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         var $mUser; # User object; only used when doing pre-save transform
 107
 108         # Temporary
 109         # These are variables reset at least once per parse regardless of $clearState
 110         var $mOptions;      # ParserOptions object
 111         var $mTitle;        # Title context, used for self-link rendering and similar things
 112         var $mOutputType;   # Output type, one of the OT_xxx constants
 113         var $ot;            # Shortcut alias, see setOutputType()
 114         var $mRevisionObject; # The revision object of the specified revision ID
 115         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 116         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 117         var $mRevisionUser; # Userto display in {{REVISIONUSER}} tag
 118         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 119
 120         /**
 121          * Constructor
 122          *
 123          * @public
 124          */
 125         function __construct( $conf = array() ) {
 126                 $this->mConf = $conf;
 127                 $this->mTagHooks = array();
 128                 $this->mTransparentTagHooks = array();
 129                 $this->mFunctionHooks = array();
 130                 $this->mFunctionTagHooks = array();
 131                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 132                 $this->mDefaultStripList = $this->mStripList = array();
 133                 $this->mUrlProtocols = wfUrlProtocols();
 134                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 135                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 136                 $this->mVarCache = array();
 137                 if ( isset( $conf['preprocessorClass'] ) ) {
 138                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 139                 } elseif ( extension_loaded( 'domxml' ) ) {
 140                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 141                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 142                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 143                 } elseif ( extension_loaded( 'dom' ) ) {
 144                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 145                 } else {
 146                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 147                 }
 148                 $this->mMarkerIndex = 0;
 149                 $this->mFirstCall = true;
 150         }
 151
 152         /**
 153          * Reduce memory usage to reduce the impact of circular references
 154          */
 155         function __destruct() {
 156                 if ( isset( $this->mLinkHolders ) ) {
 157                         $this->mLinkHolders->__destruct();
 158                 }
 159                 foreach ( $this as $name => $value ) {
 160                         unset( $this->$name );
 161                 }
 162         }
 163
 164         /**
 165          * Do various kinds of initialisation on the first call of the parser
 166          */
 167         function firstCallInit() {
 168                 if ( !$this->mFirstCall ) {
 169                         return;
 170                 }
 171                 $this->mFirstCall = false;
 172
 173                 wfProfileIn( __METHOD__ );
 174
 175                 CoreParserFunctions::register( $this );
 176                 CoreTagHooks::register( $this );
 177                 $this->initialiseVariables();
 178
 179                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 180                 wfProfileOut( __METHOD__ );
 181         }
 182
 183         /**
 184          * Clear Parser state
 185          *
 186          * @private
 187          */
 188         function clearState() {
 189                 wfProfileIn( __METHOD__ );
 190                 if ( $this->mFirstCall ) {
 191                         $this->firstCallInit();
 192                 }
 193                 $this->mOutput = new ParserOutput;
 194                 $this->mAutonumber = 0;
 195                 $this->mLastSection = '';
 196                 $this->mDTopen = false;
 197                 $this->mIncludeCount = array();
 198                 $this->mStripState = new StripState;
 199                 $this->mArgStack = false;
 200                 $this->mInPre = false;
 201                 $this->mLinkHolders = new LinkHolderArray( $this );
 202                 $this->mLinkID = 0;
 203                 $this->mRevisionObject = $this->mRevisionTimestamp =
 204                         $this->mRevisionId = $this->mRevisionUser = null;
 205                 $this->mVarCache = array();
 206                 $this->mUser = null;
 207
 208                 /**
 209                  * Prefix for temporary replacement strings for the multipass parser.
 210                  * \x07 should never appear in input as it's disallowed in XML.
 211                  * Using it at the front also gives us a little extra robustness
 212                  * since it shouldn't match when butted up against identifier-like
 213                  * string constructs.
 214                  *
 215                  * Must not consist of all title characters, or else it will change
 216                  * the behaviour of <nowiki> in a link.
 217                  */
 218                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 219                 # Changed to \x7f to allow XML double-parsing -- TS
 220                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 221
 222
 223                 # Clear these on every parse, bug 4549
 224                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 225
 226                 $this->mShowToc = true;
 227                 $this->mForceTocPosition = false;
 228                 $this->mIncludeSizes = array(
 229                         'post-expand' => 0,
 230                         'arg' => 0,
 231                 );
 232                 $this->mPPNodeCount = 0;
 233                 $this->mDefaultSort = false;
 234                 $this->mHeadings = array();
 235                 $this->mDoubleUnderscores = array();
 236                 $this->mExpensiveFunctionCount = 0;
 237
 238                 # Fix cloning
 239                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 240                         $this->mPreprocessor = null;
 241                 }
 242
 243                 wfRunHooks( 'ParserClearState', array( &$this ) );
 244                 wfProfileOut( __METHOD__ );
 245         }
 246
 247         /**
 248          * Convert wikitext to HTML
 249          * Do not call this function recursively.
 250          *
 251          * @param $text String: text we want to parse
 252          * @param $title A title object
 253          * @param $options ParserOptions
 254          * @param $linestart boolean
 255          * @param $clearState boolean
 256          * @param $revid Int: number to pass in {{REVISIONID}}
 257          * @return ParserOutput a ParserOutput
 258          */
 259         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 260                 /**
 261                  * First pass--just handle <nowiki> sections, pass the rest off
 262                  * to internalParse() which does all the real work.
 263                  */
 264
 265                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 266                 $fname = __METHOD__.'-' . wfGetCaller();
 267                 wfProfileIn( __METHOD__ );
 268                 wfProfileIn( $fname );
 269
 270                 if ( $clearState ) {
 271                         $this->clearState();
 272                 }
 273
 274                 $options->resetUsage();
 275                 $this->mOptions = $options;
 276                 $this->setTitle( $title ); # Page title has to be set for the pre-processor
 277
 278                 $oldRevisionId = $this->mRevisionId;
 279                 $oldRevisionObject = $this->mRevisionObject;
 280                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 281                 $oldRevisionUser = $this->mRevisionUser;
 282                 if ( $revid !== null ) {
 283                         $this->mRevisionId = $revid;
 284                         $this->mRevisionObject = null;
 285                         $this->mRevisionTimestamp = null;
 286                         $this->mRevisionUser = null;
 287                 }
 288                 $this->setOutputType( self::OT_HTML );
 289                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 290                 # No more strip!
 291                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 292                 $text = $this->internalParse( $text );
 293
 294                 $text = $this->mStripState->unstripGeneral( $text );
 295
 296                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 297                 $fixtags = array(
 298                         # french spaces, last one Guillemet-left
 299                         # only if there is something before the space
 300                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 301                         # french spaces, Guillemet-right
 302                         '/(\\302\\253) /' => '\\1&#160;',
 303                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 304                 );
 305                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 306
 307                 $text = $this->doBlockLevels( $text, $linestart );
 308
 309                 $this->replaceLinkHolders( $text );
 310
 311                 /**
 312                  * The page doesn't get language converted if
 313                  * a) It's disabled
 314                  * b) Content isn't converted
 315                  * c) It's a conversion table
 316                  */
 317                 if ( !( $wgDisableLangConversion
 318                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 319                                 || $this->mTitle->isConversionTable() ) ) {
 320
 321                         # The position of the convert() call should not be changed. it
 322                         # assumes that the links are all replaced and the only thing left
 323                         # is the <nowiki> mark.
 324
 325                         $text = $wgContLang->convert( $text );
 326                 }
 327
 328                 /**
 329                  * A page get its title converted except:
 330                  * a) Language conversion is globally disabled
 331                  * b) Title convert is globally disabled
 332                  * c) The page is a redirect page
 333                  * d) User request with a "linkconvert" set to "no"
 334                  * e) A "nocontentconvert" magic word has been set
 335                  * f) A "notitleconvert" magic word has been set
 336                  * g) User sets "noconvertlink" in his/her preference
 337                  *
 338                  * Note that if a user tries to set a title in a conversion
 339                  * rule but content conversion was not done, then the parser
 340                  * won't pick it up.  This is probably expected behavior.
 341                  */
 342                 if ( !( $wgDisableLangConversion
 343                                 || $wgDisableTitleConversion
 344                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 345                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 346                                 || $this->mOutput->getDisplayTitle() !== false ) )
 347                 {
 348                         $convruletitle = $wgContLang->getConvRuleTitle();
 349                         if ( $convruletitle ) {
 350                                 $this->mOutput->setTitleText( $convruletitle );
 351                         } else {
 352                                 $titleText = $wgContLang->convertTitle( $title );
 353                                 $this->mOutput->setTitleText( $titleText );
 354                         }
 355                 }
 356
 357                 $text = $this->mStripState->unstripNoWiki( $text );
 358
 359                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 360
 361 //!JF Move to its own function
 362
 363                 $uniq_prefix = $this->mUniqPrefix;
 364                 $matches = array();
 365                 $elements = array_keys( $this->mTransparentTagHooks );
 366                 $text = $this->extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 367
 368                 foreach ( $matches as $marker => $data ) {
 369                         list( $element, $content, $params, $tag ) = $data;
 370                         $tagName = strtolower( $element );
 371                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 372                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 373                         } else {
 374                                 $output = $tag;
 375                         }
 376                         $this->mStripState->general->setPair( $marker, $output );
 377                 }
 378                 $text = $this->mStripState->unstripGeneral( $text );
 379
 380                 $text = Sanitizer::normalizeCharReferences( $text );
 381
 382                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 383                         $text = MWTidy::tidy( $text );
 384                 } else {
 385                         # attempt to sanitize at least some nesting problems
 386                         # (bug #2702 and quite a few others)
 387                         $tidyregs = array(
 388                                 # ''Something [http://www.cool.com cool''] -->
 389                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 390                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 391                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 392                                 # fix up an anchor inside another anchor, only
 393                                 # at least for a single single nested link (bug 3695)
 394                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 395                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 396                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 397                                 # contains a div, so fix it up here; replace
 398                                 # div with escaped text
 399                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 400                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 401                                 # remove empty italic or bold tag pairs, some
 402                                 # introduced by rules above
 403                                 '/<([bi])><\/\\1>/' => '',
 404                         );
 405
 406                         $text = preg_replace(
 407                                 array_keys( $tidyregs ),
 408                                 array_values( $tidyregs ),
 409                                 $text );
 410                 }
 411                 global $wgExpensiveParserFunctionLimit;
 412                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 413                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 414                 }
 415
 416                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 417
 418                 # Information on include size limits, for the benefit of users who try to skirt them
 419                 if ( $this->mOptions->getEnableLimitReport() ) {
 420                         $max = $this->mOptions->getMaxIncludeSize();
 421                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 422                         $limitReport =
 423                                 "NewPP limit report\n" .
 424                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 425                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 426                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 427                                 $PFreport;
 428                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 429                         $text .= "\n<!-- \n$limitReport-->\n";
 430                 }
 431                 $this->mOutput->setText( $text );
 432
 433                 $this->mRevisionId = $oldRevisionId;
 434                 $this->mRevisionObject = $oldRevisionObject;
 435                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 436                 $this->mRevisionUser = $oldRevisionUser;
 437                 wfProfileOut( $fname );
 438                 wfProfileOut( __METHOD__ );
 439
 440                 return $this->mOutput;
 441         }
 442
 443         /**
 444          * Recursive parser entry point that can be called from an extension tag
 445          * hook.
 446          *
 447          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 448          *
 449          * @param $text String: text extension wants to have parsed
 450          * @param $frame PPFrame: The frame to use for expanding any template variables
 451          */
 452         function recursiveTagParse( $text, $frame=false ) {
 453                 wfProfileIn( __METHOD__ );
 454                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 455                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 456                 $text = $this->internalParse( $text, false, $frame );
 457                 wfProfileOut( __METHOD__ );
 458                 return $text;
 459         }
 460
 461         /**
 462          * Expand templates and variables in the text, producing valid, static wikitext.
 463          * Also removes comments.
 464          */
 465         function preprocess( $text, $title, $options, $revid = null ) {
 466                 wfProfileIn( __METHOD__ );
 467                 $this->clearState();
 468                 $this->setOutputType( self::OT_PREPROCESS );
 469                 $options->resetUsage();
 470                 $this->mOptions = $options;
 471                 $this->setTitle( $title );
 472                 if ( $revid !== null ) {
 473                         $this->mRevisionId = $revid;
 474                 }
 475                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 476                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 477                 $text = $this->replaceVariables( $text );
 478                 $text = $this->mStripState->unstripBoth( $text );
 479                 wfProfileOut( __METHOD__ );
 480                 return $text;
 481         }
 482
 483         /**
 484          * Process the wikitext for the ?preload= feature. (bug 5210)
 485          *
 486          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 487          * comments, templates, arguments, tags hooks and parser functions are untouched.
 488          */
 489         public function getPreloadText( $text, $title, $options ) {
 490                 # Parser (re)initialisation
 491                 $this->clearState();
 492                 $this->setOutputType( self::OT_PLAIN );
 493                 $options->resetUsage();
 494                 $this->mOptions = $options;
 495                 $this->setTitle( $title );
 496
 497                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 498                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 499                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 500         }
 501
 502         /**
 503          * Get a random string
 504          *
 505          * @private
 506          * @static
 507          */
 508         static private function getRandomString() {
 509                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 510         }
 511
 512         /**
 513          * Set the current user.
 514          * Should only be used when doing pre-save transform.
 515          *
 516          * @param $user Mixed: User object or null (to reset)
 517          */
 518         function setUser( $user ) {
 519                 $this->mUser = $user;
 520         }
 521
 522         /**
 523          * Accessor for mUniqPrefix.
 524          *
 525          * @return String
 526          */
 527         public function uniqPrefix() {
 528                 if ( !isset( $this->mUniqPrefix ) ) {
 529                         # @todo Fixme: this is probably *horribly wrong*
 530                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 531                         # if this is called for a parser cache hit, the parser may not
 532                         # have ever been initialized in the first place.
 533                         # Not really sure what the heck is supposed to be going on here.
 534                         return '';
 535                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 536                 }
 537                 return $this->mUniqPrefix;
 538         }
 539
 540         /**
 541          * Set the context title
 542          */
 543         function setTitle( $t ) {
 544                 if ( !$t || $t instanceof FakeTitle ) {
 545                         $t = Title::newFromText( 'NO TITLE' );
 546                 }
 547
 548                 if ( strval( $t->getFragment() ) !== '' ) {
 549                         # Strip the fragment to avoid various odd effects
 550                         $this->mTitle = clone $t;
 551                         $this->mTitle->setFragment( '' );
 552                 } else {
 553                         $this->mTitle = $t;
 554                 }
 555         }
 556
 557         /**
 558          * Accessor for the Title object
 559          *
 560          * @return Title object
 561          */
 562         function getTitle() {
 563                 return $this->mTitle;
 564         }
 565
 566         /**
 567          * Accessor/mutator for the Title object
 568          *
 569          * @param $x New Title object or null to just get the current one
 570          * @return Title object
 571          */
 572         function Title( $x = null ) {
 573                 return wfSetVar( $this->mTitle, $x );
 574         }
 575
 576         /**
 577          * Set the output type
 578          *
 579          * @param $ot Integer: new value
 580          */
 581         function setOutputType( $ot ) {
 582                 $this->mOutputType = $ot;
 583                 # Shortcut alias
 584                 $this->ot = array(
 585                         'html' => $ot == self::OT_HTML,
 586                         'wiki' => $ot == self::OT_WIKI,
 587                         'pre' => $ot == self::OT_PREPROCESS,
 588                         'plain' => $ot == self::OT_PLAIN,
 589                 );
 590         }
 591
 592         /**
 593          * Accessor/mutator for the output type
 594          *
 595          * @param $x New value or null to just get the current one
 596          * @return Integer
 597          */
 598         function OutputType( $x = null ) {
 599                 return wfSetVar( $this->mOutputType, $x );
 600         }
 601
 602         /**
 603          * Get the ParserOutput object
 604          *
 605          * @return ParserOutput object
 606          */
 607         function getOutput() {
 608                 return $this->mOutput;
 609         }
 610
 611         /**
 612          * Get the ParserOptions object
 613          *
 614          * @return ParserOptions object
 615          */
 616         function getOptions() {
 617                 return $this->mOptions;
 618         }
 619
 620         /**
 621          * Accessor/mutator for the ParserOptions object
 622          *
 623          * @param $x New value or null to just get the current one
 624          * @return Current ParserOptions object
 625          */
 626         function Options( $x = null ) {
 627                 return wfSetVar( $this->mOptions, $x );
 628         }
 629
 630         function nextLinkID() {
 631                 return $this->mLinkID++;
 632         }
 633
 634         function getFunctionLang() {
 635                 global $wgLang, $wgContLang;
 636
 637                 $target = $this->mOptions->getTargetLanguage();
 638                 if ( $target !== null ) {
 639                         return $target;
 640                 } else {
 641                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 642                 }
 643         }
 644
 645         /**
 646          * Get a User object either from $this->mUser, if set, or from the
 647          * ParserOptions object otherwise
 648          *
 649          * @return User object
 650          */
 651         function getUser() {
 652                 if ( !is_null( $this->mUser ) ) {
 653                         return $this->mUser;
 654                 }
 655                 return $this->mOptions->getUser();
 656         }
 657
 658         /**
 659          * Get a preprocessor object
 660          *
 661          * @return Preprocessor instance
 662          */
 663         function getPreprocessor() {
 664                 if ( !isset( $this->mPreprocessor ) ) {
 665                         $class = $this->mPreprocessorClass;
 666                         $this->mPreprocessor = new $class( $this );
 667                 }
 668                 return $this->mPreprocessor;
 669         }
 670
 671         /**
 672          * Replaces all occurrences of HTML-style comments and the given tags
 673          * in the text with a random marker and returns the next text. The output
 674          * parameter $matches will be an associative array filled with data in
 675          * the form:
 676          *   'UNIQ-xxxxx' => array(
 677          *     'element',
 678          *     'tag content',
 679          *     array( 'param' => 'x' ),
 680          *     '<element param="x">tag content</element>' ) )
 681          *
 682          * @param $elements list of element names. Comments are always extracted.
 683          * @param $text Source text string.
 684          * @param $matches Out parameter, Array: extracted tags
 685          * @param $uniq_prefix
 686          * @return String: stripped text
 687          *
 688          * @static
 689          */
 690         public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 691                 static $n = 1;
 692                 $stripped = '';
 693                 $matches = array();
 694
 695                 $taglist = implode( '|', $elements );
 696                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 697
 698                 while ( $text != '' ) {
 699                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 700                         $stripped .= $p[0];
 701                         if ( count( $p ) < 5 ) {
 702                                 break;
 703                         }
 704                         if ( count( $p ) > 5 ) {
 705                                 # comment
 706                                 $element    = $p[4];
 707                                 $attributes = '';
 708                                 $close      = '';
 709                                 $inside     = $p[5];
 710                         } else {
 711                                 # tag
 712                                 $element    = $p[1];
 713                                 $attributes = $p[2];
 714                                 $close      = $p[3];
 715                                 $inside     = $p[4];
 716                         }
 717
 718                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 719                         $stripped .= $marker;
 720
 721                         if ( $close === '/>' ) {
 722                                 # Empty element tag, <tag />
 723                                 $content = null;
 724                                 $text = $inside;
 725                                 $tail = null;
 726                         } else {
 727                                 if ( $element === '!--' ) {
 728                                         $end = '/(-->)/';
 729                                 } else {
 730                                         $end = "/(<\\/$element\\s*>)/i";
 731                                 }
 732                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 733                                 $content = $q[0];
 734                                 if ( count( $q ) < 3 ) {
 735                                         # No end tag -- let it run out to the end of the text.
 736                                         $tail = '';
 737                                         $text = '';
 738                                 } else {
 739                                         $tail = $q[1];
 740                                         $text = $q[2];
 741                                 }
 742                         }
 743
 744                         $matches[$marker] = array( $element,
 745                                 $content,
 746                                 Sanitizer::decodeTagAttributes( $attributes ),
 747                                 "<$element$attributes$close$content$tail" );
 748                 }
 749                 return $stripped;
 750         }
 751
 752         /**
 753          * Get a list of strippable XML-like elements
 754          */
 755         function getStripList() {
 756                 return $this->mStripList;
 757         }
 758
 759         /**
 760          * @deprecated use replaceVariables
 761          */
 762         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 763                 return $text;
 764         }
 765
 766         /**
 767          * Restores pre, math, and other extensions removed by strip()
 768          *
 769          * always call unstripNoWiki() after this one
 770          * @private
 771          * @deprecated use $this->mStripState->unstrip()
 772          */
 773         function unstrip( $text, $state ) {
 774                 return $state->unstripGeneral( $text );
 775         }
 776
 777         /**
 778          * Always call this after unstrip() to preserve the order
 779          *
 780          * @private
 781          * @deprecated use $this->mStripState->unstrip()
 782          */
 783         function unstripNoWiki( $text, $state ) {
 784                 return $state->unstripNoWiki( $text );
 785         }
 786
 787         /**
 788          * @deprecated use $this->mStripState->unstripBoth()
 789          */
 790         function unstripForHTML( $text ) {
 791                 return $this->mStripState->unstripBoth( $text );
 792         }
 793
 794         /**
 795          * Add an item to the strip state
 796          * Returns the unique tag which must be inserted into the stripped text
 797          * The tag will be replaced with the original text in unstrip()
 798          *
 799          * @private
 800          */
 801         function insertStripItem( $text ) {
 802                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 803                 $this->mMarkerIndex++;
 804                 $this->mStripState->general->setPair( $rnd, $text );
 805                 return $rnd;
 806         }
 807
 808         /**
 809          * Interface with html tidy
 810          * @deprecated Use MWTidy::tidy()
 811          */
 812         public static function tidy( $text ) {
 813                 wfDeprecated( __METHOD__ );
 814                 return MWTidy::tidy( $text );
 815         }
 816
 817         /**
 818          * parse the wiki syntax used to render tables
 819          *
 820          * @private
 821          */
 822         function doTableStuff( $text ) {
 823                 wfProfileIn( __METHOD__ );
 824
 825                 $lines = StringUtils::explode( "\n", $text );
 826                 $out = '';
 827                 $td_history = array(); # Is currently a td tag open?
