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  *  $wgMaxArticleSize*
  42  *
  43  *  * only within ParserOptions
  44  * </pre>
  45  *
  46  * @ingroup Parser
  47  */
  48 class Parser {
  49         /**
  50          * Update this version number when the ParserOutput format
  51          * changes in an incompatible way, so the parser cache
  52          * can automatically discard old data.
  53          */
  54         const VERSION = '1.6.4';
  55
  56         # Flags for Parser::setFunctionHook
  57         # Also available as global constants from Defines.php
  58         const SFH_NO_HASH = 1;
  59         const SFH_OBJECT_ARGS = 2;
  60
  61         # Constants needed for external link processing
  62         # Everything except bracket, space, or control characters
  63         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  64         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  65                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  66
  67         # State constants for the definition list colon extraction
  68         const COLON_STATE_TEXT = 0;
  69         const COLON_STATE_TAG = 1;
  70         const COLON_STATE_TAGSTART = 2;
  71         const COLON_STATE_CLOSETAG = 3;
  72         const COLON_STATE_TAGSLASH = 4;
  73         const COLON_STATE_COMMENT = 5;
  74         const COLON_STATE_COMMENTDASH = 6;
  75         const COLON_STATE_COMMENTDASHDASH = 7;
  76
  77         # Flags for preprocessToDom
  78         const PTD_FOR_INCLUSION = 1;
  79
  80         # Allowed values for $this->mOutputType
  81         # Parameter to startExternalParse().
  82         const OT_HTML = 1; # like parse()
  83         const OT_WIKI = 2; # like preSaveTransform()
  84         const OT_PREPROCESS = 3; # like preprocess()
  85         const OT_MSG = 3;
  86         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  87
  88         # Marker Suffix needs to be accessible staticly.
  89         const MARKER_SUFFIX = "-QINU\x7f";
  90
  91         # Persistent:
  92         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  93         var $mSubstWords, $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex;
  94         var $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList;
  95         var $mVarCache, $mConf, $mFunctionTagHooks;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mLinkHolders, $mLinkID;
 102         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 103         var $mTplExpandCache; # empty-frame expansion cache
 104         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 105         var $mExpensiveFunctionCount; # number of expensive parser function calls
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions;      # ParserOptions object
 110         var $mTitle;        # Title context, used for self-link rendering and similar things
 111         var $mOutputType;   # Output type, one of the OT_xxx constants
 112         var $ot;            # Shortcut alias, see setOutputType()
 113         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 114         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 115         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 116
 117         /**
 118          * Constructor
 119          *
 120          * @public
 121          */
 122         function __construct( $conf = array() ) {
 123                 $this->mConf = $conf;
 124                 $this->mTagHooks = array();
 125                 $this->mTransparentTagHooks = array();
 126                 $this->mFunctionHooks = array();
 127                 $this->mFunctionTagHooks = array();
 128                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 129                 $this->mDefaultStripList = $this->mStripList = array();
 130                 $this->mUrlProtocols = wfUrlProtocols();
 131                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 132                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 133                 $this->mVarCache = array();
 134                 if ( isset( $conf['preprocessorClass'] ) ) {
 135                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 136                 } elseif ( extension_loaded( 'domxml' ) ) {
 137                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 138                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 139                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 140                 } elseif ( extension_loaded( 'dom' ) ) {
 141                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 142                 } else {
 143                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 144                 }
 145                 $this->mMarkerIndex = 0;
 146                 $this->mFirstCall = true;
 147         }
 148
 149         /**
 150          * Reduce memory usage to reduce the impact of circular references
 151          */
 152         function __destruct() {
 153                 if ( isset( $this->mLinkHolders ) ) {
 154                         $this->mLinkHolders->__destruct();
 155                 }
 156                 foreach ( $this as $name => $value ) {
 157                         unset( $this->$name );
 158                 }
 159         }
 160
 161         /**
 162          * Do various kinds of initialisation on the first call of the parser
 163          */
 164         function firstCallInit() {
 165                 if ( !$this->mFirstCall ) {
 166                         return;
 167                 }
 168                 $this->mFirstCall = false;
 169
 170                 wfProfileIn( __METHOD__ );
 171
 172                 CoreParserFunctions::register( $this );
 173                 CoreTagHooks::register( $this );
 174                 $this->initialiseVariables();
 175
 176                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 177                 wfProfileOut( __METHOD__ );
 178         }
 179
 180         /**
 181          * Clear Parser state
 182          *
 183          * @private
 184          */
 185         function clearState() {
 186                 wfProfileIn( __METHOD__ );
 187                 if ( $this->mFirstCall ) {
 188                         $this->firstCallInit();
 189                 }
 190                 $this->mOutput = new ParserOutput;
 191                 $this->mAutonumber = 0;
 192                 $this->mLastSection = '';
 193                 $this->mDTopen = false;
 194                 $this->mIncludeCount = array();
 195                 $this->mStripState = new StripState;
 196                 $this->mArgStack = false;
 197                 $this->mInPre = false;
 198                 $this->mLinkHolders = new LinkHolderArray( $this );
 199                 $this->mLinkID = 0;
 200                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 201                 $this->mVarCache = array();
 202
 203                 /**
 204                  * Prefix for temporary replacement strings for the multipass parser.
 205                  * \x07 should never appear in input as it's disallowed in XML.
 206                  * Using it at the front also gives us a little extra robustness
 207                  * since it shouldn't match when butted up against identifier-like
 208                  * string constructs.
 209                  *
 210                  * Must not consist of all title characters, or else it will change
 211                  * the behaviour of <nowiki> in a link.
 212                  */
 213                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 214                 # Changed to \x7f to allow XML double-parsing -- TS
 215                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 216
 217
 218                 # Clear these on every parse, bug 4549
 219                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 220
 221                 $this->mShowToc = true;
 222                 $this->mForceTocPosition = false;
 223                 $this->mIncludeSizes = array(
 224                         'post-expand' => 0,
 225                         'arg' => 0,
 226                 );
 227                 $this->mPPNodeCount = 0;
 228                 $this->mDefaultSort = false;
 229                 $this->mHeadings = array();
 230                 $this->mDoubleUnderscores = array();
 231                 $this->mExpensiveFunctionCount = 0;
 232
 233                 # Fix cloning
 234                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 235                         $this->mPreprocessor = null;
 236                 }
 237
 238                 wfRunHooks( 'ParserClearState', array( &$this ) );
 239                 wfProfileOut( __METHOD__ );
 240         }
 241
 242         /**
 243          * Convert wikitext to HTML
 244          * Do not call this function recursively.
 245          *
 246          * @param $text String: text we want to parse
 247          * @param $title A title object
 248          * @param $options ParserOptions
 249          * @param $linestart boolean
 250          * @param $clearState boolean
 251          * @param $revid Int: number to pass in {{REVISIONID}}
 252          * @return ParserOutput a ParserOutput
 253          */
 254         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 255                 /**
 256                  * First pass--just handle <nowiki> sections, pass the rest off
 257                  * to internalParse() which does all the real work.
 258                  */
 259
 260                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 261                 $fname = __METHOD__.'-' . wfGetCaller();
 262                 wfProfileIn( __METHOD__ );
 263                 wfProfileIn( $fname );
 264
 265                 if ( $clearState ) {
 266                         $this->clearState();
 267                 }
 268
 269                 $this->mOptions = $options;
 270                 $this->setTitle( $title ); # Page title has to be set for the pre-processor
 271
 272                 $oldRevisionId = $this->mRevisionId;
 273                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 274                 if ( $revid !== null ) {
 275                         $this->mRevisionId = $revid;
 276                         $this->mRevisionTimestamp = null;
 277                 }
 278                 $this->setOutputType( self::OT_HTML );
 279                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 280                 # No more strip!
 281                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 282                 $text = $this->internalParse( $text );
 283
 284                 $text = $this->mStripState->unstripGeneral( $text );
 285
 286                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 287                 $fixtags = array(
 288                         # french spaces, last one Guillemet-left
 289                         # only if there is something before the space
 290                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 291                         # french spaces, Guillemet-right
 292                         '/(\\302\\253) /' => '\\1&#160;',
 293                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 294                 );
 295                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 296
 297                 $text = $this->doBlockLevels( $text, $linestart );
 298
 299                 $this->replaceLinkHolders( $text );
 300
 301                 /**
 302                  * The page doesn't get language converted if
 303                  * a) It's disabled
 304                  * b) Content isn't converted
 305                  * c) It's a conversion table
 306                  */
 307                 if ( !( $wgDisableLangConversion
 308                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 309                                 || $this->mTitle->isConversionTable() ) ) {
 310
 311                         # The position of the convert() call should not be changed. it
 312                         # assumes that the links are all replaced and the only thing left
 313                         # is the <nowiki> mark.
 314
 315                         $text = $wgContLang->convert( $text );
 316                 }
 317
 318                 /**
 319                  * A page get its title converted except:
 320                  * a) Language conversion is globally disabled
 321                  * b) Title convert is globally disabled
 322                  * c) The page is a redirect page
 323                  * d) User request with a "linkconvert" set to "no"
 324                  * e) A "nocontentconvert" magic word has been set
 325                  * f) A "notitleconvert" magic word has been set
 326                  * g) User sets "noconvertlink" in his/her preference
 327                  *
 328                  * Note that if a user tries to set a title in a conversion
 329                  * rule but content conversion was not done, then the parser
 330                  * won't pick it up.  This is probably expected behavior.
 331                  */
 332                 if ( !( $wgDisableLangConversion
 333                                 || $wgDisableTitleConversion
 334                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 335                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 336                                 || $this->mOutput->getDisplayTitle() !== false ) )
 337                 {
 338                         $convruletitle = $wgContLang->getConvRuleTitle();
 339                         if ( $convruletitle ) {
 340                                 $this->mOutput->setTitleText( $convruletitle );
 341                         } else {
 342                                 $titleText = $wgContLang->convertTitle( $title );
 343                                 $this->mOutput->setTitleText( $titleText );
 344                         }
 345                 }
 346
 347                 $text = $this->mStripState->unstripNoWiki( $text );
 348
 349                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 350
 351 //!JF Move to its own function
 352
 353                 $uniq_prefix = $this->mUniqPrefix;
 354                 $matches = array();
 355                 $elements = array_keys( $this->mTransparentTagHooks );
 356                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 357
 358                 foreach ( $matches as $marker => $data ) {
 359                         list( $element, $content, $params, $tag ) = $data;
 360                         $tagName = strtolower( $element );
 361                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 362                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 363                         } else {
 364                                 $output = $tag;
 365                         }
 366                         $this->mStripState->general->setPair( $marker, $output );
 367                 }
 368                 $text = $this->mStripState->unstripGeneral( $text );
 369
 370                 $text = Sanitizer::normalizeCharReferences( $text );
 371
 372                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 373                         $text = MWTidy::tidy( $text );
 374                 } else {
 375                         # attempt to sanitize at least some nesting problems
 376                         # (bug #2702 and quite a few others)
 377                         $tidyregs = array(
 378                                 # ''Something [http://www.cool.com cool''] -->
 379                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 380                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 381                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 382                                 # fix up an anchor inside another anchor, only
 383                                 # at least for a single single nested link (bug 3695)
 384                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 385                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 386                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 387                                 # contains a div, so fix it up here; replace
 388                                 # div with escaped text
 389                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 390                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 391                                 # remove empty italic or bold tag pairs, some
 392                                 # introduced by rules above
 393                                 '/<([bi])><\/\\1>/' => '',
 394                         );
 395
 396                         $text = preg_replace(
 397                                 array_keys( $tidyregs ),
 398                                 array_values( $tidyregs ),
 399                                 $text );
 400                 }
 401                 global $wgExpensiveParserFunctionLimit;
 402                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 403                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 404                 }
 405
 406                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 407
 408                 # Information on include size limits, for the benefit of users who try to skirt them
 409                 if ( $this->mOptions->getEnableLimitReport() ) {
 410                         $max = $this->mOptions->getMaxIncludeSize();
 411                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 412                         $limitReport =
 413                                 "NewPP limit report\n" .
 414                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 415                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 416                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 417                                 $PFreport;
 418                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 419                         $text .= "\n<!-- \n$limitReport-->\n";
 420                 }
 421                 $this->mOutput->setText( $text );
 422
 423                 $this->mRevisionId = $oldRevisionId;
 424                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 425                 wfProfileOut( $fname );
 426                 wfProfileOut( __METHOD__ );
 427
 428                 return $this->mOutput;
 429         }
 430
 431         /**
 432          * Recursive parser entry point that can be called from an extension tag
 433          * hook.
 434          *
 435          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 436          *
 437          * @param $text String: text extension wants to have parsed
 438          * @param $frame PPFrame: The frame to use for expanding any template variables
 439          */
 440         function recursiveTagParse( $text, $frame=false ) {
 441                 wfProfileIn( __METHOD__ );
 442                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 443                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 444                 $text = $this->internalParse( $text, false, $frame );
 445                 wfProfileOut( __METHOD__ );
 446                 return $text;
 447         }
 448
 449         /**
 450          * Expand templates and variables in the text, producing valid, static wikitext.
 451          * Also removes comments.
 452          */
 453         function preprocess( $text, $title, $options, $revid = null ) {
 454                 wfProfileIn( __METHOD__ );
 455                 $this->clearState();
 456                 $this->setOutputType( self::OT_PREPROCESS );
 457                 $this->mOptions = $options;
 458                 $this->setTitle( $title );
 459                 if ( $revid !== null ) {
 460                         $this->mRevisionId = $revid;
 461                 }
 462                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 463                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 464                 $text = $this->replaceVariables( $text );
 465                 $text = $this->mStripState->unstripBoth( $text );
 466                 wfProfileOut( __METHOD__ );
 467                 return $text;
 468         }
 469
 470         /**
 471          * Process the wikitext for the ?preload= feature. (bug 5210)
 472          *
 473          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 474          * comments, templates, arguments, tags hooks and parser functions are untouched.
 475          */
 476         public function getPreloadText( $text, $title, $options ) {
 477                 # Parser (re)initialisation
 478                 $this->clearState();
 479                 $this->setOutputType( self::OT_PLAIN );
 480                 $this->mOptions = $options;
 481                 $this->setTitle( $title );
 482
 483                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 484                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 485                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 486         }
 487
 488         /**
 489          * Get a random string
 490          *
 491          * @private
 492          * @static
 493          */
 494         function getRandomString() {
 495                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 496         }
 497
 498         /**
 499          * Accessor for mUniqPrefix.
 500          *
 501          * @return String
 502          */
 503         public function uniqPrefix() {
 504                 if ( !isset( $this->mUniqPrefix ) ) {
 505                         # @todo Fixme: this is probably *horribly wrong*
 506                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 507                         # if this is called for a parser cache hit, the parser may not
 508                         # have ever been initialized in the first place.
 509                         # Not really sure what the heck is supposed to be going on here.
 510                         return '';
 511                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 512                 }
 513                 return $this->mUniqPrefix;
 514         }
 515
 516         /**
 517          * Set the context title
 518          */
 519         function setTitle( $t ) {
 520                 if ( !$t || $t instanceof FakeTitle ) {
 521                         $t = Title::newFromText( 'NO TITLE' );
 522                 }
 523
 524                 if ( strval( $t->getFragment() ) !== '' ) {
 525                         # Strip the fragment to avoid various odd effects
 526                         $this->mTitle = clone $t;
 527                         $this->mTitle->setFragment( '' );
 528                 } else {
 529                         $this->mTitle = $t;
 530                 }
 531         }
 532
 533         /**
 534          * Accessor for the Title object
 535          *
 536          * @return Title object
 537          */
 538         function &getTitle() {
 539                 return $this->mTitle;
 540         }
 541
 542         /**
 543          * Accessor/mutator for the Title object
 544          *
 545          * @param $x New Title object or null to just get the current one
 546          * @return Title object
 547          */
 548         function Title( $x = null ) {
 549                 return wfSetVar( $this->mTitle, $x );
 550         }
 551
 552         /**
 553          * Set the output type
 554          *
 555          * @param $ot Integer: new value
 556          */
 557         function setOutputType( $ot ) {
 558                 $this->mOutputType = $ot;
 559                 # Shortcut alias
 560                 $this->ot = array(
 561                         'html' => $ot == self::OT_HTML,
 562                         'wiki' => $ot == self::OT_WIKI,
 563                         'pre' => $ot == self::OT_PREPROCESS,
 564                         'plain' => $ot == self::OT_PLAIN,
 565                 );
 566         }
 567
 568         /**
 569          * Accessor/mutator for the output type
 570          *
 571          * @param $x New value or null to just get the current one
 572          * @return Integer
 573          */
 574         function OutputType( $x = null ) {
 575                 return wfSetVar( $this->mOutputType, $x );
 576         }
 577
 578         /**
 579          * Get the ParserOutput object
 580          *
 581          * @return ParserOutput object
 582          */
 583         function getOutput() {
 584                 return $this->mOutput;
 585         }
 586
 587         /**
 588          * Get the ParserOptions object
 589          *
 590          * @return ParserOptions object
 591          */
 592         function getOptions() {
 593                 return $this->mOptions;
 594         }
 595
 596         /**
 597          * Accessor/mutator for the ParserOptions object
 598          *
 599          * @param $x New value or null to just get the current one
 600          * @return Current ParserOptions object
 601          */
 602         function Options( $x = null ) {
 603                 return wfSetVar( $this->mOptions, $x );
 604         }
 605
 606         function nextLinkID() {
 607                 return $this->mLinkID++;
 608         }
 609
 610         function getFunctionLang() {
 611                 global $wgLang, $wgContLang;
 612
 613                 $target = $this->mOptions->getTargetLanguage();
 614                 if ( $target !== null ) {
 615                         return $target;
 616                 } else {
 617                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 618                 }
 619         }
 620
 621         /**
 622          * Get a preprocessor object
 623          *
 624          * @return Preprocessor instance
 625          */
 626         function getPreprocessor() {
 627                 if ( !isset( $this->mPreprocessor ) ) {
 628                         $class = $this->mPreprocessorClass;
 629                         $this->mPreprocessor = new $class( $this );
 630                 }
 631                 return $this->mPreprocessor;
 632         }
 633
 634         /**
 635          * Replaces all occurrences of HTML-style comments and the given tags
 636          * in the text with a random marker and returns the next text. The output
 637          * parameter $matches will be an associative array filled with data in
 638          * the form:
 639          *   'UNIQ-xxxxx' => array(
 640          *     'element',
 641          *     'tag content',
 642          *     array( 'param' => 'x' ),
 643          *     '<element param="x">tag content</element>' ) )
 644          *
 645          * @param $elements list of element names. Comments are always extracted.
 646          * @param $text Source text string.
 647          * @param $matches Out parameter, Array: extarcted tags
 648          * @param $uniq_prefix
 649          * @return String: stripped text
 650          *
 651          * @static
 652          */
 653         public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 654                 static $n = 1;
 655                 $stripped = '';
 656                 $matches = array();
 657
 658                 $taglist = implode( '|', $elements );
 659                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 660
 661                 while ( $text != '' ) {
 662                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 663                         $stripped .= $p[0];
 664                         if ( count( $p ) < 5 ) {
 665                                 break;
 666                         }
 667                         if ( count( $p ) > 5 ) {
 668                                 # comment
 669                                 $element    = $p[4];
 670                                 $attributes = '';
 671                                 $close      = '';
 672                                 $inside     = $p[5];
 673                         } else {
 674                                 # tag
 675                                 $element    = $p[1];
 676                                 $attributes = $p[2];
 677                                 $close      = $p[3];
 678                                 $inside     = $p[4];
 679                         }
 680
 681                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 682                         $stripped .= $marker;
 683
 684                         if ( $close === '/>' ) {
 685                                 # Empty element tag, <tag />
 686                                 $content = null;
 687                                 $text = $inside;
 688                                 $tail = null;
 689                         } else {
 690                                 if ( $element === '!--' ) {
 691                                         $end = '/(-->)/';
 692                                 } else {
 693                                         $end = "/(<\\/$element\\s*>)/i";
 694                                 }
 695                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 696                                 $content = $q[0];
 697                                 if ( count( $q ) < 3 ) {
 698                                         # No end tag -- let it run out to the end of the text.
 699                                         $tail = '';
 700                                         $text = '';
 701                                 } else {
 702                                         $tail = $q[1];
 703                                         $text = $q[2];
 704                                 }
 705                         }
 706
 707                         $matches[$marker] = array( $element,
 708                                 $content,
 709                                 Sanitizer::decodeTagAttributes( $attributes ),
 710                                 "<$element$attributes$close$content$tail" );
 711                 }
 712                 return $stripped;
 713         }
 714
 715         /**
 716          * Get a list of strippable XML-like elements
 717          */
 718         function getStripList() {
 719                 return $this->mStripList;
 720         }
 721
 722         /**
 723          * @deprecated use replaceVariables
 724          */
 725         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 726                 return $text;
 727         }
 728
 729         /**
 730          * Restores pre, math, and other extensions removed by strip()
 731          *
 732          * always call unstripNoWiki() after this one
 733          * @private
 734          * @deprecated use $this->mStripState->unstrip()
 735          */
 736         function unstrip( $text, $state ) {
 737                 return $state->unstripGeneral( $text );
 738         }
 739
 740         /**
 741          * Always call this after unstrip() to preserve the order
 742          *
 743          * @private
 744          * @deprecated use $this->mStripState->unstrip()
 745          */
 746         function unstripNoWiki( $text, $state ) {
 747                 return $state->unstripNoWiki( $text );
 748         }
 749
 750         /**
 751          * @deprecated use $this->mStripState->unstripBoth()
 752          */
 753         function unstripForHTML( $text ) {
 754                 return $this->mStripState->unstripBoth( $text );
 755         }
 756
 757         /**
 758          * Add an item to the strip state
 759          * Returns the unique tag which must be inserted into the stripped text
 760          * The tag will be replaced with the original text in unstrip()
 761          *
 762          * @private
 763          */
 764         function insertStripItem( $text ) {
 765                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 766                 $this->mMarkerIndex++;
 767                 $this->mStripState->general->setPair( $rnd, $text );
 768                 return $rnd;
 769         }
 770
 771         /**
 772          * Interface with html tidy
 773          * @deprecated Use MWTidy::tidy()
 774          */
 775         public static function tidy( $text ) {
 776                 wfDeprecated( __METHOD__ );
 777                 return MWTidy::tidy( $text );
 778         }
 779
 780         /**
 781          * parse the wiki syntax used to render tables
 782          *
 783          * @private
 784          */
 785         function doTableStuff( $text ) {
 786                 wfProfileIn( __METHOD__ );
 787
 788                 $lines = StringUtils::explode( "\n", $text );
 789                 $text = null;
 790                 $out = '';
 791                 $td_history = array(); # Is currently a td tag open?
