is the timestamp saved as hardcoded wikitext to
3960                  * the database, we use $wgContLang here in order to give
3961                  * everyone the same signature and use the default one rather
3962                  * than the one selected in each user's preferences.
3963                  *
3964                  * (see also bug 12815)
3965                  */
3966                 $ts = $this->mOptions->getTimestamp();
3967                 if ( isset( $wgLocaltimezone ) ) {
3968                         $tz = $wgLocaltimezone;
3969                 } else {
3970                         $tz = date_default_timezone_get();
3971                 }
3972
3973                 $unixts = wfTimestamp( TS_UNIX, $ts );
3974                 $oldtz = date_default_timezone_get();
3975                 date_default_timezone_set( $tz );
3976                 $ts = date( 'YmdHis', $unixts );
3977                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
3978
3979                 /* Allow translation of timezones trough wiki. date() can return
3980                  * whatever crap the system uses, localised or not, so we cannot
3981                  * ship premade translations.
3982                  */
3983                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
3984                 $value = wfMsgForContent( $key );
3985                 if ( !wfEmptyMsg( $key, $value ) ) $tzMsg = $value;
3986
3987                 date_default_timezone_set( $oldtz );
3988
3989                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
3990
3991                 # Variable replacement
3992                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3993                 $text = $this->replaceVariables( $text );
3994
3995                 # Signatures
3996                 $sigText = $this->getUserSig( $user );
3997                 $text = strtr( $text, array(
3998                         '~~~~~' => $d,
3999                         '~~~~' => "$sigText $d",
4000                         '~~~' => $sigText
4001                 ) );
4002
4003                 # Context links: [[|name]] and [[name (context)|]]
4004                 #
4005                 global $wgLegalTitleChars;
4006                 $tc = "[$wgLegalTitleChars]";
4007                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4008
4009                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4010                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4011                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4012                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4013
4014                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4015                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4016                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4017                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4018
4019                 $t = $this->mTitle->getText();
4020                 $m = array();
4021                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4022                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4023                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4024                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4025                 } else {
4026                         # if there's no context, don't bother duplicating the title
4027                         $text = preg_replace( $p2, '[[\\1]]', $text );
4028                 }
4029
4030                 # Trim trailing whitespace
4031                 $text = rtrim( $text );
4032
4033                 return $text;
4034         }
4035
4036         /**
4037          * Fetch the user's signature text, if any, and normalize to
4038          * validated, ready-to-insert wikitext.
4039          * If you have pre-fetched the nickname or the fancySig option, you can
4040          * specify them here to save a database query.
4041          *
4042          * @param User $user
4043          * @return string
4044          */
4045         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4046                 global $wgMaxSigChars;
4047
4048                 $username = $user->getName();
4049
4050                 // If not given, retrieve from the user object.
4051                 if ( $nickname === false )
4052                         $nickname = $user->getOption( 'nickname' );
4053
4054                 if ( is_null( $fancySig ) )
4055                         $fancySig = $user->getBoolOption( 'fancysig' );
4056
4057                 $nickname = $nickname == null ? $username : $nickname;
4058
4059                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4060                         $nickname = $username;
4061                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4062                 } elseif( $fancySig !== false ) {
4063                         # Sig. might contain markup; validate this
4064                         if( $this->validateSig( $nickname ) !== false ) {
4065                                 # Validated; clean up (if needed) and return it
4066                                 return $this->cleanSig( $nickname, true );
4067                         } else {
4068                                 # Failed to validate; fall back to the default
4069                                 $nickname = $username;
4070                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4071                         }
4072                 }
4073
4074                 // Make sure nickname doesnt get a sig in a sig
4075                 $nickname = $this->cleanSigInSig( $nickname );
4076
4077                 # If we're still here, make it a link to the user page
4078                 $userText = wfEscapeWikiText( $username );
4079                 $nickText = wfEscapeWikiText( $nickname );
4080                 if ( $user->isAnon() )  {
4081                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4082                 } else {
4083                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4084                 }
4085         }
4086
4087         /**
4088          * Check that the user's signature contains no bad XML
4089          *
4090          * @param string $text
4091          * @return mixed An expanded string, or false if invalid.