 828                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 829                 $tr_history = array(); # Is currently a tr tag open?
 830                 $tr_attributes = array(); # history of tr attributes
 831                 $has_opened_tr = array(); # Did this table open a <tr> element?
 832                 $indent_level = 0; # indent level of the table
 833
 834                 foreach ( $lines as $outLine ) {
 835                         $line = trim( $outLine );
 836
 837                         if ( $line === '' ) { # empty line, go to next line
 838                                 $out .= $outLine."\n";
 839                                 continue;
 840                         }
 841
 842                         $first_character = $line[0];
 843                         $matches = array();
 844
 845                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 846                                 # First check if we are starting a new table
 847                                 $indent_level = strlen( $matches[1] );
 848
 849                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 850                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 851
 852                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 853                                 array_push( $td_history , false );
 854                                 array_push( $last_tag_history , '' );
 855                                 array_push( $tr_history , false );
 856                                 array_push( $tr_attributes , '' );
 857                                 array_push( $has_opened_tr , false );
 858                         } elseif ( count( $td_history ) == 0 ) {
 859                                 # Don't do any of the following
 860                                 $out .= $outLine."\n";
 861                                 continue;
 862                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 863                                 # We are ending a table
 864                                 $line = '</table>' . substr( $line , 2 );
 865                                 $last_tag = array_pop( $last_tag_history );
 866
 867                                 if ( !array_pop( $has_opened_tr ) ) {
 868                                         $line = "<tr><td></td></tr>{$line}";
 869                                 }
 870
 871                                 if ( array_pop( $tr_history ) ) {
 872                                         $line = "</tr>{$line}";
 873                                 }
 874
 875                                 if ( array_pop( $td_history ) ) {
 876                                         $line = "</{$last_tag}>{$line}";
 877                                 }
 878                                 array_pop( $tr_attributes );
 879                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 880                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 881                                 # Now we have a table row
 882                                 $line = preg_replace( '#^\|-+#', '', $line );
 883
 884                                 # Whats after the tag is now only attributes
 885                                 $attributes = $this->mStripState->unstripBoth( $line );
 886                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 887                                 array_pop( $tr_attributes );
 888                                 array_push( $tr_attributes, $attributes );
 889
 890                                 $line = '';
 891                                 $last_tag = array_pop( $last_tag_history );
 892                                 array_pop( $has_opened_tr );
 893                                 array_push( $has_opened_tr , true );
 894
 895                                 if ( array_pop( $tr_history ) ) {
 896                                         $line = '</tr>';
 897                                 }
 898
 899                                 if ( array_pop( $td_history ) ) {
 900                                         $line = "</{$last_tag}>{$line}";
 901                                 }
 902
 903                                 $outLine = $line;
 904                                 array_push( $tr_history , false );
 905                                 array_push( $td_history , false );
 906                                 array_push( $last_tag_history , '' );
 907                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 908                                 # This might be cell elements, td, th or captions
 909                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 910                                         $first_character = '+';
 911                                         $line = substr( $line , 1 );
 912                                 }
 913
 914                                 $line = substr( $line , 1 );
 915
 916                                 if ( $first_character === '!' ) {
 917                                         $line = str_replace( '!!' , '||' , $line );
 918                                 }
 919
 920                                 # Split up multiple cells on the same line.
 921                                 # FIXME : This can result in improper nesting of tags processed
 922                                 # by earlier parser steps, but should avoid splitting up eg
 923                                 # attribute values containing literal "||".
 924                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 925
 926                                 $outLine = '';
 927
 928                                 # Loop through each table cell
 929                                 foreach ( $cells as $cell ) {
 930                                         $previous = '';
 931                                         if ( $first_character !== '+' ) {
 932                                                 $tr_after = array_pop( $tr_attributes );
 933                                                 if ( !array_pop( $tr_history ) ) {
 934                                                         $previous = "<tr{$tr_after}>\n";
 935                                                 }
 936                                                 array_push( $tr_history , true );
 937                                                 array_push( $tr_attributes , '' );
 938                                                 array_pop( $has_opened_tr );
 939                                                 array_push( $has_opened_tr , true );
 940                                         }
 941
 942                                         $last_tag = array_pop( $last_tag_history );
 943
 944                                         if ( array_pop( $td_history ) ) {
 945                                                 $previous = "</{$last_tag}>\n{$previous}";
 946                                         }
 947
 948                                         if ( $first_character === '|' ) {
 949                                                 $last_tag = 'td';
 950                                         } elseif ( $first_character === '!' ) {
 951                                                 $last_tag = 'th';
 952                                         } elseif ( $first_character === '+' ) {
 953                                                 $last_tag = 'caption';
 954                                         } else {
 955                                                 $last_tag = '';
 956                                         }
 957
 958                                         array_push( $last_tag_history , $last_tag );
 959
 960                                         # A cell could contain both parameters and data
 961                                         $cell_data = explode( '|' , $cell , 2 );
 962
 963                                         # Bug 553: Note that a '|' inside an invalid link should not
 964                                         # be mistaken as delimiting cell parameters
 965                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 966                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 967                                         } elseif ( count( $cell_data ) == 1 ) {
 968                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 969                                         } else {
 970                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 971                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 972                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 973                                         }
 974
 975                                         $outLine .= $cell;
 976                                         array_push( $td_history , true );
 977                                 }
 978                         }
 979                         $out .= $outLine . "\n";
 980                 }
 981
 982                 # Closing open td, tr && table
 983                 while ( count( $td_history ) > 0 ) {
 984                         if ( array_pop( $td_history ) ) {
 985                                 $out .= "</td>\n";
 986                         }
 987                         if ( array_pop( $tr_history ) ) {
 988                                 $out .= "</tr>\n";
 989                         }
 990                         if ( !array_pop( $has_opened_tr ) ) {
 991                                 $out .= "<tr><td></td></tr>\n" ;
 992                         }
 993
 994                         $out .= "</table>\n";
 995                 }
 996
 997                 # Remove trailing line-ending (b/c)
 998                 if ( substr( $out, -1 ) === "\n" ) {
 999                         $out = substr( $out, 0, -1 );
1000                 }
1001
1002                 # special case: don't return empty table
1003                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1004                         $out = '';
1005                 }
1006
1007                 wfProfileOut( __METHOD__ );
1008
1009                 return $out;
1010         }
1011
1012         /**
1013          * Helper function for parse() that transforms wiki markup into
1014          * HTML. Only called for $mOutputType == self::OT_HTML.
1015          *
1016          * @private
1017          */
1018         function internalParse( $text, $isMain = true, $frame=false ) {
1019                 wfProfileIn( __METHOD__ );
1020
1021                 $origText = $text;
1022
1023                 # Hook to suspend the parser in this state
1024                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1025                         wfProfileOut( __METHOD__ );
1026                         return $text ;
1027                 }
1028
1029                 # if $frame is provided, then use $frame for replacing any variables
1030                 if ( $frame ) {
1031                         # use frame depth to infer how include/noinclude tags should be handled
1032                         # depth=0 means this is the top-level document; otherwise it's an included document
1033                         if ( !$frame->depth ) {
1034                                 $flag = 0;
1035                         } else {
1036                                 $flag = Parser::PTD_FOR_INCLUSION;
1037                         }
1038                         $dom = $this->preprocessToDom( $text, $flag );
1039                         $text = $frame->expand( $dom );
1040                 } else {
1041                         # if $frame is not provided, then use old-style replaceVariables
1042                         $text = $this->replaceVariables( $text );
1043                 }
1044
1045                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1046                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1047
1048                 # Tables need to come after variable replacement for things to work
1049                 # properly; putting them before other transformations should keep
1050                 # exciting things like link expansions from showing up in surprising
1051                 # places.
1052                 $text = $this->doTableStuff( $text );
1053
1054                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1055
1056                 $text = $this->doDoubleUnderscore( $text );
1057
1058                 $text = $this->doHeadings( $text );
1059                 if ( $this->mOptions->getUseDynamicDates() ) {
1060                         $df = DateFormatter::getInstance();
1061                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1062                 }
1063                 $text = $this->replaceInternalLinks( $text );
1064                 $text = $this->doAllQuotes( $text );
1065                 $text = $this->replaceExternalLinks( $text );
1066
1067                 # replaceInternalLinks may sometimes leave behind
1068                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1069                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1070
1071                 $text = $this->doMagicLinks( $text );
1072                 $text = $this->formatHeadings( $text, $origText, $isMain );
1073
1074                 wfProfileOut( __METHOD__ );
1075                 return $text;
1076         }
1077
1078         /**
1079          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1080          * magic external links.
1081          *
1082          * DML
1083          * @private
1084          */
1085         function doMagicLinks( $text ) {
1086                 wfProfileIn( __METHOD__ );
1087                 $prots = $this->mUrlProtocols;
1088                 $urlChar = self::EXT_LINK_URL_CLASS;
1089                 $text = preg_replace_callback(
1090                         '!(?:                           # Start cases
1091                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1092                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1093                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1094                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1095                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1096                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1097                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1098                                     [0-9Xx]                 # check digit
1099                                     \b)
1100                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1101                 wfProfileOut( __METHOD__ );
1102                 return $text;
1103         }
1104
1105         function magicLinkCallback( $m ) {
1106                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1107                         # Skip anchor
1108                         return $m[0];
1109                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1110                         # Skip HTML element
1111                         return $m[0];
1112                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1113                         # Free external link
1114                         return $this->makeFreeExternalLink( $m[0] );
1115                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1116                         # RFC or PMID
1117                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1118                                 $keyword = 'RFC';
1119                                 $urlmsg = 'rfcurl';
1120                                 $CssClass = 'mw-magiclink-rfc';
1121                                 $id = $m[4];
1122                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1123                                 $keyword = 'PMID';
1124                                 $urlmsg = 'pubmedurl';
1125                                 $CssClass = 'mw-magiclink-pmid';
1126                                 $id = $m[4];
1127                         } else {
1128                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1129                                         substr( $m[0], 0, 20 ) . '"' );
1130                         }
1131                         $url = wfMsgForContent( $urlmsg, $id);
1132                         $sk = $this->mOptions->getSkin( $this->mTitle );
1133                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1134                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1135                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1136                         # ISBN
1137                         $isbn = $m[5];
1138                         $num = strtr( $isbn, array(
1139                                 '-' => '',
1140                                 ' ' => '',
1141                                 'x' => 'X',
1142                         ));
1143                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1144                         return'<a href="' .
1145                                 $titleObj->escapeLocalUrl() .
1146                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1147                 } else {
1148                         return $m[0];
1149                 }
1150         }
1151
1152         /**
1153          * Make a free external link, given a user-supplied URL
1154          * @return HTML
1155          * @private
1156          */
1157         function makeFreeExternalLink( $url ) {
1158                 global $wgContLang;
1159                 wfProfileIn( __METHOD__ );
1160
1161                 $sk = $this->mOptions->getSkin( $this->mTitle );
1162                 $trail = '';
1163
1164                 # The characters '<' and '>' (which were escaped by
1165                 # removeHTMLtags()) should not be included in
1166                 # URLs, per RFC 2396.
1167                 $m2 = array();
1168                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1169                         $trail = substr( $url, $m2[0][1] ) . $trail;
1170                         $url = substr( $url, 0, $m2[0][1] );
1171                 }
1172
1173                 # Move trailing punctuation to $trail
1174                 $sep = ',;\.:!?';
1175                 # If there is no left bracket, then consider right brackets fair game too
1176                 if ( strpos( $url, '(' ) === false ) {
1177                         $sep .= ')';
1178                 }
1179
1180                 $numSepChars = strspn( strrev( $url ), $sep );
1181                 if ( $numSepChars ) {
1182                         $trail = substr( $url, -$numSepChars ) . $trail;
1183                         $url = substr( $url, 0, -$numSepChars );
1184                 }
1185
1186                 $url = Sanitizer::cleanUrl( $url );
1187
1188                 # Is this an external image?
1189                 $text = $this->maybeMakeExternalImage( $url );
1190                 if ( $text === false ) {
1191                         # Not an image, make a link
1192                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1193                                 $this->getExternalLinkAttribs( $url ) );
1194                         # Register it in the output object...
1195                         # Replace unnecessary URL escape codes with their equivalent characters
1196                         $pasteurized = self::replaceUnusualEscapes( $url );
1197                         $this->mOutput->addExternalLink( $pasteurized );
1198                 }
1199                 wfProfileOut( __METHOD__ );
1200                 return $text . $trail;
1201         }
1202
1203
1204         /**
1205          * Parse headers and return html
1206          *
1207          * @private
1208          */
1209         function doHeadings( $text ) {
1210                 wfProfileIn( __METHOD__ );
1211                 for ( $i = 6; $i >= 1; --$i ) {
1212                         $h = str_repeat( '=', $i );
1213                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1214                           "<h$i>\\1</h$i>", $text );
1215                 }
1216                 wfProfileOut( __METHOD__ );
1217                 return $text;
1218         }
1219
1220         /**
1221          * Replace single quotes with HTML markup
1222          * @private
1223          * @return string the altered text
1224          */
1225         function doAllQuotes( $text ) {
1226                 wfProfileIn( __METHOD__ );
1227                 $outtext = '';
1228                 $lines = StringUtils::explode( "\n", $text );
1229                 foreach ( $lines as $line ) {
1230                         $outtext .= $this->doQuotes( $line ) . "\n";
1231                 }
1232                 $outtext = substr( $outtext, 0,-1 );
1233                 wfProfileOut( __METHOD__ );
1234                 return $outtext;
1235         }
1236
1237         /**
1238          * Helper function for doAllQuotes()
1239          */
1240         public function doQuotes( $text ) {
1241                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1242                 if ( count( $arr ) == 1 ) {
1243                         return $text;
1244                 } else {
1245                         # First, do some preliminary work. This may shift some apostrophes from
1246                         # being mark-up to being text. It also counts the number of occurrences
1247                         # of bold and italics mark-ups.
1248                         $numbold = 0;
1249                         $numitalics = 0;
1250                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1251                                 if ( ( $i % 2 ) == 1 ) {
1252                                         # If there are ever four apostrophes, assume the first is supposed to
1253                                         # be text, and the remaining three constitute mark-up for bold text.
1254                                         if ( strlen( $arr[$i] ) == 4 ) {
1255                                                 $arr[$i-1] .= "'";
1256                                                 $arr[$i] = "'''";
1257                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1258                                                 # If there are more than 5 apostrophes in a row, assume they're all
1259                                                 # text except for the last 5.
1260                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1261                                                 $arr[$i] = "'''''";
1262                                         }
1263                                         # Count the number of occurrences of bold and italics mark-ups.
1264                                         # We are not counting sequences of five apostrophes.
1265                                         if ( strlen( $arr[$i] ) == 2 ) {
1266                                                 $numitalics++;
1267                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1268                                                 $numbold++;
1269                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1270                                                 $numitalics++;
1271                                                 $numbold++;
1272                                         }
1273                                 }
1274                         }
1275
1276                         # If there is an odd number of both bold and italics, it is likely
1277                         # that one of the bold ones was meant to be an apostrophe followed
1278                         # by italics. Which one we cannot know for certain, but it is more
1279                         # likely to be one that has a single-letter word before it.
1280                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1281                                 $i = 0;
1282                                 $firstsingleletterword = -1;
1283                                 $firstmultiletterword = -1;
1284                                 $firstspace = -1;
1285                                 foreach ( $arr as $r ) {
1286                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1287                                                 $x1 = substr( $arr[$i-1], -1 );
1288                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1289                                                 if ( $x1 === ' ' ) {
1290                                                         if ( $firstspace == -1 ) {
1291                                                                 $firstspace = $i;
1292                                                         }
1293                                                 } elseif ( $x2 === ' ') {
1294                                                         if ( $firstsingleletterword == -1 ) {
1295                                                                 $firstsingleletterword = $i;
1296                                                         }
1297                                                 } else {
1298                                                         if ( $firstmultiletterword == -1 ) {
1299                                                                 $firstmultiletterword = $i;
1300                                                         }
1301                                                 }
1302                                         }
1303                                         $i++;
1304                                 }
1305
1306                                 # If there is a single-letter word, use it!
1307                                 if ( $firstsingleletterword > -1 ) {
1308                                         $arr[$firstsingleletterword] = "''";
1309                                         $arr[$firstsingleletterword-1] .= "'";
1310                                 } elseif ( $firstmultiletterword > -1 ) {
1311                                         # If not, but there's a multi-letter word, use that one.
1312                                         $arr[$firstmultiletterword] = "''";
1313                                         $arr[$firstmultiletterword-1] .= "'";
1314                                 } elseif ( $firstspace > -1 ) {
1315                                         # ... otherwise use the first one that has neither.
1316                                         # (notice that it is possible for all three to be -1 if, for example,
1317                                         # there is only one pentuple-apostrophe in the line)
1318                                         $arr[$firstspace] = "''";
1319                                         $arr[$firstspace-1] .= "'";
1320                                 }
1321                         }
1322
1323                         # Now let's actually convert our apostrophic mush to HTML!
1324                         $output = '';
1325                         $buffer = '';
1326                         $state = '';
1327                         $i = 0;
1328                         foreach ( $arr as $r ) {
1329                                 if ( ( $i % 2 ) == 0 ) {
1330                                         if ( $state === 'both' ) {
1331                                                 $buffer .= $r;
1332                                         } else {
1333                                                 $output .= $r;
1334                                         }
1335                                 } else {
1336                                         if ( strlen( $r ) == 2 ) {
1337                                                 if ( $state === 'i' ) {
1338                                                         $output .= '</i>'; $state = '';
1339                                                 } elseif ( $state === 'bi' ) {
1340                                                         $output .= '</i>'; $state = 'b';
1341                                                 } elseif ( $state === 'ib' ) {
1342                                                         $output .= '</b></i><b>'; $state = 'b';
1343                                                 } elseif ( $state === 'both' ) {
1344                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1345                                                 } else { # $state can be 'b' or ''
1346                                                         $output .= '<i>'; $state .= 'i';
1347                                                 }
1348                                         } elseif ( strlen( $r ) == 3 ) {
1349                                                 if ( $state === 'b' ) {
1350                                                         $output .= '</b>'; $state = '';
1351                                                 } elseif ( $state === 'bi' ) {
1352                                                         $output .= '</i></b><i>'; $state = 'i';
1353                                                 } elseif ( $state === 'ib' ) {
1354                                                         $output .= '</b>'; $state = 'i';
1355                                                 } elseif ( $state === 'both' ) {
1356                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1357                                                 } else { # $state can be 'i' or ''
1358                                                         $output .= '<b>'; $state .= 'b';
1359                                                 }
1360                                         } elseif ( strlen( $r ) == 5 ) {
1361                                                 if ( $state === 'b' ) {
1362                                                         $output .= '</b><i>'; $state = 'i';
1363                                                 } elseif ( $state === 'i' ) {
1364                                                         $output .= '</i><b>'; $state = 'b';
1365                                                 } elseif ( $state === 'bi' ) {
1366                                                         $output .= '</i></b>'; $state = '';
1367                                                 } elseif ( $state === 'ib' ) {
1368                                                         $output .= '</b></i>'; $state = '';
1369                                                 } elseif ( $state === 'both' ) {
1370                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1371                                                 } else { # ($state == '')
1372                                                         $buffer = ''; $state = 'both';
1373                                                 }
1374                                         }
1375                                 }
1376                                 $i++;
1377                         }
1378                         # Now close all remaining tags.  Notice that the order is important.
1379                         if ( $state === 'b' || $state === 'ib' ) {
1380                                 $output .= '</b>';
1381                         }
1382                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1383                                 $output .= '</i>';
1384                         }
1385                         if ( $state === 'bi' ) {
1386                                 $output .= '</b>';
1387                         }
1388                         # There might be lonely ''''', so make sure we have a buffer
1389                         if ( $state === 'both' && $buffer ) {
1390                                 $output .= '<b><i>'.$buffer.'</i></b>';
1391                         }
1392                         return $output;
1393                 }
1394         }
1395
1396         /**
1397          * Replace external links (REL)
1398          *
1399          * Note: this is all very hackish and the order of execution matters a lot.
1400          * Make sure to run maintenance/parserTests.php if you change this code.