 792                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 793                 $tr_history = array(); # Is currently a tr tag open?
 794                 $tr_attributes = array(); # history of tr attributes
 795                 $has_opened_tr = array(); # Did this table open a <tr> element?
 796                 $indent_level = 0; # indent level of the table
 797
 798                 foreach ( $lines as $outLine ) {
 799                         $line = trim( $outLine );
 800
 801                         if ( $line === '' ) { # empty line, go to next line
 802                                 $out .= $outLine."\n";
 803                                 continue;
 804                         }
 805
 806                         $first_character = $line[0];
 807                         $matches = array();
 808
 809                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 810                                 # First check if we are starting a new table
 811                                 $indent_level = strlen( $matches[1] );
 812
 813                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 814                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 815
 816                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 817                                 array_push( $td_history , false );
 818                                 array_push( $last_tag_history , '' );
 819                                 array_push( $tr_history , false );
 820                                 array_push( $tr_attributes , '' );
 821                                 array_push( $has_opened_tr , false );
 822                         } elseif ( count( $td_history ) == 0 ) {
 823                                 # Don't do any of the following
 824                                 $out .= $outLine."\n";
 825                                 continue;
 826                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 827                                 # We are ending a table
 828                                 $line = '</table>' . substr( $line , 2 );
 829                                 $last_tag = array_pop( $last_tag_history );
 830
 831                                 if ( !array_pop( $has_opened_tr ) ) {
 832                                         $line = "<tr><td></td></tr>{$line}";
 833                                 }
 834
 835                                 if ( array_pop( $tr_history ) ) {
 836                                         $line = "</tr>{$line}";
 837                                 }
 838
 839                                 if ( array_pop( $td_history ) ) {
 840                                         $line = "</{$last_tag}>{$line}";
 841                                 }
 842                                 array_pop( $tr_attributes );
 843                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 844                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 845                                 # Now we have a table row
 846                                 $line = preg_replace( '#^\|-+#', '', $line );
 847
 848                                 # Whats after the tag is now only attributes
 849                                 $attributes = $this->mStripState->unstripBoth( $line );
 850                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 851                                 array_pop( $tr_attributes );
 852                                 array_push( $tr_attributes, $attributes );
 853
 854                                 $line = '';
 855                                 $last_tag = array_pop( $last_tag_history );
 856                                 array_pop( $has_opened_tr );
 857                                 array_push( $has_opened_tr , true );
 858
 859                                 if ( array_pop( $tr_history ) ) {
 860                                         $line = '</tr>';
 861                                 }
 862
 863                                 if ( array_pop( $td_history ) ) {
 864                                         $line = "</{$last_tag}>{$line}";
 865                                 }
 866
 867                                 $outLine = $line;
 868                                 array_push( $tr_history , false );
 869                                 array_push( $td_history , false );
 870                                 array_push( $last_tag_history , '' );
 871                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 872                                 # This might be cell elements, td, th or captions
 873                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 874                                         $first_character = '+';
 875                                         $line = substr( $line , 1 );
 876                                 }
 877
 878                                 $line = substr( $line , 1 );
 879
 880                                 if ( $first_character === '!' ) {
 881                                         $line = str_replace( '!!' , '||' , $line );
 882                                 }
 883
 884                                 # Split up multiple cells on the same line.
 885                                 # FIXME : This can result in improper nesting of tags processed
 886                                 # by earlier parser steps, but should avoid splitting up eg
 887                                 # attribute values containing literal "||".
 888                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 889
 890                                 $outLine = '';
 891
 892                                 # Loop through each table cell
 893                                 foreach ( $cells as $cell ) {
 894                                         $previous = '';
 895                                         if ( $first_character !== '+' ) {
 896                                                 $tr_after = array_pop( $tr_attributes );
 897                                                 if ( !array_pop( $tr_history ) ) {
 898                                                         $previous = "<tr{$tr_after}>\n";
 899                                                 }
 900                                                 array_push( $tr_history , true );
 901                                                 array_push( $tr_attributes , '' );
 902                                                 array_pop( $has_opened_tr );
 903                                                 array_push( $has_opened_tr , true );
 904                                         }
 905
 906                                         $last_tag = array_pop( $last_tag_history );
 907
 908                                         if ( array_pop( $td_history ) ) {
 909                                                 $previous = "</{$last_tag}>{$previous}";
 910                                         }
 911
 912                                         if ( $first_character === '|' ) {
 913                                                 $last_tag = 'td';
 914                                         } elseif ( $first_character === '!' ) {
 915                                                 $last_tag = 'th';
 916                                         } elseif ( $first_character === '+' ) {
 917                                                 $last_tag = 'caption';
 918                                         } else {
 919                                                 $last_tag = '';
 920                                         }
 921
 922                                         array_push( $last_tag_history , $last_tag );
 923
 924                                         # A cell could contain both parameters and data
 925                                         $cell_data = explode( '|' , $cell , 2 );
 926
 927                                         # Bug 553: Note that a '|' inside an invalid link should not
 928                                         # be mistaken as delimiting cell parameters
 929                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 930                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 931                                         } elseif ( count( $cell_data ) == 1 ) {
 932                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 933                                         } else {
 934                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 935                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 936                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 937                                         }
 938
 939                                         $outLine .= $cell;
 940                                         array_push( $td_history , true );
 941                                 }
 942                         }
 943                         $out .= $outLine . "\n";
 944                 }
 945
 946                 # Closing open td, tr && table
 947                 while ( count( $td_history ) > 0 ) {
 948                         if ( array_pop( $td_history ) ) {
 949                                 $out .= "</td>\n";
 950                         }
 951                         if ( array_pop( $tr_history ) ) {
 952                                 $out .= "</tr>\n";
 953                         }
 954                         if ( !array_pop( $has_opened_tr ) ) {
 955                                 $out .= "<tr><td></td></tr>\n" ;
 956                         }
 957
 958                         $out .= "</table>\n";
 959                 }
 960
 961                 # Remove trailing line-ending (b/c)
 962                 if ( substr( $out, -1 ) === "\n" ) {
 963                         $out = substr( $out, 0, -1 );
 964                 }
 965
 966                 # special case: don't return empty table
 967                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 968                         $out = '';
 969                 }
 970
 971                 wfProfileOut( __METHOD__ );
 972
 973                 return $out;
 974         }
 975
 976         /**
 977          * Helper function for parse() that transforms wiki markup into
 978          * HTML. Only called for $mOutputType == self::OT_HTML.
 979          *
 980          * @private
 981          */
 982         function internalParse( $text, $isMain = true, $frame=false ) {
 983                 wfProfileIn( __METHOD__ );
 984
 985                 $origText = $text;
 986
 987                 # Hook to suspend the parser in this state
 988                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 989                         wfProfileOut( __METHOD__ );
 990                         return $text ;
 991                 }
 992
 993                 # if $frame is provided, then use $frame for replacing any variables
 994                 if ( $frame ) {
 995                         # use frame depth to infer how include/noinclude tags should be handled
 996                         # depth=0 means this is the top-level document; otherwise it's an included document
 997                         if ( !$frame->depth ) {
 998                                 $flag = 0;
 999                         } else {
1000                                 $flag = Parser::PTD_FOR_INCLUSION;
1001                         }
1002                         $dom = $this->preprocessToDom( $text, $flag );
1003                         $text = $frame->expand( $dom );
1004                 } else {
1005                         # if $frame is not provided, then use old-style replaceVariables
1006                         $text = $this->replaceVariables( $text );
1007                 }
1008
1009                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1010                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1011
1012                 # Tables need to come after variable replacement for things to work
1013                 # properly; putting them before other transformations should keep
1014                 # exciting things like link expansions from showing up in surprising
1015                 # places.
1016                 $text = $this->doTableStuff( $text );
1017
1018                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1019
1020                 $text = $this->doDoubleUnderscore( $text );
1021
1022                 $text = $this->doHeadings( $text );
1023                 if ( $this->mOptions->getUseDynamicDates() ) {
1024                         $df = DateFormatter::getInstance();
1025                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1026                 }
1027                 $text = $this->doAllQuotes( $text );
1028                 $text = $this->replaceInternalLinks( $text );
1029                 $text = $this->replaceExternalLinks( $text );
1030
1031                 # replaceInternalLinks may sometimes leave behind
1032                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1033                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1034
1035                 $text = $this->doMagicLinks( $text );
1036                 $text = $this->formatHeadings( $text, $origText, $isMain );
1037
1038                 wfProfileOut( __METHOD__ );
1039                 return $text;
1040         }
1041
1042         /**
1043          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1044          * magic external links.
1045          *
1046          * DML
1047          * @private
1048          */
1049         function doMagicLinks( $text ) {
1050                 wfProfileIn( __METHOD__ );
1051                 $prots = $this->mUrlProtocols;
1052                 $urlChar = self::EXT_LINK_URL_CLASS;
1053                 $text = preg_replace_callback(
1054                         '!(?:                           # Start cases
1055                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1056                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1057                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1058                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1059                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1060                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1061                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1062                                     [0-9Xx]                 # check digit
1063                                     \b)
1064                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1065                 wfProfileOut( __METHOD__ );
1066                 return $text;
1067         }
1068
1069         function magicLinkCallback( $m ) {
1070                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1071                         # Skip anchor
1072                         return $m[0];
1073                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1074                         # Skip HTML element
1075                         return $m[0];
1076                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1077                         # Free external link
1078                         return $this->makeFreeExternalLink( $m[0] );
1079                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1080                         # RFC or PMID
1081                         $CssClass = '';
1082                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1083                                 $keyword = 'RFC';
1084                                 $urlmsg = 'rfcurl';
1085                                 $CssClass = 'mw-magiclink-rfc';
1086                                 $id = $m[4];
1087                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1088                                 $keyword = 'PMID';
1089                                 $urlmsg = 'pubmedurl';
1090                                 $CssClass = 'mw-magiclink-pmid';
1091                                 $id = $m[4];
1092                         } else {
1093                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1094                                         substr( $m[0], 0, 20 ) . '"' );
1095                         }
1096                         $url = wfMsg( $urlmsg, $id);
1097                         $sk = $this->mOptions->getSkin();
1098                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1099                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1100                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1101                         # ISBN
1102                         $isbn = $m[5];
1103                         $num = strtr( $isbn, array(
1104                                 '-' => '',
1105                                 ' ' => '',
1106                                 'x' => 'X',
1107                         ));
1108                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1109                         return'<a href="' .
1110                                 $titleObj->escapeLocalUrl() .
1111                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1112                 } else {
1113                         return $m[0];
1114                 }
1115         }
1116
1117         /**
1118          * Make a free external link, given a user-supplied URL
1119          * @return HTML
1120          * @private
1121          */
1122         function makeFreeExternalLink( $url ) {
1123                 global $wgContLang;
1124                 wfProfileIn( __METHOD__ );
1125
1126                 $sk = $this->mOptions->getSkin();
1127                 $trail = '';
1128
1129                 # The characters '<' and '>' (which were escaped by
1130                 # removeHTMLtags()) should not be included in
1131                 # URLs, per RFC 2396.
1132                 $m2 = array();
1133                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1134                         $trail = substr( $url, $m2[0][1] ) . $trail;
1135                         $url = substr( $url, 0, $m2[0][1] );
1136                 }
1137
1138                 # Move trailing punctuation to $trail
1139                 $sep = ',;\.:!?';
1140                 # If there is no left bracket, then consider right brackets fair game too
1141                 if ( strpos( $url, '(' ) === false ) {
1142                         $sep .= ')';
1143                 }
1144
1145                 $numSepChars = strspn( strrev( $url ), $sep );
1146                 if ( $numSepChars ) {
1147                         $trail = substr( $url, -$numSepChars ) . $trail;
1148                         $url = substr( $url, 0, -$numSepChars );
1149                 }
1150
1151                 $url = Sanitizer::cleanUrl( $url );
1152
1153                 # Is this an external image?
1154                 $text = $this->maybeMakeExternalImage( $url );
1155                 if ( $text === false ) {
1156                         # Not an image, make a link
1157                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1158                                 $this->getExternalLinkAttribs( $url ) );
1159                         # Register it in the output object...
1160                         # Replace unnecessary URL escape codes with their equivalent characters
1161                         $pasteurized = self::replaceUnusualEscapes( $url );
1162                         $this->mOutput->addExternalLink( $pasteurized );
1163                 }
1164                 wfProfileOut( __METHOD__ );
1165                 return $text . $trail;
1166         }
1167
1168
1169         /**
1170          * Parse headers and return html
1171          *
1172          * @private
1173          */
1174         function doHeadings( $text ) {
1175                 wfProfileIn( __METHOD__ );
1176                 for ( $i = 6; $i >= 1; --$i ) {
1177                         $h = str_repeat( '=', $i );
1178                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1179                           "<h$i>\\1</h$i>", $text );
1180                 }
1181                 wfProfileOut( __METHOD__ );
1182                 return $text;
1183         }
1184
1185         /**
1186          * Replace single quotes with HTML markup
1187          * @private
1188          * @return string the altered text
1189          */
1190         function doAllQuotes( $text ) {
1191                 wfProfileIn( __METHOD__ );
1192                 $outtext = '';
1193                 $lines = StringUtils::explode( "\n", $text );
1194                 foreach ( $lines as $line ) {
1195                         $outtext .= $this->doQuotes( $line ) . "\n";
1196                 }
1197                 $outtext = substr( $outtext, 0,-1 );
1198                 wfProfileOut( __METHOD__ );
1199                 return $outtext;
1200         }
1201
1202         /**
1203          * Helper function for doAllQuotes()
1204          */
1205         public function doQuotes( $text ) {
1206                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1207                 if ( count( $arr ) == 1 ) {
1208                         return $text;
1209                 } else {
1210                         # First, do some preliminary work. This may shift some apostrophes from
1211                         # being mark-up to being text. It also counts the number of occurrences
1212                         # of bold and italics mark-ups.
1213                         $i = 0;
1214                         $numbold = 0;
1215                         $numitalics = 0;
1216                         foreach ( $arr as $r ) {
1217                                 if ( ( $i % 2 ) == 1 ) {
1218                                         # If there are ever four apostrophes, assume the first is supposed to
1219                                         # be text, and the remaining three constitute mark-up for bold text.
1220                                         if ( strlen( $arr[$i] ) == 4 ) {
1221                                                 $arr[$i-1] .= "'";
1222                                                 $arr[$i] = "'''";
1223                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1224                                                 # If there are more than 5 apostrophes in a row, assume they're all
1225                                                 # text except for the last 5.
1226                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1227                                                 $arr[$i] = "'''''";
1228                                         }
1229                                         # Count the number of occurrences of bold and italics mark-ups.
1230                                         # We are not counting sequences of five apostrophes.
1231                                         if ( strlen( $arr[$i] ) == 2 ) {
1232                                                 $numitalics++;
1233                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1234                                                 $numbold++;
1235                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1236                                                 $numitalics++;
1237                                                 $numbold++;
1238                                         }
1239                                 }
1240                                 $i++;
1241                         }
1242
1243                         # If there is an odd number of both bold and italics, it is likely
1244                         # that one of the bold ones was meant to be an apostrophe followed
1245                         # by italics. Which one we cannot know for certain, but it is more
1246                         # likely to be one that has a single-letter word before it.
1247                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1248                                 $i = 0;
1249                                 $firstsingleletterword = -1;
1250                                 $firstmultiletterword = -1;
1251                                 $firstspace = -1;
1252                                 foreach ( $arr as $r ) {
1253                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1254                                                 $x1 = substr( $arr[$i-1], -1 );
1255                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1256                                                 if ( $x1 === ' ' ) {
1257                                                         if ( $firstspace == -1 ) {
1258                                                                 $firstspace = $i;
1259                                                         }
1260                                                 } elseif ( $x2 === ' ') {
1261                                                         if ( $firstsingleletterword == -1 ) {
1262                                                                 $firstsingleletterword = $i;
1263                                                         }
1264                                                 } else {
1265                                                         if ( $firstmultiletterword == -1 ) {
1266                                                                 $firstmultiletterword = $i;
1267                                                         }
1268                                                 }
1269                                         }
1270                                         $i++;
1271                                 }
1272
1273                                 # If there is a single-letter word, use it!
1274                                 if ( $firstsingleletterword > -1 ) {
1275                                         $arr[$firstsingleletterword] = "''";
1276                                         $arr[$firstsingleletterword-1] .= "'";
1277                                 } elseif ( $firstmultiletterword > -1 ) {
1278                                         # If not, but there's a multi-letter word, use that one.
1279                                         $arr[$firstmultiletterword] = "''";
1280                                         $arr[$firstmultiletterword-1] .= "'";
1281                                 } elseif ( $firstspace > -1 ) {
1282                                         # ... otherwise use the first one that has neither.
1283                                         # (notice that it is possible for all three to be -1 if, for example,
1284                                         # there is only one pentuple-apostrophe in the line)
1285                                         $arr[$firstspace] = "''";
1286                                         $arr[$firstspace-1] .= "'";
1287                                 }
1288                         }
1289
1290                         # Now let's actually convert our apostrophic mush to HTML!
1291                         $output = '';
1292                         $buffer = '';
1293                         $state = '';
1294                         $i = 0;
1295                         foreach ( $arr as $r ) {
1296                                 if ( ( $i % 2 ) == 0 ) {
1297                                         if ( $state === 'both' ) {
1298                                                 $buffer .= $r;
1299                                         } else {
1300                                                 $output .= $r;
1301                                         }
1302                                 } else {
1303                                         if ( strlen( $r ) == 2 ) {
1304                                                 if ( $state === 'i' ) {
1305                                                         $output .= '</i>'; $state = '';
1306                                                 } elseif ( $state === 'bi' ) {
1307                                                         $output .= '</i>'; $state = 'b';
1308                                                 } elseif ( $state === 'ib' ) {
1309                                                         $output .= '</b></i><b>'; $state = 'b';
1310                                                 } elseif ( $state === 'both' ) {
1311                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1312                                                 } else { # $state can be 'b' or ''
1313                                                         $output .= '<i>'; $state .= 'i';
1314                                                 }
1315                                         } elseif ( strlen( $r ) == 3 ) {
1316                                                 if ( $state === 'b' ) {
1317                                                         $output .= '</b>'; $state = '';
1318                                                 } elseif ( $state === 'bi' ) {
1319                                                         $output .= '</i></b><i>'; $state = 'i';
1320                                                 } elseif ( $state === 'ib' ) {
1321                                                         $output .= '</b>'; $state = 'i';
1322                                                 } elseif ( $state === 'both' ) {
1323                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1324                                                 } else { # $state can be 'i' or ''
1325                                                         $output .= '<b>'; $state .= 'b';
1326                                                 }
1327                                         } elseif ( strlen( $r ) == 5 ) {
1328                                                 if ( $state === 'b' ) {
1329                                                         $output .= '</b><i>'; $state = 'i';
1330                                                 } elseif ( $state === 'i' ) {
1331                                                         $output .= '</i><b>'; $state = 'b';
1332                                                 } elseif ( $state === 'bi' ) {
1333                                                         $output .= '</i></b>'; $state = '';
1334                                                 } elseif ( $state === 'ib' ) {
1335                                                         $output .= '</b></i>'; $state = '';
1336                                                 } elseif ( $state === 'both' ) {
1337                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1338                                                 } else { # ($state == '')
1339                                                         $buffer = ''; $state = 'both';
1340                                                 }
1341                                         }
1342                                 }
1343                                 $i++;
1344                         }
1345                         # Now close all remaining tags.  Notice that the order is important.
1346                         if ( $state === 'b' || $state === 'ib' ) {
1347                                 $output .= '</b>';
1348                         }
1349                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1350                                 $output .= '</i>';
1351                         }
1352                         if ( $state === 'bi' ) {
1353                                 $output .= '</b>';
1354                         }
1355                         # There might be lonely ''''', so make sure we have a buffer
1356                         if ( $state === 'both' && $buffer ) {
1357                                 $output .= '<b><i>'.$buffer.'</i></b>';
1358                         }
1359                         return $output;
1360                 }
1361         }
1362
1363         /**
1364          * Replace external links (REL)
1365          *
1366          * Note: this is all very hackish and the order of execution matters a lot.
1367          * Make sure to run maintenance/parserTests.php if you change this code.
1368          *
1369          * @private
1370          */
1371         function replaceExternalLinks( $text ) {
1372                 global $wgContLang;
1373                 wfProfileIn( __METHOD__ );
1374
1375                 $sk = $this->mOptions->getSkin();
1376
1377                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1378                 $s = array_shift( $bits );
1379
1380                 $i = 0;
1381                 while ( $i<count( $bits ) ) {
1382                         $url = $bits[$i++];
1383                         $protocol = $bits[$i++];
1384                         $text = $bits[$i++];
1385                         $trail = $bits[$i++];
1386
1387                         # The characters '<' and '>' (which were escaped by
1388                         # removeHTMLtags()) should not be included in
1389                         # URLs, per RFC 2396.
