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