1401          *
1402          * @private
1403          */
1404         function replaceExternalLinks( $text ) {
1405                 global $wgContLang;
1406                 wfProfileIn( __METHOD__ );
1407
1408                 $sk = $this->mOptions->getSkin( $this->mTitle );
1409
1410                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1411                 $s = array_shift( $bits );
1412
1413                 $i = 0;
1414                 while ( $i<count( $bits ) ) {
1415                         $url = $bits[$i++];
1416                         $protocol = $bits[$i++];
1417                         $text = $bits[$i++];
1418                         $trail = $bits[$i++];
1419
1420                         # The characters '<' and '>' (which were escaped by
1421                         # removeHTMLtags()) should not be included in
1422                         # URLs, per RFC 2396.
1423                         $m2 = array();
1424                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1425                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1426                                 $url = substr( $url, 0, $m2[0][1] );
1427                         }
1428
1429                         # If the link text is an image URL, replace it with an <img> tag
1430                         # This happened by accident in the original parser, but some people used it extensively
1431                         $img = $this->maybeMakeExternalImage( $text );
1432                         if ( $img !== false ) {
1433                                 $text = $img;
1434                         }
1435
1436                         $dtrail = '';
1437
1438                         # Set linktype for CSS - if URL==text, link is essentially free
1439                         $linktype = ( $text === $url ) ? 'free' : 'text';
1440
1441                         # No link text, e.g. [http://domain.tld/some.link]
1442                         if ( $text == '' ) {
1443                                 # Autonumber if allowed. See bug #5918
1444                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1445                                         $langObj = $this->getFunctionLang();
1446                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1447                                         $linktype = 'autonumber';
1448                                 } else {
1449                                         # Otherwise just use the URL
1450                                         $text = htmlspecialchars( $url );
1451                                         $linktype = 'free';
1452                                 }
1453                         } else {
1454                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1455                                 # Check for trail
1456                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1457                         }
1458
1459                         $text = $wgContLang->markNoConversion( $text );
1460
1461                         $url = Sanitizer::cleanUrl( $url );
1462
1463                         # Use the encoded URL
1464                         # This means that users can paste URLs directly into the text
1465                         # Funny characters like ö aren't valid in URLs anyway
1466                         # This was changed in August 2004
1467                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1468                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1469
1470                         # Register link in the output object.
1471                         # Replace unnecessary URL escape codes with the referenced character
1472                         # This prevents spammers from hiding links from the filters
1473                         $pasteurized = self::replaceUnusualEscapes( $url );
1474                         $this->mOutput->addExternalLink( $pasteurized );
1475                 }
1476
1477                 wfProfileOut( __METHOD__ );
1478                 return $s;
1479         }
1480
1481         /**
1482          * Get an associative array of additional HTML attributes appropriate for a
1483          * particular external link.  This currently may include rel => nofollow
1484          * (depending on configuration, namespace, and the URL's domain) and/or a
1485          * target attribute (depending on configuration).
1486          *
1487          * @param $url String: optional URL, to extract the domain from for rel =>
1488          *   nofollow if appropriate
1489          * @return Array: associative array of HTML attributes
1490          */
1491         function getExternalLinkAttribs( $url = false ) {
1492                 $attribs = array();
1493                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1494                 $ns = $this->mTitle->getNamespace();
1495                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1496                         $attribs['rel'] = 'nofollow';
1497
1498                         global $wgNoFollowDomainExceptions;
1499                         if ( $wgNoFollowDomainExceptions ) {
1500                                 $bits = wfParseUrl( $url );
1501                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1502                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1503                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1504                                                         unset( $attribs['rel'] );
1505                                                         break;
1506                                                 }
1507                                         }
1508                                 }
1509                         }
1510                 }
1511                 if ( $this->mOptions->getExternalLinkTarget() ) {
1512                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1513                 }
1514                 return $attribs;
1515         }
1516
1517
1518         /**
1519          * Replace unusual URL escape codes with their equivalent characters
1520          *
1521          * @param $url String
1522          * @return String
1523          *
1524          * @todo  This can merge genuinely required bits in the path or query string,
1525          *        breaking legit URLs. A proper fix would treat the various parts of
1526          *        the URL differently; as a workaround, just use the output for
1527          *        statistical records, not for actual linking/output.
1528          */
1529         static function replaceUnusualEscapes( $url ) {
1530                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1531                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1532         }
1533
1534         /**
1535          * Callback function used in replaceUnusualEscapes().
1536          * Replaces unusual URL escape codes with their equivalent character
1537          */
1538         private static function replaceUnusualEscapesCallback( $matches ) {
1539                 $char = urldecode( $matches[0] );
1540                 $ord = ord( $char );
1541                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1542                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1543                         # No, shouldn't be escaped
1544                         return $char;
1545                 } else {
1546                         # Yes, leave it escaped
1547                         return $matches[0];
1548                 }
1549         }
1550
1551         /**
1552          * make an image if it's allowed, either through the global
1553          * option, through the exception, or through the on-wiki whitelist
1554          * @private
1555          */
1556         function maybeMakeExternalImage( $url ) {
1557                 $sk = $this->mOptions->getSkin( $this->mTitle );
1558                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1559                 $imagesexception = !empty( $imagesfrom );
1560                 $text = false;
1561                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1562                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1563                         $imagematch = false;
1564                         foreach ( $imagesfrom as $match ) {
1565                                 if ( strpos( $url, $match ) === 0 ) {
1566                                         $imagematch = true;
1567                                         break;
1568                                 }
1569                         }
1570                 } elseif ( $imagesexception ) {
1571                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1572                 } else {
1573                         $imagematch = false;
1574                 }
1575                 if ( $this->mOptions->getAllowExternalImages()
1576                      || ( $imagesexception && $imagematch ) ) {
1577                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1578                                 # Image found
1579                                 $text = $sk->makeExternalImage( $url );
1580                         }
1581                 }
1582                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1583                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1584                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1585                         foreach ( $whitelist as $entry ) {
1586                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1587                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1588                                         continue;
1589                                 }
1590                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1591                                         # Image matches a whitelist entry
1592                                         $text = $sk->makeExternalImage( $url );
1593                                         break;
1594                                 }
1595                         }
1596                 }
1597                 return $text;
1598         }
1599
1600         /**
1601          * Process [[ ]] wikilinks
1602          * @return String: processed text
1603          *
1604          * @private
1605          */
1606         function replaceInternalLinks( $s ) {
1607                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1608                 return $s;
1609         }
1610
1611         /**
1612          * Process [[ ]] wikilinks (RIL)
1613          * @return LinkHolderArray
1614          *
1615          * @private
1616          */
1617         function replaceInternalLinks2( &$s ) {
1618                 global $wgContLang;
1619
1620                 wfProfileIn( __METHOD__ );
1621
1622                 wfProfileIn( __METHOD__.'-setup' );
1623                 static $tc = FALSE, $e1, $e1_img;
1624                 # the % is needed to support urlencoded titles as well
1625                 if ( !$tc ) {
1626                         $tc = Title::legalChars() . '#%';
1627                         # Match a link having the form [[namespace:link|alternate]]trail
1628                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1629                         # Match cases where there is no "]]", which might still be images
1630                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1631                 }
1632
1633                 $sk = $this->mOptions->getSkin( $this->mTitle );
1634                 $holders = new LinkHolderArray( $this );
1635
1636                 # split the entire text string on occurences of [[
1637                 $a = StringUtils::explode( '[[', ' ' . $s );
1638                 # get the first element (all text up to first [[), and remove the space we added
1639                 $s = $a->current();
1640                 $a->next();
1641                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1642                 $s = substr( $s, 1 );
1643
1644                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1645                 $e2 = null;
1646                 if ( $useLinkPrefixExtension ) {
1647                         # Match the end of a line for a word that's not followed by whitespace,
1648                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1649                         $e2 = wfMsgForContent( 'linkprefix' );
1650                 }
1651
1652                 if ( is_null( $this->mTitle ) ) {
1653                         wfProfileOut( __METHOD__.'-setup' );
1654                         wfProfileOut( __METHOD__ );
1655                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1656                 }
1657                 $nottalk = !$this->mTitle->isTalkPage();
1658
1659                 if ( $useLinkPrefixExtension ) {
1660                         $m = array();
1661                         if ( preg_match( $e2, $s, $m ) ) {
1662                                 $first_prefix = $m[2];
1663                         } else {
1664                                 $first_prefix = false;
1665                         }
1666                 } else {
1667                         $prefix = '';
1668                 }
1669
1670                 if ( $wgContLang->hasVariants() ) {
1671                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1672                 } else {
1673                         $selflink = array( $this->mTitle->getPrefixedText() );
1674                 }
1675                 $useSubpages = $this->areSubpagesAllowed();
1676                 wfProfileOut( __METHOD__.'-setup' );
1677
1678                 # Loop for each link
1679                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1680                         # Check for excessive memory usage
1681                         if ( $holders->isBig() ) {
1682                                 # Too big
1683                                 # Do the existence check, replace the link holders and clear the array
1684                                 $holders->replace( $s );
1685                                 $holders->clear();
1686                         }
1687
1688                         if ( $useLinkPrefixExtension ) {
1689                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1690                                 if ( preg_match( $e2, $s, $m ) ) {
1691                                         $prefix = $m[2];
1692                                         $s = $m[1];
1693                                 } else {
1694                                         $prefix='';
1695                                 }
1696                                 # first link
1697                                 if ( $first_prefix ) {
1698                                         $prefix = $first_prefix;
1699                                         $first_prefix = false;
1700                                 }
1701                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1702                         }
1703
1704                         $might_be_img = false;
1705
1706                         wfProfileIn( __METHOD__."-e1" );
1707                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1708                                 $text = $m[2];
1709                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1710                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1711                                 # the real problem is with the $e1 regex
1712                                 # See bug 1300.
1713                                 #
1714                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1715                                 # and no image is in sight. See bug 2095.
1716                                 #
1717                                 if ( $text !== '' &&
1718                                         substr( $m[3], 0, 1 ) === ']' &&
1719                                         strpos( $text, '[' ) !== false
1720                                 )
1721                                 {
1722                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1723                                         $m[3] = substr( $m[3], 1 );
1724                                 }
1725                                 # fix up urlencoded title texts
1726                                 if ( strpos( $m[1], '%' ) !== false ) {
1727                                         # Should anchors '#' also be rejected?
1728                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode( $m[1] ) );
1729                                 }
1730                                 $trail = $m[3];
1731                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1732                                 $might_be_img = true;
1733                                 $text = $m[2];
1734                                 if ( strpos( $m[1], '%' ) !== false ) {
1735                                         $m[1] = urldecode( $m[1] );
1736                                 }
1737                                 $trail = "";
1738                         } else { # Invalid form; output directly
1739                                 $s .= $prefix . '[[' . $line ;
1740                                 wfProfileOut( __METHOD__."-e1" );
1741                                 continue;
1742                         }
1743                         wfProfileOut( __METHOD__."-e1" );
1744                         wfProfileIn( __METHOD__."-misc" );
1745
1746                         # Don't allow internal links to pages containing
1747                         # PROTO: where PROTO is a valid URL protocol; these
1748                         # should be external links.
1749                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1750                                 $s .= $prefix . '[[' . $line ;
1751                                 wfProfileOut( __METHOD__."-misc" );
1752                                 continue;
1753                         }
1754
1755                         # Make subpage if necessary
1756                         if ( $useSubpages ) {
1757                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1758                         } else {
1759                                 $link = $m[1];
1760                         }
1761
1762                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1763                         if ( !$noforce ) {
1764                                 # Strip off leading ':'
1765                                 $link = substr( $link, 1 );
1766                         }
1767
1768                         wfProfileOut( __METHOD__."-misc" );
1769                         wfProfileIn( __METHOD__."-title" );
1770                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1771                         if ( $nt === null ) {
1772                                 $s .= $prefix . '[[' . $line;
1773                                 wfProfileOut( __METHOD__."-title" );
1774                                 continue;
1775                         }
1776
1777                         $ns = $nt->getNamespace();
1778                         $iw = $nt->getInterWiki();
1779                         wfProfileOut( __METHOD__."-title" );
1780
1781                         if ( $might_be_img ) { # if this is actually an invalid link
1782                                 wfProfileIn( __METHOD__."-might_be_img" );
1783                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1784                                         $found = false;
1785                                         while ( true ) {
1786                                                 # look at the next 'line' to see if we can close it there
1787                                                 $a->next();
1788                                                 $next_line = $a->current();
1789                                                 if ( $next_line === false || $next_line === null ) {
1790                                                         break;
1791                                                 }
1792                                                 $m = explode( ']]', $next_line, 3 );
1793                                                 if ( count( $m ) == 3 ) {
1794                                                         # the first ]] closes the inner link, the second the image
1795                                                         $found = true;
1796                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1797                                                         $trail = $m[2];
1798                                                         break;
1799                                                 } elseif ( count( $m ) == 2 ) {
1800                                                         # if there's exactly one ]] that's fine, we'll keep looking
1801                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1802                                                 } else {
1803                                                         # if $next_line is invalid too, we need look no further
1804                                                         $text .= '[[' . $next_line;
1805                                                         break;
1806                                                 }
1807                                         }
1808                                         if ( !$found ) {
1809                                                 # we couldn't find the end of this imageLink, so output it raw
1810                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1811                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1812                                                 $s .= "{$prefix}[[$link|$text";
1813                                                 # note: no $trail, because without an end, there *is* no trail
1814                                                 wfProfileOut( __METHOD__."-might_be_img" );
1815                                                 continue;
1816                                         }
1817                                 } else { # it's not an image, so output it raw
1818                                         $s .= "{$prefix}[[$link|$text";
1819                                         # note: no $trail, because without an end, there *is* no trail
1820                                         wfProfileOut( __METHOD__."-might_be_img" );
1821                                         continue;
1822                                 }
1823                                 wfProfileOut( __METHOD__."-might_be_img" );
1824                         }
1825
1826                         $wasblank = ( $text  == '' );
1827                         if ( $wasblank ) {
1828                                 $text = $link;
1829                         } else {
1830                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1831                                 # [[Lista d''e paise d''o munno]] -> <a href="">Lista d''e paise d''o munno</a>
1832                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']] -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1833                                 $text = $this->doQuotes($text);
1834                         }
1835
1836                         # Link not escaped by : , create the various objects
1837                         if ( $noforce ) {
1838
1839                                 # Interwikis
1840                                 wfProfileIn( __METHOD__."-interwiki" );
1841                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1842                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1843                                         $s = rtrim( $s . $prefix );
1844                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1845                                         wfProfileOut( __METHOD__."-interwiki" );
1846                                         continue;
1847                                 }
1848                                 wfProfileOut( __METHOD__."-interwiki" );
1849
1850                                 if ( $ns == NS_FILE ) {
1851                                         wfProfileIn( __METHOD__."-image" );
1852                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1853                                                 if ( $wasblank ) {
1854                                                         # if no parameters were passed, $text
1855                                                         # becomes something like "File:Foo.png",
1856                                                         # which we don't want to pass on to the
1857                                                         # image generator
1858                                                         $text = '';
1859                                                 } else {
1860                                                         # recursively parse links inside the image caption
1861                                                         # actually, this will parse them in any other parameters, too,
1862                                                         # but it might be hard to fix that, and it doesn't matter ATM
1863                                                         $text = $this->replaceExternalLinks( $text );
1864                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1865                                                 }
1866                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1867                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1868                                         } else {
1869                                                 $s .= $prefix . $trail;
1870                                         }
1871                                         $this->mOutput->addImage( $nt->getDBkey() );
1872                                         wfProfileOut( __METHOD__."-image" );
1873                                         continue;
1874
1875                                 }
1876
1877                                 if ( $ns == NS_CATEGORY ) {
1878                                         wfProfileIn( __METHOD__."-category" );
1879                                         $s = rtrim( $s . "\n" ); # bug 87
1880
1881                                         if ( $wasblank ) {
1882                                                 $sortkey = $this->getDefaultSort();
1883                                         } else {
1884                                                 $sortkey = $text;
1885                                         }
1886                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1887                                         $sortkey = str_replace( "\n", '', $sortkey );
1888                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1889                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1890
1891                                         /**
1892                                          * Strip the whitespace Category links produce, see bug 87
1893                                          * @todo We might want to use trim($tmp, "\n") here.
1894                                          */
1895                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1896
1897                                         wfProfileOut( __METHOD__."-category" );
1898                                         continue;
1899                                 }
1900                         }
1901
1902                         # Self-link checking
1903                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1904                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1905                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1906                                         continue;
1907                                 }
1908                         }
1909
1910                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1911                         # FIXME: Should do batch file existence checks, see comment below
1912                         if ( $ns == NS_MEDIA ) {
1913                                 wfProfileIn( __METHOD__."-media" );
1914                                 # Give extensions a chance to select the file revision for us
1915                                 $skip = $time = false;
1916                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1917                                 if ( $skip ) {
1918                                         $link = $sk->link( $nt );
1919                                 } else {
1920                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1921                                 }
1922                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1923                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1924                                 $this->mOutput->addImage( $nt->getDBkey() );
1925                                 wfProfileOut( __METHOD__."-media" );
1926                                 continue;
1927                         }
1928
1929                         wfProfileIn( __METHOD__."-always_known" );
1930                         # Some titles, such as valid special pages or files in foreign repos, should
1931                         # be shown as bluelinks even though they're not included in the page table
1932                         #
1933                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1934                         # batch file existence checks for NS_FILE and NS_MEDIA
1935                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1936                                 $this->mOutput->addLink( $nt );
1937                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1938                         } else {
1939                                 # Links will be added to the output link list after checking
1940                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1941                         }
1942                         wfProfileOut( __METHOD__."-always_known" );
1943                 }
1944                 wfProfileOut( __METHOD__ );
1945                 return $holders;
1946         }
1947
1948         /**
1949          * Make a link placeholder. The text returned can be later resolved to a real link with
1950          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1951          * parsing of interwiki links, and secondly to allow all existence checks and
1952          * article length checks (for stub links) to be bundled into a single query.
1953          *
1954          * @deprecated
1955          */
1956         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1957                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1958         }
1959
1960         /**
1961          * Render a forced-blue link inline; protect against double expansion of
1962          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1963          * Since this little disaster has to split off the trail text to avoid
1964          * breaking URLs in the following text without breaking trails on the
1965          * wiki links, it's been made into a horrible function.
1966          *
1967          * @param $nt Title
1968          * @param $text String
1969          * @param $query String
1970          * @param $trail String
1971          * @param $prefix String
1972          * @return String: HTML-wikitext mix oh yuck
1973          */
1974         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1975                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1976                 $sk = $this->mOptions->getSkin( $this->mTitle );
1977                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1978                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1979                 return $this->armorLinks( $link ) . $trail;
1980         }
1981
1982         /**
1983          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1984          * going to go through further parsing steps before inline URL expansion.
1985          *
1986          * Not needed quite as much as it used to be since free links are a bit
1987          * more sensible these days. But bracketed links are still an issue.
1988          *
1989          * @param $text String: more-or-less HTML
1990          * @return String: less-or-more HTML with NOPARSE bits
1991          */
1992         function armorLinks( $text ) {
1993                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1994                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1995         }
1996
1997         /**
1998          * Return true if subpage links should be expanded on this page.
1999          * @return Boolean
2000          */
2001         function areSubpagesAllowed() {
2002                 # Some namespaces don't allow subpages
2003                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2004         }
2005
2006         /**
2007          * Handle link to subpage if necessary
2008          *
2009          * @param $target String: the source of the link
2010          * @param &$text String: the link text, modified as necessary
2011          * @return string the full name of the link
2012          * @private
2013          */
2014         function maybeDoSubpageLink( $target, &$text ) {
2015                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2016         }
2017
2018         /**#@+
2019          * Used by doBlockLevels()
2020          * @private
2021          */
2022         function closeParagraph() {
2023                 $result = '';
2024                 if ( $this->mLastSection != '' ) {
2025                         $result = '</' . $this->mLastSection  . ">\n";
2026                 }
2027                 $this->mInPre = false;
2028                 $this->mLastSection = '';
2029                 return $result;
2030         }
2031
2032         /**
2033          * getCommon() returns the length of the longest common substring
2034          * of both arguments, starting at the beginning of both.
2035          * @private
2036          */
2037         function getCommon( $st1, $st2 ) {
2038                 $fl = strlen( $st1 );
2039                 $shorter = strlen( $st2 );
2040                 if ( $fl < $shorter ) {
2041                         $shorter = $fl;
2042                 }
2043
2044                 for ( $i = 0; $i < $shorter; ++$i ) {
2045                         if ( $st1{$i} != $st2{$i} ) {
2046                                 break;
2047                         }
2048                 }
2049                 return $i;
2050         }
2051
2052         /**
2053          * These next three functions open, continue, and close the list
2054          * element appropriate to the prefix character passed into them.