1390                         $m2 = array();
1391                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1392                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1393                                 $url = substr( $url, 0, $m2[0][1] );
1394                         }
1395
1396                         # If the link text is an image URL, replace it with an <img> tag
1397                         # This happened by accident in the original parser, but some people used it extensively
1398                         $img = $this->maybeMakeExternalImage( $text );
1399                         if ( $img !== false ) {
1400                                 $text = $img;
1401                         }
1402
1403                         $dtrail = '';
1404
1405                         # Set linktype for CSS - if URL==text, link is essentially free
1406                         $linktype = ( $text === $url ) ? 'free' : 'text';
1407
1408                         # No link text, e.g. [http://domain.tld/some.link]
1409                         if ( $text == '' ) {
1410                                 # Autonumber if allowed. See bug #5918
1411                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1412                                         $langObj = $this->getFunctionLang();
1413                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1414                                         $linktype = 'autonumber';
1415                                 } else {
1416                                         # Otherwise just use the URL
1417                                         $text = htmlspecialchars( $url );
1418                                         $linktype = 'free';
1419                                 }
1420                         } else {
1421                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1422                                 # Check for trail
1423                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1424                         }
1425
1426                         $text = $wgContLang->markNoConversion( $text );
1427
1428                         $url = Sanitizer::cleanUrl( $url );
1429
1430                         # Use the encoded URL
1431                         # This means that users can paste URLs directly into the text
1432                         # Funny characters like ö aren't valid in URLs anyway
1433                         # This was changed in August 2004
1434                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1435                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1436
1437                         # Register link in the output object.
1438                         # Replace unnecessary URL escape codes with the referenced character
1439                         # This prevents spammers from hiding links from the filters
1440                         $pasteurized = self::replaceUnusualEscapes( $url );
1441                         $this->mOutput->addExternalLink( $pasteurized );
1442                 }
1443
1444                 wfProfileOut( __METHOD__ );
1445                 return $s;
1446         }
1447
1448         /**
1449          * Get an associative array of additional HTML attributes appropriate for a
1450          * particular external link.  This currently may include rel => nofollow
1451          * (depending on configuration, namespace, and the URL's domain) and/or a
1452          * target attribute (depending on configuration).
1453          *
1454          * @param $url String: optional URL, to extract the domain from for rel =>
1455          *   nofollow if appropriate
1456          * @return Array: associative array of HTML attributes
1457          */
1458         function getExternalLinkAttribs( $url = false ) {
1459                 $attribs = array();
1460                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1461                 $ns = $this->mTitle->getNamespace();
1462                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1463                         $attribs['rel'] = 'nofollow';
1464
1465                         global $wgNoFollowDomainExceptions;
1466                         if ( $wgNoFollowDomainExceptions ) {
1467                                 $bits = wfParseUrl( $url );
1468                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1469                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1470                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1471                                                         unset( $attribs['rel'] );
1472                                                         break;
1473                                                 }
1474                                         }
1475                                 }
1476                         }
1477                 }
1478                 if ( $this->mOptions->getExternalLinkTarget() ) {
1479                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1480                 }
1481                 return $attribs;
1482         }
1483
1484
1485         /**
1486          * Replace unusual URL escape codes with their equivalent characters
1487          *
1488          * @param $url String
1489          * @return String
1490          *
1491          * @todo  This can merge genuinely required bits in the path or query string,
1492          *        breaking legit URLs. A proper fix would treat the various parts of
1493          *        the URL differently; as a workaround, just use the output for
1494          *        statistical records, not for actual linking/output.
1495          */
1496         static function replaceUnusualEscapes( $url ) {
1497                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1498                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1499         }
1500
1501         /**
1502          * Callback function used in replaceUnusualEscapes().
1503          * Replaces unusual URL escape codes with their equivalent character
1504          */
1505         private static function replaceUnusualEscapesCallback( $matches ) {
1506                 $char = urldecode( $matches[0] );
1507                 $ord = ord( $char );
1508                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1509                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1510                         # No, shouldn't be escaped
1511                         return $char;
1512                 } else {
1513                         # Yes, leave it escaped
1514                         return $matches[0];
1515                 }
1516         }
1517
1518         /**
1519          * make an image if it's allowed, either through the global
1520          * option, through the exception, or through the on-wiki whitelist
1521          * @private
1522          */
1523         function maybeMakeExternalImage( $url ) {
1524                 $sk = $this->mOptions->getSkin();
1525                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1526                 $imagesexception = !empty( $imagesfrom );
1527                 $text = false;
1528                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1529                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1530                         $imagematch = false;
1531                         foreach ( $imagesfrom as $match ) {
1532                                 if ( strpos( $url, $match ) === 0 ) {
1533                                         $imagematch = true;
1534                                         break;
1535                                 }
1536                         }
1537                 } elseif ( $imagesexception ) {
1538                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1539                 } else {
1540                         $imagematch = false;
1541                 }
1542                 if ( $this->mOptions->getAllowExternalImages()
1543                      || ( $imagesexception && $imagematch ) ) {
1544                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1545                                 # Image found
1546                                 $text = $sk->makeExternalImage( $url );
1547                         }
1548                 }
1549                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1550                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1551                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1552                         foreach ( $whitelist as $entry ) {
1553                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1554                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1555                                         continue;
1556                                 }
1557                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1558                                         # Image matches a whitelist entry
1559                                         $text = $sk->makeExternalImage( $url );
1560                                         break;
1561                                 }
1562                         }
1563                 }
1564                 return $text;
1565         }
1566
1567         /**
1568          * Process [[ ]] wikilinks
1569          * @return processed text
1570          *
1571          * @private
1572          */
1573         function replaceInternalLinks( $s ) {
1574                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1575                 return $s;
1576         }
1577
1578         /**
1579          * Process [[ ]] wikilinks (RIL)
1580          * @return LinkHolderArray
1581          *
1582          * @private
1583          */
1584         function replaceInternalLinks2( &$s ) {
1585                 global $wgContLang;
1586
1587                 wfProfileIn( __METHOD__ );
1588
1589                 wfProfileIn( __METHOD__.'-setup' );
1590                 static $tc = FALSE, $e1, $e1_img;
1591                 # the % is needed to support urlencoded titles as well
1592                 if ( !$tc ) {
1593                         $tc = Title::legalChars() . '#%';
1594                         # Match a link having the form [[namespace:link|alternate]]trail
1595                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1596                         # Match cases where there is no "]]", which might still be images
1597                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1598                 }
1599
1600                 $sk = $this->mOptions->getSkin();
1601                 $holders = new LinkHolderArray( $this );
1602
1603                 # split the entire text string on occurences of [[
1604                 $a = StringUtils::explode( '[[', ' ' . $s );
1605                 # get the first element (all text up to first [[), and remove the space we added
1606                 $s = $a->current();
1607                 $a->next();
1608                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1609                 $s = substr( $s, 1 );
1610
1611                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1612                 $e2 = null;
1613                 if ( $useLinkPrefixExtension ) {
1614                         # Match the end of a line for a word that's not followed by whitespace,
1615                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1616                         $e2 = wfMsgForContent( 'linkprefix' );
1617                 }
1618
1619                 if ( is_null( $this->mTitle ) ) {
1620                         wfProfileOut( __METHOD__.'-setup' );
1621                         wfProfileOut( __METHOD__ );
1622                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1623                 }
1624                 $nottalk = !$this->mTitle->isTalkPage();
1625
1626                 if ( $useLinkPrefixExtension ) {
1627                         $m = array();
1628                         if ( preg_match( $e2, $s, $m ) ) {
1629                                 $first_prefix = $m[2];
1630                         } else {
1631                                 $first_prefix = false;
1632                         }
1633                 } else {
1634                         $prefix = '';
1635                 }
1636
1637                 if ( $wgContLang->hasVariants() ) {
1638                         $selflink = $wgContLang->convertLinkToAllVariants( $this->mTitle->getPrefixedText() );
1639                 } else {
1640                         $selflink = array( $this->mTitle->getPrefixedText() );
1641                 }
1642                 $useSubpages = $this->areSubpagesAllowed();
1643                 wfProfileOut( __METHOD__.'-setup' );
1644
1645                 # Loop for each link
1646                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1647                         # Check for excessive memory usage
1648                         if ( $holders->isBig() ) {
1649                                 # Too big
1650                                 # Do the existence check, replace the link holders and clear the array
1651                                 $holders->replace( $s );
1652                                 $holders->clear();
1653                         }
1654
1655                         if ( $useLinkPrefixExtension ) {
1656                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1657                                 if ( preg_match( $e2, $s, $m ) ) {
1658                                         $prefix = $m[2];
1659                                         $s = $m[1];
1660                                 } else {
1661                                         $prefix='';
1662                                 }
1663                                 # first link
1664                                 if ( $first_prefix ) {
1665                                         $prefix = $first_prefix;
1666                                         $first_prefix = false;
1667                                 }
1668                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1669                         }
1670
1671                         $might_be_img = false;
1672
1673                         wfProfileIn( __METHOD__."-e1" );
1674                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1675                                 $text = $m[2];
1676                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1677                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1678                                 # the real problem is with the $e1 regex
1679                                 # See bug 1300.
1680                                 #
1681                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1682                                 # and no image is in sight. See bug 2095.
1683                                 #
1684                                 if ( $text !== '' &&
1685                                         substr( $m[3], 0, 1 ) === ']' &&
1686                                         strpos( $text, '[' ) !== false
1687                                 )
1688                                 {
1689                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1690                                         $m[3] = substr( $m[3], 1 );
1691                                 }
1692                                 # fix up urlencoded title texts
1693                                 if ( strpos( $m[1], '%' ) !== false ) {
1694                                         # Should anchors '#' also be rejected?
1695                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode( $m[1] ) );
1696                                 }
1697                                 $trail = $m[3];
1698                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1699                                 $might_be_img = true;
1700                                 $text = $m[2];
1701                                 if ( strpos( $m[1], '%' ) !== false ) {
1702                                         $m[1] = urldecode( $m[1] );
1703                                 }
1704                                 $trail = "";
1705                         } else { # Invalid form; output directly
1706                                 $s .= $prefix . '[[' . $line ;
1707                                 wfProfileOut( __METHOD__."-e1" );
1708                                 continue;
1709                         }
1710                         wfProfileOut( __METHOD__."-e1" );
1711                         wfProfileIn( __METHOD__."-misc" );
1712
1713                         # Don't allow internal links to pages containing
1714                         # PROTO: where PROTO is a valid URL protocol; these
1715                         # should be external links.
1716                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1717                                 $s .= $prefix . '[[' . $line ;
1718                                 wfProfileOut( __METHOD__."-misc" );
1719                                 continue;
1720                         }
1721
1722                         # Make subpage if necessary
1723                         if ( $useSubpages ) {
1724                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1725                         } else {
1726                                 $link = $m[1];
1727                         }
1728
1729                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1730                         if ( !$noforce ) {
1731                                 # Strip off leading ':'
1732                                 $link = substr( $link, 1 );
1733                         }
1734
1735                         wfProfileOut( __METHOD__."-misc" );
1736                         wfProfileIn( __METHOD__."-title" );
1737                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1738                         if ( $nt === null ) {
1739                                 $s .= $prefix . '[[' . $line;
1740                                 wfProfileOut( __METHOD__."-title" );
1741                                 continue;
1742                         }
1743
1744                         $ns = $nt->getNamespace();
1745                         $iw = $nt->getInterWiki();
1746                         wfProfileOut( __METHOD__."-title" );
1747
1748                         if ( $might_be_img ) { # if this is actually an invalid link
1749                                 wfProfileIn( __METHOD__."-might_be_img" );
1750                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1751                                         $found = false;
1752                                         while ( true ) {
1753                                                 # look at the next 'line' to see if we can close it there
1754                                                 $a->next();
1755                                                 $next_line = $a->current();
1756                                                 if ( $next_line === false || $next_line === null ) {
1757                                                         break;
1758                                                 }
1759                                                 $m = explode( ']]', $next_line, 3 );
1760                                                 if ( count( $m ) == 3 ) {
1761                                                         # the first ]] closes the inner link, the second the image
1762                                                         $found = true;
1763                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1764                                                         $trail = $m[2];
1765                                                         break;
1766                                                 } elseif ( count( $m ) == 2 ) {
1767                                                         # if there's exactly one ]] that's fine, we'll keep looking
1768                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1769                                                 } else {
1770                                                         # if $next_line is invalid too, we need look no further
1771                                                         $text .= '[[' . $next_line;
1772                                                         break;
1773                                                 }
1774                                         }
1775                                         if ( !$found ) {
1776                                                 # we couldn't find the end of this imageLink, so output it raw
1777                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1778                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1779                                                 $s .= "{$prefix}[[$link|$text";
1780                                                 # note: no $trail, because without an end, there *is* no trail
1781                                                 wfProfileOut( __METHOD__."-might_be_img" );
1782                                                 continue;
1783                                         }
1784                                 } else { # it's not an image, so output it raw
1785                                         $s .= "{$prefix}[[$link|$text";
1786                                         # note: no $trail, because without an end, there *is* no trail
1787                                         wfProfileOut( __METHOD__."-might_be_img" );
1788                                         continue;
1789                                 }
1790                                 wfProfileOut( __METHOD__."-might_be_img" );
1791                         }
1792
1793                         $wasblank = ( $text  == '' );
1794                         if ( $wasblank ) {
1795                                 $text = $link;
1796                         }
1797
1798                         # Link not escaped by : , create the various objects
1799                         if ( $noforce ) {
1800
1801                                 # Interwikis
1802                                 wfProfileIn( __METHOD__."-interwiki" );
1803                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1804                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1805                                         $s = rtrim( $s . $prefix );
1806                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1807                                         wfProfileOut( __METHOD__."-interwiki" );
1808                                         continue;
1809                                 }
1810                                 wfProfileOut( __METHOD__."-interwiki" );
1811
1812                                 if ( $ns == NS_FILE ) {
1813                                         wfProfileIn( __METHOD__."-image" );
1814                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1815                                                 if ( $wasblank ) {
1816                                                         # if no parameters were passed, $text
1817                                                         # becomes something like "File:Foo.png",
1818                                                         # which we don't want to pass on to the
1819                                                         # image generator
1820                                                         $text = '';
1821                                                 } else {
1822                                                         # recursively parse links inside the image caption
1823                                                         # actually, this will parse them in any other parameters, too,
1824                                                         # but it might be hard to fix that, and it doesn't matter ATM
1825                                                         $text = $this->replaceExternalLinks( $text );
1826                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1827                                                 }
1828                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1829                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1830                                         } else {
1831                                                 $s .= $prefix . $trail;
1832                                         }
1833                                         $this->mOutput->addImage( $nt->getDBkey() );
1834                                         wfProfileOut( __METHOD__."-image" );
1835                                         continue;
1836
1837                                 }
1838
1839                                 if ( $ns == NS_CATEGORY ) {
1840                                         wfProfileIn( __METHOD__."-category" );
1841                                         $s = rtrim( $s . "\n" ); # bug 87
1842
1843                                         if ( $wasblank ) {
1844                                                 $sortkey = $this->getDefaultSort();
1845                                         } else {
1846                                                 $sortkey = $text;
1847                                         }
1848                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1849                                         $sortkey = str_replace( "\n", '', $sortkey );
1850                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1851                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1852
1853                                         /**
1854                                          * Strip the whitespace Category links produce, see bug 87
1855                                          * @todo We might want to use trim($tmp, "\n") here.
1856                                          */
1857                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '': $prefix . $trail;
1858
1859                                         wfProfileOut( __METHOD__."-category" );
1860                                         continue;
1861                                 }
1862                         }
1863
1864                         # Self-link checking
1865                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1866                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1867                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1868                                         continue;
1869                                 }
1870                         }
1871
1872                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1873                         # FIXME: Should do batch file existence checks, see comment below
1874                         if ( $ns == NS_MEDIA ) {
1875                                 wfProfileIn( __METHOD__."-media" );
1876                                 # Give extensions a chance to select the file revision for us
1877                                 $skip = $time = false;
1878                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1879                                 if ( $skip ) {
1880                                         $link = $sk->link( $nt );
1881                                 } else {
1882                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1883                                 }
1884                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1885                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1886                                 $this->mOutput->addImage( $nt->getDBkey() );
1887                                 wfProfileOut( __METHOD__."-media" );
1888                                 continue;
1889                         }
1890
1891                         wfProfileIn( __METHOD__."-always_known" );
1892                         # Some titles, such as valid special pages or files in foreign repos, should
1893                         # be shown as bluelinks even though they're not included in the page table
1894                         #
1895                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1896                         # batch file existence checks for NS_FILE and NS_MEDIA
1897                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1898                                 $this->mOutput->addLink( $nt );
1899                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1900                         } else {
1901                                 # Links will be added to the output link list after checking
1902                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1903                         }
1904                         wfProfileOut( __METHOD__."-always_known" );
1905                 }
1906                 wfProfileOut( __METHOD__ );
1907                 return $holders;
1908         }
1909
1910         /**
1911          * Make a link placeholder. The text returned can be later resolved to a real link with
1912          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1913          * parsing of interwiki links, and secondly to allow all existence checks and
1914          * article length checks (for stub links) to be bundled into a single query.
1915          *
1916          * @deprecated
1917          */
1918         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1919                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1920         }
1921
1922         /**
1923          * Render a forced-blue link inline; protect against double expansion of
1924          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1925          * Since this little disaster has to split off the trail text to avoid
1926          * breaking URLs in the following text without breaking trails on the
1927          * wiki links, it's been made into a horrible function.
1928          *
1929          * @param $nt Title
1930          * @param $text String
1931          * @param $query String
1932          * @param $trail String
1933          * @param $prefix String
1934          * @return String: HTML-wikitext mix oh yuck
1935          */
1936         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1937                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1938                 $sk = $this->mOptions->getSkin();
1939                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1940                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1941                 return $this->armorLinks( $link ) . $trail;
1942         }
1943
1944         /**
1945          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1946          * going to go through further parsing steps before inline URL expansion.
1947          *
1948          * Not needed quite as much as it used to be since free links are a bit
1949          * more sensible these days. But bracketed links are still an issue.
1950          *
1951          * @param $text String: more-or-less HTML
1952          * @return String: less-or-more HTML with NOPARSE bits
1953          */
1954         function armorLinks( $text ) {
1955                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1956                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1957         }
1958
1959         /**
1960          * Return true if subpage links should be expanded on this page.
1961          * @return Boolean
1962          */
1963         function areSubpagesAllowed() {
1964                 # Some namespaces don't allow subpages
1965                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1966         }
1967
1968         /**
1969          * Handle link to subpage if necessary
1970          *
1971          * @param $target String: the source of the link
1972          * @param &$text String: the link text, modified as necessary
1973          * @return string the full name of the link
1974          * @private
1975          */
1976         function maybeDoSubpageLink( $target, &$text ) {
1977                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1978         }
1979
1980         /**#@+
1981          * Used by doBlockLevels()
1982          * @private
1983          */
1984         function closeParagraph() {
1985                 $result = '';
1986                 if ( $this->mLastSection != '' ) {
1987                         $result = '</' . $this->mLastSection  . ">\n";
1988                 }
1989                 $this->mInPre = false;
1990                 $this->mLastSection = '';
1991                 return $result;
1992         }
1993
1994         /**
1995          * getCommon() returns the length of the longest common substring
1996          * of both arguments, starting at the beginning of both.
1997          * @private
1998          */
1999         function getCommon( $st1, $st2 ) {
2000                 $fl = strlen( $st1 );
2001                 $shorter = strlen( $st2 );
2002                 if ( $fl < $shorter ) {
2003                         $shorter = $fl;
2004                 }
2005
2006                 for ( $i = 0; $i < $shorter; ++$i ) {
2007                         if ( $st1{$i} != $st2{$i} ) {
2008                                 break;
2009                         }
2010                 }
2011                 return $i;
2012         }
2013
2014         /**
2015          * These next three functions open, continue, and close the list
2016          * element appropriate to the prefix character passed into them.
2017          * @private
2018          */
2019         function openList( $char ) {
2020                 $result = $this->closeParagraph();
2021
2022                 if ( '*' === $char ) {
2023                         $result .= '<ul><li>';
2024                 } elseif ( '#' === $char ) {
2025                         $result .= '<ol><li>';
2026                 } elseif ( ':' === $char ) {
2027                         $result .= '<dl><dd>';
2028                 } elseif ( ';' === $char ) {
2029                         $result .= '<dl><dt>';
2030                         $this->mDTopen = true;
2031                 } else {
2032                         $result = '<!-- ERR 1 -->';
2033                 }
2034
2035                 return $result;
2036         }
2037
2038         /**
2039          * TODO: document
2040          * @param $char String
2041          * @private
2042          */
2043         function nextItem( $char ) {
2044                 if ( '*' === $char || '#' === $char ) {
2045                         return '</li><li>';
2046                 } elseif ( ':' === $char || ';' === $char ) {
2047                         $close = '</dd>';
2048                         if ( $this->mDTopen ) {
2049                                 $close = '</dt>';
2050                         }
2051                         if ( ';' === $char ) {
2052                                 $this->mDTopen = true;
2053                                 return $close . '<dt>';
2054                         } else {
2055                                 $this->mDTopen = false;
2056                                 return $close . '<dd>';
2057                         }
2058                 }
2059                 return '<!-- ERR 2 -->';
2060         }
2061
2062         /**
2063          * TODO: document
2064          * @param $char String
2065          * @private
2066          */
2067         function closeList( $char ) {
2068                 if ( '*' === $char ) {
2069                         $text = '</li></ul>';
2070                 } elseif ( '#' === $char ) {
2071                         $text = '</li></ol>';
2072                 } elseif ( ':' === $char ) {
2073                         if ( $this->mDTopen ) {
2074                                 $this->mDTopen = false;
2075                                 $text = '</dt></dl>';
2076                         } else {
2077                                 $text = '</dd></dl>';
2078                         }
2079                 } else {
2080                         return '<!-- ERR 3 -->';
2081                 }
2082                 return $text."\n";
2083         }
2084         /**#@-*/
2085
2086         /**
2087          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2088          *
2089          * @param $text String
2090          * @param $linestart Boolean: whether or not this is at the start of a line.