4092          */
4093         function validateSig( $text ) {
4094                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4095         }
4096
4097         /**
4098          * Clean up signature text
4099          *
4100          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4101          * 2) Substitute all transclusions
4102          *
4103          * @param string $text
4104          * @param $parsing Whether we're cleaning (preferences save) or parsing
4105          * @return string Signature text
4106          */
4107         function cleanSig( $text, $parsing = false ) {
4108                 if ( !$parsing ) {
4109                         global $wgTitle;
4110                         $this->clearState();
4111                         $this->setTitle( $wgTitle );
4112                         $this->mOptions = new ParserOptions;
4113                         $this->setOutputType = self::OT_PREPROCESS;
4114                 }
4115
4116                 # Option to disable this feature
4117                 if ( !$this->mOptions->getCleanSignatures() ) {
4118                         return $text;
4119                 }
4120
4121                 # FIXME: regex doesn't respect extension tags or nowiki
4122                 #  => Move this logic to braceSubstitution()
4123                 $substWord = MagicWord::get( 'subst' );
4124                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4125                 $substText = '{{' . $substWord->getSynonym( 0 );
4126
4127                 $text = preg_replace( $substRegex, $substText, $text );
4128                 $text = $this->cleanSigInSig( $text );
4129                 $dom = $this->preprocessToDom( $text );
4130                 $frame = $this->getPreprocessor()->newFrame();
4131                 $text = $frame->expand( $dom );
4132
4133                 if ( !$parsing ) {
4134                         $text = $this->mStripState->unstripBoth( $text );
4135                 }
4136
4137                 return $text;
4138         }
4139
4140         /**
4141          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4142          * @param string $text
4143          * @return string Signature text with /~{3,5}/ removed
4144          */
4145         function cleanSigInSig( $text ) {
4146                 $text = preg_replace( '/~{3,5}/', '', $text );
4147                 return $text;
4148         }
4149
4150         /**
4151          * Set up some variables which are usually set up in parse()
4152          * so that an external function can call some class members with confidence
4153          * @public
4154          */
4155         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4156                 $this->setTitle( $title );
4157                 $this->mOptions = $options;
4158                 $this->setOutputType( $outputType );
4159                 if ( $clearState ) {
4160                         $this->clearState();
4161                 }
4162         }
4163
4164         /**
4165          * Wrapper for preprocess()
4166          *
4167          * @param string $text the text to preprocess
4168          * @param ParserOptions $options  options
4169          * @return string
4170          * @public
4171          */
4172         function transformMsg( $text, $options ) {
4173                 global $wgTitle;
4174                 static $executing = false;
4175
4176                 # Guard against infinite recursion
4177                 if ( $executing ) {
4178                         return $text;
4179                 }
4180                 $executing = true;
4181
4182                 wfProfileIn(__METHOD__);
4183                 $text = $this->preprocess( $text, $wgTitle, $options );
4184
4185                 $executing = false;
4186                 wfProfileOut(__METHOD__);
4187                 return $text;
4188         }
4189
4190         /**
4191          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4192          * The callback should have the following form:
4193          *    function myParserHook( $text, $params, &$parser ) { ... }
4194          *
4195          * Transform and return $text. Use $parser for any required context, e.g. use
4196          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4197          *
4198          * @public
4199          *
4200          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4201          * @param mixed $callback The callback function (and object) to use for the tag
4202          *
4203          * @return The old value of the mTagHooks array associated with the hook
4204          */
4205         function setHook( $tag, $callback ) {
4206                 $tag = strtolower( $tag );
4207                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4208                 $this->mTagHooks[$tag] = $callback;
4209                 if( !in_array( $tag, $this->mStripList ) ) {
4210                         $this->mStripList[] = $tag;
4211                 }
4212
4213                 return $oldVal;
4214         }
4215
4216         function setTransparentTagHook( $tag, $callback ) {
4217                 $tag = strtolower( $tag );
4218                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4219                 $this->mTransparentTagHooks[$tag] = $callback;
4220
4221                 return $oldVal;
4222         }
4223
4224         /**
4225          * Remove all tag hooks
4226          */
4227         function clearTagHooks() {
4228                 $this->mTagHooks = array();
4229                 $this->mStripList = $this->mDefaultStripList;
4230         }
4231
4232         /**
4233          * Create a function, e.g. {{sum:1|2|3}}
4234          * The callback function should have the form:
4235          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4236          *
4237          * Or with SFH_OBJECT_ARGS:
4238          *    function myParserFunction( $parser, $frame, $args ) { ... }
4239          *
4240          * The callback may either return the text result of the function, or an array with the text
4241          * in element 0, and a number of flags in the other elements. The names of the flags are
4242          * specified in the keys. Valid flags are:
4243          *   found                     The text returned is valid, stop processing the template. This
4244          *                             is on by default.
4245          *   nowiki                    Wiki markup in the return value should be escaped
4246          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4247          *
4248          * @public
4249          *
4250          * @param string $id The magic word ID
4251          * @param mixed $callback The callback function (and object) to use
4252          * @param integer $flags a combination of the following flags:
4253          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4254          *
4255          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4256          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4257          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4258          *     the arguments, and to control the way they are expanded.
4259          *
4260          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4261          *     arguments, for instance:
4262          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4263          *
4264          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4265          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4266          *     working if/when this is changed.
4267          *
4268          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4269          *     expansion.
4270          *
4271          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4272          *     about the methods available in PPFrame and PPNode.