2055          * @private
2056          */
2057         function openList( $char ) {
2058                 $result = $this->closeParagraph();
2059
2060                 if ( '*' === $char ) {
2061                         $result .= '<ul><li>';
2062                 } elseif ( '#' === $char ) {
2063                         $result .= '<ol><li>';
2064                 } elseif ( ':' === $char ) {
2065                         $result .= '<dl><dd>';
2066                 } elseif ( ';' === $char ) {
2067                         $result .= '<dl><dt>';
2068                         $this->mDTopen = true;
2069                 } else {
2070                         $result = '<!-- ERR 1 -->';
2071                 }
2072
2073                 return $result;
2074         }
2075
2076         /**
2077          * TODO: document
2078          * @param $char String
2079          * @private
2080          */
2081         function nextItem( $char ) {
2082                 if ( '*' === $char || '#' === $char ) {
2083                         return '</li><li>';
2084                 } elseif ( ':' === $char || ';' === $char ) {
2085                         $close = '</dd>';
2086                         if ( $this->mDTopen ) {
2087                                 $close = '</dt>';
2088                         }
2089                         if ( ';' === $char ) {
2090                                 $this->mDTopen = true;
2091                                 return $close . '<dt>';
2092                         } else {
2093                                 $this->mDTopen = false;
2094                                 return $close . '<dd>';
2095                         }
2096                 }
2097                 return '<!-- ERR 2 -->';
2098         }
2099
2100         /**
2101          * TODO: document
2102          * @param $char String
2103          * @private
2104          */
2105         function closeList( $char ) {
2106                 if ( '*' === $char ) {
2107                         $text = '</li></ul>';
2108                 } elseif ( '#' === $char ) {
2109                         $text = '</li></ol>';
2110                 } elseif ( ':' === $char ) {
2111                         if ( $this->mDTopen ) {
2112                                 $this->mDTopen = false;
2113                                 $text = '</dt></dl>';
2114                         } else {
2115                                 $text = '</dd></dl>';
2116                         }
2117                 } else {
2118                         return '<!-- ERR 3 -->';
2119                 }
2120                 return $text."\n";
2121         }
2122         /**#@-*/
2123
2124         /**
2125          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2126          *
2127          * @param $text String
2128          * @param $linestart Boolean: whether or not this is at the start of a line.
2129          * @private
2130          * @return string the lists rendered as HTML
2131          */
2132         function doBlockLevels( $text, $linestart ) {
2133                 wfProfileIn( __METHOD__ );
2134
2135                 # Parsing through the text line by line.  The main thing
2136                 # happening here is handling of block-level elements p, pre,
2137                 # and making lists from lines starting with * # : etc.
2138                 #
2139                 $textLines = StringUtils::explode( "\n", $text );
2140
2141                 $lastPrefix = $output = '';
2142                 $this->mDTopen = $inBlockElem = false;
2143                 $prefixLength = 0;
2144                 $paragraphStack = false;
2145
2146                 foreach ( $textLines as $oLine ) {
2147                         # Fix up $linestart
2148                         if ( !$linestart ) {
2149                                 $output .= $oLine;
2150                                 $linestart = true;
2151                                 continue;
2152                         }
2153                         # * = ul
2154                         # # = ol
2155                         # ; = dt
2156                         # : = dd
2157
2158                         $lastPrefixLength = strlen( $lastPrefix );
2159                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2160                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2161                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2162                         if ( !$this->mInPre ) {
2163                                 # Multiple prefixes may abut each other for nested lists.
2164                                 $prefixLength = strspn( $oLine, '*#:;' );
2165                                 $prefix = substr( $oLine, 0, $prefixLength );
2166
2167                                 # eh?
2168                                 # ; and : are both from definition-lists, so they're equivalent
2169                                 #  for the purposes of determining whether or not we need to open/close
2170                                 #  elements.
2171                                 $prefix2 = str_replace( ';', ':', $prefix );
2172                                 $t = substr( $oLine, $prefixLength );
2173                                 $this->mInPre = (bool)$preOpenMatch;
2174                         } else {
2175                                 # Don't interpret any other prefixes in preformatted text
2176                                 $prefixLength = 0;
2177                                 $prefix = $prefix2 = '';
2178                                 $t = $oLine;
2179                         }
2180
2181                         # List generation
2182                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2183                                 # Same as the last item, so no need to deal with nesting or opening stuff
2184                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2185                                 $paragraphStack = false;
2186
2187                                 if ( substr( $prefix, -1 ) === ';') {
2188                                         # The one nasty exception: definition lists work like this:
2189                                         # ; title : definition text
2190                                         # So we check for : in the remainder text to split up the
2191                                         # title and definition, without b0rking links.
2192                                         $term = $t2 = '';
2193                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2194                                                 $t = $t2;
2195                                                 $output .= $term . $this->nextItem( ':' );
2196                                         }
2197                                 }
2198                         } elseif ( $prefixLength || $lastPrefixLength ) {
2199                                 # We need to open or close prefixes, or both.
2200
2201                                 # Either open or close a level...
2202                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2203                                 $paragraphStack = false;
2204
2205                                 # Close all the prefixes which aren't shared.
2206                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2207                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2208                                         --$lastPrefixLength;
2209                                 }
2210
2211                                 # Continue the current prefix if appropriate.
2212                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2213                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2214                                 }
2215
2216                                 # Open prefixes where appropriate.
2217                                 while ( $prefixLength > $commonPrefixLength ) {
2218                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2219                                         $output .= $this->openList( $char );
2220
2221                                         if ( ';' === $char ) {
2222                                                 # FIXME: This is dupe of code above
2223                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2224                                                         $t = $t2;
2225                                                         $output .= $term . $this->nextItem( ':' );
2226                                                 }
2227                                         }
2228                                         ++$commonPrefixLength;
2229                                 }
2230                                 $lastPrefix = $prefix2;
2231                         }
2232
2233                         # If we have no prefixes, go to paragraph mode.
2234                         if ( 0 == $prefixLength ) {
2235                                 wfProfileIn( __METHOD__."-paragraph" );
2236                                 # No prefix (not in list)--go to paragraph mode
2237                                 # XXX: use a stack for nestable elements like span, table and div
2238                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2239                                 $closematch = preg_match(
2240                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2241                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2242                                 if ( $openmatch or $closematch ) {
2243                                         $paragraphStack = false;
2244                                         # TODO bug 5718: paragraph closed
2245                                         $output .= $this->closeParagraph();
2246                                         if ( $preOpenMatch and !$preCloseMatch ) {
2247                                                 $this->mInPre = true;
2248                                         }
2249                                         $inBlockElem = !$closematch;
2250                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2251                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2252                                                 # pre
2253                                                 if ( $this->mLastSection !== 'pre' ) {
2254                                                         $paragraphStack = false;
2255                                                         $output .= $this->closeParagraph().'<pre>';
2256                                                         $this->mLastSection = 'pre';
2257                                                 }
2258                                                 $t = substr( $t, 1 );
2259                                         } else {
2260                                                 # paragraph
2261                                                 if ( trim( $t ) === '' ) {
2262                                                         if ( $paragraphStack ) {
2263                                                                 $output .= $paragraphStack.'<br />';
2264                                                                 $paragraphStack = false;
2265                                                                 $this->mLastSection = 'p';
2266                                                         } else {
2267                                                                 if ( $this->mLastSection !== 'p' ) {
2268                                                                         $output .= $this->closeParagraph();
2269                                                                         $this->mLastSection = '';
2270                                                                         $paragraphStack = '<p>';
2271                                                                 } else {
2272                                                                         $paragraphStack = '</p><p>';
2273                                                                 }
2274                                                         }
2275                                                 } else {
2276                                                         if ( $paragraphStack ) {
2277                                                                 $output .= $paragraphStack;
2278                                                                 $paragraphStack = false;
2279                                                                 $this->mLastSection = 'p';
2280                                                         } elseif ( $this->mLastSection !== 'p' ) {
2281                                                                 $output .= $this->closeParagraph().'<p>';
2282                                                                 $this->mLastSection = 'p';
2283                                                         }
2284                                                 }
2285                                         }
2286                                 }
2287                                 wfProfileOut( __METHOD__."-paragraph" );
2288                         }
2289                         # somewhere above we forget to get out of pre block (bug 785)
2290                         if ( $preCloseMatch && $this->mInPre ) {
2291                                 $this->mInPre = false;
2292                         }
2293                         if ( $paragraphStack === false ) {
2294                                 $output .= $t."\n";
2295                         }
2296                 }
2297                 while ( $prefixLength ) {
2298                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2299                         --$prefixLength;
2300                 }
2301                 if ( $this->mLastSection != '' ) {
2302                         $output .= '</' . $this->mLastSection . '>';
2303                         $this->mLastSection = '';
2304                 }
2305
2306                 wfProfileOut( __METHOD__ );
2307                 return $output;
2308         }
2309
2310         /**
2311          * Split up a string on ':', ignoring any occurences inside tags
2312          * to prevent illegal overlapping.
2313          *
2314          * @param $str String: the string to split
2315          * @param &$before String: set to everything before the ':'
2316          * @param &$after String: set to everything after the ':'
2317          * return String: the position of the ':', or false if none found
2318          */
2319         function findColonNoLinks( $str, &$before, &$after ) {
2320                 wfProfileIn( __METHOD__ );
2321
2322                 $pos = strpos( $str, ':' );
2323                 if ( $pos === false ) {
2324                         # Nothing to find!
2325                         wfProfileOut( __METHOD__ );
2326                         return false;
2327                 }
2328
2329                 $lt = strpos( $str, '<' );
2330                 if ( $lt === false || $lt > $pos ) {
2331                         # Easy; no tag nesting to worry about
2332                         $before = substr( $str, 0, $pos );
2333                         $after = substr( $str, $pos+1 );
2334                         wfProfileOut( __METHOD__ );
2335                         return $pos;
2336                 }
2337
2338                 # Ugly state machine to walk through avoiding tags.
2339                 $state = self::COLON_STATE_TEXT;
2340                 $stack = 0;
2341                 $len = strlen( $str );
2342                 for( $i = 0; $i < $len; $i++ ) {
2343                         $c = $str{$i};
2344
2345                         switch( $state ) {
2346                         # (Using the number is a performance hack for common cases)
2347                         case 0: # self::COLON_STATE_TEXT:
2348                                 switch( $c ) {
2349                                 case "<":
2350                                         # Could be either a <start> tag or an </end> tag
2351                                         $state = self::COLON_STATE_TAGSTART;
2352                                         break;
2353                                 case ":":
2354                                         if ( $stack == 0 ) {
2355                                                 # We found it!
2356                                                 $before = substr( $str, 0, $i );
2357                                                 $after = substr( $str, $i + 1 );
2358                                                 wfProfileOut( __METHOD__ );
2359                                                 return $i;
2360                                         }
2361                                         # Embedded in a tag; don't break it.
2362                                         break;
2363                                 default:
2364                                         # Skip ahead looking for something interesting
2365                                         $colon = strpos( $str, ':', $i );
2366                                         if ( $colon === false ) {
2367                                                 # Nothing else interesting
2368                                                 wfProfileOut( __METHOD__ );
2369                                                 return false;
2370                                         }
2371                                         $lt = strpos( $str, '<', $i );
2372                                         if ( $stack === 0 ) {
2373                                                 if ( $lt === false || $colon < $lt ) {
2374                                                         # We found it!
2375                                                         $before = substr( $str, 0, $colon );
2376                                                         $after = substr( $str, $colon + 1 );
2377                                                         wfProfileOut( __METHOD__ );
2378                                                         return $i;
2379                                                 }
2380                                         }
2381                                         if ( $lt === false ) {
2382                                                 # Nothing else interesting to find; abort!
2383                                                 # We're nested, but there's no close tags left. Abort!
2384                                                 break 2;
2385                                         }
2386                                         # Skip ahead to next tag start
2387                                         $i = $lt;
2388                                         $state = self::COLON_STATE_TAGSTART;
2389                                 }
2390                                 break;
2391                         case 1: # self::COLON_STATE_TAG:
2392                                 # In a <tag>
2393                                 switch( $c ) {
2394                                 case ">":
2395                                         $stack++;
2396                                         $state = self::COLON_STATE_TEXT;
2397                                         break;
2398                                 case "/":
2399                                         # Slash may be followed by >?
2400                                         $state = self::COLON_STATE_TAGSLASH;
2401                                         break;
2402                                 default:
2403                                         # ignore
2404                                 }
2405                                 break;
2406                         case 2: # self::COLON_STATE_TAGSTART:
2407                                 switch( $c ) {
2408                                 case "/":
2409                                         $state = self::COLON_STATE_CLOSETAG;
2410                                         break;
2411                                 case "!":
2412                                         $state = self::COLON_STATE_COMMENT;
2413                                         break;
2414                                 case ">":
2415                                         # Illegal early close? This shouldn't happen D:
2416                                         $state = self::COLON_STATE_TEXT;
2417                                         break;
2418                                 default:
2419                                         $state = self::COLON_STATE_TAG;
2420                                 }
2421                                 break;
2422                         case 3: # self::COLON_STATE_CLOSETAG:
2423                                 # In a </tag>
2424                                 if ( $c === ">" ) {
2425                                         $stack--;
2426                                         if ( $stack < 0 ) {
2427                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2428                                                 wfProfileOut( __METHOD__ );
2429                                                 return false;
2430                                         }
2431                                         $state = self::COLON_STATE_TEXT;
2432                                 }
2433                                 break;
2434                         case self::COLON_STATE_TAGSLASH:
2435                                 if ( $c === ">" ) {
2436                                         # Yes, a self-closed tag <blah/>
2437                                         $state = self::COLON_STATE_TEXT;
2438                                 } else {
2439                                         # Probably we're jumping the gun, and this is an attribute
2440                                         $state = self::COLON_STATE_TAG;
2441                                 }
2442                                 break;
2443                         case 5: # self::COLON_STATE_COMMENT:
2444                                 if ( $c === "-" ) {
2445                                         $state = self::COLON_STATE_COMMENTDASH;
2446                                 }
2447                                 break;
2448                         case self::COLON_STATE_COMMENTDASH:
2449                                 if ( $c === "-" ) {
2450                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2451                                 } else {
2452                                         $state = self::COLON_STATE_COMMENT;
2453                                 }
2454                                 break;
2455                         case self::COLON_STATE_COMMENTDASHDASH:
2456                                 if ( $c === ">" ) {
2457                                         $state = self::COLON_STATE_TEXT;
2458                                 } else {
2459                                         $state = self::COLON_STATE_COMMENT;
2460                                 }
2461                                 break;
2462                         default:
2463                                 throw new MWException( "State machine error in " . __METHOD__ );
2464                         }
2465                 }
2466                 if ( $stack > 0 ) {
2467                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2468                         return false;
2469                 }
2470                 wfProfileOut( __METHOD__ );
2471                 return false;
2472         }
2473
2474         /**
2475          * Return value of a magic variable (like PAGENAME)
2476          *
2477          * @private
2478          */
2479         function getVariableValue( $index, $frame=false ) {
2480                 global $wgContLang, $wgSitename, $wgServer;
2481                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2482
2483                 /**
2484                  * Some of these require message or data lookups and can be
2485                  * expensive to check many times.
2486                  */
2487                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2488                         if ( isset( $this->mVarCache[$index] ) ) {
2489                                 return $this->mVarCache[$index];
2490                         }
2491                 }
2492
2493                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2494                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2495
2496                 # Use the time zone
2497                 global $wgLocaltimezone;
2498                 if ( isset( $wgLocaltimezone ) ) {
2499                         $oldtz = date_default_timezone_get();
2500                         date_default_timezone_set( $wgLocaltimezone );
2501                 }
2502
2503                 $localTimestamp = date( 'YmdHis', $ts );
2504                 $localMonth = date( 'm', $ts );
2505                 $localMonth1 = date( 'n', $ts );
2506                 $localMonthName = date( 'n', $ts );
2507                 $localDay = date( 'j', $ts );
2508                 $localDay2 = date( 'd', $ts );
2509                 $localDayOfWeek = date( 'w', $ts );
2510                 $localWeek = date( 'W', $ts );
2511                 $localYear = date( 'Y', $ts );
2512                 $localHour = date( 'H', $ts );
2513                 if ( isset( $wgLocaltimezone ) ) {
2514                         date_default_timezone_set( $oldtz );
2515                 }
2516
2517                 switch ( $index ) {
2518                         case 'currentmonth':
2519                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2520                                 break;
2521                         case 'currentmonth1':
2522                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2523                                 break;
2524                         case 'currentmonthname':
2525                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2526                                 break;
2527                         case 'currentmonthnamegen':
2528                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2529                                 break;
2530                         case 'currentmonthabbrev':
2531                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2532                                 break;
2533                         case 'currentday':
2534                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2535                                 break;
2536                         case 'currentday2':
2537                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2538                                 break;
2539                         case 'localmonth':
2540                                 $value = $wgContLang->formatNum( $localMonth );
2541                                 break;
2542                         case 'localmonth1':
2543                                 $value = $wgContLang->formatNum( $localMonth1 );
2544                                 break;
2545                         case 'localmonthname':
2546                                 $value = $wgContLang->getMonthName( $localMonthName );
2547                                 break;
2548                         case 'localmonthnamegen':
2549                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2550                                 break;
2551                         case 'localmonthabbrev':
2552                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2553                                 break;
2554                         case 'localday':
2555                                 $value = $wgContLang->formatNum( $localDay );
2556                                 break;
2557                         case 'localday2':
2558                                 $value = $wgContLang->formatNum( $localDay2 );
2559                                 break;
2560                         case 'pagename':
2561                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2562                                 break;
2563                         case 'pagenamee':
2564                                 $value = $this->mTitle->getPartialURL();
2565                                 break;
2566                         case 'fullpagename':
2567                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2568                                 break;
2569                         case 'fullpagenamee':
2570                                 $value = $this->mTitle->getPrefixedURL();
2571                                 break;
2572                         case 'subpagename':
2573                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2574                                 break;
2575                         case 'subpagenamee':
2576                                 $value = $this->mTitle->getSubpageUrlForm();
2577                                 break;
2578                         case 'basepagename':
2579                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2580                                 break;
2581                         case 'basepagenamee':
2582                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2583                                 break;
2584                         case 'talkpagename':
2585                                 if ( $this->mTitle->canTalk() ) {
2586                                         $talkPage = $this->mTitle->getTalkPage();
2587                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2588                                 } else {
2589                                         $value = '';
2590                                 }
2591                                 break;
2592                         case 'talkpagenamee':
2593                                 if ( $this->mTitle->canTalk() ) {
2594                                         $talkPage = $this->mTitle->getTalkPage();
2595                                         $value = $talkPage->getPrefixedUrl();
2596                                 } else {
2597                                         $value = '';
2598                                 }
2599                                 break;
2600                         case 'subjectpagename':
2601                                 $subjPage = $this->mTitle->getSubjectPage();
2602                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2603                                 break;
2604                         case 'subjectpagenamee':
2605                                 $subjPage = $this->mTitle->getSubjectPage();
2606                                 $value = $subjPage->getPrefixedUrl();
2607                                 break;
2608                         case 'revisionid':
2609                                 # Let the edit saving system know we should parse the page
2610                                 # *after* a revision ID has been assigned.
2611                                 $this->mOutput->setFlag( 'vary-revision' );
2612                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2613                                 $value = $this->mRevisionId;
2614                                 break;
2615                         case 'revisionday':
2616                                 # Let the edit saving system know we should parse the page
2617                                 # *after* a revision ID has been assigned. This is for null edits.
2618                                 $this->mOutput->setFlag( 'vary-revision' );
2619                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2620                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2621                                 break;
2622                         case 'revisionday2':
2623                                 # Let the edit saving system know we should parse the page
2624                                 # *after* a revision ID has been assigned. This is for null edits.
2625                                 $this->mOutput->setFlag( 'vary-revision' );
2626                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2627                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2628                                 break;
2629                         case 'revisionmonth':
2630                                 # Let the edit saving system know we should parse the page
2631                                 # *after* a revision ID has been assigned. This is for null edits.
2632                                 $this->mOutput->setFlag( 'vary-revision' );
2633                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2634                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2635                                 break;
2636                         case 'revisionmonth1':
2637                                 # Let the edit saving system know we should parse the page
2638                                 # *after* a revision ID has been assigned. This is for null edits.
2639                                 $this->mOutput->setFlag( 'vary-revision' );
2640                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2641                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2642                                 break;
2643                         case 'revisionyear':
2644                                 # Let the edit saving system know we should parse the page
2645                                 # *after* a revision ID has been assigned. This is for null edits.
2646                                 $this->mOutput->setFlag( 'vary-revision' );
2647                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2648                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2649                                 break;
2650                         case 'revisiontimestamp':
2651                                 # Let the edit saving system know we should parse the page
2652                                 # *after* a revision ID has been assigned. This is for null edits.
2653                                 $this->mOutput->setFlag( 'vary-revision' );
2654                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2655                                 $value = $this->getRevisionTimestamp();
2656                                 break;
2657                         case 'revisionuser':
2658                                 # Let the edit saving system know we should parse the page
2659                                 # *after* a revision ID has been assigned. This is for null edits.