2091          * @private
2092          * @return string the lists rendered as HTML
2093          */
2094         function doBlockLevels( $text, $linestart ) {
2095                 wfProfileIn( __METHOD__ );
2096
2097                 # Parsing through the text line by line.  The main thing
2098                 # happening here is handling of block-level elements p, pre,
2099                 # and making lists from lines starting with * # : etc.
2100                 #
2101                 $textLines = StringUtils::explode( "\n", $text );
2102
2103                 $lastPrefix = $output = '';
2104                 $this->mDTopen = $inBlockElem = false;
2105                 $prefixLength = 0;
2106                 $paragraphStack = false;
2107
2108                 foreach ( $textLines as $oLine ) {
2109                         # Fix up $linestart
2110                         if ( !$linestart ) {
2111                                 $output .= $oLine;
2112                                 $linestart = true;
2113                                 continue;
2114                         }
2115                         # * = ul
2116                         # # = ol
2117                         # ; = dt
2118                         # : = dd
2119
2120                         $lastPrefixLength = strlen( $lastPrefix );
2121                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2122                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2123                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2124                         if ( !$this->mInPre ) {
2125                                 # Multiple prefixes may abut each other for nested lists.
2126                                 $prefixLength = strspn( $oLine, '*#:;' );
2127                                 $prefix = substr( $oLine, 0, $prefixLength );
2128
2129                                 # eh?
2130                                 # ; and : are both from definition-lists, so they're equivalent
2131                                 #  for the purposes of determining whether or not we need to open/close
2132                                 #  elements.
2133                                 $prefix2 = str_replace( ';', ':', $prefix );
2134                                 $t = substr( $oLine, $prefixLength );
2135                                 $this->mInPre = (bool)$preOpenMatch;
2136                         } else {
2137                                 # Don't interpret any other prefixes in preformatted text
2138                                 $prefixLength = 0;
2139                                 $prefix = $prefix2 = '';
2140                                 $t = $oLine;
2141                         }
2142
2143                         # List generation
2144                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2145                                 # Same as the last item, so no need to deal with nesting or opening stuff
2146                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2147                                 $paragraphStack = false;
2148
2149                                 if ( substr( $prefix, -1 ) === ';') {
2150                                         # The one nasty exception: definition lists work like this:
2151                                         # ; title : definition text
2152                                         # So we check for : in the remainder text to split up the
2153                                         # title and definition, without b0rking links.
2154                                         $term = $t2 = '';
2155                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2156                                                 $t = $t2;
2157                                                 $output .= $term . $this->nextItem( ':' );
2158                                         }
2159                                 }
2160                         } elseif ( $prefixLength || $lastPrefixLength ) {
2161                                 # We need to open or close prefixes, or both.
2162
2163                                 # Either open or close a level...
2164                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2165                                 $paragraphStack = false;
2166
2167                                 # Close all the prefixes which aren't shared.
2168                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2169                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2170                                         --$lastPrefixLength;
2171                                 }
2172
2173                                 # Continue the current prefix if appropriate.
2174                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2175                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2176                                 }
2177
2178                                 # Open prefixes where appropriate.
2179                                 while ( $prefixLength > $commonPrefixLength ) {
2180                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2181                                         $output .= $this->openList( $char );
2182
2183                                         if ( ';' === $char ) {
2184                                                 # FIXME: This is dupe of code above
2185                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2186                                                         $t = $t2;
2187                                                         $output .= $term . $this->nextItem( ':' );
2188                                                 }
2189                                         }
2190                                         ++$commonPrefixLength;
2191                                 }
2192                                 $lastPrefix = $prefix2;
2193                         }
2194
2195                         # If we have no prefixes, go to paragraph mode.
2196                         if ( 0 == $prefixLength ) {
2197                                 wfProfileIn( __METHOD__."-paragraph" );
2198                                 # No prefix (not in list)--go to paragraph mode
2199                                 # XXX: use a stack for nestable elements like span, table and div
2200                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2201                                 $closematch = preg_match(
2202                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2203                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2204                                 if ( $openmatch or $closematch ) {
2205                                         $paragraphStack = false;
2206                                         # TODO bug 5718: paragraph closed
2207                                         $output .= $this->closeParagraph();
2208                                         if ( $preOpenMatch and !$preCloseMatch ) {
2209                                                 $this->mInPre = true;
2210                                         }
2211                                         if ( $closematch ) {
2212                                                 $inBlockElem = false;
2213                                         } else {
2214                                                 $inBlockElem = true;
2215                                         }
2216                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2217                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2218                                                 # pre
2219                                                 if ( $this->mLastSection !== 'pre' ) {
2220                                                         $paragraphStack = false;
2221                                                         $output .= $this->closeParagraph().'<pre>';
2222                                                         $this->mLastSection = 'pre';
2223                                                 }
2224                                                 $t = substr( $t, 1 );
2225                                         } else {
2226                                                 # paragraph
2227                                                 if ( trim( $t ) === '' ) {
2228                                                         if ( $paragraphStack ) {
2229                                                                 $output .= $paragraphStack.'<br />';
2230                                                                 $paragraphStack = false;
2231                                                                 $this->mLastSection = 'p';
2232                                                         } else {
2233                                                                 if ( $this->mLastSection !== 'p' ) {
2234                                                                         $output .= $this->closeParagraph();
2235                                                                         $this->mLastSection = '';
2236                                                                         $paragraphStack = '<p>';
2237                                                                 } else {
2238                                                                         $paragraphStack = '</p><p>';
2239                                                                 }
2240                                                         }
2241                                                 } else {
2242                                                         if ( $paragraphStack ) {
2243                                                                 $output .= $paragraphStack;
2244                                                                 $paragraphStack = false;
2245                                                                 $this->mLastSection = 'p';
2246                                                         } elseif ( $this->mLastSection !== 'p' ) {
2247                                                                 $output .= $this->closeParagraph().'<p>';
2248                                                                 $this->mLastSection = 'p';
2249                                                         }
2250                                                 }
2251                                         }
2252                                 }
2253                                 wfProfileOut( __METHOD__."-paragraph" );
2254                         }
2255                         # somewhere above we forget to get out of pre block (bug 785)
2256                         if ( $preCloseMatch && $this->mInPre ) {
2257                                 $this->mInPre = false;
2258                         }
2259                         if ( $paragraphStack === false ) {
2260                                 $output .= $t."\n";
2261                         }
2262                 }
2263                 while ( $prefixLength ) {
2264                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2265                         --$prefixLength;
2266                 }
2267                 if ( $this->mLastSection != '' ) {
2268                         $output .= '</' . $this->mLastSection . '>';
2269                         $this->mLastSection = '';
2270                 }
2271
2272                 wfProfileOut( __METHOD__ );
2273                 return $output;
2274         }
2275
2276         /**
2277          * Split up a string on ':', ignoring any occurences inside tags
2278          * to prevent illegal overlapping.
2279          *
2280          * @param $str String: the string to split
2281          * @param &$before String: set to everything before the ':'
2282          * @param &$after String: set to everything after the ':'
2283          * return String: the position of the ':', or false if none found
2284          */
2285         function findColonNoLinks( $str, &$before, &$after ) {
2286                 wfProfileIn( __METHOD__ );
2287
2288                 $pos = strpos( $str, ':' );
2289                 if ( $pos === false ) {
2290                         # Nothing to find!
2291                         wfProfileOut( __METHOD__ );
2292                         return false;
2293                 }
2294
2295                 $lt = strpos( $str, '<' );
2296                 if ( $lt === false || $lt > $pos ) {
2297                         # Easy; no tag nesting to worry about
2298                         $before = substr( $str, 0, $pos );
2299                         $after = substr( $str, $pos+1 );
2300                         wfProfileOut( __METHOD__ );
2301                         return $pos;
2302                 }
2303
2304                 # Ugly state machine to walk through avoiding tags.
2305                 $state = self::COLON_STATE_TEXT;
2306                 $stack = 0;
2307                 $len = strlen( $str );
2308                 for( $i = 0; $i < $len; $i++ ) {
2309                         $c = $str{$i};
2310
2311                         switch( $state ) {
2312                         # (Using the number is a performance hack for common cases)
2313                         case 0: # self::COLON_STATE_TEXT:
2314                                 switch( $c ) {
2315                                 case "<":
2316                                         # Could be either a <start> tag or an </end> tag
2317                                         $state = self::COLON_STATE_TAGSTART;
2318                                         break;
2319                                 case ":":
2320                                         if ( $stack == 0 ) {
2321                                                 # We found it!
2322                                                 $before = substr( $str, 0, $i );
2323                                                 $after = substr( $str, $i + 1 );
2324                                                 wfProfileOut( __METHOD__ );
2325                                                 return $i;
2326                                         }
2327                                         # Embedded in a tag; don't break it.
2328                                         break;
2329                                 default:
2330                                         # Skip ahead looking for something interesting
2331                                         $colon = strpos( $str, ':', $i );
2332                                         if ( $colon === false ) {
2333                                                 # Nothing else interesting
2334                                                 wfProfileOut( __METHOD__ );
2335                                                 return false;
2336                                         }
2337                                         $lt = strpos( $str, '<', $i );
2338                                         if ( $stack === 0 ) {
2339                                                 if ( $lt === false || $colon < $lt ) {
2340                                                         # We found it!
2341                                                         $before = substr( $str, 0, $colon );
2342                                                         $after = substr( $str, $colon + 1 );
2343                                                         wfProfileOut( __METHOD__ );
2344                                                         return $i;
2345                                                 }
2346                                         }
2347                                         if ( $lt === false ) {
2348                                                 # Nothing else interesting to find; abort!
2349                                                 # We're nested, but there's no close tags left. Abort!
2350                                                 break 2;
2351                                         }
2352                                         # Skip ahead to next tag start
2353                                         $i = $lt;
2354                                         $state = self::COLON_STATE_TAGSTART;
2355                                 }
2356                                 break;
2357                         case 1: # self::COLON_STATE_TAG:
2358                                 # In a <tag>
2359                                 switch( $c ) {
2360                                 case ">":
2361                                         $stack++;
2362                                         $state = self::COLON_STATE_TEXT;
2363                                         break;
2364                                 case "/":
2365                                         # Slash may be followed by >?
2366                                         $state = self::COLON_STATE_TAGSLASH;
2367                                         break;
2368                                 default:
2369                                         # ignore
2370                                 }
2371                                 break;
2372                         case 2: # self::COLON_STATE_TAGSTART:
2373                                 switch( $c ) {
2374                                 case "/":
2375                                         $state = self::COLON_STATE_CLOSETAG;
2376                                         break;
2377                                 case "!":
2378                                         $state = self::COLON_STATE_COMMENT;
2379                                         break;
2380                                 case ">":
2381                                         # Illegal early close? This shouldn't happen D:
2382                                         $state = self::COLON_STATE_TEXT;
2383                                         break;
2384                                 default:
2385                                         $state = self::COLON_STATE_TAG;
2386                                 }
2387                                 break;
2388                         case 3: # self::COLON_STATE_CLOSETAG:
2389                                 # In a </tag>
2390                                 if ( $c === ">" ) {
2391                                         $stack--;
2392                                         if ( $stack < 0 ) {
2393                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2394                                                 wfProfileOut( __METHOD__ );
2395                                                 return false;
2396                                         }
2397                                         $state = self::COLON_STATE_TEXT;
2398                                 }
2399                                 break;
2400                         case self::COLON_STATE_TAGSLASH:
2401                                 if ( $c === ">" ) {
2402                                         # Yes, a self-closed tag <blah/>
2403                                         $state = self::COLON_STATE_TEXT;
2404                                 } else {
2405                                         # Probably we're jumping the gun, and this is an attribute
2406                                         $state = self::COLON_STATE_TAG;
2407                                 }
2408                                 break;
2409                         case 5: # self::COLON_STATE_COMMENT:
2410                                 if ( $c === "-" ) {
2411                                         $state = self::COLON_STATE_COMMENTDASH;
2412                                 }
2413                                 break;
2414                         case self::COLON_STATE_COMMENTDASH:
2415                                 if ( $c === "-" ) {
2416                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2417                                 } else {
2418                                         $state = self::COLON_STATE_COMMENT;
2419                                 }
2420                                 break;
2421                         case self::COLON_STATE_COMMENTDASHDASH:
2422                                 if ( $c === ">" ) {
2423                                         $state = self::COLON_STATE_TEXT;
2424                                 } else {
2425                                         $state = self::COLON_STATE_COMMENT;
2426                                 }
2427                                 break;
2428                         default:
2429                                 throw new MWException( "State machine error in " . __METHOD__ );
2430                         }
2431                 }
2432                 if ( $stack > 0 ) {
2433                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2434                         return false;
2435                 }
2436                 wfProfileOut( __METHOD__ );
2437                 return false;
2438         }
2439
2440         /**
2441          * Return value of a magic variable (like PAGENAME)
2442          *
2443          * @private
2444          */
2445         function getVariableValue( $index, $frame=false ) {
2446                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2447                 global $wgScriptPath, $wgStylePath;
2448
2449                 /**
2450                  * Some of these require message or data lookups and can be
2451                  * expensive to check many times.
2452                  */
2453                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2454                         if ( isset( $this->mVarCache[$index] ) ) {
2455                                 return $this->mVarCache[$index];
2456                         }
2457                 }
2458
2459                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2460                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2461
2462                 # Use the time zone
2463                 global $wgLocaltimezone;
2464                 if ( isset( $wgLocaltimezone ) ) {
2465                         $oldtz = date_default_timezone_get();
2466                         date_default_timezone_set( $wgLocaltimezone );
2467                 }
2468
2469                 $localTimestamp = date( 'YmdHis', $ts );
2470                 $localMonth = date( 'm', $ts );
2471                 $localMonth1 = date( 'n', $ts );
2472                 $localMonthName = date( 'n', $ts );
2473                 $localDay = date( 'j', $ts );
2474                 $localDay2 = date( 'd', $ts );
2475                 $localDayOfWeek = date( 'w', $ts );
2476                 $localWeek = date( 'W', $ts );
2477                 $localYear = date( 'Y', $ts );
2478                 $localHour = date( 'H', $ts );
2479                 if ( isset( $wgLocaltimezone ) ) {
2480                         date_default_timezone_set( $oldtz );
2481                 }
2482
2483                 switch ( $index ) {
2484                         case 'currentmonth':
2485                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2486                                 break;
2487                         case 'currentmonth1':
2488                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2489                                 break;
2490                         case 'currentmonthname':
2491                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2492                                 break;
2493                         case 'currentmonthnamegen':
2494                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2495                                 break;
2496                         case 'currentmonthabbrev':
2497                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2498                                 break;
2499                         case 'currentday':
2500                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2501                                 break;
2502                         case 'currentday2':
2503                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2504                                 break;
2505                         case 'localmonth':
2506                                 $value = $wgContLang->formatNum( $localMonth );
2507                                 break;
2508                         case 'localmonth1':
2509                                 $value = $wgContLang->formatNum( $localMonth1 );
2510                                 break;
2511                         case 'localmonthname':
2512                                 $value = $wgContLang->getMonthName( $localMonthName );
2513                                 break;
2514                         case 'localmonthnamegen':
2515                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2516                                 break;
2517                         case 'localmonthabbrev':
2518                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2519                                 break;
2520                         case 'localday':
2521                                 $value = $wgContLang->formatNum( $localDay );
2522                                 break;
2523                         case 'localday2':
2524                                 $value = $wgContLang->formatNum( $localDay2 );
2525                                 break;
2526                         case 'pagename':
2527                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2528                                 break;
2529                         case 'pagenamee':
2530                                 $value = $this->mTitle->getPartialURL();
2531                                 break;
2532                         case 'fullpagename':
2533                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2534                                 break;
2535                         case 'fullpagenamee':
2536                                 $value = $this->mTitle->getPrefixedURL();
2537                                 break;
2538                         case 'subpagename':
2539                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2540                                 break;
2541                         case 'subpagenamee':
2542                                 $value = $this->mTitle->getSubpageUrlForm();
2543                                 break;
2544                         case 'basepagename':
2545                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2546                                 break;
2547                         case 'basepagenamee':
2548                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2549                                 break;
2550                         case 'talkpagename':
2551                                 if ( $this->mTitle->canTalk() ) {
2552                                         $talkPage = $this->mTitle->getTalkPage();
2553                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2554                                 } else {
2555                                         $value = '';
2556                                 }
2557                                 break;
2558                         case 'talkpagenamee':
2559                                 if ( $this->mTitle->canTalk() ) {
2560                                         $talkPage = $this->mTitle->getTalkPage();
2561                                         $value = $talkPage->getPrefixedUrl();
2562                                 } else {
2563                                         $value = '';
2564                                 }
2565                                 break;
2566                         case 'subjectpagename':
2567                                 $subjPage = $this->mTitle->getSubjectPage();
2568                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2569                                 break;
2570                         case 'subjectpagenamee':
2571                                 $subjPage = $this->mTitle->getSubjectPage();
2572                                 $value = $subjPage->getPrefixedUrl();
2573                                 break;
2574                         case 'revisionid':
2575                                 # Let the edit saving system know we should parse the page
2576                                 # *after* a revision ID has been assigned.
2577                                 $this->mOutput->setFlag( 'vary-revision' );
2578                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2579                                 $value = $this->mRevisionId;
2580                                 break;
2581                         case 'revisionday':
2582                                 # Let the edit saving system know we should parse the page
2583                                 # *after* a revision ID has been assigned. This is for null edits.
2584                                 $this->mOutput->setFlag( 'vary-revision' );
2585                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2586                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2587                                 break;
2588                         case 'revisionday2':
2589                                 # Let the edit saving system know we should parse the page
2590                                 # *after* a revision ID has been assigned. This is for null edits.
2591                                 $this->mOutput->setFlag( 'vary-revision' );
2592                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2593                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2594                                 break;
2595                         case 'revisionmonth':
2596                                 # Let the edit saving system know we should parse the page
2597                                 # *after* a revision ID has been assigned. This is for null edits.
2598                                 $this->mOutput->setFlag( 'vary-revision' );
2599                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2600                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2601                                 break;
2602                         case 'revisionmonth1':
2603                                 # Let the edit saving system know we should parse the page
2604                                 # *after* a revision ID has been assigned. This is for null edits.
2605                                 $this->mOutput->setFlag( 'vary-revision' );
2606                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2607                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2608                                 break;
2609                         case 'revisionyear':
2610                                 # Let the edit saving system know we should parse the page
2611                                 # *after* a revision ID has been assigned. This is for null edits.
2612                                 $this->mOutput->setFlag( 'vary-revision' );
2613                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2614                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2615                                 break;
2616                         case 'revisiontimestamp':
2617                                 # Let the edit saving system know we should parse the page
2618                                 # *after* a revision ID has been assigned. This is for null edits.
2619                                 $this->mOutput->setFlag( 'vary-revision' );
2620                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2621                                 $value = $this->getRevisionTimestamp();
2622                                 break;
2623                         case 'revisionuser':
2624                                 # Let the edit saving system know we should parse the page
2625                                 # *after* a revision ID has been assigned. This is for null edits.
2626                                 $this->mOutput->setFlag( 'vary-revision' );
2627                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2628                                 $value = $this->getRevisionUser();
2629                                 break;
2630                         case 'namespace':
2631                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2632                                 break;
2633                         case 'namespacee':
2634                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2635                                 break;
2636                         case 'talkspace':
2637                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2638                                 break;
2639                         case 'talkspacee':
2640                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2641                                 break;
2642                         case 'subjectspace':
2643                                 $value = $this->mTitle->getSubjectNsText();
2644                                 break;
2645                         case 'subjectspacee':
2646                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2647                                 break;
2648                         case 'currentdayname':
2649                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2650                                 break;
2651                         case 'currentyear':
2652                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2653                                 break;
2654                         case 'currenttime':
2655                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2656                                 break;
2657                         case 'currenthour':
2658                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2659                                 break;
2660                         case 'currentweek':
2661                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2662                                 # int to remove the padding
2663                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2664                                 break;
2665                         case 'currentdow':
2666                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2667                                 break;
2668                         case 'localdayname':
2669                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2670                                 break;
2671                         case 'localyear':
2672                                 $value = $wgContLang->formatNum( $localYear, true );
2673                                 break;
2674                         case 'localtime':
2675                                 $value = $wgContLang->time( $localTimestamp, false, false );
2676                                 break;
2677                         case 'localhour':
2678                                 $value = $wgContLang->formatNum( $localHour, true );
2679                                 break;
2680                         case 'localweek':
2681                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2682                                 # int to remove the padding
2683                                 $value = $wgContLang->formatNum( (int)$localWeek );
2684                                 break;
2685                         case 'localdow':
2686                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2687                                 break;
2688                         case 'numberofarticles':
2689                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2690                                 break;
2691                         case 'numberoffiles':
2692                                 $value = $wgContLang->formatNum( SiteStats::images() );
2693                                 break;
2694                         case 'numberofusers':
2695                                 $value = $wgContLang->formatNum( SiteStats::users() );
2696                                 break;
2697                         case 'numberofactiveusers':
2698                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2699                                 break;
2700                         case 'numberofpages':
2701                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2702                                 break;
2703                         case 'numberofadmins':
2704                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2705                                 break;
2706                         case 'numberofedits':
2707                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2708                                 break;
2709                         case 'numberofviews':
2710                                 $value = $wgContLang->formatNum( SiteStats::views() );
2711                                 break;
2712                         case 'currenttimestamp':
2713                                 $value = wfTimestamp( TS_MW, $ts );
2714                                 break;
2715                         case 'localtimestamp':
2716                                 $value = $localTimestamp;
2717                                 break;
2718                         case 'currentversion':
2719                                 $value = SpecialVersion::getVersion();
2720                                 break;
2721                         case 'sitename':
2722                                 return $wgSitename;
2723                         case 'server':
2724                                 return $wgServer;
2725                         case 'servername':
2726                                 return $wgServerName;
2727                         case 'scriptpath':
2728                                 return $wgScriptPath;
2729                         case 'stylepath':
2730                                 return $wgStylePath;
2731                         case 'directionmark':
2732                                 return $wgContLang->getDirMark();
2733                         case 'contentlanguage':
2734                                 global $wgContLanguageCode;
2735                                 return $wgContLanguageCode;
2736                         default:
2737                                 $ret = null;
2738                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2739                                         return $ret;
2740                                 } else {
2741                                         return null;
2742                                 }
2743                 }
2744
2745                 if ( $index )
2746                         $this->mVarCache[$index] = $value;
2747
2748                 return $value;
2749         }
2750
2751         /**
2752          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2753          *
2754          * @private
2755          */
2756         function initialiseVariables() {
2757                 wfProfileIn( __METHOD__ );
2758                 $variableIDs = MagicWord::getVariableIDs();
2759                 $substIDs = MagicWord::getSubstIDs();
2760
2761                 $this->mVariables = new MagicWordArray( $variableIDs );
2762                 $this->mSubstWords = new MagicWordArray( $substIDs );
2763                 wfProfileOut( __METHOD__ );
2764         }
2765
2766         /**
2767          * Preprocess some wikitext and return the document tree.