4273          *
4274          * @return The old callback function for this name, if any
4275          */
4276         function setFunctionHook( $id, $callback, $flags = 0 ) {
4277                 global $wgContLang;
4278
4279                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4280                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4281
4282                 # Add to function cache
4283                 $mw = MagicWord::get( $id );
4284                 if( !$mw )
4285                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4286
4287                 $synonyms = $mw->getSynonyms();
4288                 $sensitive = intval( $mw->isCaseSensitive() );
4289
4290                 foreach ( $synonyms as $syn ) {
4291                         # Case
4292                         if ( !$sensitive ) {
4293                                 $syn = $wgContLang->lc( $syn );
4294                         }
4295                         # Add leading hash
4296                         if ( !( $flags & SFH_NO_HASH ) ) {
4297                                 $syn = '#' . $syn;
4298                         }
4299                         # Remove trailing colon
4300                         if ( substr( $syn, -1, 1 ) === ':' ) {
4301                                 $syn = substr( $syn, 0, -1 );
4302                         }
4303                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4304                 }
4305                 return $oldVal;
4306         }
4307
4308         /**
4309          * Get all registered function hook identifiers
4310          *
4311          * @return array
4312          */
4313         function getFunctionHooks() {
4314                 return array_keys( $this->mFunctionHooks );
4315         }
4316
4317         /**
4318          * Create a tag function, e.g. <test>some stuff</test>.
4319          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4320          * Unlike parser functions, their content is not preprocessed.
4321          */
4322         function setFunctionTagHook( $tag, $callback, $flags ) {
4323                 $tag = strtolower( $tag );
4324                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4325                         $this->mFunctionTagHooks[$tag] : null;
4326                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4327
4328                 if( !in_array( $tag, $this->mStripList ) ) {
4329                         $this->mStripList[] = $tag;
4330                 }
4331
4332                 return $old;
4333         }
4334
4335         /**
4336          * FIXME: update documentation. makeLinkObj() is deprecated.
4337          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4338          * Placeholders created in Skin::makeLinkObj()
4339          * Returns an array of link CSS classes, indexed by PDBK.
4340          */
4341         function replaceLinkHolders( &$text, $options = 0 ) {
4342                 return $this->mLinkHolders->replace( $text );
4343         }
4344
4345         /**
4346          * Replace <!--LINK--> link placeholders with plain text of links
4347          * (not HTML-formatted).
4348          * @param string $text
4349          * @return string
4350          */
4351         function replaceLinkHoldersText( $text ) {
4352                 return $this->mLinkHolders->replaceText( $text );
4353         }
4354
4355         /**
4356          * Renders an image gallery from a text with one line per image.
4357          * text labels may be given by using |-style alternative text. E.g.
4358          *   Image:one.jpg|The number "1"
4359          *   Image:tree.jpg|A tree
4360          * given as text will return the HTML of a gallery with two images,
4361          * labeled 'The number "1"' and
4362          * 'A tree'.
4363          */
4364         function renderImageGallery( $text, $params ) {
4365                 $ig = new ImageGallery();
4366                 $ig->setContextTitle( $this->mTitle );
4367                 $ig->setShowBytes( false );
4368                 $ig->setShowFilename( false );
4369                 $ig->setParser( $this );
4370                 $ig->setHideBadImages();
4371                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4372                 $ig->useSkin( $this->mOptions->getSkin() );
4373                 $ig->mRevisionId = $this->mRevisionId;
4374
4375                 if( isset( $params['caption'] ) ) {
4376                         $caption = $params['caption'];
4377                         $caption = htmlspecialchars( $caption );
4378                         $caption = $this->replaceInternalLinks( $caption );
4379                         $ig->setCaptionHtml( $caption );
4380                 }
4381                 if( isset( $params['perrow'] ) ) {
4382                         $ig->setPerRow( $params['perrow'] );
4383                 }
4384                 if( isset( $params['widths'] ) ) {
4385                         $ig->setWidths( $params['widths'] );
4386                 }
4387                 if( isset( $params['heights'] ) ) {
4388                         $ig->setHeights( $params['heights'] );
4389                 }
4390
4391                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4392
4393                 $lines = StringUtils::explode( "\n", $text );
4394                 foreach ( $lines as $line ) {
4395                         # match lines like these:
4396                         # Image:someimage.jpg|This is some image
4397                         $matches = array();
4398                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4399                         # Skip empty lines
4400                         if ( count( $matches ) == 0 ) {
4401                                 continue;
4402                         }
4403
4404                         if ( strpos( $matches[0], '%' ) !== false )
4405                                 $matches[1] = urldecode( $matches[1] );
4406                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4407                         $nt =& $tp;
4408                         if( is_null( $nt ) ) {
4409                                 # Bogus title. Ignore these so we don't bomb out later.