2660                                 $this->mOutput->setFlag( 'vary-revision' );
2661                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2662                                 $value = $this->getRevisionUser();
2663                                 break;
2664                         case 'namespace':
2665                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2666                                 break;
2667                         case 'namespacee':
2668                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2669                                 break;
2670                         case 'talkspace':
2671                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2672                                 break;
2673                         case 'talkspacee':
2674                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2675                                 break;
2676                         case 'subjectspace':
2677                                 $value = $this->mTitle->getSubjectNsText();
2678                                 break;
2679                         case 'subjectspacee':
2680                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2681                                 break;
2682                         case 'currentdayname':
2683                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2684                                 break;
2685                         case 'currentyear':
2686                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2687                                 break;
2688                         case 'currenttime':
2689                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2690                                 break;
2691                         case 'currenthour':
2692                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2693                                 break;
2694                         case 'currentweek':
2695                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2696                                 # int to remove the padding
2697                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2698                                 break;
2699                         case 'currentdow':
2700                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2701                                 break;
2702                         case 'localdayname':
2703                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2704                                 break;
2705                         case 'localyear':
2706                                 $value = $wgContLang->formatNum( $localYear, true );
2707                                 break;
2708                         case 'localtime':
2709                                 $value = $wgContLang->time( $localTimestamp, false, false );
2710                                 break;
2711                         case 'localhour':
2712                                 $value = $wgContLang->formatNum( $localHour, true );
2713                                 break;
2714                         case 'localweek':
2715                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2716                                 # int to remove the padding
2717                                 $value = $wgContLang->formatNum( (int)$localWeek );
2718                                 break;
2719                         case 'localdow':
2720                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2721                                 break;
2722                         case 'numberofarticles':
2723                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2724                                 break;
2725                         case 'numberoffiles':
2726                                 $value = $wgContLang->formatNum( SiteStats::images() );
2727                                 break;
2728                         case 'numberofusers':
2729                                 $value = $wgContLang->formatNum( SiteStats::users() );
2730                                 break;
2731                         case 'numberofactiveusers':
2732                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2733                                 break;
2734                         case 'numberofpages':
2735                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2736                                 break;
2737                         case 'numberofadmins':
2738                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2739                                 break;
2740                         case 'numberofedits':
2741                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2742                                 break;
2743                         case 'numberofviews':
2744                                 $value = $wgContLang->formatNum( SiteStats::views() );
2745                                 break;
2746                         case 'currenttimestamp':
2747                                 $value = wfTimestamp( TS_MW, $ts );
2748                                 break;
2749                         case 'localtimestamp':
2750                                 $value = $localTimestamp;
2751                                 break;
2752                         case 'currentversion':
2753                                 $value = SpecialVersion::getVersion();
2754                                 break;
2755                         case 'articlepath':
2756                                 return $wgArticlePath;
2757                         case 'sitename':
2758                                 return $wgSitename;
2759                         case 'server':
2760                                 return $wgServer;
2761                         case 'servername':
2762                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2763                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2764                                 wfRestoreWarnings();
2765                                 return $serverName ? $serverName : $wgServer;
2766                         case 'scriptpath':
2767                                 return $wgScriptPath;
2768                         case 'stylepath':
2769                                 return $wgStylePath;
2770                         case 'directionmark':
2771                                 return $wgContLang->getDirMark();
2772                         case 'contentlanguage':
2773                                 global $wgLanguageCode;
2774                                 return $wgLanguageCode;
2775                         default:
2776                                 $ret = null;
2777                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2778                                         return $ret;
2779                                 } else {
2780                                         return null;
2781                                 }
2782                 }
2783
2784                 if ( $index )
2785                         $this->mVarCache[$index] = $value;
2786
2787                 return $value;
2788         }
2789
2790         /**
2791          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2792          *
2793          * @private
2794          */
2795         function initialiseVariables() {
2796                 wfProfileIn( __METHOD__ );
2797                 $variableIDs = MagicWord::getVariableIDs();
2798                 $substIDs = MagicWord::getSubstIDs();
2799
2800                 $this->mVariables = new MagicWordArray( $variableIDs );
2801                 $this->mSubstWords = new MagicWordArray( $substIDs );
2802                 wfProfileOut( __METHOD__ );
2803         }
2804
2805         /**
2806          * Preprocess some wikitext and return the document tree.
2807          * This is the ghost of replace_variables().
2808          *
2809          * @param $text String: The text to parse
2810          * @param $flags Integer: bitwise combination of:
2811          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2812          *                                     included. Default is to assume a direct page view.
2813          *
2814          * The generated DOM tree must depend only on the input text and the flags.
2815          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2816          *
2817          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2818          * change in the DOM tree for a given text, must be passed through the section identifier
2819          * in the section edit link and thus back to extractSections().
2820          *
2821          * The output of this function is currently only cached in process memory, but a persistent
2822          * cache may be implemented at a later date which takes further advantage of these strict
2823          * dependency requirements.
2824          *
2825          * @private
2826          */
2827         function preprocessToDom( $text, $flags = 0 ) {
2828                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2829                 return $dom;
2830         }
2831
2832         /**
2833          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2834          */
2835         public static function splitWhitespace( $s ) {
2836                 $ltrimmed = ltrim( $s );
2837                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2838                 $trimmed = rtrim( $ltrimmed );
2839                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2840                 if ( $diff > 0 ) {
2841                         $w2 = substr( $ltrimmed, -$diff );
2842                 } else {
2843                         $w2 = '';
2844                 }
2845                 return array( $w1, $trimmed, $w2 );
2846         }
2847
2848         /**
2849          * Replace magic variables, templates, and template arguments
2850          * with the appropriate text. Templates are substituted recursively,
2851          * taking care to avoid infinite loops.
2852          *
2853          * Note that the substitution depends on value of $mOutputType:
2854          *  self::OT_WIKI: only {{subst:}} templates
2855          *  self::OT_PREPROCESS: templates but not extension tags
2856          *  self::OT_HTML: all templates and extension tags
2857          *
2858          * @param $text String: the text to transform
2859          * @param $frame PPFrame Object describing the arguments passed to the template.
2860          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2861          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2862          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2863          * @private
2864          */
2865         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2866                 # Is there any text? Also, Prevent too big inclusions!
2867                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2868                         return $text;
2869                 }
2870                 wfProfileIn( __METHOD__ );
2871
2872                 if ( $frame === false ) {
2873                         $frame = $this->getPreprocessor()->newFrame();
2874                 } elseif ( !( $frame instanceof PPFrame ) ) {
2875                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2876                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2877                 }
2878
2879                 $dom = $this->preprocessToDom( $text );
2880                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2881                 $text = $frame->expand( $dom, $flags );
2882
2883                 wfProfileOut( __METHOD__ );
2884                 return $text;
2885         }
2886
2887         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2888         static function createAssocArgs( $args ) {
2889                 $assocArgs = array();
2890                 $index = 1;
2891                 foreach ( $args as $arg ) {
2892                         $eqpos = strpos( $arg, '=' );
2893                         if ( $eqpos === false ) {
2894                                 $assocArgs[$index++] = $arg;
2895                         } else {
2896                                 $name = trim( substr( $arg, 0, $eqpos ) );
2897                                 $value = trim( substr( $arg, $eqpos+1 ) );
2898                                 if ( $value === false ) {
2899                                         $value = '';
2900                                 }
2901                                 if ( $name !== false ) {
2902                                         $assocArgs[$name] = $value;
2903                                 }
2904                         }
2905                 }
2906
2907                 return $assocArgs;
2908         }
2909
2910         /**
2911          * Warn the user when a parser limitation is reached
2912          * Will warn at most once the user per limitation type
2913          *
2914          * @param $limitationType String: should be one of:
2915          *   'expensive-parserfunction' (corresponding messages:
2916          *       'expensive-parserfunction-warning',
2917          *       'expensive-parserfunction-category')
2918          *   'post-expand-template-argument' (corresponding messages:
2919          *       'post-expand-template-argument-warning',
2920          *       'post-expand-template-argument-category')
2921          *   'post-expand-template-inclusion' (corresponding messages:
2922          *       'post-expand-template-inclusion-warning',
2923          *       'post-expand-template-inclusion-category')
2924          * @param $current Current value
2925          * @param $max Maximum allowed, when an explicit limit has been
2926          *       exceeded, provide the values (optional)
2927          */
2928         function limitationWarn( $limitationType, $current=null, $max=null) {
2929                 # does no harm if $current and $max are present but are unnecessary for the message
2930                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2931                 $this->mOutput->addWarning( $warning );
2932                 $this->addTrackingCategory( "$limitationType-category" );
2933         }
2934
2935         /**
2936          * Return the text of a template, after recursively
2937          * replacing any variables or templates within the template.
2938          *
2939          * @param $piece Array: the parts of the template
2940          *  $piece['title']: the title, i.e. the part before the |
2941          *  $piece['parts']: the parameter array
2942          *  $piece['lineStart']: whether the brace was at the start of a line
2943          * @param $frame PPFrame The current frame, contains template arguments
2944          * @return String: the text of the template
2945          * @private
2946          */
2947         function braceSubstitution( $piece, $frame ) {
2948                 global $wgContLang, $wgNonincludableNamespaces;
2949                 wfProfileIn( __METHOD__ );
2950                 wfProfileIn( __METHOD__.'-setup' );
2951
2952                 # Flags
2953                 $found = false;             # $text has been filled
2954                 $nowiki = false;            # wiki markup in $text should be escaped
2955                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2956                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2957                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2958                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2959
2960                 # Title object, where $text came from
2961                 $title = null;
2962
2963                 # $part1 is the bit before the first |, and must contain only title characters.
2964                 # Various prefixes will be stripped from it later.
2965                 $titleWithSpaces = $frame->expand( $piece['title'] );
2966                 $part1 = trim( $titleWithSpaces );
2967                 $titleText = false;
2968
2969                 # Original title text preserved for various purposes
2970                 $originalTitle = $part1;
2971
2972                 # $args is a list of argument nodes, starting from index 0, not including $part1
2973                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2974                 wfProfileOut( __METHOD__.'-setup' );
2975
2976                 # SUBST
2977                 wfProfileIn( __METHOD__.'-modifiers' );
2978                 if ( !$found ) {
2979
2980                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2981
2982                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2983                         # Decide whether to expand template or keep wikitext as-is.
2984                         if ( $this->ot['wiki'] ) {
2985                                 if ( $substMatch === false ) {
2986                                         $literal = true;  # literal when in PST with no prefix
2987                                 } else {
2988                                         $literal = false; # expand when in PST with subst: or safesubst:
2989                                 }
2990                         } else {
2991                                 if ( $substMatch == 'subst' ) {
2992                                         $literal = true;  # literal when not in PST with plain subst:
2993                                 } else {
2994                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2995                                 }
2996                         }
2997                         if ( $literal ) {
2998                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2999                                 $isLocalObj = true;
3000                                 $found = true;
3001                         }
3002                 }
3003
3004                 # Variables
3005                 if ( !$found && $args->getLength() == 0 ) {
3006                         $id = $this->mVariables->matchStartToEnd( $part1 );
3007                         if ( $id !== false ) {
3008                                 $text = $this->getVariableValue( $id, $frame );
3009                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3010                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3011                                 }
3012                                 $found = true;
3013                         }
3014                 }
3015
3016                 # MSG, MSGNW and RAW
3017                 if ( !$found ) {
3018                         # Check for MSGNW:
3019                         $mwMsgnw = MagicWord::get( 'msgnw' );
3020                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3021                                 $nowiki = true;
3022                         } else {
3023                                 # Remove obsolete MSG:
3024                                 $mwMsg = MagicWord::get( 'msg' );
3025                                 $mwMsg->matchStartAndRemove( $part1 );
3026                         }
3027
3028                         # Check for RAW:
3029                         $mwRaw = MagicWord::get( 'raw' );
3030                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3031                                 $forceRawInterwiki = true;
3032                         }
3033                 }
3034                 wfProfileOut( __METHOD__.'-modifiers' );
3035
3036                 # Parser functions
3037                 if ( !$found ) {
3038                         wfProfileIn( __METHOD__ . '-pfunc' );
3039
3040                         $colonPos = strpos( $part1, ':' );
3041                         if ( $colonPos !== false ) {
3042                                 # Case sensitive functions
3043                                 $function = substr( $part1, 0, $colonPos );
3044                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3045                                         $function = $this->mFunctionSynonyms[1][$function];
3046                                 } else {
3047                                         # Case insensitive functions
3048                                         $function = $wgContLang->lc( $function );
3049                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3050                                                 $function = $this->mFunctionSynonyms[0][$function];
3051                                         } else {
3052                                                 $function = false;
3053                                         }
3054                                 }
3055                                 if ( $function ) {
3056                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3057                                         $initialArgs = array( &$this );
3058                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3059                                         if ( $flags & SFH_OBJECT_ARGS ) {
3060                                                 # Add a frame parameter, and pass the arguments as an array
3061                                                 $allArgs = $initialArgs;
3062                                                 $allArgs[] = $frame;
3063                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3064                                                         $funcArgs[] = $args->item( $i );
3065                                                 }
3066                                                 $allArgs[] = $funcArgs;
3067                                         } else {
3068                                                 # Convert arguments to plain text
3069                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3070                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3071                                                 }
3072                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3073                                         }
3074
3075                                         # Workaround for PHP bug 35229 and similar
3076                                         if ( !is_callable( $callback ) ) {
3077                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3078                                                 wfProfileOut( __METHOD__ );
3079                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3080                                         }
3081                                         $result = call_user_func_array( $callback, $allArgs );
3082                                         $found = true;
3083                                         $noparse = true;
3084                                         $preprocessFlags = 0;
3085
3086                                         if ( is_array( $result ) ) {
3087                                                 if ( isset( $result[0] ) ) {
3088                                                         $text = $result[0];
3089                                                         unset( $result[0] );
3090                                                 }
3091
3092                                                 # Extract flags into the local scope
3093                                                 # This allows callers to set flags such as nowiki, found, etc.
3094                                                 extract( $result );
3095                                         } else {
3096                                                 $text = $result;
3097                                         }
3098                                         if ( !$noparse ) {
3099                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3100                                                 $isChildObj = true;
3101                                         }
3102                                 }
3103                         }
3104                         wfProfileOut( __METHOD__ . '-pfunc' );
3105                 }
3106
3107                 # Finish mangling title and then check for loops.
3108                 # Set $title to a Title object and $titleText to the PDBK
3109                 if ( !$found ) {
3110                         $ns = NS_TEMPLATE;
3111                         # Split the title into page and subpage
3112                         $subpage = '';
3113                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3114                         if ( $subpage !== '' ) {
3115                                 $ns = $this->mTitle->getNamespace();
3116                         }
3117                         $title = Title::newFromText( $part1, $ns );
3118                         if ( $title ) {
3119                                 $titleText = $title->getPrefixedText();
3120                                 # Check for language variants if the template is not found
3121                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3122                                         $wgContLang->findVariantLink( $part1, $title, true );
3123                                 }
3124                                 # Do recursion depth check
3125                                 $limit = $this->mOptions->getMaxTemplateDepth();
3126                                 if ( $frame->depth >= $limit ) {
3127                                         $found = true;
3128                                         $text = '<span class="error">'
3129                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3130                                                 . '</span>';
3131                                 }
3132                         }
3133                 }
3134
3135                 # Load from database
3136                 if ( !$found && $title ) {
3137                         wfProfileIn( __METHOD__ . '-loadtpl' );
3138                         if ( !$title->isExternal() ) {
3139                                 if ( $title->getNamespace() == NS_SPECIAL
3140                                         && $this->mOptions->getAllowSpecialInclusion()
3141                                         && $this->ot['html'] )
3142                                 {
3143                                         $text = SpecialPage::capturePath( $title );
3144                                         if ( is_string( $text ) ) {
3145                                                 $found = true;
3146                                                 $isHTML = true;
3147                                                 $this->disableCache();
3148                                         }
3149                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3150                                         $found = false; # access denied
3151                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3152                                 } else {
3153                                         list( $text, $title ) = $this->getTemplateDom( $title );
3154                                         if ( $text !== false ) {
3155                                                 $found = true;
3156                                                 $isChildObj = true;
3157                                         }
3158                                 }
3159
3160                                 # If the title is valid but undisplayable, make a link to it
3161                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3162                                         $text = "[[:$titleText]]";
3163                                         $found = true;
3164                                 }
3165                         } elseif ( $title->isTrans() ) {
3166                                 # Interwiki transclusion
3167                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3168                                         $text = $this->interwikiTransclude( $title, 'render' );
3169                                         $isHTML = true;
3170                                 } else {
3171                                         $text = $this->interwikiTransclude( $title, 'raw' );
3172                                         # Preprocess it like a template
3173                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3174                                         $isChildObj = true;
3175                                 }
3176                                 $found = true;
3177                         }
3178
3179                         # Do infinite loop check
3180                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3181                         if ( !$frame->loopCheck( $title ) ) {
3182                                 $found = true;
3183                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3184                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3185                         }
3186                         wfProfileOut( __METHOD__ . '-loadtpl' );
3187                 }
3188
3189                 # If we haven't found text to substitute by now, we're done
3190                 # Recover the source wikitext and return it
3191                 if ( !$found ) {
3192                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3193                         wfProfileOut( __METHOD__ );
3194                         return array( 'object' => $text );
3195                 }
3196
3197                 # Expand DOM-style return values in a child frame
3198                 if ( $isChildObj ) {
3199                         # Clean up argument array
3200                         $newFrame = $frame->newChild( $args, $title );
3201
3202                         if ( $nowiki ) {
3203                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3204                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3205                                 # Expansion is eligible for the empty-frame cache
3206                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3207                                         $text = $this->mTplExpandCache[$titleText];
3208                                 } else {
3209                                         $text = $newFrame->expand( $text );
3210                                         $this->mTplExpandCache[$titleText] = $text;
3211                                 }
3212                         } else {
3213                                 # Uncached expansion
3214                                 $text = $newFrame->expand( $text );
3215                         }
3216                 }
3217                 if ( $isLocalObj && $nowiki ) {
3218                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3219                         $isLocalObj = false;
3220                 }
3221
3222                 # Replace raw HTML by a placeholder
3223                 # Add a blank line preceding, to prevent it from mucking up
3224                 # immediately preceding headings
3225                 if ( $isHTML ) {
3226                         $text = "\n\n" . $this->insertStripItem( $text );
3227                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3228                         # Escape nowiki-style return values
3229                         $text = wfEscapeWikiText( $text );
3230                 } elseif ( is_string( $text )
3231                         && !$piece['lineStart']
3232                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3233                 {
3234                         # Bug 529: if the template begins with a table or block-level
3235                         # element, it should be treated as beginning a new line.
3236                         # This behaviour is somewhat controversial.
3237                         $text = "\n" . $text;
3238                 }
3239
3240                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3241                         # Error, oversize inclusion
3242                         if ( $titleText !== false ) {
3243                                 # Make a working, properly escaped link if possible (bug 23588)
3244                                 $text = "[[:$titleText]]";
3245                         } else {
3246                                 # This will probably not be a working link, but at least it may
3247                                 # provide some hint of where the problem is
3248                                 preg_replace( '/^:/', '', $originalTitle );
3249                                 $text = "[[:$originalTitle]]";
3250                         }
3251                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3252                         $this->limitationWarn( 'post-expand-template-inclusion' );
3253                 }
3254
3255                 if ( $isLocalObj ) {
3256                         $ret = array( 'object' => $text );
3257                 } else {
3258                         $ret = array( 'text' => $text );
3259                 }
3260
3261                 wfProfileOut( __METHOD__ );
3262                 return $ret;
3263         }
3264
3265         /**
3266          * Get the semi-parsed DOM representation of a template with a given title,
3267          * and its redirect destination title. Cached.
3268          */
3269         function getTemplateDom( $title ) {
3270                 $cacheTitle = $title;
3271                 $titleText = $title->getPrefixedDBkey();
3272
3273                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3274                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3275                         $title = Title::makeTitle( $ns, $dbk );
3276                         $titleText = $title->getPrefixedDBkey();
3277                 }
3278                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3279                         return array( $this->mTplDomCache[$titleText], $title );
3280                 }
3281
3282                 # Cache miss, go to the database
3283                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3284
3285                 if ( $text === false ) {
3286                         $this->mTplDomCache[$titleText] = false;
3287                         return array( false, $title );
3288                 }
3289
3290                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3291                 $this->mTplDomCache[ $titleText ] = $dom;
3292
3293                 if ( !$title->equals( $cacheTitle ) ) {
3294                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3295                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3296                 }
3297
3298                 return array( $dom, $title );
3299         }
3300
3301         /**
3302          * Fetch the unparsed text of a template and register a reference to it.
3303          */
3304         function fetchTemplateAndTitle( $title ) {
3305                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3306                 $stuff = call_user_func( $templateCb, $title, $this );
3307                 $text = $stuff['text'];
3308                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3309                 if ( isset( $stuff['deps'] ) ) {
3310                         foreach ( $stuff['deps'] as $dep ) {
3311                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3312                         }
3313                 }
3314                 return array( $text, $finalTitle );
3315         }
3316
3317         function fetchTemplate( $title ) {
3318                 $rv = $this->fetchTemplateAndTitle( $title );
3319                 return $rv[0];
3320         }
3321
3322         /**
3323          * Static function to get a template
3324          * Can be overridden via ParserOptions::setTemplateCallback().