2768          * This is the ghost of replace_variables().
2769          *
2770          * @param $text String: The text to parse
2771          * @param $flags Integer: bitwise combination of:
2772          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2773          *                                     included. Default is to assume a direct page view.
2774          *
2775          * The generated DOM tree must depend only on the input text and the flags.
2776          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2777          *
2778          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2779          * change in the DOM tree for a given text, must be passed through the section identifier
2780          * in the section edit link and thus back to extractSections().
2781          *
2782          * The output of this function is currently only cached in process memory, but a persistent
2783          * cache may be implemented at a later date which takes further advantage of these strict
2784          * dependency requirements.
2785          *
2786          * @private
2787          */
2788         function preprocessToDom( $text, $flags = 0 ) {
2789                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2790                 return $dom;
2791         }
2792
2793         /**
2794          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2795          */
2796         public static function splitWhitespace( $s ) {
2797                 $ltrimmed = ltrim( $s );
2798                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2799                 $trimmed = rtrim( $ltrimmed );
2800                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2801                 if ( $diff > 0 ) {
2802                         $w2 = substr( $ltrimmed, -$diff );
2803                 } else {
2804                         $w2 = '';
2805                 }
2806                 return array( $w1, $trimmed, $w2 );
2807         }
2808
2809         /**
2810          * Replace magic variables, templates, and template arguments
2811          * with the appropriate text. Templates are substituted recursively,
2812          * taking care to avoid infinite loops.
2813          *
2814          * Note that the substitution depends on value of $mOutputType:
2815          *  self::OT_WIKI: only {{subst:}} templates
2816          *  self::OT_PREPROCESS: templates but not extension tags
2817          *  self::OT_HTML: all templates and extension tags
2818          *
2819          * @param $text String: the text to transform
2820          * @param $frame PPFrame Object describing the arguments passed to the template.
2821          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2822          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2823          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2824          * @private
2825          */
2826         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2827                 # Is there any text? Also, Prevent too big inclusions!
2828                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2829                         return $text;
2830                 }
2831                 wfProfileIn( __METHOD__ );
2832
2833                 if ( $frame === false ) {
2834                         $frame = $this->getPreprocessor()->newFrame();
2835                 } elseif ( !( $frame instanceof PPFrame ) ) {
2836                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2837                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2838                 }
2839
2840                 $dom = $this->preprocessToDom( $text );
2841                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2842                 $text = $frame->expand( $dom, $flags );
2843
2844                 wfProfileOut( __METHOD__ );
2845                 return $text;
2846         }
2847
2848         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2849         static function createAssocArgs( $args ) {
2850                 $assocArgs = array();
2851                 $index = 1;
2852                 foreach ( $args as $arg ) {
2853                         $eqpos = strpos( $arg, '=' );
2854                         if ( $eqpos === false ) {
2855                                 $assocArgs[$index++] = $arg;
2856                         } else {
2857                                 $name = trim( substr( $arg, 0, $eqpos ) );
2858                                 $value = trim( substr( $arg, $eqpos+1 ) );
2859                                 if ( $value === false ) {
2860                                         $value = '';
2861                                 }
2862                                 if ( $name !== false ) {
2863                                         $assocArgs[$name] = $value;
2864                                 }
2865                         }
2866                 }
2867
2868                 return $assocArgs;
2869         }
2870
2871         /**
2872          * Warn the user when a parser limitation is reached
2873          * Will warn at most once the user per limitation type
2874          *
2875          * @param $limitationType String: should be one of:
2876          *   'expensive-parserfunction' (corresponding messages:
2877          *       'expensive-parserfunction-warning',
2878          *       'expensive-parserfunction-category')
2879          *   'post-expand-template-argument' (corresponding messages:
2880          *       'post-expand-template-argument-warning',
2881          *       'post-expand-template-argument-category')
2882          *   'post-expand-template-inclusion' (corresponding messages:
2883          *       'post-expand-template-inclusion-warning',
2884          *       'post-expand-template-inclusion-category')
2885          * @param $current Current value
2886          * @param $max Maximum allowed, when an explicit limit has been
2887          *       exceeded, provide the values (optional)
2888          */
2889         function limitationWarn( $limitationType, $current=null, $max=null) {
2890                 # does no harm if $current and $max are present but are unnecessary for the message
2891                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2892                 $this->mOutput->addWarning( $warning );
2893                 $this->addTrackingCategory( "$limitationType-category" );
2894         }
2895
2896         /**
2897          * Return the text of a template, after recursively
2898          * replacing any variables or templates within the template.
2899          *
2900          * @param $piece Array: the parts of the template
2901          *  $piece['title']: the title, i.e. the part before the |
2902          *  $piece['parts']: the parameter array
2903          *  $piece['lineStart']: whether the brace was at the start of a line
2904          * @param $frame PPFrame The current frame, contains template arguments
2905          * @return String: the text of the template
2906          * @private
2907          */
2908         function braceSubstitution( $piece, $frame ) {
2909                 global $wgContLang, $wgNonincludableNamespaces;
2910                 wfProfileIn( __METHOD__ );
2911                 wfProfileIn( __METHOD__.'-setup' );
2912
2913                 # Flags
2914                 $found = false;             # $text has been filled
2915                 $nowiki = false;            # wiki markup in $text should be escaped
2916                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2917                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2918                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2919                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2920
2921                 # Title object, where $text came from
2922                 $title = null;
2923
2924                 # $part1 is the bit before the first |, and must contain only title characters.
2925                 # Various prefixes will be stripped from it later.
2926                 $titleWithSpaces = $frame->expand( $piece['title'] );
2927                 $part1 = trim( $titleWithSpaces );
2928                 $titleText = false;
2929
2930                 # Original title text preserved for various purposes
2931                 $originalTitle = $part1;
2932
2933                 # $args is a list of argument nodes, starting from index 0, not including $part1
2934                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2935                 wfProfileOut( __METHOD__.'-setup' );
2936
2937                 # SUBST
2938                 wfProfileIn( __METHOD__.'-modifiers' );
2939                 if ( !$found ) {
2940
2941                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2942
2943                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2944                         # Decide whether to expand template or keep wikitext as-is.
2945                         if ( $this->ot['wiki'] ) {
2946                                 if ( $substMatch === false ) {
2947                                         $literal = true;  # literal when in PST with no prefix
2948                                 } else {
2949                                         $literal = false; # expand when in PST with subst: or safesubst:
2950                                 }
2951                         } else {
2952                                 if ( $substMatch == 'subst' ) {
2953                                         $literal = true;  # literal when not in PST with plain subst:
2954                                 } else {
2955                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2956                                 }
2957                         }
2958                         if ( $literal ) {
2959                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2960                                 $isLocalObj = true;
2961                                 $found = true;
2962                         }
2963                 }
2964
2965                 # Variables
2966                 if ( !$found && $args->getLength() == 0 ) {
2967                         $id = $this->mVariables->matchStartToEnd( $part1 );
2968                         if ( $id !== false ) {
2969                                 $text = $this->getVariableValue( $id, $frame );
2970                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
2971                                         $this->mOutput->mContainsOldMagic = true;
2972                                 }
2973                                 $found = true;
2974                         }
2975                 }
2976
2977                 # MSG, MSGNW and RAW
2978                 if ( !$found ) {
2979                         # Check for MSGNW:
2980                         $mwMsgnw = MagicWord::get( 'msgnw' );
2981                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2982                                 $nowiki = true;
2983                         } else {
2984                                 # Remove obsolete MSG:
2985                                 $mwMsg = MagicWord::get( 'msg' );
2986                                 $mwMsg->matchStartAndRemove( $part1 );
2987                         }
2988
2989                         # Check for RAW:
2990                         $mwRaw = MagicWord::get( 'raw' );
2991                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2992                                 $forceRawInterwiki = true;
2993                         }
2994                 }
2995                 wfProfileOut( __METHOD__.'-modifiers' );
2996
2997                 # Parser functions
2998                 if ( !$found ) {
2999                         wfProfileIn( __METHOD__ . '-pfunc' );
3000
3001                         $colonPos = strpos( $part1, ':' );
3002                         if ( $colonPos !== false ) {
3003                                 # Case sensitive functions
3004                                 $function = substr( $part1, 0, $colonPos );
3005                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3006                                         $function = $this->mFunctionSynonyms[1][$function];
3007                                 } else {
3008                                         # Case insensitive functions
3009                                         $function = $wgContLang->lc( $function );
3010                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3011                                                 $function = $this->mFunctionSynonyms[0][$function];
3012                                         } else {
3013                                                 $function = false;
3014                                         }
3015                                 }
3016                                 if ( $function ) {
3017                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3018                                         $initialArgs = array( &$this );
3019                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3020                                         if ( $flags & SFH_OBJECT_ARGS ) {
3021                                                 # Add a frame parameter, and pass the arguments as an array
3022                                                 $allArgs = $initialArgs;
3023                                                 $allArgs[] = $frame;
3024                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3025                                                         $funcArgs[] = $args->item( $i );
3026                                                 }
3027                                                 $allArgs[] = $funcArgs;
3028                                         } else {
3029                                                 # Convert arguments to plain text
3030                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3031                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3032                                                 }
3033                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3034                                         }
3035
3036                                         # Workaround for PHP bug 35229 and similar
3037                                         if ( !is_callable( $callback ) ) {
3038                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3039                                                 wfProfileOut( __METHOD__ );
3040                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3041                                         }
3042                                         $result = call_user_func_array( $callback, $allArgs );
3043                                         $found = true;
3044                                         $noparse = true;
3045                                         $preprocessFlags = 0;
3046
3047                                         if ( is_array( $result ) ) {
3048                                                 if ( isset( $result[0] ) ) {
3049                                                         $text = $result[0];
3050                                                         unset( $result[0] );
3051                                                 }
3052
3053                                                 # Extract flags into the local scope
3054                                                 # This allows callers to set flags such as nowiki, found, etc.
3055                                                 extract( $result );
3056                                         } else {
3057                                                 $text = $result;
3058                                         }
3059                                         if ( !$noparse ) {
3060                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3061                                                 $isChildObj = true;
3062                                         }
3063                                 }
3064                         }
3065                         wfProfileOut( __METHOD__ . '-pfunc' );
3066                 }
3067
3068                 # Finish mangling title and then check for loops.
3069                 # Set $title to a Title object and $titleText to the PDBK
3070                 if ( !$found ) {
3071                         $ns = NS_TEMPLATE;
3072                         # Split the title into page and subpage
3073                         $subpage = '';
3074                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3075                         if ( $subpage !== '' ) {
3076                                 $ns = $this->mTitle->getNamespace();
3077                         }
3078                         $title = Title::newFromText( $part1, $ns );
3079                         if ( $title ) {
3080                                 $titleText = $title->getPrefixedText();
3081                                 # Check for language variants if the template is not found
3082                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3083                                         $wgContLang->findVariantLink( $part1, $title, true );
3084                                 }
3085                                 # Do recursion depth check
3086                                 $limit = $this->mOptions->getMaxTemplateDepth();
3087                                 if ( $frame->depth >= $limit ) {
3088                                         $found = true;
3089                                         $text = '<span class="error">'
3090                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3091                                                 . '</span>';
3092                                 }
3093                         }
3094                 }
3095
3096                 # Load from database
3097                 if ( !$found && $title ) {
3098                         wfProfileIn( __METHOD__ . '-loadtpl' );
3099                         if ( !$title->isExternal() ) {
3100                                 if ( $title->getNamespace() == NS_SPECIAL
3101                                         && $this->mOptions->getAllowSpecialInclusion()
3102                                         && $this->ot['html'] )
3103                                 {
3104                                         $text = SpecialPage::capturePath( $title );
3105                                         if ( is_string( $text ) ) {
3106                                                 $found = true;
3107                                                 $isHTML = true;
3108                                                 $this->disableCache();
3109                                         }
3110                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3111                                         $found = false; # access denied
3112                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3113                                 } else {
3114                                         list( $text, $title ) = $this->getTemplateDom( $title );
3115                                         if ( $text !== false ) {
3116                                                 $found = true;
3117                                                 $isChildObj = true;
3118                                         }
3119                                 }
3120
3121                                 # If the title is valid but undisplayable, make a link to it
3122                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3123                                         $text = "[[:$titleText]]";
3124                                         $found = true;
3125                                 }
3126                         } elseif ( $title->isTrans() ) {
3127                                 # Interwiki transclusion
3128                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3129                                         $text = $this->interwikiTransclude( $title, 'render' );
3130                                         $isHTML = true;
3131                                 } else {
3132                                         $text = $this->interwikiTransclude( $title, 'raw' );
3133                                         # Preprocess it like a template
3134                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3135                                         $isChildObj = true;
3136                                 }
3137                                 $found = true;
3138                         }
3139
3140                         # Do infinite loop check
3141                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3142                         if ( !$frame->loopCheck( $title ) ) {
3143                                 $found = true;
3144                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3145                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3146                         }
3147                         wfProfileOut( __METHOD__ . '-loadtpl' );
3148                 }
3149
3150                 # If we haven't found text to substitute by now, we're done
3151                 # Recover the source wikitext and return it
3152                 if ( !$found ) {
3153                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3154                         wfProfileOut( __METHOD__ );
3155                         return array( 'object' => $text );
3156                 }
3157
3158                 # Expand DOM-style return values in a child frame
3159                 if ( $isChildObj ) {
3160                         # Clean up argument array
3161                         $newFrame = $frame->newChild( $args, $title );
3162
3163                         if ( $nowiki ) {
3164                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3165                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3166                                 # Expansion is eligible for the empty-frame cache
3167                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3168                                         $text = $this->mTplExpandCache[$titleText];
3169                                 } else {
3170                                         $text = $newFrame->expand( $text );
3171                                         $this->mTplExpandCache[$titleText] = $text;
3172                                 }
3173                         } else {
3174                                 # Uncached expansion
3175                                 $text = $newFrame->expand( $text );
3176                         }
3177                 }
3178                 if ( $isLocalObj && $nowiki ) {
3179                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3180                         $isLocalObj = false;
3181                 }
3182
3183                 # Replace raw HTML by a placeholder
3184                 # Add a blank line preceding, to prevent it from mucking up
3185                 # immediately preceding headings
3186                 if ( $isHTML ) {
3187                         $text = "\n\n" . $this->insertStripItem( $text );
3188                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3189                         # Escape nowiki-style return values
3190                         $text = wfEscapeWikiText( $text );
3191                 } elseif ( is_string( $text )
3192                         && !$piece['lineStart']
3193                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3194                 {
3195                         # Bug 529: if the template begins with a table or block-level
3196                         # element, it should be treated as beginning a new line.
3197                         # This behaviour is somewhat controversial.
3198                         $text = "\n" . $text;
3199                 }
3200
3201                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3202                         # Error, oversize inclusion
3203                         if ( $titleText !== false ) {
3204                                 # Make a working, properly escaped link if possible (bug 23588)
3205                                 $text = "[[:$titleText]]";
3206                         } else {
3207                                 # This will probably not be a working link, but at least it may
3208                                 # provide some hint of where the problem is
3209                                 preg_replace( '/^:/', '', $originalTitle );
3210                                 $text = "[[:$originalTitle]]";
3211                         }
3212                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3213                         $this->limitationWarn( 'post-expand-template-inclusion' );
3214                 }
3215
3216                 if ( $isLocalObj ) {
3217                         $ret = array( 'object' => $text );
3218                 } else {
3219                         $ret = array( 'text' => $text );
3220                 }
3221
3222                 wfProfileOut( __METHOD__ );
3223                 return $ret;
3224         }
3225
3226         /**
3227          * Get the semi-parsed DOM representation of a template with a given title,
3228          * and its redirect destination title. Cached.
3229          */
3230         function getTemplateDom( $title ) {
3231                 $cacheTitle = $title;
3232                 $titleText = $title->getPrefixedDBkey();
3233
3234                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3235                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3236                         $title = Title::makeTitle( $ns, $dbk );
3237                         $titleText = $title->getPrefixedDBkey();
3238                 }
3239                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3240                         return array( $this->mTplDomCache[$titleText], $title );
3241                 }
3242
3243                 # Cache miss, go to the database
3244                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3245
3246                 if ( $text === false ) {
3247                         $this->mTplDomCache[$titleText] = false;
3248                         return array( false, $title );
3249                 }
3250
3251                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3252                 $this->mTplDomCache[ $titleText ] = $dom;
3253
3254                 if ( !$title->equals( $cacheTitle ) ) {
3255                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3256                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3257                 }
3258
3259                 return array( $dom, $title );
3260         }
3261
3262         /**
3263          * Fetch the unparsed text of a template and register a reference to it.
3264          */
3265         function fetchTemplateAndTitle( $title ) {
3266                 $templateCb = $this->mOptions->getTemplateCallback();
3267                 $stuff = call_user_func( $templateCb, $title, $this );
3268                 $text = $stuff['text'];
3269                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3270                 if ( isset( $stuff['deps'] ) ) {
3271                         foreach ( $stuff['deps'] as $dep ) {
3272                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3273                         }
3274                 }
3275                 return array( $text, $finalTitle );
3276         }
3277
3278         function fetchTemplate( $title ) {
3279                 $rv = $this->fetchTemplateAndTitle( $title );
3280                 return $rv[0];
3281         }
3282
3283         /**
3284          * Static function to get a template
3285          * Can be overridden via ParserOptions::setTemplateCallback().
3286          */
3287         static function statelessFetchTemplate( $title, $parser=false ) {
3288                 $text = $skip = false;
3289                 $finalTitle = $title;
3290                 $deps = array();
3291
3292                 # Loop to fetch the article, with up to 1 redirect
3293                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3294                         # Give extensions a chance to select the revision instead
3295                         $id = false; # Assume current
3296                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3297
3298                         if ( $skip ) {
3299                                 $text = false;
3300                                 $deps[] = array(
3301                                         'title' => $title,
3302                                         'page_id' => $title->getArticleID(),
3303                                         'rev_id' => null );
3304                                 break;
3305                         }
3306                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3307                         $rev_id = $rev ? $rev->getId() : 0;
3308                         # If there is no current revision, there is no page
3309                         if ( $id === false && !$rev ) {
3310                                 $linkCache = LinkCache::singleton();
3311                                 $linkCache->addBadLinkObj( $title );
3312                         }
3313
3314                         $deps[] = array(
3315                                 'title' => $title,
3316                                 'page_id' => $title->getArticleID(),
3317                                 'rev_id' => $rev_id );
3318
3319                         if ( $rev ) {
3320                                 $text = $rev->getText();
3321                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3322                                 global $wgContLang;
3323                                 $message = $wgContLang->lcfirst( $title->getText() );
3324                                 $text = wfMsgForContentNoTrans( $message );
3325                                 if ( wfEmptyMsg( $message, $text ) ) {
3326                                         $text = false;
3327                                         break;
3328                                 }
3329                         } else {
3330                                 break;
3331                         }
3332                         if ( $text === false ) {
3333                                 break;
3334                         }
3335                         # Redirect?
3336                         $finalTitle = $title;
3337                         $title = Title::newFromRedirect( $text );
3338                 }
3339                 return array(
3340                         'text' => $text,
3341                         'finalTitle' => $finalTitle,
3342                         'deps' => $deps );
3343         }
3344
3345         /**
3346          * Transclude an interwiki link.