4410                                 continue;
4411                         }
4412                         if ( isset( $matches[3] ) ) {
4413                                 $label = $matches[3];
4414                         } else {
4415                                 $label = '';
4416                         }
4417
4418                         $html = $this->recursiveTagParse( trim( $label ) );
4419
4420                         $ig->add( $nt, $html );
4421
4422                         # Only add real images (bug #5586)
4423                         if ( $nt->getNamespace() == NS_FILE ) {
4424                                 $this->mOutput->addImage( $nt->getDBkey() );
4425                         }
4426                 }
4427                 return $ig->toHTML();
4428         }
4429
4430         function getImageParams( $handler ) {
4431                 if ( $handler ) {
4432                         $handlerClass = get_class( $handler );
4433                 } else {
4434                         $handlerClass = '';
4435                 }
4436                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4437                         // Initialise static lists
4438                         static $internalParamNames = array(
4439                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4440                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4441                                         'bottom', 'text-bottom' ),
4442                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4443                                         'upright', 'border', 'link', 'alt' ),
4444                         );
4445                         static $internalParamMap;
4446                         if ( !$internalParamMap ) {
4447                                 $internalParamMap = array();
4448                                 foreach ( $internalParamNames as $type => $names ) {
4449                                         foreach ( $names as $name ) {
4450                                                 $magicName = str_replace( '-', '_', "img_$name" );
4451                                                 $internalParamMap[$magicName] = array( $type, $name );
4452                                         }
4453                                 }
4454                         }
4455
4456                         // Add handler params
4457                         $paramMap = $internalParamMap;
4458                         if ( $handler ) {
4459                                 $handlerParamMap = $handler->getParamMap();
4460                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4461                                         $paramMap[$magic] = array( 'handler', $paramName );
4462                                 }
4463                         }
4464                         $this->mImageParams[$handlerClass] = $paramMap;
4465                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4466                 }
4467                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4468         }
4469
4470         /**
4471          * Parse image options text and use it to make an image
4472          * @param Title $title
4473          * @param string $options
4474          * @param LinkHolderArray $holders
4475          */
4476         function makeImage( $title, $options, $holders = false ) {
4477                 # Check if the options text is of the form "options|alt text"
4478                 # Options are:
4479                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4480                 #  * left       no resizing, just left align. label is used for alt= only
4481                 #  * right      same, but right aligned
4482                 #  * none       same, but not aligned
4483                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4484                 #  * center     center the image
4485                 #  * frame      Keep original image size, no magnify-button.
4486                 #  * framed     Same as "frame"
4487                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4488                 #  * upright    reduce width for upright images, rounded to full __0 px
4489                 #  * border     draw a 1px border around the image
4490                 #  * alt        Text for HTML alt attribute (defaults to empty)
4491                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4492                 # vertical-align values (no % or length right now):
4493                 #  * baseline
4494                 #  * sub
4495                 #  * super
4496                 #  * top
4497                 #  * text-top
4498                 #  * middle
4499                 #  * bottom
4500                 #  * text-bottom
4501
4502                 $parts = StringUtils::explode( "|", $options );
4503                 $sk = $this->mOptions->getSkin();
4504
4505                 # Give extensions a chance to select the file revision for us
4506                 $skip = $time = $descQuery = false;
4507                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4508
4509                 if ( $skip ) {
4510                         return $sk->link( $title );
4511                 }
4512
4513                 # Get the file
4514                 $imagename = $title->getDBkey();
4515                 $file = wfFindFile( $title, array( 'time' => $time ) );
4516                 # Get parameter map
4517                 $handler = $file ? $file->getHandler() : false;
4518
4519                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4520
4521                 # Process the input parameters
4522                 $caption = '';
4523                 $params = array( 'frame' => array(), 'handler' => array(),
4524                         'horizAlign' => array(), 'vertAlign' => array() );
4525                 foreach( $parts as $part ) {
4526                         $part = trim( $part );
4527                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4528                         $validated = false;
4529                         if( isset( $paramMap[$magicName] ) ) {
4530                                 list( $type, $paramName ) = $paramMap[$magicName];
4531
4532                                 // Special case; width and height come in one variable together
4533                                 if( $type === 'handler' && $paramName === 'width' ) {
4534                                         $m = array();
4535                                         # (bug 13500) In both cases (width/height and width only),
4536                                         # permit trailing "px" for backward compatibility.
4537                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4538                                                 $width = intval( $m[1] );
4539                                                 $height = intval( $m[2] );
4540                                                 if ( $handler->validateParam( 'width', $width ) ) {
4541                                                         $params[$type]['width'] = $width;
4542                                                         $validated = true;
4543                                                 }
4544                                                 if ( $handler->validateParam( 'height', $height ) ) {
4545                                                         $params[$type]['height'] = $height;
4546                                                         $validated = true;
4547                                                 }
4548                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4549                                                 $width = intval( $value );
4550                                                 if ( $handler->validateParam( 'width', $width ) ) {
4551                                                         $params[$type]['width'] = $width;
4552                                                         $validated = true;
4553                                                 }
4554                                         } // else no validation -- bug 13436
4555                                 } else {
4556                                         if ( $type === 'handler' ) {
4557                                                 # Validate handler parameter
4558                                                 $validated = $handler->validateParam( $paramName, $value );
4559                                         } else {
4560                                                 # Validate internal parameters
4561                                                 switch( $paramName ) {
4562                                                 case 'manualthumb':
4563                                                 case 'alt':
4564                                                         // @todo Fixme: possibly check validity here for
4565                                                         // manualthumb? downstream behavior seems odd with
4566                                                         // missing manual thumbs.