3325          */
3326         static function statelessFetchTemplate( $title, $parser=false ) {
3327                 $text = $skip = false;
3328                 $finalTitle = $title;
3329                 $deps = array();
3330
3331                 # Loop to fetch the article, with up to 1 redirect
3332                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3333                         # Give extensions a chance to select the revision instead
3334                         $id = false; # Assume current
3335                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3336
3337                         if ( $skip ) {
3338                                 $text = false;
3339                                 $deps[] = array(
3340                                         'title' => $title,
3341                                         'page_id' => $title->getArticleID(),
3342                                         'rev_id' => null );
3343                                 break;
3344                         }
3345                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3346                         $rev_id = $rev ? $rev->getId() : 0;
3347                         # If there is no current revision, there is no page
3348                         if ( $id === false && !$rev ) {
3349                                 $linkCache = LinkCache::singleton();
3350                                 $linkCache->addBadLinkObj( $title );
3351                         }
3352
3353                         $deps[] = array(
3354                                 'title' => $title,
3355                                 'page_id' => $title->getArticleID(),
3356                                 'rev_id' => $rev_id );
3357
3358                         if ( $rev ) {
3359                                 $text = $rev->getText();
3360                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3361                                 global $wgContLang;
3362                                 $message = $wgContLang->lcfirst( $title->getText() );
3363                                 $text = wfMsgForContentNoTrans( $message );
3364                                 if ( wfEmptyMsg( $message, $text ) ) {
3365                                         $text = false;
3366                                         break;
3367                                 }
3368                         } else {
3369                                 break;
3370                         }
3371                         if ( $text === false ) {
3372                                 break;
3373                         }
3374                         # Redirect?
3375                         $finalTitle = $title;
3376                         $title = Title::newFromRedirect( $text );
3377                 }
3378                 return array(
3379                         'text' => $text,
3380                         'finalTitle' => $finalTitle,
3381                         'deps' => $deps );
3382         }
3383
3384         /**
3385          * Transclude an interwiki link.
3386          */
3387         function interwikiTransclude( $title, $action ) {
3388                 global $wgEnableScaryTranscluding;
3389
3390                 if ( !$wgEnableScaryTranscluding ) {
3391                         return wfMsgForContent('scarytranscludedisabled');
3392                 }
3393
3394                 $url = $title->getFullUrl( "action=$action" );
3395
3396                 if ( strlen( $url ) > 255 ) {
3397                         return wfMsgForContent( 'scarytranscludetoolong' );
3398                 }
3399                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3400         }
3401
3402         function fetchScaryTemplateMaybeFromCache( $url ) {
3403                 global $wgTranscludeCacheExpiry;
3404                 $dbr = wfGetDB( DB_SLAVE );
3405                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3406                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3407                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3408                 if ( $obj ) {
3409                         return $obj->tc_contents;
3410                 }
3411
3412                 $text = Http::get( $url );
3413                 if ( !$text ) {
3414                         return wfMsgForContent( 'scarytranscludefailed', $url );
3415                 }
3416
3417                 $dbw = wfGetDB( DB_MASTER );
3418                 $dbw->replace( 'transcache', array('tc_url'), array(
3419                         'tc_url' => $url,
3420                         'tc_time' => $dbw->timestamp( time() ),
3421                         'tc_contents' => $text)
3422                 );
3423                 return $text;
3424         }
3425
3426
3427         /**
3428          * Triple brace replacement -- used for template arguments
3429          * @private
3430          */
3431         function argSubstitution( $piece, $frame ) {
3432                 wfProfileIn( __METHOD__ );
3433
3434                 $error = false;
3435                 $parts = $piece['parts'];
3436                 $nameWithSpaces = $frame->expand( $piece['title'] );
3437                 $argName = trim( $nameWithSpaces );
3438                 $object = false;
3439                 $text = $frame->getArgument( $argName );
3440                 if (  $text === false && $parts->getLength() > 0
3441                   && (
3442                     $this->ot['html']
3443                     || $this->ot['pre']
3444                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3445                   )
3446                 ) {
3447                         # No match in frame, use the supplied default
3448                         $object = $parts->item( 0 )->getChildren();
3449                 }
3450                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3451                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3452                         $this->limitationWarn( 'post-expand-template-argument' );
3453                 }
3454
3455                 if ( $text === false && $object === false ) {
3456                         # No match anywhere
3457                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3458                 }
3459                 if ( $error !== false ) {
3460                         $text .= $error;
3461                 }
3462                 if ( $object !== false ) {
3463                         $ret = array( 'object' => $object );
3464                 } else {
3465                         $ret = array( 'text' => $text );
3466                 }
3467
3468                 wfProfileOut( __METHOD__ );
3469                 return $ret;
3470         }
3471
3472         /**
3473          * Return the text to be used for a given extension tag.
3474          * This is the ghost of strip().
3475          *
3476          * @param $params Associative array of parameters:
3477          *     name       PPNode for the tag name
3478          *     attr       PPNode for unparsed text where tag attributes are thought to be
3479          *     attributes Optional associative array of parsed attributes
3480          *     inner      Contents of extension element
3481          *     noClose    Original text did not have a close tag
3482          * @param $frame PPFrame
3483          */
3484         function extensionSubstitution( $params, $frame ) {
3485                 $name = $frame->expand( $params['name'] );
3486                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3487                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3488                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3489
3490                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3491                         ( $this->ot['html'] || $this->ot['pre'] );
3492                 if ( $isFunctionTag ) {
3493                         $markerType = 'none';
3494                 } else {
3495                         $markerType = 'general';
3496                 }
3497                 if ( $this->ot['html'] || $isFunctionTag ) {
3498                         $name = strtolower( $name );
3499                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3500                         if ( isset( $params['attributes'] ) ) {
3501                                 $attributes = $attributes + $params['attributes'];
3502                         }
3503
3504                         if ( isset( $this->mTagHooks[$name] ) ) {
3505                                 # Workaround for PHP bug 35229 and similar
3506                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3507                                         throw new MWException( "Tag hook for $name is not callable\n" );
3508                                 }
3509                                 $output = call_user_func_array( $this->mTagHooks[$name],
3510                                         array( $content, $attributes, $this, $frame ) );
3511                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3512                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3513                                 if ( !is_callable( $callback ) ) {
3514                                         throw new MWException( "Tag hook for $name is not callable\n" );
3515                                 }
3516
3517                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3518                         } else {
3519                                 $output = '<span class="error">Invalid tag extension name: ' .
3520                                         htmlspecialchars( $name ) . '</span>';
3521                         }
3522
3523                         if ( is_array( $output ) ) {
3524                                 # Extract flags to local scope (to override $markerType)
3525                                 $flags = $output;
3526                                 $output = $flags[0];
3527                                 unset( $flags[0] );
3528                                 extract( $flags );
3529                         }
3530                 } else {
3531                         if ( is_null( $attrText ) ) {
3532                                 $attrText = '';
3533                         }
3534                         if ( isset( $params['attributes'] ) ) {
3535                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3536                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3537                                                 htmlspecialchars( $attrValue ) . '"';
3538                                 }
3539                         }
3540                         if ( $content === null ) {
3541                                 $output = "<$name$attrText/>";
3542                         } else {
3543                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3544                                 $output = "<$name$attrText>$content$close";
3545                         }
3546                 }
3547
3548                 if ( $markerType === 'none' ) {
3549                         return $output;
3550                 } elseif ( $markerType === 'nowiki' ) {
3551                         $this->mStripState->nowiki->setPair( $marker, $output );
3552                 } elseif ( $markerType === 'general' ) {
3553                         $this->mStripState->general->setPair( $marker, $output );
3554                 } else {
3555                         throw new MWException( __METHOD__.': invalid marker type' );
3556                 }
3557                 return $marker;
3558         }
3559
3560         /**
3561          * Increment an include size counter
3562          *
3563          * @param $type String: the type of expansion
3564          * @param $size Integer: the size of the text
3565          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3566          */
3567         function incrementIncludeSize( $type, $size ) {
3568                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3569                         return false;
3570                 } else {
3571                         $this->mIncludeSizes[$type] += $size;
3572                         return true;
3573                 }
3574         }
3575
3576         /**
3577          * Increment the expensive function count
3578          *
3579          * @return Boolean: false if the limit has been exceeded
3580          */
3581         function incrementExpensiveFunctionCount() {
3582                 global $wgExpensiveParserFunctionLimit;
3583                 $this->mExpensiveFunctionCount++;
3584                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3585                         return true;
3586                 }
3587                 return false;
3588         }
3589
3590         /**
3591          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3592          * Fills $this->mDoubleUnderscores, returns the modified text
3593          */
3594         function doDoubleUnderscore( $text ) {
3595                 wfProfileIn( __METHOD__ );
3596
3597                 # The position of __TOC__ needs to be recorded
3598                 $mw = MagicWord::get( 'toc' );
3599                 if ( $mw->match( $text ) ) {
3600                         $this->mShowToc = true;
3601                         $this->mForceTocPosition = true;
3602
3603                         # Set a placeholder. At the end we'll fill it in with the TOC.
3604                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3605
3606                         # Only keep the first one.
3607                         $text = $mw->replace( '', $text );
3608                 }
3609
3610                 # Now match and remove the rest of them
3611                 $mwa = MagicWord::getDoubleUnderscoreArray();
3612                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3613
3614                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3615                         $this->mOutput->mNoGallery = true;
3616                 }
3617                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3618                         $this->mShowToc = false;
3619                 }
3620                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3621                         $this->addTrackingCategory( 'hidden-category-category' );
3622                 }
3623                 # (bug 8068) Allow control over whether robots index a page.
3624                 #
3625                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3626                 # is not desirable, the last one on the page should win.
3627                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3628                         $this->mOutput->setIndexPolicy( 'noindex' );
3629                         $this->addTrackingCategory( 'noindex-category' );
3630                 }
3631                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3632                         $this->mOutput->setIndexPolicy( 'index' );
3633                         $this->addTrackingCategory( 'index-category' );
3634                 }
3635
3636                 # Cache all double underscores in the database
3637                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3638                         $this->mOutput->setProperty( $key, '' );
3639                 }
3640
3641                 wfProfileOut( __METHOD__ );
3642                 return $text;
3643         }
3644
3645         /**
3646          * Add a tracking category, getting the title from a system message,
3647          * or print a debug message if the title is invalid.
3648          *
3649          * @param $msg String: message key
3650          * @return Boolean: whether the addition was successful
3651          */
3652         protected function addTrackingCategory( $msg ) {
3653                 $cat = wfMsgForContent( $msg );
3654
3655                 # Allow tracking categories to be disabled by setting them to "-"
3656                 if ( $cat === '-' ) {
3657                         return false;
3658                 }
3659
3660                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3661                 if ( $containerCategory ) {
3662                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3663                         return true;
3664                 } else {
3665                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3666                         return false;
3667                 }
3668         }
3669
3670         /**
3671          * This function accomplishes several tasks:
3672          * 1) Auto-number headings if that option is enabled
3673          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3674          * 3) Add a Table of contents on the top for users who have enabled the option
3675          * 4) Auto-anchor headings
3676          *
3677          * It loops through all headlines, collects the necessary data, then splits up the
3678          * string and re-inserts the newly formatted headlines.
3679          *
3680          * @param $text String
3681          * @param $origText String: original, untouched wikitext
3682          * @param $isMain Boolean
3683          * @private
3684          */
3685         function formatHeadings( $text, $origText, $isMain=true ) {
3686                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3687
3688                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3689
3690                 # Inhibit editsection links if requested in the page
3691                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3692                         $showEditLink = 0;
3693                 } else {
3694                         $showEditLink = $this->mOptions->getEditSection();
3695                 }
3696
3697                 # Get all headlines for numbering them and adding funky stuff like [edit]
3698                 # links - this is for later, but we need the number of headlines right now
3699                 $matches = array();
3700                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3701
3702                 # if there are fewer than 4 headlines in the article, do not show TOC
3703                 # unless it's been explicitly enabled.
3704                 $enoughToc = $this->mShowToc &&
3705                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3706
3707                 # Allow user to stipulate that a page should have a "new section"
3708                 # link added via __NEWSECTIONLINK__
3709                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3710                         $this->mOutput->setNewSection( true );
3711                 }
3712
3713                 # Allow user to remove the "new section"
3714                 # link via __NONEWSECTIONLINK__
3715                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3716                         $this->mOutput->hideNewSection( true );
3717                 }
3718
3719                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3720                 # override above conditions and always show TOC above first header
3721                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3722                         $this->mShowToc = true;
3723                         $enoughToc = true;
3724                 }
3725
3726                 # We need this to perform operations on the HTML
3727                 $sk = $this->mOptions->getSkin( $this->mTitle );
3728
3729                 # headline counter
3730                 $headlineCount = 0;
3731                 $numVisible = 0;
3732
3733                 # Ugh .. the TOC should have neat indentation levels which can be
3734                 # passed to the skin functions. These are determined here
3735                 $toc = '';
3736                 $full = '';
3737                 $head = array();
3738                 $sublevelCount = array();
3739                 $levelCount = array();
3740                 $level = 0;
3741                 $prevlevel = 0;
3742                 $toclevel = 0;
3743                 $prevtoclevel = 0;
3744                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3745                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3746                 $oldType = $this->mOutputType;
3747                 $this->setOutputType( self::OT_WIKI );
3748                 $frame = $this->getPreprocessor()->newFrame();
3749                 $root = $this->preprocessToDom( $origText );
3750                 $node = $root->getFirstChild();
3751                 $byteOffset = 0;
3752                 $tocraw = array();
3753                 $refers = array();
3754
3755                 foreach ( $matches[3] as $headline ) {
3756                         $isTemplate = false;
3757                         $titleText = false;
3758                         $sectionIndex = false;
3759                         $numbering = '';
3760                         $markerMatches = array();
3761                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3762                                 $serial = $markerMatches[1];
3763                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3764                                 $isTemplate = ( $titleText != $baseTitleText );
3765                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3766                         }
3767
3768                         if ( $toclevel ) {
3769                                 $prevlevel = $level;
3770                         }
3771                         $level = $matches[1][$headlineCount];
3772
3773                         if ( $level > $prevlevel ) {
3774                                 # Increase TOC level
3775                                 $toclevel++;
3776                                 $sublevelCount[$toclevel] = 0;
3777                                 if ( $toclevel<$wgMaxTocLevel ) {
3778                                         $prevtoclevel = $toclevel;
3779                                         $toc .= $sk->tocIndent();
3780                                         $numVisible++;
3781                                 }
3782                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3783                                 # Decrease TOC level, find level to jump to
3784
3785                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3786                                         if ( $levelCount[$i] == $level ) {
3787                                                 # Found last matching level
3788                                                 $toclevel = $i;
3789                                                 break;
3790                                         } elseif ( $levelCount[$i] < $level ) {
3791                                                 # Found first matching level below current level
3792                                                 $toclevel = $i + 1;
3793                                                 break;
3794                                         }
3795                                 }
3796                                 if ( $i == 0 ) {
3797                                         $toclevel = 1;
3798                                 }
3799                                 if ( $toclevel<$wgMaxTocLevel ) {
3800                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3801                                                 # Unindent only if the previous toc level was shown :p
3802                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3803                                                 $prevtoclevel = $toclevel;
3804                                         } else {
3805                                                 $toc .= $sk->tocLineEnd();
3806                                         }
3807                                 }
3808                         } else {
3809                                 # No change in level, end TOC line
3810                                 if ( $toclevel<$wgMaxTocLevel ) {
3811                                         $toc .= $sk->tocLineEnd();
3812                                 }
3813                         }
3814
3815                         $levelCount[$toclevel] = $level;
3816
3817                         # count number of headlines for each level
3818                         @$sublevelCount[$toclevel]++;
3819                         $dot = 0;
3820                         for( $i = 1; $i <= $toclevel; $i++ ) {
3821                                 if ( !empty( $sublevelCount[$i] ) ) {
3822                                         if ( $dot ) {
3823                                                 $numbering .= '.';
3824                                         }
3825                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3826                                         $dot = 1;
3827                                 }
3828                         }
3829
3830                         # The safe header is a version of the header text safe to use for links
3831                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3832                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3833
3834                         # Remove link placeholders by the link text.
3835                         #     <!--LINK number-->
3836                         # turns into
3837                         #     link text with suffix
3838                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3839
3840                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3841                         $tocline = preg_replace(
3842                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3843                                 array( '',                          '<$1>' ),
3844                                 $safeHeadline
3845                         );
3846                         $tocline = trim( $tocline );
3847
3848                         # For the anchor, strip out HTML-y stuff period
3849                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3850                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
3851
3852                         # Save headline for section edit hint before it's escaped
3853                         $headlineHint = $safeHeadline;
3854
3855                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3856                                 # For reverse compatibility, provide an id that's
3857                                 # HTML4-compatible, like we used to.
3858                                 #
3859                                 # It may be worth noting, academically, that it's possible for
3860                                 # the legacy anchor to conflict with a non-legacy headline
3861                                 # anchor on the page.  In this case likely the "correct" thing
3862                                 # would be to either drop the legacy anchors or make sure
3863                                 # they're numbered first.  However, this would require people
3864                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3865                                 # manually, so let's not bother worrying about it.
3866                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3867                                         array( 'noninitial', 'legacy' ) );
3868                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3869
3870                                 if ( $legacyHeadline == $safeHeadline ) {
3871                                         # No reason to have both (in fact, we can't)
3872                                         $legacyHeadline = false;
3873                                 }
3874                         } else {
3875                                 $legacyHeadline = false;
3876                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3877                                         'noninitial' );
3878                         }
3879
3880                         # HTML names must be case-insensitively unique (bug 10721).
3881                         # This does not apply to Unicode characters per
3882                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
3883                         # FIXME: We may be changing them depending on the current locale.
3884                         $arrayKey = strtolower( $safeHeadline );
3885                         if ( $legacyHeadline === false ) {
3886                                 $legacyArrayKey = false;
3887                         } else {
3888                                 $legacyArrayKey = strtolower( $legacyHeadline );
3889                         }
3890
3891                         # count how many in assoc. array so we can track dupes in anchors
3892                         if ( isset( $refers[$arrayKey] ) ) {
3893                                 $refers[$arrayKey]++;
3894                         } else {
3895                                 $refers[$arrayKey] = 1;
3896                         }
3897                         if ( isset( $refers[$legacyArrayKey] ) ) {
3898                                 $refers[$legacyArrayKey]++;
3899                         } else {
3900                                 $refers[$legacyArrayKey] = 1;
3901                         }
3902
3903                         # Don't number the heading if it is the only one (looks silly)
3904                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3905                                 # the two are different if the line contains a link
3906                                 $headline = $numbering . ' ' . $headline;
3907                         }
3908
3909                         # Create the anchor for linking from the TOC to the section
3910                         $anchor = $safeHeadline;
3911                         $legacyAnchor = $legacyHeadline;
3912                         if ( $refers[$arrayKey] > 1 ) {
3913                                 $anchor .= '_' . $refers[$arrayKey];
3914                         }
3915                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3916                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3917                         }
3918                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3919                                 $toc .= $sk->tocLine( $anchor, $tocline,
3920                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3921                         }
3922
3923                         # Add the section to the section tree
3924                         # Find the DOM node for this header
3925                         while ( $node && !$isTemplate ) {
3926                                 if ( $node->getName() === 'h' ) {
3927                                         $bits = $node->splitHeading();
3928                                         if ( $bits['i'] == $sectionIndex ) {
3929                                                 break;
3930                                         }
3931                                 }
3932                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3933                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3934                                 $node = $node->getNextSibling();
3935                         }
3936                         $tocraw[] = array(
3937                                 'toclevel' => $toclevel,
3938                                 'level' => $level,
3939                                 'line' => $tocline,
3940                                 'number' => $numbering,
3941                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3942                                 'fromtitle' => $titleText,
3943                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3944                                 'anchor' => $anchor,
3945                         );
3946
3947                         # give headline the correct <h#> tag
3948                         if ( $showEditLink && $sectionIndex !== false ) {
3949                                 if ( $isTemplate ) {
3950                                         # Put a T flag in the section identifier, to indicate to extractSections()
3951                                         # that sections inside <includeonly> should be counted.