3347          */
3348         function interwikiTransclude( $title, $action ) {
3349                 global $wgEnableScaryTranscluding;
3350
3351                 if ( !$wgEnableScaryTranscluding ) {
3352                         return wfMsg('scarytranscludedisabled');
3353                 }
3354
3355                 $url = $title->getFullUrl( "action=$action" );
3356
3357                 if ( strlen( $url ) > 255 ) {
3358                         return wfMsg( 'scarytranscludetoolong' );
3359                 }
3360                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3361         }
3362
3363         function fetchScaryTemplateMaybeFromCache( $url ) {
3364                 global $wgTranscludeCacheExpiry;
3365                 $dbr = wfGetDB( DB_SLAVE );
3366                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3367                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3368                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3369                 if ( $obj ) {
3370                         return $obj->tc_contents;
3371                 }
3372
3373                 $text = Http::get( $url );
3374                 if ( !$text ) {
3375                         return wfMsg( 'scarytranscludefailed', $url );
3376                 }
3377
3378                 $dbw = wfGetDB( DB_MASTER );
3379                 $dbw->replace( 'transcache', array('tc_url'), array(
3380                         'tc_url' => $url,
3381                         'tc_time' => $dbw->timestamp( time() ),
3382                         'tc_contents' => $text)
3383                 );
3384                 return $text;
3385         }
3386
3387
3388         /**
3389          * Triple brace replacement -- used for template arguments
3390          * @private
3391          */
3392         function argSubstitution( $piece, $frame ) {
3393                 wfProfileIn( __METHOD__ );
3394
3395                 $error = false;
3396                 $parts = $piece['parts'];
3397                 $nameWithSpaces = $frame->expand( $piece['title'] );
3398                 $argName = trim( $nameWithSpaces );
3399                 $object = false;
3400                 $text = $frame->getArgument( $argName );
3401                 if (  $text === false && $parts->getLength() > 0
3402                   && (
3403                     $this->ot['html']
3404                     || $this->ot['pre']
3405                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3406                   )
3407                 ) {
3408                         # No match in frame, use the supplied default
3409                         $object = $parts->item( 0 )->getChildren();
3410                 }
3411                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3412                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3413                         $this->limitationWarn( 'post-expand-template-argument' );
3414                 }
3415
3416                 if ( $text === false && $object === false ) {
3417                         # No match anywhere
3418                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3419                 }
3420                 if ( $error !== false ) {
3421                         $text .= $error;
3422                 }
3423                 if ( $object !== false ) {
3424                         $ret = array( 'object' => $object );
3425                 } else {
3426                         $ret = array( 'text' => $text );
3427                 }
3428
3429                 wfProfileOut( __METHOD__ );
3430                 return $ret;
3431         }
3432
3433         /**
3434          * Return the text to be used for a given extension tag.
3435          * This is the ghost of strip().
3436          *
3437          * @param $params Associative array of parameters:
3438          *     name       PPNode for the tag name
3439          *     attr       PPNode for unparsed text where tag attributes are thought to be
3440          *     attributes Optional associative array of parsed attributes
3441          *     inner      Contents of extension element
3442          *     noClose    Original text did not have a close tag
3443          * @param $frame PPFrame
3444          */
3445         function extensionSubstitution( $params, $frame ) {
3446                 global $wgRawHtml, $wgContLang;
3447
3448                 $name = $frame->expand( $params['name'] );
3449                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3450                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3451                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3452
3453                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3454                         ( $this->ot['html'] || $this->ot['pre'] );
3455                 if ( $isFunctionTag ) {
3456                         $markerType = 'none';
3457                 } else {
3458                         $markerType = 'general';
3459                 }
3460                 if ( $this->ot['html'] || $isFunctionTag ) {
3461                         $name = strtolower( $name );
3462                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3463                         if ( isset( $params['attributes'] ) ) {
3464                                 $attributes = $attributes + $params['attributes'];
3465                         }
3466
3467                         if ( isset( $this->mTagHooks[$name] ) ) {
3468                                 # Workaround for PHP bug 35229 and similar
3469                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3470                                         throw new MWException( "Tag hook for $name is not callable\n" );
3471                                 }
3472                                 $output = call_user_func_array( $this->mTagHooks[$name],
3473                                         array( $content, $attributes, $this, $frame ) );
3474                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3475                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3476                                 if ( !is_callable( $callback ) ) {
3477                                         throw new MWException( "Tag hook for $name is not callable\n" );
3478                                 }
3479
3480                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3481                         } else {
3482                                 $output = '<span class="error">Invalid tag extension name: ' .
3483                                         htmlspecialchars( $name ) . '</span>';
3484                         }
3485
3486                         if ( is_array( $output ) ) {
3487                                 # Extract flags to local scope (to override $markerType)
3488                                 $flags = $output;
3489                                 $output = $flags[0];
3490                                 unset( $flags[0] );
3491                                 extract( $flags );
3492                         }
3493                 } else {
3494                         if ( is_null( $attrText ) ) {
3495                                 $attrText = '';
3496                         }
3497                         if ( isset( $params['attributes'] ) ) {
3498                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3499                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3500                                                 htmlspecialchars( $attrValue ) . '"';
3501                                 }
3502                         }
3503                         if ( $content === null ) {
3504                                 $output = "<$name$attrText/>";
3505                         } else {
3506                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3507                                 $output = "<$name$attrText>$content$close";
3508                         }
3509                 }
3510
3511                 if ( $markerType === 'none' ) {
3512                         return $output;
3513                 } elseif ( $markerType === 'nowiki' ) {
3514                         $this->mStripState->nowiki->setPair( $marker, $output );
3515                 } elseif ( $markerType === 'general' ) {
3516                         $this->mStripState->general->setPair( $marker, $output );
3517                 } else {
3518                         throw new MWException( __METHOD__.': invalid marker type' );
3519                 }
3520                 return $marker;
3521         }
3522
3523         /**
3524          * Increment an include size counter
3525          *
3526          * @param $type String: the type of expansion
3527          * @param $size Integer: the size of the text
3528          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3529          */
3530         function incrementIncludeSize( $type, $size ) {
3531                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3532                         return false;
3533                 } else {
3534                         $this->mIncludeSizes[$type] += $size;
3535                         return true;
3536                 }
3537         }
3538
3539         /**
3540          * Increment the expensive function count
3541          *
3542          * @return Boolean: false if the limit has been exceeded
3543          */
3544         function incrementExpensiveFunctionCount() {
3545                 global $wgExpensiveParserFunctionLimit;
3546                 $this->mExpensiveFunctionCount++;
3547                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3548                         return true;
3549                 }
3550                 return false;
3551         }
3552
3553         /**
3554          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3555          * Fills $this->mDoubleUnderscores, returns the modified text
3556          */
3557         function doDoubleUnderscore( $text ) {
3558                 wfProfileIn( __METHOD__ );
3559
3560                 # The position of __TOC__ needs to be recorded
3561                 $mw = MagicWord::get( 'toc' );
3562                 if ( $mw->match( $text ) ) {
3563                         $this->mShowToc = true;
3564                         $this->mForceTocPosition = true;
3565
3566                         # Set a placeholder. At the end we'll fill it in with the TOC.
3567                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3568
3569                         # Only keep the first one.
3570                         $text = $mw->replace( '', $text );
3571                 }
3572
3573                 # Now match and remove the rest of them
3574                 $mwa = MagicWord::getDoubleUnderscoreArray();
3575                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3576
3577                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3578                         $this->mOutput->mNoGallery = true;
3579                 }
3580                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3581                         $this->mShowToc = false;
3582                 }
3583                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3584                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3585                         $this->addTrackingCategory( 'hidden-category-category' );
3586                 }
3587                 # (bug 8068) Allow control over whether robots index a page.
3588                 #
3589                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3590                 # is not desirable, the last one on the page should win.
3591                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3592                         $this->mOutput->setIndexPolicy( 'noindex' );
3593                         $this->addTrackingCategory( 'noindex-category' );
3594                 }
3595                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3596                         $this->mOutput->setIndexPolicy( 'index' );
3597                         $this->addTrackingCategory( 'index-category' );
3598                 }
3599
3600                 wfProfileOut( __METHOD__ );
3601                 return $text;
3602         }
3603
3604         /**
3605          * Add a tracking category, getting the title from a system message,
3606          * or print a debug message if the title is invalid.
3607          *
3608          * @param $msg String: message key
3609          * @return Boolean: whether the addition was successful
3610          */
3611         protected function addTrackingCategory( $msg ) {
3612                 $cat = wfMsgForContent( $msg );
3613
3614                 # Allow tracking categories to be disabled by setting them to "-"
3615                 if ( $cat === '-' ) {
3616                         return false;
3617                 }
3618
3619                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3620                 if ( $containerCategory ) {
3621                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3622                         return true;
3623                 } else {
3624                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3625                         return false;
3626                 }
3627         }
3628
3629         /**
3630          * This function accomplishes several tasks:
3631          * 1) Auto-number headings if that option is enabled
3632          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3633          * 3) Add a Table of contents on the top for users who have enabled the option
3634          * 4) Auto-anchor headings
3635          *
3636          * It loops through all headlines, collects the necessary data, then splits up the
3637          * string and re-inserts the newly formatted headlines.
3638          *
3639          * @param $text String
3640          * @param $origText String: original, untouched wikitext
3641          * @param $isMain Boolean
3642          * @private
3643          */
3644         function formatHeadings( $text, $origText, $isMain=true ) {
3645                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3646
3647                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3648                 $showEditLink = $this->mOptions->getEditSection();
3649
3650                 # Do not call quickUserCan unless necessary
3651                 if ( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3652                         $showEditLink = 0;
3653                 }
3654
3655                 # Inhibit editsection links if requested in the page
3656                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3657                         $showEditLink = 0;
3658                 }
3659
3660                 # Get all headlines for numbering them and adding funky stuff like [edit]
3661                 # links - this is for later, but we need the number of headlines right now
3662                 $matches = array();
3663                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3664
3665                 # if there are fewer than 4 headlines in the article, do not show TOC
3666                 # unless it's been explicitly enabled.
3667                 $enoughToc = $this->mShowToc &&
3668                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3669
3670                 # Allow user to stipulate that a page should have a "new section"
3671                 # link added via __NEWSECTIONLINK__
3672                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3673                         $this->mOutput->setNewSection( true );
3674                 }
3675
3676                 # Allow user to remove the "new section"
3677                 # link via __NONEWSECTIONLINK__
3678                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3679                         $this->mOutput->hideNewSection( true );
3680                 }
3681
3682                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3683                 # override above conditions and always show TOC above first header
3684                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3685                         $this->mShowToc = true;
3686                         $enoughToc = true;
3687                 }
3688
3689                 # We need this to perform operations on the HTML
3690                 $sk = $this->mOptions->getSkin();
3691
3692                 # headline counter
3693                 $headlineCount = 0;
3694                 $numVisible = 0;
3695
3696                 # Ugh .. the TOC should have neat indentation levels which can be
3697                 # passed to the skin functions. These are determined here
3698                 $toc = '';
3699                 $full = '';
3700                 $head = array();
3701                 $sublevelCount = array();
3702                 $levelCount = array();
3703                 $toclevel = 0;
3704                 $level = 0;
3705                 $prevlevel = 0;
3706                 $toclevel = 0;
3707                 $prevtoclevel = 0;
3708                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3709                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3710                 $oldType = $this->mOutputType;
3711                 $this->setOutputType( self::OT_WIKI );
3712                 $frame = $this->getPreprocessor()->newFrame();
3713                 $root = $this->preprocessToDom( $origText );
3714                 $node = $root->getFirstChild();
3715                 $byteOffset = 0;
3716                 $tocraw = array();
3717
3718                 foreach ( $matches[3] as $headline ) {
3719                         $isTemplate = false;
3720                         $titleText = false;
3721                         $sectionIndex = false;
3722                         $numbering = '';
3723                         $markerMatches = array();
3724                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3725                                 $serial = $markerMatches[1];
3726                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3727                                 $isTemplate = ( $titleText != $baseTitleText );
3728                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3729                         }
3730
3731                         if ( $toclevel ) {
3732                                 $prevlevel = $level;
3733                                 $prevtoclevel = $toclevel;
3734                         }
3735                         $level = $matches[1][$headlineCount];
3736
3737                         if ( $level > $prevlevel ) {
3738                                 # Increase TOC level
3739                                 $toclevel++;
3740                                 $sublevelCount[$toclevel] = 0;
3741                                 if ( $toclevel<$wgMaxTocLevel ) {
3742                                         $prevtoclevel = $toclevel;
3743                                         $toc .= $sk->tocIndent();
3744                                         $numVisible++;
3745                                 }
3746                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3747                                 # Decrease TOC level, find level to jump to
3748
3749                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3750                                         if ( $levelCount[$i] == $level ) {
3751                                                 # Found last matching level
3752                                                 $toclevel = $i;
3753                                                 break;
3754                                         } elseif ( $levelCount[$i] < $level ) {
3755                                                 # Found first matching level below current level
3756                                                 $toclevel = $i + 1;
3757                                                 break;
3758                                         }
3759                                 }
3760                                 if ( $i == 0 ) {
3761                                         $toclevel = 1;
3762                                 }
3763                                 if ( $toclevel<$wgMaxTocLevel ) {
3764                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3765                                                 # Unindent only if the previous toc level was shown :p
3766                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3767                                                 $prevtoclevel = $toclevel;
3768                                         } else {
3769                                                 $toc .= $sk->tocLineEnd();
3770                                         }
3771                                 }
3772                         } else {
3773                                 # No change in level, end TOC line
3774                                 if ( $toclevel<$wgMaxTocLevel ) {
3775                                         $toc .= $sk->tocLineEnd();
3776                                 }
3777                         }
3778
3779                         $levelCount[$toclevel] = $level;
3780
3781                         # count number of headlines for each level
3782                         @$sublevelCount[$toclevel]++;
3783                         $dot = 0;
3784                         for( $i = 1; $i <= $toclevel; $i++ ) {
3785                                 if ( !empty( $sublevelCount[$i] ) ) {
3786                                         if ( $dot ) {
3787                                                 $numbering .= '.';
3788                                         }
3789                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3790                                         $dot = 1;
3791                                 }
3792                         }
3793
3794                         # The safe header is a version of the header text safe to use for links
3795                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3796                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3797
3798                         # Remove link placeholders by the link text.
3799                         #     <!--LINK number-->
3800                         # turns into
3801                         #     link text with suffix
3802                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3803
3804                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3805                         $tocline = preg_replace(
3806                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3807                                 array( '',                          '<$1>' ),
3808                                 $safeHeadline
3809                         );
3810                         $tocline = trim( $tocline );
3811
3812                         # For the anchor, strip out HTML-y stuff period
3813                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3814                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3815                         $safeHeadline = trim( $safeHeadline );
3816
3817                         # Save headline for section edit hint before it's escaped
3818                         $headlineHint = $safeHeadline;
3819
3820                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3821                                 # For reverse compatibility, provide an id that's
3822                                 # HTML4-compatible, like we used to.
3823                                 #
3824                                 # It may be worth noting, academically, that it's possible for
3825                                 # the legacy anchor to conflict with a non-legacy headline
3826                                 # anchor on the page.  In this case likely the "correct" thing
3827                                 # would be to either drop the legacy anchors or make sure
3828                                 # they're numbered first.  However, this would require people
3829                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3830                                 # manually, so let's not bother worrying about it.
3831                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3832                                         array( 'noninitial', 'legacy' ) );
3833                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3834
3835                                 if ( $legacyHeadline == $safeHeadline ) {
3836                                         # No reason to have both (in fact, we can't)
3837                                         $legacyHeadline = false;
3838                                 }
3839                         } else {
3840                                 $legacyHeadline = false;
3841                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3842                                         'noninitial' );
3843                         }
3844
3845                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3846                         # Does this apply to Unicode characters?  Because we aren't
3847                         # handling those here.
3848                         $arrayKey = strtolower( $safeHeadline );
3849                         if ( $legacyHeadline === false ) {
3850                                 $legacyArrayKey = false;
3851                         } else {
3852                                 $legacyArrayKey = strtolower( $legacyHeadline );
3853                         }
3854
3855                         # count how many in assoc. array so we can track dupes in anchors
3856                         if ( isset( $refers[$arrayKey] ) ) {
3857                                 $refers[$arrayKey]++;
3858                         } else {
3859                                 $refers[$arrayKey] = 1;
3860                         }
3861                         if ( isset( $refers[$legacyArrayKey] ) ) {
3862                                 $refers[$legacyArrayKey]++;
3863                         } else {
3864                                 $refers[$legacyArrayKey] = 1;
3865                         }
3866
3867                         # Don't number the heading if it is the only one (looks silly)
3868                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3869                                 # the two are different if the line contains a link
3870                                 $headline = $numbering . ' ' . $headline;
3871                         }
3872
3873                         # Create the anchor for linking from the TOC to the section
3874                         $anchor = $safeHeadline;
3875                         $legacyAnchor = $legacyHeadline;
3876                         if ( $refers[$arrayKey] > 1 ) {
3877                                 $anchor .= '_' . $refers[$arrayKey];
3878                         }
3879                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3880                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3881                         }
3882                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3883                                 $toc .= $sk->tocLine( $anchor, $tocline,
3884                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3885                         }
3886
3887                         # Add the section to the section tree
3888                         # Find the DOM node for this header
3889                         while ( $node && !$isTemplate ) {
3890                                 if ( $node->getName() === 'h' ) {
3891                                         $bits = $node->splitHeading();
3892                                         if ( $bits['i'] == $sectionIndex )
3893                                                 break;
3894                                 }
3895                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3896                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3897                                 $node = $node->getNextSibling();
3898                         }
3899                         $tocraw[] = array(
3900                                 'toclevel' => $toclevel,
3901                                 'level' => $level,
3902                                 'line' => $tocline,
3903                                 'number' => $numbering,
3904                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3905                                 'fromtitle' => $titleText,
3906                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3907                                 'anchor' => $anchor,
3908                         );
3909
3910                         # give headline the correct <h#> tag
3911                         if ( $showEditLink && $sectionIndex !== false ) {
3912                                 if ( $isTemplate ) {
3913                                         # Put a T flag in the section identifier, to indicate to extractSections()
3914                                         # that sections inside <includeonly> should be counted.
3915                                         $editlink = $sk->doEditSectionLink( Title::newFromText( $titleText ), "T-$sectionIndex" );
3916                                 } else {
3917                                         $editlink = $sk->doEditSectionLink( $this->mTitle, $sectionIndex, $headlineHint );
3918                                 }
3919                         } else {
3920                                 $editlink = '';
3921                         }
3922                         $head[$headlineCount] = $sk->makeHeadline( $level,
3923                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3924                                 $editlink, $legacyAnchor );
3925
3926                         $headlineCount++;
3927                 }
3928
3929                 $this->setOutputType( $oldType );
3930
3931                 # Never ever show TOC if no headers
3932                 if ( $numVisible < 1 ) {
3933                         $enoughToc = false;
3934                 }
3935
3936                 if ( $enoughToc ) {
3937                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3938                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3939                         }
3940                         $toc = $sk->tocList( $toc );
3941                         $this->mOutput->setTOCHTML( $toc );
3942                 }
3943
3944                 if ( $isMain ) {
3945                         $this->mOutput->setSections( $tocraw );
3946                 }
3947
3948                 # split up and insert constructed headlines
3949
3950                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3951                 $i = 0;
3952
3953                 foreach ( $blocks as $block ) {
3954                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3955                                 # This is the [edit] link that appears for the top block of text when
3956                                 # section editing is enabled
3957
3958                                 # Disabled because it broke block formatting
3959                                 # For example, a bullet point in the top line
3960                                 # $full .= $sk->editSectionLink(0);
3961                         }
3962                         $full .= $block;
3963                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3964                                 # Top anchor now in skin
3965                                 $full = $full.$toc;
3966                         }
3967
3968                         if ( !empty( $head[$i] ) ) {
3969                                 $full .= $head[$i];
3970                         }
3971                         $i++;
3972                 }
3973                 if ( $this->mForceTocPosition ) {
3974                         return str_replace( '<!--MWTOC-->', $toc, $full );
3975                 } else {
3976                         return $full;
3977                 }
3978         }
3979
3980         /**
3981          * Merge $tree2 into $tree1 by replacing the section with index
3982          * $section in $tree1 and its descendants with the sections in $tree2.
3983          * Note that in the returned section tree, only the 'index' and
3984          * 'byteoffset' fields are guaranteed to be correct.
3985          *
3986          * @param $tree1 Array: section tree from ParserOutput::getSectons()
3987         *  @param $tree2 Array: section tree
3988          * @param $section Integer: section index
3989          * @param $title Title: Title both section trees come from
3990          * @param $len2 Integer: length of the original wikitext for $tree2
3991          * @return Array: merged section tree
3992          */
3993         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3994                 global $wgContLang;
3995                 $newTree = array();
3996                 $targetLevel = false;
3997                 $merged = false;
3998                 $lastLevel = 1;
3999                 $nextIndex = 1;
4000                 $numbering = array( 0 );
4001                 $titletext = $title->getPrefixedDBkey();
4002                 foreach ( $tree1 as $s ) {
4003                         if ( $targetLevel !== false ) {
4004                                 if ( $s['level'] <= $targetLevel ) {
4005                                         # We've skipped enough
4006                                         $targetLevel = false;
4007                                 } else {
4008                                         continue;
4009                                 }
4010                         }
4011                         if ( $s['index'] != $section ||
4012                                         $s['fromtitle'] != $titletext ) {
4013                                 self::incrementNumbering( $numbering,
4014                                         $s['toclevel'], $lastLevel );
4015
4016                                 # Rewrite index, byteoffset and number
4017                                 if ( $s['fromtitle'] == $titletext ) {
4018                                         $s['index'] = $nextIndex++;
4019                                         if ( $merged ) {
4020                                                 $s['byteoffset'] += $len2;
4021                                         }
4022                                 }
4023                                 $s['number']  = implode( '.', array_map(
4024                                         array( $wgContLang, 'formatnum' ),
4025                                         $numbering ) );
4026                                 $lastLevel = $s['toclevel'];
4027                                 $newTree[] = $s;
4028                         } else {
4029                                 # We're at $section
4030                                 # Insert sections from $tree2 here
4031                                 foreach ( $tree2 as $s2 ) {
4032                                         # Rewrite the fields in $s2
4033                                         # before inserting it
4034                                         $s2['toclevel'] += $s['toclevel'] - 1;
4035                                         $s2['level'] += $s['level'] - 1;
4036                                         $s2['index'] = $nextIndex++;
4037                                         $s2['byteoffset'] += $s['byteoffset'];
4038
4039                                         self::incrementNumbering( $numbering,
4040                                                 $s2['toclevel'], $lastLevel );
4041                                         $s2['number']  = implode( '.', array_map(
4042                                                 array( $wgContLang, 'formatnum' ),
4043                                                 $numbering ) );
4044                                         $lastLevel = $s2['toclevel'];
4045                                         $newTree[] = $s2;
4046                                 }
4047                                 # Skip all descendants of $section in $tree1
4048                                 $targetLevel = $s['level'];
4049                                 $merged = true;
4050                         }
4051                 }
4052                 return $newTree;
4053         }
4054
4055         /**
4056          * Increment a section number. Helper function for mergeSectionTrees()
4057          *
4058          * @param $number Array representing a section number
4059          * @param $level Integer: current TOC level (depth)
4060          * @param $lastLevel Integer: level of previous TOC entry
4061          */
4062         private static function incrementNumbering( &$number, $level, $lastLevel ) {
4063                 if ( $level > $lastLevel ) {
4064                         $number[$level - 1] = 1;
4065                 } elseif ( $level < $lastLevel ) {
4066                         foreach ( $number as $key => $unused )
4067                                 if ( $key >= $level ) {
4068                                         unset( $number[$key] );
4069                                 }
4070                         $number[$level - 1]++;
4071                 } else {
4072                         $number[$level - 1]++;
4073                 }
4074         }
4075
4076         /**
4077          * Transform wiki markup when saving a page by doing \r\n -> \n
4078          * conversion, substitting signatures, {{subst:}} templates, etc.