4567                                                         $validated = true;
4568                                                         $value = $this->stripAltText( $value, $holders );
4569                                                         break;
4570                                                 case 'link':
4571                                                         $chars = self::EXT_LINK_URL_CLASS;
4572                                                         $prots = $this->mUrlProtocols;
4573                                                         if ( $value === '' ) {
4574                                                                 $paramName = 'no-link';
4575                                                                 $value = true;
4576                                                                 $validated = true;
4577                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4578                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4579                                                                         $paramName = 'link-url';
4580                                                                         $this->mOutput->addExternalLink( $value );
4581                                                                         $validated = true;
4582                                                                 }
4583                                                         } else {
4584                                                                 $linkTitle = Title::newFromText( $value );
4585                                                                 if ( $linkTitle ) {
4586                                                                         $paramName = 'link-title';
4587                                                                         $value = $linkTitle;
4588                                                                         $this->mOutput->addLink( $linkTitle );
4589                                                                         $validated = true;
4590                                                                 }
4591                                                         }
4592                                                         break;
4593                                                 default:
4594                                                         // Most other things appear to be empty or numeric...
4595                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4596                                                 }
4597                                         }
4598
4599                                         if ( $validated ) {
4600                                                 $params[$type][$paramName] = $value;
4601                                         }
4602                                 }
4603                         }
4604                         if ( !$validated ) {
4605                                 $caption = $part;
4606                         }
4607                 }
4608
4609                 # Process alignment parameters
4610                 if ( $params['horizAlign'] ) {
4611                         $params['frame']['align'] = key( $params['horizAlign'] );
4612                 }
4613                 if ( $params['vertAlign'] ) {
4614                         $params['frame']['valign'] = key( $params['vertAlign'] );
4615                 }
4616
4617                 $params['frame']['caption'] = $caption;
4618
4619                 # Will the image be presented in a frame, with the caption below?
4620                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4621                                  isset( $params['frame']['framed'] ) ||
4622                                  isset( $params['frame']['thumbnail'] ) ||
4623                                  isset( $params['frame']['manualthumb'] );
4624
4625                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4626                 # came to also set the caption, ordinary text after the image -- which
4627                 # makes no sense, because that just repeats the text multiple times in
4628                 # screen readers.  It *also* came to set the title attribute.
4629                 #
4630                 # Now that we have an alt attribute, we should not set the alt text to
4631                 # equal the caption: that's worse than useless, it just repeats the
4632                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4633                 # use the unnamed parameter for alt text as well, just for the time be-
4634                 # ing, if the unnamed param is set and the alt param is not.
4635                 #
4636                 # For the future, we need to figure out if we want to tweak this more,
4637                 # e.g., introducing a title= parameter for the title; ignoring the un-
4638                 # named parameter entirely for images without a caption; adding an ex-
4639                 # plicit caption= parameter and preserving the old magic unnamed para-
4640                 # meter for BC; ...
4641                 if ( $imageIsFramed ) { # Framed image
4642                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4643                                 # No caption or alt text, add the filename as the alt text so
4644                                 # that screen readers at least get some description of the image
4645                                 $params['frame']['alt'] = $title->getText();
4646                         }
4647                         # Do not set $params['frame']['title'] because tooltips don't make sense
4648                         # for framed images
4649                 } else { # Inline image
4650                         if ( !isset( $params['frame']['alt'] ) ) {
4651                                 # No alt text, use the "caption" for the alt text
4652                                 if ( $caption !== '') {
4653                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4654                                 } else {
4655                                         # No caption, fall back to using the filename for the
4656                                         # alt text
4657                                         $params['frame']['alt'] = $title->getText();
4658                                 }
4659                         }
4660                         # Use the "caption" for the tooltip text
4661                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4662                 }
4663
4664                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4665
4666                 # Linker does the rest
4667                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4668
4669                 # Give the handler a chance to modify the parser object
4670                 if ( $handler ) {
4671                         $handler->parserTransformHook( $this, $file );
4672                 }
4673
4674                 return $ret;
4675         }
4676
4677         protected function stripAltText( $caption, $holders ) {
4678                 # Strip bad stuff out of the title (tooltip).  We can't just use
4679                 # replaceLinkHoldersText() here, because if this function is called
4680                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4681                 if ( $holders ) {
4682                         $tooltip = $holders->replaceText( $caption );
4683                 } else {
4684                         $tooltip = $this->replaceLinkHoldersText( $caption );
4685                 }
4686
4687                 # make sure there are no placeholders in thumbnail attributes
4688                 # that are later expanded to html- so expand them now and
4689                 # remove the tags
4690                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4691                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4692
4693                 return $tooltip;
4694         }
4695
4696         /**
4697          * Set a flag in the output object indicating that the content is dynamic and
4698          * shouldn't be cached.
4699          */
4700         function disableCache() {
4701                 wfDebug( "Parser output marked as uncacheable.\n" );
4702                 $this->mOutput->mCacheTime = -1;
4703         }
4704
4705         /**#@+
4706          * Callback from the Sanitizer for expanding items found in HTML attribute
4707          * values, so they can be safely tested and escaped.
4708          * @param string $text
4709          * @param PPFrame $frame
4710          * @return string
4711          * @private
4712          */
4713         function attributeStripCallback( &$text, $frame = false ) {
4714                 $text = $this->replaceVariables( $text, $frame );
4715                 $text = $this->mStripState->unstripBoth( $text );
4716                 return $text;
4717         }
4718
4719         /**#@-*/
4720
4721         /**#@+
4722          * Accessor/mutator
4723          */
4724         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4725         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4726         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4727         /**#@-*/
4728
4729         /**#@+
4730          * Accessor
4731          */
4732         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4733         /**#@-*/
4734
4735
4736         /**
4737          * Break wikitext input into sections, and either pull or replace
4738          * some particular section's text.