3952                                         $editlink = $sk->doEditSectionLink( Title::newFromText( $titleText ), "T-$sectionIndex", null, $this->mOptions->getUserLang() );
3953                                 } else {
3954                                         $editlink = $sk->doEditSectionLink( $this->mTitle, $sectionIndex, $headlineHint, $this->mOptions->getUserLang() );
3955                                 }
3956                         } else {
3957                                 $editlink = '';
3958                         }
3959                         $head[$headlineCount] = $sk->makeHeadline( $level,
3960                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3961                                 $editlink, $legacyAnchor );
3962
3963                         $headlineCount++;
3964                 }
3965
3966                 $this->setOutputType( $oldType );
3967
3968                 # Never ever show TOC if no headers
3969                 if ( $numVisible < 1 ) {
3970                         $enoughToc = false;
3971                 }
3972
3973                 if ( $enoughToc ) {
3974                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3975                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3976                         }
3977                         $toc = $sk->tocList( $toc );
3978                         $this->mOutput->setTOCHTML( $toc );
3979                 }
3980
3981                 if ( $isMain ) {
3982                         $this->mOutput->setSections( $tocraw );
3983                 }
3984
3985                 # split up and insert constructed headlines
3986
3987                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3988                 $i = 0;
3989
3990                 foreach ( $blocks as $block ) {
3991                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3992                                 # This is the [edit] link that appears for the top block of text when
3993                                 # section editing is enabled
3994
3995                                 # Disabled because it broke block formatting
3996                                 # For example, a bullet point in the top line
3997                                 # $full .= $sk->editSectionLink(0);
3998                         }
3999                         $full .= $block;
4000                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4001                                 # Top anchor now in skin
4002                                 $full = $full.$toc;
4003                         }
4004
4005                         if ( !empty( $head[$i] ) ) {
4006                                 $full .= $head[$i];
4007                         }
4008                         $i++;
4009                 }
4010                 if ( $this->mForceTocPosition ) {
4011                         return str_replace( '<!--MWTOC-->', $toc, $full );
4012                 } else {
4013                         return $full;
4014                 }
4015         }
4016
4017         /**
4018          * Transform wiki markup when saving a page by doing \r\n -> \n
4019          * conversion, substitting signatures, {{subst:}} templates, etc.
4020          *
4021          * @param $text String: the text to transform
4022          * @param $title Title: the Title object for the current article
4023          * @param $user User: the User object describing the current user
4024          * @param $options ParserOptions: parsing options
4025          * @param $clearState Boolean: whether to clear the parser state first
4026          * @return String: the altered wiki markup
4027          */
4028         public function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
4029                 $options->resetUsage();
4030                 $this->mOptions = $options;
4031                 $this->setTitle( $title );
4032                 $this->setUser( $user );
4033                 $this->setOutputType( self::OT_WIKI );
4034
4035                 if ( $clearState ) {
4036                         $this->clearState();
4037                 }
4038
4039                 $pairs = array(
4040                         "\r\n" => "\n",
4041                 );
4042                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4043                 $text = $this->pstPass2( $text, $user );
4044                 $text = $this->mStripState->unstripBoth( $text );
4045
4046                 $this->setUser( null ); #Reset
4047
4048                 return $text;
4049         }
4050
4051         /**
4052          * Pre-save transform helper function
4053          * @private
4054          */
4055         function pstPass2( $text, $user ) {
4056                 global $wgContLang, $wgLocaltimezone;
4057
4058                 # Note: This is the timestamp saved as hardcoded wikitext to
4059                 # the database, we use $wgContLang here in order to give
4060                 # everyone the same signature and use the default one rather
4061                 # than the one selected in each user's preferences.
4062                 # (see also bug 12815)
4063                 $ts = $this->mOptions->getTimestamp();
4064                 if ( isset( $wgLocaltimezone ) ) {
4065                         $tz = $wgLocaltimezone;
4066                 } else {
4067                         $tz = date_default_timezone_get();
4068                 }
4069
4070                 $unixts = wfTimestamp( TS_UNIX, $ts );
4071                 $oldtz = date_default_timezone_get();
4072                 date_default_timezone_set( $tz );
4073                 $ts = date( 'YmdHis', $unixts );
4074                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4075
4076                 # Allow translation of timezones through wiki. date() can return
4077                 # whatever crap the system uses, localised or not, so we cannot
4078                 # ship premade translations.
4079                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4080                 $value = wfMsgForContent( $key );
4081                 if ( !wfEmptyMsg( $key, $value ) ) {
4082                         $tzMsg = $value;
4083                 }
4084
4085                 date_default_timezone_set( $oldtz );
4086
4087                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4088
4089                 # Variable replacement
4090                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4091                 $text = $this->replaceVariables( $text );
4092
4093                 # Signatures
4094                 $sigText = $this->getUserSig( $user );
4095                 $text = strtr( $text, array(
4096                         '~~~~~' => $d,
4097                         '~~~~' => "$sigText $d",
4098                         '~~~' => $sigText
4099                 ) );
4100
4101                 # Context links: [[|name]] and [[name (context)|]]
4102                 global $wgLegalTitleChars;
4103                 $tc = "[$wgLegalTitleChars]";
4104                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4105
4106                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4107                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4108                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4109                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4110
4111                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4112                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4113                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4114                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4115
4116                 $t = $this->mTitle->getText();
4117                 $m = array();
4118                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4119                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4120                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4121                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4122                 } else {
4123                         # if there's no context, don't bother duplicating the title
4124                         $text = preg_replace( $p2, '[[\\1]]', $text );
4125                 }
4126
4127                 # Trim trailing whitespace
4128                 $text = rtrim( $text );
4129
4130                 return $text;
4131         }
4132
4133         /**
4134          * Fetch the user's signature text, if any, and normalize to
4135          * validated, ready-to-insert wikitext.
4136          * If you have pre-fetched the nickname or the fancySig option, you can
4137          * specify them here to save a database query.
4138          *
4139          * @param $user User
4140          * @param $nickname String: nickname to use or false to use user's default nickname
4141          * @param $fancySig Boolean: whether the nicknname is the complete signature
4142          *                  or null to use default value
4143          * @return string
4144          */
4145         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4146                 global $wgMaxSigChars;
4147
4148                 $username = $user->getName();
4149
4150                 # If not given, retrieve from the user object.
4151                 if ( $nickname === false )
4152                         $nickname = $user->getOption( 'nickname' );
4153
4154                 if ( is_null( $fancySig ) ) {
4155                         $fancySig = $user->getBoolOption( 'fancysig' );
4156                 }
4157
4158                 $nickname = $nickname == null ? $username : $nickname;
4159
4160                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4161                         $nickname = $username;
4162                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4163                 } elseif ( $fancySig !== false ) {
4164                         # Sig. might contain markup; validate this
4165                         if ( $this->validateSig( $nickname ) !== false ) {
4166                                 # Validated; clean up (if needed) and return it
4167                                 return $this->cleanSig( $nickname, true );
4168                         } else {
4169                                 # Failed to validate; fall back to the default
4170                                 $nickname = $username;
4171                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4172                         }
4173                 }
4174
4175                 # Make sure nickname doesnt get a sig in a sig
4176                 $nickname = $this->cleanSigInSig( $nickname );
4177
4178                 # If we're still here, make it a link to the user page
4179                 $userText = wfEscapeWikiText( $username );
4180                 $nickText = wfEscapeWikiText( $nickname );
4181                 if ( $user->isAnon() )  {
4182                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4183                 } else {
4184                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4185                 }
4186         }
4187
4188         /**
4189          * Check that the user's signature contains no bad XML
4190          *
4191          * @param $text String
4192          * @return mixed An expanded string, or false if invalid.
4193          */
4194         function validateSig( $text ) {
4195                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4196         }
4197
4198         /**
4199          * Clean up signature text
4200          *
4201          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4202          * 2) Substitute all transclusions
4203          *
4204          * @param $text String
4205          * @param $parsing Whether we're cleaning (preferences save) or parsing
4206          * @return String: signature text
4207          */
4208         function cleanSig( $text, $parsing = false ) {
4209                 if ( !$parsing ) {
4210                         global $wgTitle;
4211                         $this->clearState();
4212                         $this->setTitle( $wgTitle );
4213                         $this->mOptions = new ParserOptions;
4214                         $this->setOutputType = self::OT_PREPROCESS;
4215                 }
4216
4217                 # Option to disable this feature
4218                 if ( !$this->mOptions->getCleanSignatures() ) {
4219                         return $text;
4220                 }
4221
4222                 # FIXME: regex doesn't respect extension tags or nowiki
4223                 #  => Move this logic to braceSubstitution()
4224                 $substWord = MagicWord::get( 'subst' );
4225                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4226                 $substText = '{{' . $substWord->getSynonym( 0 );
4227
4228                 $text = preg_replace( $substRegex, $substText, $text );
4229                 $text = $this->cleanSigInSig( $text );
4230                 $dom = $this->preprocessToDom( $text );
4231                 $frame = $this->getPreprocessor()->newFrame();
4232                 $text = $frame->expand( $dom );
4233
4234                 if ( !$parsing ) {
4235                         $text = $this->mStripState->unstripBoth( $text );
4236                 }
4237
4238                 return $text;
4239         }
4240
4241         /**
4242          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4243          *
4244          * @param $text String
4245          * @return String: signature text with /~{3,5}/ removed
4246          */
4247         function cleanSigInSig( $text ) {
4248                 $text = preg_replace( '/~{3,5}/', '', $text );
4249                 return $text;
4250         }
4251
4252         /**
4253          * Set up some variables which are usually set up in parse()
4254          * so that an external function can call some class members with confidence
4255          */
4256         public function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4257                 $this->setTitle( $title );
4258                 $options->resetUsage();
4259                 $this->mOptions = $options;
4260                 $this->setOutputType( $outputType );
4261                 if ( $clearState ) {
4262                         $this->clearState();
4263                 }
4264         }
4265
4266         /**
4267          * Wrapper for preprocess()
4268          *
4269          * @param $text String: the text to preprocess
4270          * @param $options ParserOptions: options
4271          * @return String
4272          */
4273         public function transformMsg( $text, $options ) {
4274                 global $wgTitle;
4275                 static $executing = false;
4276
4277                 # Guard against infinite recursion
4278                 if ( $executing ) {
4279                         return $text;
4280                 }
4281                 $executing = true;
4282
4283                 wfProfileIn( __METHOD__ );
4284                 $text = $this->preprocess( $text, $wgTitle, $options );
4285
4286                 $executing = false;
4287                 wfProfileOut( __METHOD__ );
4288                 return $text;
4289         }
4290
4291         /**
4292          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4293          * The callback should have the following form:
4294          *    function myParserHook( $text, $params, $parser ) { ... }
4295          *
4296          * Transform and return $text. Use $parser for any required context, e.g. use
4297          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4298          *
4299          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4300          * @param $callback Mixed: the callback function (and object) to use for the tag
4301          * @return The old value of the mTagHooks array associated with the hook
4302          */
4303         public function setHook( $tag, $callback ) {
4304                 $tag = strtolower( $tag );
4305                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4306                 $this->mTagHooks[$tag] = $callback;
4307                 if ( !in_array( $tag, $this->mStripList ) ) {
4308                         $this->mStripList[] = $tag;
4309                 }
4310
4311                 return $oldVal;
4312         }
4313
4314         function setTransparentTagHook( $tag, $callback ) {
4315                 $tag = strtolower( $tag );
4316                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4317                 $this->mTransparentTagHooks[$tag] = $callback;
4318
4319                 return $oldVal;
4320         }
4321
4322         /**
4323          * Remove all tag hooks
4324          */
4325         function clearTagHooks() {
4326                 $this->mTagHooks = array();
4327                 $this->mStripList = $this->mDefaultStripList;
4328         }
4329
4330         /**
4331          * Create a function, e.g. {{sum:1|2|3}}
4332          * The callback function should have the form:
4333          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4334          *
4335          * Or with SFH_OBJECT_ARGS:
4336          *    function myParserFunction( $parser, $frame, $args ) { ... }
4337          *
4338          * The callback may either return the text result of the function, or an array with the text
4339          * in element 0, and a number of flags in the other elements. The names of the flags are
4340          * specified in the keys. Valid flags are:
4341          *   found                     The text returned is valid, stop processing the template. This
4342          *                             is on by default.
4343          *   nowiki                    Wiki markup in the return value should be escaped
4344          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4345          *
4346          * @param $id String: The magic word ID
4347          * @param $callback Mixed: the callback function (and object) to use
4348          * @param $flags Integer: a combination of the following flags:
4349          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4350          *
4351          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4352          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4353          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4354          *     the arguments, and to control the way they are expanded.
4355          *
4356          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4357          *     arguments, for instance:
4358          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4359          *
4360          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4361          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4362          *     working if/when this is changed.
4363          *
4364          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4365          *     expansion.
4366          *
4367          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4368          *     about the methods available in PPFrame and PPNode.
4369          *
4370          * @return The old callback function for this name, if any
4371          */
4372         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4373                 global $wgContLang;
4374
4375                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4376                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4377
4378                 # Add to function cache
4379                 $mw = MagicWord::get( $id );
4380                 if ( !$mw )
4381                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4382
4383                 $synonyms = $mw->getSynonyms();
4384                 $sensitive = intval( $mw->isCaseSensitive() );
4385
4386                 foreach ( $synonyms as $syn ) {
4387                         # Case
4388                         if ( !$sensitive ) {
4389                                 $syn = $wgContLang->lc( $syn );
4390                         }
4391                         # Add leading hash
4392                         if ( !( $flags & SFH_NO_HASH ) ) {
4393                                 $syn = '#' . $syn;
4394                         }
4395                         # Remove trailing colon
4396                         if ( substr( $syn, -1, 1 ) === ':' ) {
4397                                 $syn = substr( $syn, 0, -1 );
4398                         }
4399                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4400                 }
4401                 return $oldVal;
4402         }
4403
4404         /**
4405          * Get all registered function hook identifiers
4406          *
4407          * @return Array
4408          */
4409         function getFunctionHooks() {
4410                 return array_keys( $this->mFunctionHooks );
4411         }
4412
4413         /**
4414          * Create a tag function, e.g. <test>some stuff</test>.
4415          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4416          * Unlike parser functions, their content is not preprocessed.
4417          */
4418         function setFunctionTagHook( $tag, $callback, $flags ) {
4419                 $tag = strtolower( $tag );
4420                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4421                         $this->mFunctionTagHooks[$tag] : null;
4422                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4423
4424                 if ( !in_array( $tag, $this->mStripList ) ) {
4425                         $this->mStripList[] = $tag;
4426                 }
4427
4428                 return $old;
4429         }
4430
4431         /**
4432          * FIXME: update documentation. makeLinkObj() is deprecated.
4433          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4434          * Placeholders created in Skin::makeLinkObj()
4435          * Returns an array of link CSS classes, indexed by PDBK.
4436          */
4437         function replaceLinkHolders( &$text, $options = 0 ) {
4438                 return $this->mLinkHolders->replace( $text );
4439         }
4440
4441         /**
4442          * Replace <!--LINK--> link placeholders with plain text of links
4443          * (not HTML-formatted).
4444          *
4445          * @param $text String
4446          * @return String
4447          */
4448         function replaceLinkHoldersText( $text ) {
4449                 return $this->mLinkHolders->replaceText( $text );
4450         }
4451
4452         /**
4453          * Renders an image gallery from a text with one line per image.
4454          * text labels may be given by using |-style alternative text. E.g.
4455          *   Image:one.jpg|The number "1"
4456          *   Image:tree.jpg|A tree
4457          * given as text will return the HTML of a gallery with two images,
4458          * labeled 'The number "1"' and
4459          * 'A tree'.
4460          */
4461         function renderImageGallery( $text, $params ) {
4462                 $ig = new ImageGallery();
4463                 $ig->setContextTitle( $this->mTitle );
4464                 $ig->setShowBytes( false );
4465                 $ig->setShowFilename( false );
4466                 $ig->setParser( $this );
4467                 $ig->setHideBadImages();
4468                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4469                 $ig->useSkin( $this->mOptions->getSkin( $this->mTitle ) );
4470                 $ig->mRevisionId = $this->mRevisionId;
4471
4472                 if ( isset( $params['showfilename'] ) ) {
4473                         $ig->setShowFilename( true );
4474                 } else {
4475                         $ig->setShowFilename( false );
4476                 }
4477                 if ( isset( $params['caption'] ) ) {
4478                         $caption = $params['caption'];
4479                         $caption = htmlspecialchars( $caption );
4480                         $caption = $this->replaceInternalLinks( $caption );
4481                         $ig->setCaptionHtml( $caption );
4482                 }
4483                 if ( isset( $params['perrow'] ) ) {
4484                         $ig->setPerRow( $params['perrow'] );
4485                 }
4486                 if ( isset( $params['widths'] ) ) {
4487                         $ig->setWidths( $params['widths'] );
4488                 }
4489                 if ( isset( $params['heights'] ) ) {
4490                         $ig->setHeights( $params['heights'] );
4491                 }
4492
4493                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4494
4495                 $lines = StringUtils::explode( "\n", $text );
4496                 foreach ( $lines as $line ) {
4497                         # match lines like these:
4498                         # Image:someimage.jpg|This is some image
4499                         $matches = array();
4500                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4501                         # Skip empty lines
4502                         if ( count( $matches ) == 0 ) {
4503                                 continue;
4504                         }
4505
4506                         if ( strpos( $matches[0], '%' ) !== false ) {
4507                                 $matches[1] = urldecode( $matches[1] );
4508                         }
4509                         $tp = Title::newFromText( $matches[1] );
4510                         $nt =& $tp;
4511                         if ( is_null( $nt ) ) {
4512                                 # Bogus title. Ignore these so we don't bomb out later.
4513                                 continue;
4514                         }
4515                         if ( isset( $matches[3] ) ) {
4516                                 $label = $matches[3];
4517                         } else {
4518                                 $label = '';
4519                         }
4520
4521                         $html = $this->recursiveTagParse( trim( $label ) );
4522
4523                         $ig->add( $nt, $html );
4524
4525                         # Only add real images (bug #5586)
4526                         if ( $nt->getNamespace() == NS_FILE ) {
4527                                 $this->mOutput->addImage( $nt->getDBkey() );
4528                         }
4529                 }
4530                 return $ig->toHTML();
4531         }
4532
4533         function getImageParams( $handler ) {
4534                 if ( $handler ) {
4535                         $handlerClass = get_class( $handler );
4536                 } else {
4537                         $handlerClass = '';
4538                 }
4539                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4540                         # Initialise static lists
4541                         static $internalParamNames = array(
4542                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4543                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4544                                         'bottom', 'text-bottom' ),
4545                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4546                                         'upright', 'border', 'link', 'alt' ),
4547                         );
4548                         static $internalParamMap;
4549                         if ( !$internalParamMap ) {
4550                                 $internalParamMap = array();
4551                                 foreach ( $internalParamNames as $type => $names ) {
4552                                         foreach ( $names as $name ) {
4553                                                 $magicName = str_replace( '-', '_', "img_$name" );
4554                                                 $internalParamMap[$magicName] = array( $type, $name );
4555                                         }
4556                                 }
4557                         }
4558
4559                         # Add handler params
4560                         $paramMap = $internalParamMap;
4561                         if ( $handler ) {
4562                                 $handlerParamMap = $handler->getParamMap();
4563                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4564                                         $paramMap[$magic] = array( 'handler', $paramName );
4565                                 }
4566                         }
4567                         $this->mImageParams[$handlerClass] = $paramMap;
4568                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4569                 }
4570                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4571         }
4572
4573         /**
4574          * Parse image options text and use it to make an image
4575          *
4576          * @param $title Title
4577          * @param $options String
4578          * @param $holders LinkHolderArray
4579          */
4580         function makeImage( $title, $options, $holders = false ) {
4581                 # Check if the options text is of the form "options|alt text"
4582                 # Options are:
4583                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4584                 #  * left       no resizing, just left align. label is used for alt= only
4585                 #  * right      same, but right aligned
4586                 #  * none       same, but not aligned
4587                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4588                 #  * center     center the image
4589                 #  * frame      Keep original image size, no magnify-button.
4590                 #  * framed     Same as "frame"
4591                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4592                 #  * upright    reduce width for upright images, rounded to full __0 px
4593                 #  * border     draw a 1px border around the image
4594                 #  * alt        Text for HTML alt attribute (defaults to empty)
4595                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4596                 # vertical-align values (no % or length right now):
4597                 #  * baseline
4598                 #  * sub
4599                 #  * super
4600                 #  * top
4601                 #  * text-top
4602                 #  * middle
4603                 #  * bottom
4604                 #  * text-bottom
4605
4606                 $parts = StringUtils::explode( "|", $options );
4607                 $sk = $this->mOptions->getSkin( $this->mTitle );
4608
4609                 # Give extensions a chance to select the file revision for us
4610                 $skip = $time = $descQuery = false;
4611                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4612
4613                 if ( $skip ) {
4614                         return $sk->link( $title );
4615                 }
4616
4617                 # Get the file
4618                 $file = wfFindFile( $title, array( 'time' => $time ) );
4619                 # Get parameter map
4620                 $handler = $file ? $file->getHandler() : false;
4621
4622                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4623
4624                 # Process the input parameters
4625                 $caption = '';
4626                 $params = array( 'frame' => array(), 'handler' => array(),
4627                         'horizAlign' => array(), 'vertAlign' => array() );
4628                 foreach ( $parts as $part ) {
4629                         $part = trim( $part );
4630                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4631                         $validated = false;
4632                         if ( isset( $paramMap[$magicName] ) ) {
4633                                 list( $type, $paramName ) = $paramMap[$magicName];
4634
4635                                 # Special case; width and height come in one variable together
4636                                 if ( $type === 'handler' && $paramName === 'width' ) {
4637                                         $m = array();
4638                                         # (bug 13500) In both cases (width/height and width only),
4639                                         # permit trailing "px" for backward compatibility.