4079          *
4080          * @param $text String: the text to transform
4081          * @param &$title Title: the Title object for the current article
4082          * @param $user User: the User object describing the current user
4083          * @param $options ParserOptions: parsing options
4084          * @param $clearState Boolean: whether to clear the parser state first
4085          * @return String: the altered wiki markup
4086          */
4087         public function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
4088                 $this->mOptions = $options;
4089                 $this->setTitle( $title );
4090                 $this->setOutputType( self::OT_WIKI );
4091
4092                 if ( $clearState ) {
4093                         $this->clearState();
4094                 }
4095
4096                 $pairs = array(
4097                         "\r\n" => "\n",
4098                 );
4099                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4100                 $text = $this->pstPass2( $text, $user );
4101                 $text = $this->mStripState->unstripBoth( $text );
4102                 return $text;
4103         }
4104
4105         /**
4106          * Pre-save transform helper function
4107          * @private
4108          */
4109         function pstPass2( $text, $user ) {
4110                 global $wgContLang, $wgLocaltimezone;
4111
4112                 # Note: This is the timestamp saved as hardcoded wikitext to
4113                 # the database, we use $wgContLang here in order to give
4114                 # everyone the same signature and use the default one rather
4115                 # than the one selected in each user's preferences.
4116                 # (see also bug 12815)
4117                 $ts = $this->mOptions->getTimestamp();
4118                 if ( isset( $wgLocaltimezone ) ) {
4119                         $tz = $wgLocaltimezone;
4120                 } else {
4121                         $tz = date_default_timezone_get();
4122                 }
4123
4124                 $unixts = wfTimestamp( TS_UNIX, $ts );
4125                 $oldtz = date_default_timezone_get();
4126                 date_default_timezone_set( $tz );
4127                 $ts = date( 'YmdHis', $unixts );
4128                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4129
4130                 # Allow translation of timezones through wiki. date() can return
4131                 # whatever crap the system uses, localised or not, so we cannot
4132                 # ship premade translations.
4133                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4134                 $value = wfMsgForContent( $key );
4135                 if ( !wfEmptyMsg( $key, $value ) ) {
4136                         $tzMsg = $value;
4137                 }
4138
4139                 date_default_timezone_set( $oldtz );
4140
4141                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4142
4143                 # Variable replacement
4144                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4145                 $text = $this->replaceVariables( $text );
4146
4147                 # Signatures
4148                 $sigText = $this->getUserSig( $user );
4149                 $text = strtr( $text, array(
4150                         '~~~~~' => $d,
4151                         '~~~~' => "$sigText $d",
4152                         '~~~' => $sigText
4153                 ) );
4154
4155                 # Context links: [[|name]] and [[name (context)|]]
4156                 global $wgLegalTitleChars;
4157                 $tc = "[$wgLegalTitleChars]";
4158                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4159
4160                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4161                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4162                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4163                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4164
4165                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4166                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4167                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4168                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4169
4170                 $t = $this->mTitle->getText();
4171                 $m = array();
4172                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4173                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4174                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4175                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4176                 } else {
4177                         # if there's no context, don't bother duplicating the title
4178                         $text = preg_replace( $p2, '[[\\1]]', $text );
4179                 }
4180
4181                 # Trim trailing whitespace
4182                 $text = rtrim( $text );
4183
4184                 return $text;
4185         }
4186
4187         /**
4188          * Fetch the user's signature text, if any, and normalize to
4189          * validated, ready-to-insert wikitext.
4190          * If you have pre-fetched the nickname or the fancySig option, you can
4191          * specify them here to save a database query.
4192          *
4193          * @param $user User
4194          * @param $nickname String: nickname to use or false to use user's default nickname
4195          * @param $fancySig Boolean: whether the nicknname is the complete signature
4196          *                  or null to use default value
4197          * @return string
4198          */
4199         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4200                 global $wgMaxSigChars;
4201
4202                 $username = $user->getName();
4203
4204                 # If not given, retrieve from the user object.
4205                 if ( $nickname === false )
4206                         $nickname = $user->getOption( 'nickname' );
4207
4208                 if ( is_null( $fancySig ) ) {
4209                         $fancySig = $user->getBoolOption( 'fancysig' );
4210                 }
4211
4212                 $nickname = $nickname == null ? $username : $nickname;
4213
4214                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4215                         $nickname = $username;
4216                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4217                 } elseif ( $fancySig !== false ) {
4218                         # Sig. might contain markup; validate this
4219                         if ( $this->validateSig( $nickname ) !== false ) {
4220                                 # Validated; clean up (if needed) and return it
4221                                 return $this->cleanSig( $nickname, true );
4222                         } else {
4223                                 # Failed to validate; fall back to the default
4224                                 $nickname = $username;
4225                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4226                         }
4227                 }
4228
4229                 # Make sure nickname doesnt get a sig in a sig
4230                 $nickname = $this->cleanSigInSig( $nickname );
4231
4232                 # If we're still here, make it a link to the user page
4233                 $userText = wfEscapeWikiText( $username );
4234                 $nickText = wfEscapeWikiText( $nickname );
4235                 if ( $user->isAnon() )  {
4236                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4237                 } else {
4238                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4239                 }
4240         }
4241
4242         /**
4243          * Check that the user's signature contains no bad XML
4244          *
4245          * @param $text String
4246          * @return mixed An expanded string, or false if invalid.
4247          */
4248         function validateSig( $text ) {
4249                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4250         }
4251
4252         /**
4253          * Clean up signature text
4254          *
4255          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4256          * 2) Substitute all transclusions
4257          *
4258          * @param $text String
4259          * @param $parsing Whether we're cleaning (preferences save) or parsing
4260          * @return String: signature text
4261          */
4262         function cleanSig( $text, $parsing = false ) {
4263                 if ( !$parsing ) {
4264                         global $wgTitle;
4265                         $this->clearState();
4266                         $this->setTitle( $wgTitle );
4267                         $this->mOptions = new ParserOptions;
4268                         $this->setOutputType = self::OT_PREPROCESS;
4269                 }
4270
4271                 # Option to disable this feature
4272                 if ( !$this->mOptions->getCleanSignatures() ) {
4273                         return $text;
4274                 }
4275
4276                 # FIXME: regex doesn't respect extension tags or nowiki
4277                 #  => Move this logic to braceSubstitution()
4278                 $substWord = MagicWord::get( 'subst' );
4279                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4280                 $substText = '{{' . $substWord->getSynonym( 0 );
4281
4282                 $text = preg_replace( $substRegex, $substText, $text );
4283                 $text = $this->cleanSigInSig( $text );
4284                 $dom = $this->preprocessToDom( $text );
4285                 $frame = $this->getPreprocessor()->newFrame();
4286                 $text = $frame->expand( $dom );
4287
4288                 if ( !$parsing ) {
4289                         $text = $this->mStripState->unstripBoth( $text );
4290                 }
4291
4292                 return $text;
4293         }
4294
4295         /**
4296          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4297          *
4298          * @param $text String
4299          * @return String: signature text with /~{3,5}/ removed
4300          */
4301         function cleanSigInSig( $text ) {
4302                 $text = preg_replace( '/~{3,5}/', '', $text );
4303                 return $text;
4304         }
4305
4306         /**
4307          * Set up some variables which are usually set up in parse()
4308          * so that an external function can call some class members with confidence
4309          */
4310         public function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4311                 $this->setTitle( $title );
4312                 $this->mOptions = $options;
4313                 $this->setOutputType( $outputType );
4314                 if ( $clearState ) {
4315                         $this->clearState();
4316                 }
4317         }
4318
4319         /**
4320          * Wrapper for preprocess()
4321          *
4322          * @param $text String: the text to preprocess
4323          * @param $options ParserOptions: options
4324          * @return String
4325          */
4326         public function transformMsg( $text, $options ) {
4327                 global $wgTitle;
4328                 static $executing = false;
4329
4330                 # Guard against infinite recursion
4331                 if ( $executing ) {
4332                         return $text;
4333                 }
4334                 $executing = true;
4335
4336                 wfProfileIn( __METHOD__ );
4337                 $text = $this->preprocess( $text, $wgTitle, $options );
4338
4339                 $executing = false;
4340                 wfProfileOut( __METHOD__ );
4341                 return $text;
4342         }
4343
4344         /**
4345          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4346          * The callback should have the following form:
4347          *    function myParserHook( $text, $params, $parser ) { ... }
4348          *
4349          * Transform and return $text. Use $parser for any required context, e.g. use
4350          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4351          *
4352          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4353          * @param $callback Mixed: the callback function (and object) to use for the tag
4354          * @return The old value of the mTagHooks array associated with the hook
4355          */
4356         public function setHook( $tag, $callback ) {
4357                 $tag = strtolower( $tag );
4358                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4359                 $this->mTagHooks[$tag] = $callback;
4360                 if ( !in_array( $tag, $this->mStripList ) ) {
4361                         $this->mStripList[] = $tag;
4362                 }
4363
4364                 return $oldVal;
4365         }
4366
4367         function setTransparentTagHook( $tag, $callback ) {
4368                 $tag = strtolower( $tag );
4369                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4370                 $this->mTransparentTagHooks[$tag] = $callback;
4371
4372                 return $oldVal;
4373         }
4374
4375         /**
4376          * Remove all tag hooks
4377          */
4378         function clearTagHooks() {
4379                 $this->mTagHooks = array();
4380                 $this->mStripList = $this->mDefaultStripList;
4381         }
4382
4383         /**
4384          * Create a function, e.g. {{sum:1|2|3}}
4385          * The callback function should have the form:
4386          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4387          *
4388          * Or with SFH_OBJECT_ARGS:
4389          *    function myParserFunction( $parser, $frame, $args ) { ... }
4390          *
4391          * The callback may either return the text result of the function, or an array with the text
4392          * in element 0, and a number of flags in the other elements. The names of the flags are
4393          * specified in the keys. Valid flags are:
4394          *   found                     The text returned is valid, stop processing the template. This
4395          *                             is on by default.
4396          *   nowiki                    Wiki markup in the return value should be escaped
4397          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4398          *
4399          * @param $id String: The magic word ID
4400          * @param $callback Mixed: the callback function (and object) to use
4401          * @param $flags Integer: a combination of the following flags:
4402          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4403          *
4404          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4405          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4406          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4407          *     the arguments, and to control the way they are expanded.
4408          *
4409          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4410          *     arguments, for instance:
4411          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4412          *
4413          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4414          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4415          *     working if/when this is changed.
4416          *
4417          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4418          *     expansion.
4419          *
4420          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4421          *     about the methods available in PPFrame and PPNode.
4422          *
4423          * @return The old callback function for this name, if any
4424          */
4425         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4426                 global $wgContLang;
4427
4428                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4429                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4430
4431                 # Add to function cache
4432                 $mw = MagicWord::get( $id );
4433                 if ( !$mw )
4434                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4435
4436                 $synonyms = $mw->getSynonyms();
4437                 $sensitive = intval( $mw->isCaseSensitive() );
4438
4439                 foreach ( $synonyms as $syn ) {
4440                         # Case
4441                         if ( !$sensitive ) {
4442                                 $syn = $wgContLang->lc( $syn );
4443                         }
4444                         # Add leading hash
4445                         if ( !( $flags & SFH_NO_HASH ) ) {
4446                                 $syn = '#' . $syn;
4447                         }
4448                         # Remove trailing colon
4449                         if ( substr( $syn, -1, 1 ) === ':' ) {
4450                                 $syn = substr( $syn, 0, -1 );
4451                         }
4452                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4453                 }
4454                 return $oldVal;
4455         }
4456
4457         /**
4458          * Get all registered function hook identifiers
4459          *
4460          * @return Array
4461          */
4462         function getFunctionHooks() {
4463                 return array_keys( $this->mFunctionHooks );
4464         }
4465
4466         /**
4467          * Create a tag function, e.g. <test>some stuff</test>.
4468          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4469          * Unlike parser functions, their content is not preprocessed.
4470          */
4471         function setFunctionTagHook( $tag, $callback, $flags ) {
4472                 $tag = strtolower( $tag );
4473                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4474                         $this->mFunctionTagHooks[$tag] : null;
4475                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4476
4477                 if ( !in_array( $tag, $this->mStripList ) ) {
4478                         $this->mStripList[] = $tag;
4479                 }
4480
4481                 return $old;
4482         }
4483
4484         /**
4485          * FIXME: update documentation. makeLinkObj() is deprecated.
4486          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4487          * Placeholders created in Skin::makeLinkObj()
4488          * Returns an array of link CSS classes, indexed by PDBK.
4489          */
4490         function replaceLinkHolders( &$text, $options = 0 ) {
4491                 return $this->mLinkHolders->replace( $text );
4492         }
4493
4494         /**
4495          * Replace <!--LINK--> link placeholders with plain text of links
4496          * (not HTML-formatted).
4497          *
4498          * @param $text String
4499          * @return String
4500          */
4501         function replaceLinkHoldersText( $text ) {
4502                 return $this->mLinkHolders->replaceText( $text );
4503         }
4504
4505         /**
4506          * Renders an image gallery from a text with one line per image.
4507          * text labels may be given by using |-style alternative text. E.g.
4508          *   Image:one.jpg|The number "1"
4509          *   Image:tree.jpg|A tree
4510          * given as text will return the HTML of a gallery with two images,
4511          * labeled 'The number "1"' and
4512          * 'A tree'.
4513          */
4514         function renderImageGallery( $text, $params ) {
4515                 $ig = new ImageGallery();
4516                 $ig->setContextTitle( $this->mTitle );
4517                 $ig->setShowBytes( false );
4518                 $ig->setShowFilename( false );
4519                 $ig->setParser( $this );
4520                 $ig->setHideBadImages();
4521                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4522                 $ig->useSkin( $this->mOptions->getSkin() );
4523                 $ig->mRevisionId = $this->mRevisionId;
4524
4525                 if ( isset( $params['showfilename'] ) ) {
4526                         $ig->setShowFilename( true );
4527                 } else {
4528                         $ig->setShowFilename( false );
4529                 }
4530                 if ( isset( $params['caption'] ) ) {
4531                         $caption = $params['caption'];
4532                         $caption = htmlspecialchars( $caption );
4533                         $caption = $this->replaceInternalLinks( $caption );
4534                         $ig->setCaptionHtml( $caption );
4535                 }
4536                 if ( isset( $params['perrow'] ) ) {
4537                         $ig->setPerRow( $params['perrow'] );
4538                 }
4539                 if ( isset( $params['widths'] ) ) {
4540                         $ig->setWidths( $params['widths'] );
4541                 }
4542                 if ( isset( $params['heights'] ) ) {
4543                         $ig->setHeights( $params['heights'] );
4544                 }
4545
4546                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4547
4548                 $lines = StringUtils::explode( "\n", $text );
4549                 foreach ( $lines as $line ) {
4550                         # match lines like these:
4551                         # Image:someimage.jpg|This is some image
4552                         $matches = array();
4553                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4554                         # Skip empty lines
4555                         if ( count( $matches ) == 0 ) {
4556                                 continue;
4557                         }
4558
4559                         if ( strpos( $matches[0], '%' ) !== false ) {
4560                                 $matches[1] = urldecode( $matches[1] );
4561                         }
4562                         $tp = Title::newFromText( $matches[1] );
4563                         $nt =& $tp;
4564                         if ( is_null( $nt ) ) {
4565                                 # Bogus title. Ignore these so we don't bomb out later.
4566                                 continue;
4567                         }
4568                         if ( isset( $matches[3] ) ) {
4569                                 $label = $matches[3];
4570                         } else {
4571                                 $label = '';
4572                         }
4573
4574                         $html = $this->recursiveTagParse( trim( $label ) );
4575
4576                         $ig->add( $nt, $html );
4577
4578                         # Only add real images (bug #5586)
4579                         if ( $nt->getNamespace() == NS_FILE ) {
4580                                 $this->mOutput->addImage( $nt->getDBkey() );
4581                         }
4582                 }
4583                 return $ig->toHTML();
4584         }
4585
4586         function getImageParams( $handler ) {
4587                 if ( $handler ) {
4588                         $handlerClass = get_class( $handler );
4589                 } else {
4590                         $handlerClass = '';
4591                 }
4592                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4593                         # Initialise static lists
4594                         static $internalParamNames = array(
4595                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4596                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4597                                         'bottom', 'text-bottom' ),
4598                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4599                                         'upright', 'border', 'link', 'alt' ),
4600                         );
4601                         static $internalParamMap;
4602                         if ( !$internalParamMap ) {
4603                                 $internalParamMap = array();
4604                                 foreach ( $internalParamNames as $type => $names ) {
4605                                         foreach ( $names as $name ) {
4606                                                 $magicName = str_replace( '-', '_', "img_$name" );
4607                                                 $internalParamMap[$magicName] = array( $type, $name );
4608                                         }
4609                                 }
4610                         }
4611
4612                         # Add handler params
4613                         $paramMap = $internalParamMap;
4614                         if ( $handler ) {
4615                                 $handlerParamMap = $handler->getParamMap();
4616                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4617                                         $paramMap[$magic] = array( 'handler', $paramName );
4618                                 }
4619                         }
4620                         $this->mImageParams[$handlerClass] = $paramMap;
4621                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4622                 }
4623                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4624         }
4625
4626         /**
4627          * Parse image options text and use it to make an image
4628          *
4629          * @param $title Title
4630          * @param $options String
4631          * @param $holders LinkHolderArray
4632          */
4633         function makeImage( $title, $options, $holders = false ) {
4634                 # Check if the options text is of the form "options|alt text"
4635                 # Options are:
4636                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4637                 #  * left       no resizing, just left align. label is used for alt= only
4638                 #  * right      same, but right aligned
4639                 #  * none       same, but not aligned
4640                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4641                 #  * center     center the image
4642                 #  * frame      Keep original image size, no magnify-button.
4643                 #  * framed     Same as "frame"
4644                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4645                 #  * upright    reduce width for upright images, rounded to full __0 px
4646                 #  * border     draw a 1px border around the image
4647                 #  * alt        Text for HTML alt attribute (defaults to empty)
4648                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4649                 # vertical-align values (no % or length right now):
4650                 #  * baseline
4651                 #  * sub
4652                 #  * super
4653                 #  * top
4654                 #  * text-top
4655                 #  * middle
4656                 #  * bottom
4657                 #  * text-bottom
4658
4659                 $parts = StringUtils::explode( "|", $options );
4660                 $sk = $this->mOptions->getSkin();
4661
4662                 # Give extensions a chance to select the file revision for us
4663                 $skip = $time = $descQuery = false;
4664                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4665
4666                 if ( $skip ) {
4667                         return $sk->link( $title );
4668                 }
4669
4670                 # Get the file
4671                 $imagename = $title->getDBkey();
4672                 $file = wfFindFile( $title, array( 'time' => $time ) );
4673                 # Get parameter map
4674                 $handler = $file ? $file->getHandler() : false;
4675
4676                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4677
4678                 # Process the input parameters
4679                 $caption = '';
4680                 $params = array( 'frame' => array(), 'handler' => array(),
4681                         'horizAlign' => array(), 'vertAlign' => array() );
4682                 foreach ( $parts as $part ) {
4683                         $part = trim( $part );
4684                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4685                         $validated = false;
4686                         if ( isset( $paramMap[$magicName] ) ) {
4687                                 list( $type, $paramName ) = $paramMap[$magicName];
4688
4689                                 # Special case; width and height come in one variable together
4690                                 if ( $type === 'handler' && $paramName === 'width' ) {
4691                                         $m = array();
4692                                         # (bug 13500) In both cases (width/height and width only),
4693                                         # permit trailing "px" for backward compatibility.