4739          *
4740          * External callers should use the getSection and replaceSection methods.
4741          *
4742          * @param string $text Page wikitext
4743          * @param string $section A section identifier string of the form:
4744          *   <flag1> - <flag2> - ... - <section number>
4745          *
4746          * Currently the only recognised flag is "T", which means the target section number
4747          * was derived during a template inclusion parse, in other words this is a template
4748          * section edit link. If no flags are given, it was an ordinary section edit link.
4749          * This flag is required to avoid a section numbering mismatch when a section is
4750          * enclosed by <includeonly> (bug 6563).
4751          *
4752          * The section number 0 pulls the text before the first heading; other numbers will
4753          * pull the given section along with its lower-level subsections. If the section is
4754          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4755          *
4756          * @param string $mode One of "get" or "replace"
4757          * @param string $newText Replacement text for section data.
4758          * @return string for "get", the extracted section text.
4759          *                for "replace", the whole page with the section replaced.
4760          */
4761         private function extractSections( $text, $section, $mode, $newText='' ) {
4762                 global $wgTitle;
4763                 $this->clearState();
4764                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4765                 $this->mOptions = new ParserOptions;
4766                 $this->setOutputType( self::OT_PLAIN );
4767                 $outText = '';
4768                 $frame = $this->getPreprocessor()->newFrame();
4769
4770                 // Process section extraction flags
4771                 $flags = 0;
4772                 $sectionParts = explode( '-', $section );
4773                 $sectionIndex = array_pop( $sectionParts );
4774                 foreach ( $sectionParts as $part ) {
4775                         if ( $part === 'T' ) {
4776                                 $flags |= self::PTD_FOR_INCLUSION;
4777                         }
4778                 }
4779                 // Preprocess the text
4780                 $root = $this->preprocessToDom( $text, $flags );
4781
4782                 // <h> nodes indicate section breaks
4783                 // They can only occur at the top level, so we can find them by iterating the root's children
4784                 $node = $root->getFirstChild();
4785
4786                 // Find the target section
4787                 if ( $sectionIndex == 0 ) {
4788                         // Section zero doesn't nest, level=big
4789                         $targetLevel = 1000;
4790                 } else {
4791             while ( $node ) {
4792                 if ( $node->getName() === 'h' ) {
4793                     $bits = $node->splitHeading();
4794                                         if ( $bits['i'] == $sectionIndex ) {
4795                                         $targetLevel = $bits['level'];
4796                                                 break;
4797                                         }
4798                                 }
4799                                 if ( $mode === 'replace' ) {
4800                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4801                                 }
4802                                 $node = $node->getNextSibling();
4803                         }
4804                 }
4805
4806                 if ( !$node ) {
4807                         // Not found
4808                         if ( $mode === 'get' ) {
4809                                 return $newText;
4810                         } else {
4811                                 return $text;
4812                         }
4813                 }
4814
4815                 // Find the end of the section, including nested sections
4816                 do {
4817                         if ( $node->getName() === 'h' ) {
4818                                 $bits = $node->splitHeading();
4819                                 $curLevel = $bits['level'];
4820                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4821                                         break;
4822                                 }
4823                         }
4824                         if ( $mode === 'get' ) {
4825                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4826                         }
4827                         $node = $node->getNextSibling();
4828                 } while ( $node );
4829
4830                 // Write out the remainder (in replace mode only)
4831                 if ( $mode === 'replace' ) {
4832                         // Output the replacement text
4833                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4834                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4835                         // Only add trailing whitespace if there is newText
4836                         if($newText != "") {
4837                                 $outText .= $newText . "\n\n";
4838                         }
4839
4840                         while ( $node ) {
4841                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4842                                 $node = $node->getNextSibling();
4843                         }
4844                 }
4845
4846                 if ( is_string( $outText ) ) {
4847                         // Re-insert stripped tags
4848                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4849                 }
4850
4851                 return $outText;
4852         }
4853
4854         /**
4855          * This function returns the text of a section, specified by a number ($section).
4856          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4857          * the first section before any such heading (section 0).
4858          *
4859          * If a section contains subsections, these are also returned.
4860          *
4861          * @param string $text text to look in
4862          * @param string $section section identifier
4863          * @param string $deftext default to return if section is not found
4864          * @return string text of the requested section
4865          */
4866         public function getSection( $text, $section, $deftext='' ) {
4867                 return $this->extractSections( $text, $section, "get", $deftext );
4868         }
4869
4870         public function replaceSection( $oldtext, $section, $text ) {
4871                 return $this->extractSections( $oldtext, $section, "replace", $text );
4872         }
4873
4874         /**
4875          * Get the timestamp associated with the current revision, adjusted for
4876          * the default server-local timestamp
4877          */
4878         function getRevisionTimestamp() {
4879                 if ( is_null( $this->mRevisionTimestamp ) ) {
4880                         wfProfileIn( __METHOD__ );
4881                         global $wgContLang;
4882                         $dbr = wfGetDB( DB_SLAVE );
4883                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4884                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4885
4886                         // Normalize timestamp to internal MW format for timezone processing.
4887                         // This has the added side-effect of replacing a null value with
4888                         // the current time, which gives us more sensible behavior for
4889                         // previews.
4890                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4891
4892                         // The cryptic '' timezone parameter tells to use the site-default
4893                         // timezone offset instead of the user settings.
4894                         //
4895                         // Since this value will be saved into the parser cache, served
4896                         // to other users, and potentially even used inside links and such,
4897                         // it needs to be consistent for all visitors.
4898                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4899
4900                         wfProfileOut( __METHOD__ );
4901                 }
4902                 return $this->mRevisionTimestamp;
4903         }
4904
4905         /**
4906          * Get the name of the user that edited the last revision
4907          */
4908         function getRevisionUser() {
4909                 // if this template is subst: the revision id will be blank,
4910                 // so just use the current user's name
4911                 if( $this->mRevisionId ) {
4912                         $revision = Revision::newFromId( $this->mRevisionId );
4913                         $revuser = $revision->getUserText();
4914                 } else {
4915                         global $wgUser;
4916                         $revuser = $wgUser->getName();
4917                 }
4918                 return $revuser;
4919         }
4920
4921         /**
4922          * Mutator for $mDefaultSort
4923          *
4924          * @param $sort New value
4925          */
4926         public function setDefaultSort( $sort ) {
4927                 $this->mDefaultSort = $sort;
4928         }
4929
4930         /**
4931          * Accessor for $mDefaultSort
4932          * Will use the title/prefixed title if none is set
4933          *
4934          * @return string
4935          */
4936         public function getDefaultSort() {
4937                 global $wgCategoryPrefixedDefaultSortkey;
4938                 if( $this->mDefaultSort !== false ) {
4939                         return $this->mDefaultSort;
4940                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4941                         !$wgCategoryPrefixedDefaultSortkey) {
4942                         return $this->mTitle->getText();
4943                 } else {
4944                         return $this->mTitle->getPrefixedText();
4945                 }
4946         }
4947
4948         /**
4949          * Accessor for $mDefaultSort
4950          * Unlike getDefaultSort(), will return false if none is set
4951          *
4952          * @return string or false
4953          */
4954         public function getCustomDefaultSort() {
4955                 return $this->mDefaultSort;
4956         }
4957
4958         /**
4959          * Try to guess the section anchor name based on a wikitext fragment
4960          * presumably extracted from a heading, for example "Header" from
4961          * "== Header ==".
4962          */
4963         public function guessSectionNameFromWikiText( $text ) {
4964                 # Strip out wikitext links(they break the anchor)
4965                 $text = $this->stripSectionName( $text );
4966                 $headline = Sanitizer::decodeCharReferences( $text );
4967                 # strip out HTML
4968                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4969                 $headline = trim( $headline );
4970                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4971                 $replacearray = array(
4972                         '%3A' => ':',
4973                         '%' => '.'
4974                 );
4975                 return str_replace(
4976                         array_keys( $replacearray ),
4977                         array_values( $replacearray ),
4978                         $sectionanchor );
4979         }
4980
4981         /**
4982          * Strips a text string of wikitext for use in a section anchor
4983          *
4984          * Accepts a text string and then removes all wikitext from the
4985          * string and leaves only the resultant text (i.e. the result of
4986          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4987          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4988          * to create valid section anchors by mimicing the output of the
4989          * parser when headings are parsed.
4990          *
4991          * @param $text string Text string to be stripped of wikitext
4992          * for use in a Section anchor
4993          * @return Filtered text string
4994          */
4995         public function stripSectionName( $text ) {
4996                 # Strip internal link markup
4997                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4998                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4999
5000                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5001                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5002                 # on how many empty links there are on the page - need to figure that out.
5003                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5004
5005                 # Parse wikitext quotes (italics & bold)
5006                 $text = $this->doQuotes($text);
5007
5008                 # Strip HTML tags
5009                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5010                 return $text;
5011         }
5012
5013         function srvus( $text ) {
5014                 return $this->testSrvus( $text, $this->mOutputType );
5015         }
5016
5017         /**
5018          * strip/replaceVariables/unstrip for preprocessor regression testing
5019          */
5020         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5021                 $this->clearState();
5022                 if ( ! ( $title instanceof Title ) ) {
5023                         $title = Title::newFromText( $title );
5024                 }
5025                 $this->mTitle = $title;
5026                 $this->mOptions = $options;
5027                 $this->setOutputType( $outputType );
5028                 $text = $this->replaceVariables( $text );
5029                 $text = $this->mStripState->unstripBoth( $text );
5030                 $text = Sanitizer::removeHTMLtags( $text );
5031                 return $text;
5032         }
5033
5034         function testPst( $text, $title, $options ) {
5035                 global $wgUser;
5036                 if ( ! ( $title instanceof Title ) ) {
5037                         $title = Title::newFromText( $title );
5038                 }
5039                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5040         }
5041
5042         function testPreprocess( $text, $title, $options ) {
5043                 if ( ! ( $title instanceof Title ) ) {
5044                         $title = Title::newFromText( $title );
5045                 }
5046                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5047         }
5048
5049         function markerSkipCallback( $s, $callback ) {
5050                 $i = 0;
5051                 $out = '';
5052                 while ( $i < strlen( $s ) ) {
5053                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5054                         if ( $markerStart === false ) {
5055                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5056                                 break;
5057                         } else {
5058                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5059                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5060                                 if ( $markerEnd === false ) {
5061                                         $out .= substr( $s, $markerStart );
5062                                         break;
5063                                 } else {
5064                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5065                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5066                                         $i = $markerEnd;
5067                                 }
5068                         }
5069                 }
5070                 return $out;
5071         }
5072
5073         function serialiseHalfParsedText( $text ) {
5074                 $data = array();
5075                 $data['text'] = $text;
5076
5077                 // First, find all strip markers, and store their
5078                 //  data in an array.