4694                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4695                                                 $width = intval( $m[1] );
4696                                                 $height = intval( $m[2] );
4697                                                 if ( $handler->validateParam( 'width', $width ) ) {
4698                                                         $params[$type]['width'] = $width;
4699                                                         $validated = true;
4700                                                 }
4701                                                 if ( $handler->validateParam( 'height', $height ) ) {
4702                                                         $params[$type]['height'] = $height;
4703                                                         $validated = true;
4704                                                 }
4705                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4706                                                 $width = intval( $value );
4707                                                 if ( $handler->validateParam( 'width', $width ) ) {
4708                                                         $params[$type]['width'] = $width;
4709                                                         $validated = true;
4710                                                 }
4711                                         } # else no validation -- bug 13436
4712                                 } else {
4713                                         if ( $type === 'handler' ) {
4714                                                 # Validate handler parameter
4715                                                 $validated = $handler->validateParam( $paramName, $value );
4716                                         } else {
4717                                                 # Validate internal parameters
4718                                                 switch( $paramName ) {
4719                                                 case 'manualthumb':
4720                                                 case 'alt':
4721                                                         # @todo Fixme: possibly check validity here for
4722                                                         # manualthumb? downstream behavior seems odd with
4723                                                         # missing manual thumbs.
4724                                                         $validated = true;
4725                                                         $value = $this->stripAltText( $value, $holders );
4726                                                         break;
4727                                                 case 'link':
4728                                                         $chars = self::EXT_LINK_URL_CLASS;
4729                                                         $prots = $this->mUrlProtocols;
4730                                                         if ( $value === '' ) {
4731                                                                 $paramName = 'no-link';
4732                                                                 $value = true;
4733                                                                 $validated = true;
4734                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4735                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4736                                                                         $paramName = 'link-url';
4737                                                                         $this->mOutput->addExternalLink( $value );
4738                                                                         $validated = true;
4739                                                                 }
4740                                                         } else {
4741                                                                 $linkTitle = Title::newFromText( $value );
4742                                                                 if ( $linkTitle ) {
4743                                                                         $paramName = 'link-title';
4744                                                                         $value = $linkTitle;
4745                                                                         $this->mOutput->addLink( $linkTitle );
4746                                                                         $validated = true;
4747                                                                 }
4748                                                         }
4749                                                         break;
4750                                                 default:
4751                                                         # Most other things appear to be empty or numeric...
4752                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4753                                                 }
4754                                         }
4755
4756                                         if ( $validated ) {
4757                                                 $params[$type][$paramName] = $value;
4758                                         }
4759                                 }
4760                         }
4761                         if ( !$validated ) {
4762                                 $caption = $part;
4763                         }
4764                 }
4765
4766                 # Process alignment parameters
4767                 if ( $params['horizAlign'] ) {
4768                         $params['frame']['align'] = key( $params['horizAlign'] );
4769                 }
4770                 if ( $params['vertAlign'] ) {
4771                         $params['frame']['valign'] = key( $params['vertAlign'] );
4772                 }
4773
4774                 $params['frame']['caption'] = $caption;
4775
4776                 # Will the image be presented in a frame, with the caption below?
4777                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4778                                  isset( $params['frame']['framed'] ) ||
4779                                  isset( $params['frame']['thumbnail'] ) ||
4780                                  isset( $params['frame']['manualthumb'] );
4781
4782                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4783                 # came to also set the caption, ordinary text after the image -- which
4784                 # makes no sense, because that just repeats the text multiple times in
4785                 # screen readers.  It *also* came to set the title attribute.
4786                 #
4787                 # Now that we have an alt attribute, we should not set the alt text to
4788                 # equal the caption: that's worse than useless, it just repeats the
4789                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4790                 # use the unnamed parameter for alt text as well, just for the time be-
4791                 # ing, if the unnamed param is set and the alt param is not.
4792                 #
4793                 # For the future, we need to figure out if we want to tweak this more,
4794                 # e.g., introducing a title= parameter for the title; ignoring the un-
4795                 # named parameter entirely for images without a caption; adding an ex-
4796                 # plicit caption= parameter and preserving the old magic unnamed para-
4797                 # meter for BC; ...
4798                 if ( $imageIsFramed ) { # Framed image
4799                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4800                                 # No caption or alt text, add the filename as the alt text so
4801                                 # that screen readers at least get some description of the image
4802                                 $params['frame']['alt'] = $title->getText();
4803                         }
4804                         # Do not set $params['frame']['title'] because tooltips don't make sense
4805                         # for framed images
4806                 } else { # Inline image
4807                         if ( !isset( $params['frame']['alt'] ) ) {
4808                                 # No alt text, use the "caption" for the alt text
4809                                 if ( $caption !== '') {
4810                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4811                                 } else {
4812                                         # No caption, fall back to using the filename for the
4813                                         # alt text
4814                                         $params['frame']['alt'] = $title->getText();
4815                                 }
4816                         }
4817                         # Use the "caption" for the tooltip text
4818                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4819                 }
4820
4821                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4822
4823                 # Linker does the rest
4824                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4825
4826                 # Give the handler a chance to modify the parser object
4827                 if ( $handler ) {
4828                         $handler->parserTransformHook( $this, $file );
4829                 }
4830
4831                 return $ret;
4832         }
4833
4834         protected function stripAltText( $caption, $holders ) {
4835                 # Strip bad stuff out of the title (tooltip).  We can't just use
4836                 # replaceLinkHoldersText() here, because if this function is called
4837                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4838                 if ( $holders ) {
4839                         $tooltip = $holders->replaceText( $caption );
4840                 } else {
4841                         $tooltip = $this->replaceLinkHoldersText( $caption );
4842                 }
4843
4844                 # make sure there are no placeholders in thumbnail attributes
4845                 # that are later expanded to html- so expand them now and
4846                 # remove the tags
4847                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4848                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4849
4850                 return $tooltip;
4851         }
4852
4853         /**
4854          * Set a flag in the output object indicating that the content is dynamic and
4855          * shouldn't be cached.
4856          */
4857         function disableCache() {
4858                 wfDebug( "Parser output marked as uncacheable.\n" );
4859                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
4860                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
4861         }
4862
4863         /**
4864          * Callback from the Sanitizer for expanding items found in HTML attribute
4865          * values, so they can be safely tested and escaped.
4866          *
4867          * @param $text String
4868          * @param $frame PPFrame
4869          * @return String
4870          * @private
4871          */
4872         function attributeStripCallback( &$text, $frame = false ) {
4873                 $text = $this->replaceVariables( $text, $frame );
4874                 $text = $this->mStripState->unstripBoth( $text );
4875                 return $text;
4876         }
4877
4878         /**
4879          * Accessor
4880          */
4881         function getTags() {
4882                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4883         }
4884
4885         /**
4886          * Break wikitext input into sections, and either pull or replace
4887          * some particular section's text.
4888          *
4889          * External callers should use the getSection and replaceSection methods.
4890          *
4891          * @param $text String: Page wikitext
4892          * @param $section String: a section identifier string of the form:
4893          *   <flag1> - <flag2> - ... - <section number>
4894          *
4895          * Currently the only recognised flag is "T", which means the target section number
4896          * was derived during a template inclusion parse, in other words this is a template
4897          * section edit link. If no flags are given, it was an ordinary section edit link.
4898          * This flag is required to avoid a section numbering mismatch when a section is
4899          * enclosed by <includeonly> (bug 6563).
4900          *
4901          * The section number 0 pulls the text before the first heading; other numbers will
4902          * pull the given section along with its lower-level subsections. If the section is
4903          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4904          *
4905          * @param $mode String: one of "get" or "replace"
4906          * @param $newText String: replacement text for section data.
4907          * @return String: for "get", the extracted section text.
4908          *                 for "replace", the whole page with the section replaced.
4909          */
4910         private function extractSections( $text, $section, $mode, $newText='' ) {
4911                 global $wgTitle;
4912                 $this->clearState();
4913                 $this->setTitle( $wgTitle ); # not generally used but removes an ugly failure mode
4914                 $this->mOptions = new ParserOptions;
4915                 $this->setOutputType( self::OT_PLAIN );
4916                 $outText = '';
4917                 $frame = $this->getPreprocessor()->newFrame();
4918
4919                 # Process section extraction flags
4920                 $flags = 0;
4921                 $sectionParts = explode( '-', $section );
4922                 $sectionIndex = array_pop( $sectionParts );
4923                 foreach ( $sectionParts as $part ) {
4924                         if ( $part === 'T' ) {
4925                                 $flags |= self::PTD_FOR_INCLUSION;
4926                         }
4927                 }
4928                 # Preprocess the text
4929                 $root = $this->preprocessToDom( $text, $flags );
4930
4931                 # <h> nodes indicate section breaks
4932                 # They can only occur at the top level, so we can find them by iterating the root's children
4933                 $node = $root->getFirstChild();
4934
4935                 # Find the target section
4936                 if ( $sectionIndex == 0 ) {
4937                         # Section zero doesn't nest, level=big
4938                         $targetLevel = 1000;
4939                 } else {
4940                         while ( $node ) {
4941                                 if ( $node->getName() === 'h' ) {
4942                                         $bits = $node->splitHeading();
4943                                         if ( $bits['i'] == $sectionIndex ) {
4944                                                 $targetLevel = $bits['level'];
4945                                                 break;
4946                                         }
4947                                 }
4948                                 if ( $mode === 'replace' ) {
4949                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4950                                 }
4951                                 $node = $node->getNextSibling();
4952                         }
4953                 }
4954
4955                 if ( !$node ) {
4956                         # Not found
4957                         if ( $mode === 'get' ) {
4958                                 return $newText;
4959                         } else {
4960                                 return $text;
4961                         }
4962                 }
4963
4964                 # Find the end of the section, including nested sections
4965                 do {
4966                         if ( $node->getName() === 'h' ) {
4967                                 $bits = $node->splitHeading();
4968                                 $curLevel = $bits['level'];
4969                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4970                                         break;
4971                                 }
4972                         }
4973                         if ( $mode === 'get' ) {
4974                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4975                         }
4976                         $node = $node->getNextSibling();
4977                 } while ( $node );
4978
4979                 # Write out the remainder (in replace mode only)
4980                 if ( $mode === 'replace' ) {
4981                         # Output the replacement text
4982                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4983                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4984                         # Only add trailing whitespace if there is newText
4985                         if ( $newText != "" ) {
4986                                 $outText .= $newText . "\n\n";
4987                         }
4988
4989                         while ( $node ) {
4990                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4991                                 $node = $node->getNextSibling();
4992                         }
4993                 }
4994
4995                 if ( is_string( $outText ) ) {
4996                         # Re-insert stripped tags
4997                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4998                 }
4999
5000                 return $outText;
5001         }
5002
5003         /**
5004          * This function returns the text of a section, specified by a number ($section).
5005          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5006          * the first section before any such heading (section 0).
5007          *
5008          * If a section contains subsections, these are also returned.
5009          *
5010          * @param $text String: text to look in
5011          * @param $section String: section identifier
5012          * @param $deftext String: default to return if section is not found
5013          * @return string text of the requested section
5014          */
5015         public function getSection( $text, $section, $deftext='' ) {
5016                 return $this->extractSections( $text, $section, "get", $deftext );
5017         }
5018
5019         public function replaceSection( $oldtext, $section, $text ) {
5020                 return $this->extractSections( $oldtext, $section, "replace", $text );
5021         }
5022
5023         /**
5024          * Get the ID of the revision we are parsing
5025          *
5026          * @return Mixed: integer or null
5027          */
5028         function getRevisionId() {
5029                 return $this->mRevisionId;
5030         }
5031
5032         /**
5033          * Get the timestamp associated with the current revision, adjusted for
5034          * the default server-local timestamp
5035          */
5036         function getRevisionTimestamp() {
5037                 if ( is_null( $this->mRevisionTimestamp ) ) {
5038                         wfProfileIn( __METHOD__ );
5039                         global $wgContLang;
5040                         $dbr = wfGetDB( DB_SLAVE );
5041                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
5042                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
5043
5044                         # Normalize timestamp to internal MW format for timezone processing.
5045                         # This has the added side-effect of replacing a null value with
5046                         # the current time, which gives us more sensible behavior for
5047                         # previews.
5048                         $timestamp = wfTimestamp( TS_MW, $timestamp );
5049
5050                         # The cryptic '' timezone parameter tells to use the site-default
5051                         # timezone offset instead of the user settings.
5052                         #
5053                         # Since this value will be saved into the parser cache, served
5054                         # to other users, and potentially even used inside links and such,
5055                         # it needs to be consistent for all visitors.
5056                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5057
5058                         wfProfileOut( __METHOD__ );
5059                 }
5060                 return $this->mRevisionTimestamp;
5061         }
5062
5063         /**
5064          * Get the name of the user that edited the last revision
5065          *
5066          * @return String: user name
5067          */
5068         function getRevisionUser() {
5069                 # if this template is subst: the revision id will be blank,
5070                 # so just use the current user's name
5071                 if ( $this->mRevisionId ) {
5072                         $revision = Revision::newFromId( $this->mRevisionId );
5073                         $revuser = $revision->getUserText();
5074                 } else {
5075                         global $wgUser;
5076                         $revuser = $wgUser->getName();
5077                 }
5078                 return $revuser;
5079         }
5080
5081         /**
5082          * Mutator for $mDefaultSort
5083          *
5084          * @param $sort New value
5085          */
5086         public function setDefaultSort( $sort ) {
5087                 $this->mDefaultSort = $sort;
5088         }
5089
5090         /**
5091          * Accessor for $mDefaultSort
5092          * Will use the title/prefixed title if none is set
5093          *
5094          * @return string
5095          */
5096         public function getDefaultSort() {
5097                 global $wgCategoryPrefixedDefaultSortkey;
5098                 if ( $this->mDefaultSort !== false ) {
5099                         return $this->mDefaultSort;
5100                 } elseif ( $this->mTitle->getNamespace() == NS_CATEGORY ||
5101                         !$wgCategoryPrefixedDefaultSortkey )
5102                 {
5103                         return $this->mTitle->getText();
5104                 } else {
5105                         return $this->mTitle->getPrefixedText();
5106                 }
5107         }
5108
5109         /**
5110          * Accessor for $mDefaultSort
5111          * Unlike getDefaultSort(), will return false if none is set
5112          *
5113          * @return string or false
5114          */
5115         public function getCustomDefaultSort() {
5116                 return $this->mDefaultSort;
5117         }
5118
5119         /**
5120          * Try to guess the section anchor name based on a wikitext fragment
5121          * presumably extracted from a heading, for example "Header" from
5122          * "== Header ==".
5123          */
5124         public function guessSectionNameFromWikiText( $text ) {
5125                 # Strip out wikitext links(they break the anchor)
5126                 $text = $this->stripSectionName( $text );
5127                 $headline = Sanitizer::decodeCharReferences( $text );
5128                 # strip out HTML
5129                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5130                 $headline = trim( $headline );
5131                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5132                 $replacearray = array(
5133                         '%3A' => ':',
5134                         '%' => '.'
5135                 );
5136                 return str_replace(
5137                         array_keys( $replacearray ),
5138                         array_values( $replacearray ),
5139                         $sectionanchor );
5140         }
5141
5142         /**
5143          * Strips a text string of wikitext for use in a section anchor
5144          *
5145          * Accepts a text string and then removes all wikitext from the
5146          * string and leaves only the resultant text (i.e. the result of
5147          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5148          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5149          * to create valid section anchors by mimicing the output of the
5150          * parser when headings are parsed.
5151          *
5152          * @param $text String: text string to be stripped of wikitext
5153          * for use in a Section anchor
5154          * @return Filtered text string
5155          */
5156         public function stripSectionName( $text ) {
5157                 # Strip internal link markup
5158                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5159                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5160
5161                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5162                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5163                 # on how many empty links there are on the page - need to figure that out.
5164                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5165
5166                 # Parse wikitext quotes (italics & bold)
5167                 $text = $this->doQuotes( $text );
5168
5169                 # Strip HTML tags
5170                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5171                 return $text;
5172         }
5173
5174         function srvus( $text ) {
5175                 return $this->testSrvus( $text, $this->mOutputType );
5176         }
5177
5178         /**
5179          * strip/replaceVariables/unstrip for preprocessor regression testing
5180          */
5181         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5182                 $this->clearState();
5183                 if ( !$title instanceof Title ) {
5184                         $title = Title::newFromText( $title );
5185                 }
5186                 $this->mTitle = $title;
5187                 $this->mOptions = $options;
5188                 $this->setOutputType( $outputType );
5189                 $text = $this->replaceVariables( $text );
5190                 $text = $this->mStripState->unstripBoth( $text );
5191                 $text = Sanitizer::removeHTMLtags( $text );
5192                 return $text;
5193         }
5194
5195         function testPst( $text, $title, $options ) {
5196                 global $wgUser;
5197                 if ( !$title instanceof Title ) {
5198                         $title = Title::newFromText( $title );
5199                 }
5200                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5201         }
5202
5203         function testPreprocess( $text, $title, $options ) {
5204                 if ( !$title instanceof Title ) {
5205                         $title = Title::newFromText( $title );
5206                 }
5207                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5208         }
5209
5210         function markerSkipCallback( $s, $callback ) {
5211                 $i = 0;
5212                 $out = '';
5213                 while ( $i < strlen( $s ) ) {
5214                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5215                         if ( $markerStart === false ) {
5216                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5217                                 break;
5218                         } else {
5219                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5220                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5221                                 if ( $markerEnd === false ) {
5222                                         $out .= substr( $s, $markerStart );
5223                                         break;
5224                                 } else {
5225                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5226                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5227                                         $i = $markerEnd;
5228                                 }
5229                         }
5230                 }
5231                 return $out;
5232         }
5233
5234         function serialiseHalfParsedText( $text ) {
5235                 $data = array();
5236                 $data['text'] = $text;
5237
5238                 # First, find all strip markers, and store their
5239                 #  data in an array.
5240                 $stripState = new StripState;
5241                 $pos = 0;
5242                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5243                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5244                 {
5245                         $end_pos += strlen( self::MARKER_SUFFIX );
5246                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5247
5248                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5249                                 $replaceArray = $stripState->general;
5250                                 $stripText = $this->mStripState->general->data[$marker];
5251                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5252                                 $replaceArray = $stripState->nowiki;
5253                                 $stripText = $this->mStripState->nowiki->data[$marker];
5254                         } else {
5255                                 throw new MWException( "Hanging strip marker: '$marker'." );
5256                         }
5257
5258                         $replaceArray->setPair( $marker, $stripText );
5259                         $pos = $end_pos;
5260                 }
5261                 $data['stripstate'] = $stripState;
5262
5263                 # Now, find all of our links, and store THEIR
5264                 #  data in an array! :)
5265                 $links = array( 'internal' => array(), 'interwiki' => array() );
5266                 $pos = 0;
5267
5268                 # Internal links
5269                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5270                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5271
5272                         $ns = trim( $ns );
5273                         if ( empty( $links['internal'][$ns] ) ) {
5274                                 $links['internal'][$ns] = array();
5275                         }
5276
5277                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5278                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5279                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5280                 }
5281
5282                 $pos = 0;
5283
5284                 # Interwiki links
5285                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5286                         $data = substr( $text, $start_pos );
5287                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5288                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5289                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5290                 }
5291
5292                 $data['linkholder'] = $links;
5293
5294                 return $data;
5295         }
5296
5297         /**
5298          * TODO: document
5299          * @param $data Array
5300          * @param $intPrefix String unique identifying prefix
5301          * @return String
5302          */
5303         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5304                 if ( !$intPrefix ) {
5305                         $intPrefix = $this->getRandomString();
5306                 }
5307
5308                 # First, extract the strip state.
5309                 $stripState = $data['stripstate'];
5310                 $this->mStripState->general->merge( $stripState->general );
5311                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5312
5313                 # Now, extract the text, and renumber links
5314                 $text = $data['text'];
5315                 $links = $data['linkholder'];
5316
5317                 # Internal...
5318                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5319                         foreach ( $nsLinks as $key => $entry ) {
5320                                 $newKey = $intPrefix . '-' . $key;
5321                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5322
5323                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5324                         }
5325                 }
5326
5327                 # Interwiki...
5328                 foreach ( $links['interwiki'] as $key => $entry ) {
5329                         $newKey = "$intPrefix-$key";
5330                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5331
5332                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5333                 }
5334
5335                 # Should be good to go.
5336                 return $text;
5337         }
5338 }
5339
5340 /**
5341  * @todo document, briefly.
5342  * @ingroup Parser
5343  */
5344 class StripState {
5345         var $general, $nowiki;
5346
5347         function __construct() {
5348                 $this->general = new ReplacementArray;
5349                 $this->nowiki = new ReplacementArray;
5350         }
5351
5352         function unstripGeneral( $text ) {
5353                 wfProfileIn( __METHOD__ );
5354                 do {
5355                         $oldText = $text;
5356                         $text = $this->general->replace( $text );
5357                 } while ( $text !== $oldText );
5358                 wfProfileOut( __METHOD__ );
5359                 return $text;
5360         }
5361
5362         function unstripNoWiki( $text ) {
5363                 wfProfileIn( __METHOD__ );
5364                 do {
5365                         $oldText = $text;
5366                         $text = $this->nowiki->replace( $text );
5367                 } while ( $text !== $oldText );
5368                 wfProfileOut( __METHOD__ );
5369                 return $text;
5370         }
5371
5372         function unstripBoth( $text ) {
5373                 wfProfileIn( __METHOD__ );
5374                 do {
5375                         $oldText = $text;
5376                         $text = $this->general->replace( $text );
5377                         $text = $this->nowiki->replace( $text );
5378                 } while ( $text !== $oldText );
5379                 wfProfileOut( __METHOD__ );
5380                 return $text;
5381         }
5382 }
5383
5384 /**
5385  * @todo document, briefly.
5386  * @ingroup Parser
5387  */
5388 class OnlyIncludeReplacer {
5389         var $output = '';
5390
5391         function replace( $matches ) {
5392                 if ( substr( $matches[1], -1 ) === "\n" ) {
5393                         $this->output .= substr( $matches[1], 0, -1 );
5394                 } else {
5395                         $this->output .= $matches[1];
5396                 }
5397         }
5398 }