5079                 $stripState = new StripState;
5080                 $pos = 0;
5081                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
5082                         $end_pos += strlen( self::MARKER_SUFFIX );
5083                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5084
5085                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5086                                 $replaceArray = $stripState->general;
5087                                 $stripText = $this->mStripState->general->data[$marker];
5088                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5089                                 $replaceArray = $stripState->nowiki;
5090                                 $stripText = $this->mStripState->nowiki->data[$marker];
5091                         } else {
5092                                 throw new MWException( "Hanging strip marker: '$marker'." );
5093                         }
5094
5095                         $replaceArray->setPair( $marker, $stripText );
5096                         $pos = $end_pos;
5097                 }
5098                 $data['stripstate'] = $stripState;
5099
5100                 // Now, find all of our links, and store THEIR
5101                 //  data in an array! :)
5102                 $links = array( 'internal' => array(), 'interwiki' => array() );
5103                 $pos = 0;
5104
5105                 // Internal links
5106                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5107                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5108
5109                         $ns = trim($ns);
5110                         if (empty( $links['internal'][$ns] )) {
5111                                 $links['internal'][$ns] = array();
5112                         }
5113
5114                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5115                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5116                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5117                 }
5118
5119                 $pos = 0;
5120
5121                 // Interwiki links
5122                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5123                         $data = substr( $text, $start_pos );
5124                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5125                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5126                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5127                 }
5128
5129                 $data['linkholder'] = $links;
5130
5131                 return $data;
5132         }
5133
5134         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5135                 if (!$intPrefix)
5136                         $intPrefix = $this->getRandomString();
5137
5138                 // First, extract the strip state.
5139                 $stripState = $data['stripstate'];
5140                 $this->mStripState->general->merge( $stripState->general );
5141                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5142
5143                 // Now, extract the text, and renumber links
5144                 $text = $data['text'];
5145                 $links = $data['linkholder'];
5146
5147                 // Internal...
5148                 foreach( $links['internal'] as $ns => $nsLinks ) {
5149                         foreach( $nsLinks as $key => $entry ) {
5150                                 $newKey = $intPrefix . '-' . $key;
5151                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5152
5153                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5154                         }
5155                 }
5156
5157                 // Interwiki...
5158                 foreach( $links['interwiki'] as $key => $entry ) {
5159                         $newKey = "$intPrefix-$key";
5160                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5161
5162                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5163                 }
5164
5165                 // Should be good to go.
5166                 return $text;
5167         }
5168 }
5169
5170 /**
5171  * @todo document, briefly.
5172  * @ingroup Parser
5173  */
5174 class StripState {
5175         var $general, $nowiki;
5176
5177         function __construct() {
5178                 $this->general = new ReplacementArray;
5179                 $this->nowiki = new ReplacementArray;
5180         }
5181
5182         function unstripGeneral( $text ) {
5183                 wfProfileIn( __METHOD__ );
5184                 do {
5185                         $oldText = $text;
5186                         $text = $this->general->replace( $text );
5187                 } while ( $text !== $oldText );
5188                 wfProfileOut( __METHOD__ );
5189                 return $text;
5190         }
5191
5192         function unstripNoWiki( $text ) {
5193                 wfProfileIn( __METHOD__ );
5194                 do {
5195                         $oldText = $text;
5196                         $text = $this->nowiki->replace( $text );
5197                 } while ( $text !== $oldText );
5198                 wfProfileOut( __METHOD__ );
5199                 return $text;
5200         }
5201
5202         function unstripBoth( $text ) {
5203                 wfProfileIn( __METHOD__ );
5204                 do {
5205                         $oldText = $text;
5206                         $text = $this->general->replace( $text );
5207                         $text = $this->nowiki->replace( $text );
5208                 } while ( $text !== $oldText );
5209                 wfProfileOut( __METHOD__ );
5210                 return $text;
5211         }
5212 }
5213
5214 /**
5215  * @todo document, briefly.
5216  * @ingroup Parser
5217  */
5218 class OnlyIncludeReplacer {
5219         var $output = '';
5220
5221         function replace( $matches ) {
5222                 if ( substr( $matches[1], -1 ) === "\n" ) {
5223                         $this->output .= substr( $matches[1], 0, -1 );
5224                 } else {
5225                         $this->output .= $matches[1];
5226                 }
5227         }
5228 }