4640                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4641                                                 $width = intval( $m[1] );
4642                                                 $height = intval( $m[2] );
4643                                                 if ( $handler->validateParam( 'width', $width ) ) {
4644                                                         $params[$type]['width'] = $width;
4645                                                         $validated = true;
4646                                                 }
4647                                                 if ( $handler->validateParam( 'height', $height ) ) {
4648                                                         $params[$type]['height'] = $height;
4649                                                         $validated = true;
4650                                                 }
4651                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4652                                                 $width = intval( $value );
4653                                                 if ( $handler->validateParam( 'width', $width ) ) {
4654                                                         $params[$type]['width'] = $width;
4655                                                         $validated = true;
4656                                                 }
4657                                         } # else no validation -- bug 13436
4658                                 } else {
4659                                         if ( $type === 'handler' ) {
4660                                                 # Validate handler parameter
4661                                                 $validated = $handler->validateParam( $paramName, $value );
4662                                         } else {
4663                                                 # Validate internal parameters
4664                                                 switch( $paramName ) {
4665                                                 case 'manualthumb':
4666                                                 case 'alt':
4667                                                         # @todo Fixme: possibly check validity here for
4668                                                         # manualthumb? downstream behavior seems odd with
4669                                                         # missing manual thumbs.
4670                                                         $validated = true;
4671                                                         $value = $this->stripAltText( $value, $holders );
4672                                                         break;
4673                                                 case 'link':
4674                                                         $chars = self::EXT_LINK_URL_CLASS;
4675                                                         $prots = $this->mUrlProtocols;
4676                                                         if ( $value === '' ) {
4677                                                                 $paramName = 'no-link';
4678                                                                 $value = true;
4679                                                                 $validated = true;
4680                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4681                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4682                                                                         $paramName = 'link-url';
4683                                                                         $this->mOutput->addExternalLink( $value );
4684                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4685                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4686                                                                         }
4687                                                                         $validated = true;
4688                                                                 }
4689                                                         } else {
4690                                                                 $linkTitle = Title::newFromText( $value );
4691                                                                 if ( $linkTitle ) {
4692                                                                         $paramName = 'link-title';
4693                                                                         $value = $linkTitle;
4694                                                                         $this->mOutput->addLink( $linkTitle );
4695                                                                         $validated = true;
4696                                                                 }
4697                                                         }
4698                                                         break;
4699                                                 default:
4700                                                         # Most other things appear to be empty or numeric...
4701                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4702                                                 }
4703                                         }
4704
4705                                         if ( $validated ) {
4706                                                 $params[$type][$paramName] = $value;
4707                                         }
4708                                 }
4709                         }
4710                         if ( !$validated ) {
4711                                 $caption = $part;
4712                         }
4713                 }
4714
4715                 # Process alignment parameters
4716                 if ( $params['horizAlign'] ) {
4717                         $params['frame']['align'] = key( $params['horizAlign'] );
4718                 }
4719                 if ( $params['vertAlign'] ) {
4720                         $params['frame']['valign'] = key( $params['vertAlign'] );
4721                 }
4722
4723                 $params['frame']['caption'] = $caption;
4724
4725                 # Will the image be presented in a frame, with the caption below?
4726                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4727                                  isset( $params['frame']['framed'] ) ||
4728                                  isset( $params['frame']['thumbnail'] ) ||
4729                                  isset( $params['frame']['manualthumb'] );
4730
4731                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4732                 # came to also set the caption, ordinary text after the image -- which
4733                 # makes no sense, because that just repeats the text multiple times in
4734                 # screen readers.  It *also* came to set the title attribute.
4735                 #
4736                 # Now that we have an alt attribute, we should not set the alt text to
4737                 # equal the caption: that's worse than useless, it just repeats the
4738                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4739                 # use the unnamed parameter for alt text as well, just for the time be-
4740                 # ing, if the unnamed param is set and the alt param is not.
4741                 #
4742                 # For the future, we need to figure out if we want to tweak this more,
4743                 # e.g., introducing a title= parameter for the title; ignoring the un-
4744                 # named parameter entirely for images without a caption; adding an ex-
4745                 # plicit caption= parameter and preserving the old magic unnamed para-
4746                 # meter for BC; ...
4747                 if ( $imageIsFramed ) { # Framed image
4748                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4749                                 # No caption or alt text, add the filename as the alt text so
4750                                 # that screen readers at least get some description of the image
4751                                 $params['frame']['alt'] = $title->getText();
4752                         }
4753                         # Do not set $params['frame']['title'] because tooltips don't make sense
4754                         # for framed images
4755                 } else { # Inline image
4756                         if ( !isset( $params['frame']['alt'] ) ) {
4757                                 # No alt text, use the "caption" for the alt text
4758                                 if ( $caption !== '') {
4759                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4760                                 } else {
4761                                         # No caption, fall back to using the filename for the
4762                                         # alt text
4763                                         $params['frame']['alt'] = $title->getText();
4764                                 }
4765                         }
4766                         # Use the "caption" for the tooltip text
4767                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4768                 }
4769
4770                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4771
4772                 # Linker does the rest
4773                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery, $this->mOptions->getThumbSize() );
4774
4775                 # Give the handler a chance to modify the parser object
4776                 if ( $handler ) {
4777                         $handler->parserTransformHook( $this, $file );
4778                 }
4779
4780                 return $ret;
4781         }
4782
4783         protected function stripAltText( $caption, $holders ) {
4784                 # Strip bad stuff out of the title (tooltip).  We can't just use
4785                 # replaceLinkHoldersText() here, because if this function is called
4786                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4787                 if ( $holders ) {
4788                         $tooltip = $holders->replaceText( $caption );
4789                 } else {
4790                         $tooltip = $this->replaceLinkHoldersText( $caption );
4791                 }
4792
4793                 # make sure there are no placeholders in thumbnail attributes
4794                 # that are later expanded to html- so expand them now and
4795                 # remove the tags
4796                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4797                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4798
4799                 return $tooltip;
4800         }
4801
4802         /**
4803          * Set a flag in the output object indicating that the content is dynamic and
4804          * shouldn't be cached.
4805          */
4806         function disableCache() {
4807                 wfDebug( "Parser output marked as uncacheable.\n" );
4808                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
4809                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
4810         }
4811
4812         /**
4813          * Callback from the Sanitizer for expanding items found in HTML attribute
4814          * values, so they can be safely tested and escaped.
4815          *
4816          * @param $text String
4817          * @param $frame PPFrame
4818          * @return String
4819          * @private
4820          */
4821         function attributeStripCallback( &$text, $frame = false ) {
4822                 $text = $this->replaceVariables( $text, $frame );
4823                 $text = $this->mStripState->unstripBoth( $text );
4824                 return $text;
4825         }
4826
4827         /**
4828          * Accessor
4829          */
4830         function getTags() {
4831                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4832         }
4833
4834         /**
4835          * Break wikitext input into sections, and either pull or replace
4836          * some particular section's text.
4837          *
4838          * External callers should use the getSection and replaceSection methods.
4839          *
4840          * @param $text String: Page wikitext
4841          * @param $section String: a section identifier string of the form:
4842          *   <flag1> - <flag2> - ... - <section number>
4843          *
4844          * Currently the only recognised flag is "T", which means the target section number
4845          * was derived during a template inclusion parse, in other words this is a template
4846          * section edit link. If no flags are given, it was an ordinary section edit link.
4847          * This flag is required to avoid a section numbering mismatch when a section is
4848          * enclosed by <includeonly> (bug 6563).
4849          *
4850          * The section number 0 pulls the text before the first heading; other numbers will
4851          * pull the given section along with its lower-level subsections. If the section is
4852          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4853          *
4854          * @param $mode String: one of "get" or "replace"
4855          * @param $newText String: replacement text for section data.
4856          * @return String: for "get", the extracted section text.
4857          *                 for "replace", the whole page with the section replaced.
4858          */
4859         private function extractSections( $text, $section, $mode, $newText='' ) {
4860                 global $wgTitle;
4861                 $this->clearState();
4862                 $this->setTitle( $wgTitle ); # not generally used but removes an ugly failure mode
4863                 $this->mOptions = new ParserOptions;
4864                 $this->setOutputType( self::OT_PLAIN );
4865                 $outText = '';
4866                 $frame = $this->getPreprocessor()->newFrame();
4867
4868                 # Process section extraction flags
4869                 $flags = 0;
4870                 $sectionParts = explode( '-', $section );
4871                 $sectionIndex = array_pop( $sectionParts );
4872                 foreach ( $sectionParts as $part ) {
4873                         if ( $part === 'T' ) {
4874                                 $flags |= self::PTD_FOR_INCLUSION;
4875                         }
4876                 }
4877                 # Preprocess the text
4878                 $root = $this->preprocessToDom( $text, $flags );
4879
4880                 # <h> nodes indicate section breaks
4881                 # They can only occur at the top level, so we can find them by iterating the root's children
4882                 $node = $root->getFirstChild();
4883
4884                 # Find the target section
4885                 if ( $sectionIndex == 0 ) {
4886                         # Section zero doesn't nest, level=big
4887                         $targetLevel = 1000;
4888                 } else {
4889                         while ( $node ) {
4890                                 if ( $node->getName() === 'h' ) {
4891                                         $bits = $node->splitHeading();
4892                                         if ( $bits['i'] == $sectionIndex ) {
4893                                                 $targetLevel = $bits['level'];
4894                                                 break;
4895                                         }
4896                                 }
4897                                 if ( $mode === 'replace' ) {
4898                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4899                                 }
4900                                 $node = $node->getNextSibling();
4901                         }
4902                 }
4903
4904                 if ( !$node ) {
4905                         # Not found
4906                         if ( $mode === 'get' ) {
4907                                 return $newText;
4908                         } else {
4909                                 return $text;
4910                         }
4911                 }
4912
4913                 # Find the end of the section, including nested sections
4914                 do {
4915                         if ( $node->getName() === 'h' ) {
4916                                 $bits = $node->splitHeading();
4917                                 $curLevel = $bits['level'];
4918                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4919                                         break;
4920                                 }
4921                         }
4922                         if ( $mode === 'get' ) {
4923                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4924                         }
4925                         $node = $node->getNextSibling();
4926                 } while ( $node );
4927
4928                 # Write out the remainder (in replace mode only)
4929                 if ( $mode === 'replace' ) {
4930                         # Output the replacement text
4931                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4932                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4933                         # Only add trailing whitespace if there is newText
4934                         if ( $newText != "" ) {
4935                                 $outText .= $newText . "\n\n";
4936                         }
4937
4938                         while ( $node ) {
4939                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4940                                 $node = $node->getNextSibling();
4941                         }
4942                 }
4943
4944                 if ( is_string( $outText ) ) {
4945                         # Re-insert stripped tags
4946                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4947                 }
4948
4949                 return $outText;
4950         }
4951
4952         /**
4953          * This function returns the text of a section, specified by a number ($section).
4954          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4955          * the first section before any such heading (section 0).
4956          *
4957          * If a section contains subsections, these are also returned.
4958          *
4959          * @param $text String: text to look in
4960          * @param $section String: section identifier
4961          * @param $deftext String: default to return if section is not found
4962          * @return string text of the requested section
4963          */
4964         public function getSection( $text, $section, $deftext='' ) {
4965                 return $this->extractSections( $text, $section, "get", $deftext );
4966         }
4967
4968         public function replaceSection( $oldtext, $section, $text ) {
4969                 return $this->extractSections( $oldtext, $section, "replace", $text );
4970         }
4971
4972         /**
4973          * Get the ID of the revision we are parsing
4974          *
4975          * @return Mixed: integer or null
4976          */
4977         function getRevisionId() {
4978                 return $this->mRevisionId;
4979         }
4980
4981         /**
4982          * Get the revision object for $this->mRevisionId
4983          *
4984          * @return either a Revision object or null
4985          */
4986         protected function getRevisionObject() {
4987                 if ( !is_null( $this->mRevisionObject ) ) {
4988                         return $this->mRevisionObject;
4989                 }
4990                 if ( is_null( $this->mRevisionId ) ) {
4991                         return null;
4992                 }
4993
4994                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
4995                 return $this->mRevisionObject;
4996         }
4997
4998         /**
4999          * Get the timestamp associated with the current revision, adjusted for
5000          * the default server-local timestamp
5001          */
5002         function getRevisionTimestamp() {
5003                 if ( is_null( $this->mRevisionTimestamp ) ) {
5004                         wfProfileIn( __METHOD__ );
5005
5006                         $revObject = $this->getRevisionObject();
5007                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5008
5009                         if( $timestamp !== false ) {
5010                                 global $wgContLang;
5011
5012                                 # The cryptic '' timezone parameter tells to use the site-default
5013                                 # timezone offset instead of the user settings.
5014                                 #
5015                                 # Since this value will be saved into the parser cache, served
5016                                 # to other users, and potentially even used inside links and such,
5017                                 # it needs to be consistent for all visitors.
5018                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5019                         }
5020
5021                         wfProfileOut( __METHOD__ );
5022                 }
5023                 return $this->mRevisionTimestamp;
5024         }
5025
5026         /**
5027          * Get the name of the user that edited the last revision
5028          *
5029          * @return String: user name
5030          */
5031         function getRevisionUser() {
5032                 if( is_null( $this->mRevisionUser ) ) {
5033                         $revObject = $this->getRevisionObject();
5034
5035                         # if this template is subst: the revision id will be blank,
5036                         # so just use the current user's name
5037                         if( $revObject ) {
5038                                 $this->mRevisionUser = $revObject->getUserText();
5039                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5040                                 $this->mRevisionUser = $this->getUser()->getName();
5041                         }
5042                 }
5043                 return $this->mRevisionUser;
5044         }
5045
5046         /**
5047          * Mutator for $mDefaultSort
5048          *
5049          * @param $sort New value
5050          */
5051         public function setDefaultSort( $sort ) {
5052                 $this->mDefaultSort = $sort;
5053                 $this->mOutput->setProperty( 'defaultsort', $sort );
5054         }
5055
5056         /**
5057          * Accessor for $mDefaultSort
5058          * Will use the title/prefixed title if none is set
5059          *
5060          * @return string
5061          */
5062         public function getDefaultSort() {
5063                 if ( $this->mDefaultSort !== false ) {
5064                         return $this->mDefaultSort;
5065                 } else {
5066                         return $this->mTitle->getCategorySortkey();
5067                 }
5068         }
5069
5070         /**
5071          * Accessor for $mDefaultSort
5072          * Unlike getDefaultSort(), will return false if none is set
5073          *
5074          * @return string or false
5075          */
5076         public function getCustomDefaultSort() {
5077                 return $this->mDefaultSort;
5078         }
5079
5080         /**
5081          * Try to guess the section anchor name based on a wikitext fragment
5082          * presumably extracted from a heading, for example "Header" from
5083          * "== Header ==".
5084          */
5085         public function guessSectionNameFromWikiText( $text ) {
5086                 # Strip out wikitext links(they break the anchor)
5087                 $text = $this->stripSectionName( $text );
5088                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5089                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5090         }
5091
5092         /**
5093          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5094          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5095          * as something other than UTF-8 (apparently?), resulting in breakage.
5096          *
5097          * @param $text String: The section name
5098          * @return string An anchor
5099          */
5100         public function guessLegacySectionNameFromWikiText( $text ) {
5101                 # Strip out wikitext links(they break the anchor)
5102                 $text = $this->stripSectionName( $text );
5103                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5104                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5105         }
5106
5107         /**
5108          * Strips a text string of wikitext for use in a section anchor
5109          *
5110          * Accepts a text string and then removes all wikitext from the
5111          * string and leaves only the resultant text (i.e. the result of
5112          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5113          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5114          * to create valid section anchors by mimicing the output of the
5115          * parser when headings are parsed.
5116          *
5117          * @param $text String: text string to be stripped of wikitext
5118          * for use in a Section anchor
5119          * @return Filtered text string
5120          */
5121         public function stripSectionName( $text ) {
5122                 # Strip internal link markup
5123                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5124                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5125
5126                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5127                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5128                 # on how many empty links there are on the page - need to figure that out.
5129                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5130
5131                 # Parse wikitext quotes (italics & bold)
5132                 $text = $this->doQuotes( $text );
5133
5134                 # Strip HTML tags
5135                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5136                 return $text;
5137         }
5138
5139         /**
5140          * strip/replaceVariables/unstrip for preprocessor regression testing
5141          */
5142         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5143                 $this->clearState();
5144                 if ( !$title instanceof Title ) {
5145                         $title = Title::newFromText( $title );
5146                 }
5147                 $this->mTitle = $title;
5148                 $options->resetUsage();
5149                 $this->mOptions = $options;
5150                 $this->setOutputType( $outputType );
5151                 $text = $this->replaceVariables( $text );
5152                 $text = $this->mStripState->unstripBoth( $text );
5153                 $text = Sanitizer::removeHTMLtags( $text );
5154                 return $text;
5155         }
5156
5157         function testPst( $text, $title, $options ) {
5158                 global $wgUser;
5159                 if ( !$title instanceof Title ) {
5160                         $title = Title::newFromText( $title );
5161                 }
5162                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5163         }
5164
5165         function testPreprocess( $text, $title, $options ) {
5166                 if ( !$title instanceof Title ) {
5167                         $title = Title::newFromText( $title );
5168                 }
5169                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5170         }
5171
5172         function markerSkipCallback( $s, $callback ) {
5173                 $i = 0;
5174                 $out = '';
5175                 while ( $i < strlen( $s ) ) {
5176                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5177                         if ( $markerStart === false ) {
5178                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5179                                 break;
5180                         } else {
5181                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5182                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5183                                 if ( $markerEnd === false ) {
5184                                         $out .= substr( $s, $markerStart );
5185                                         break;
5186                                 } else {
5187                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5188                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5189                                         $i = $markerEnd;
5190                                 }
5191                         }
5192                 }
5193                 return $out;
5194         }
5195
5196         function serialiseHalfParsedText( $text ) {
5197                 $data = array();
5198                 $data['text'] = $text;
5199
5200                 # First, find all strip markers, and store their
5201                 #  data in an array.
5202                 $stripState = new StripState;
5203                 $pos = 0;
5204                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5205                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5206                 {
5207                         $end_pos += strlen( self::MARKER_SUFFIX );
5208                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5209
5210                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5211                                 $replaceArray = $stripState->general;
5212                                 $stripText = $this->mStripState->general->data[$marker];
5213                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5214                                 $replaceArray = $stripState->nowiki;
5215                                 $stripText = $this->mStripState->nowiki->data[$marker];
5216                         } else {
5217                                 throw new MWException( "Hanging strip marker: '$marker'." );
5218                         }
5219
5220                         $replaceArray->setPair( $marker, $stripText );
5221                         $pos = $end_pos;
5222                 }
5223                 $data['stripstate'] = $stripState;
5224
5225                 # Now, find all of our links, and store THEIR
5226                 #  data in an array! :)
5227                 $links = array( 'internal' => array(), 'interwiki' => array() );
5228                 $pos = 0;
5229
5230                 # Internal links
5231                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5232                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5233
5234                         $ns = trim( $ns );
5235                         if ( empty( $links['internal'][$ns] ) ) {
5236                                 $links['internal'][$ns] = array();
5237                         }
5238
5239                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5240                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5241                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5242                 }
5243
5244                 $pos = 0;
5245
5246                 # Interwiki links
5247                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5248                         $data = substr( $text, $start_pos );
5249                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5250                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5251                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5252                 }
5253
5254                 $data['linkholder'] = $links;
5255
5256                 return $data;
5257         }
5258
5259         /**
5260          * TODO: document
5261          * @param $data Array
5262          * @param $intPrefix String unique identifying prefix
5263          * @return String
5264          */
5265         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5266                 if ( !$intPrefix ) {
5267                         $intPrefix = self::getRandomString();
5268                 }
5269
5270                 # First, extract the strip state.
5271                 $stripState = $data['stripstate'];
5272                 $this->mStripState->general->merge( $stripState->general );
5273                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5274
5275                 # Now, extract the text, and renumber links
5276                 $text = $data['text'];
5277                 $links = $data['linkholder'];
5278
5279                 # Internal...
5280                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5281                         foreach ( $nsLinks as $key => $entry ) {
5282                                 $newKey = $intPrefix . '-' . $key;
5283                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5284
5285                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5286                         }
5287                 }
5288
5289                 # Interwiki...
5290                 foreach ( $links['interwiki'] as $key => $entry ) {
5291                         $newKey = "$intPrefix-$key";
5292                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5293
5294                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5295                 }
5296
5297                 # Should be good to go.
5298                 return $text;
5299         }
5300 }
5301
5302 /**
5303  * @todo document, briefly.
5304  * @ingroup Parser
5305  */
5306 class StripState {
5307         var $general, $nowiki;
5308
5309         function __construct() {
5310                 $this->general = new ReplacementArray;
5311                 $this->nowiki = new ReplacementArray;
5312         }
5313
5314         function unstripGeneral( $text ) {
5315                 wfProfileIn( __METHOD__ );
5316                 do {
5317                         $oldText = $text;
5318                         $text = $this->general->replace( $text );
5319                 } while ( $text !== $oldText );
5320                 wfProfileOut( __METHOD__ );
5321                 return $text;
5322         }
5323
5324         function unstripNoWiki( $text ) {
5325                 wfProfileIn( __METHOD__ );
5326                 do {
5327                         $oldText = $text;
5328                         $text = $this->nowiki->replace( $text );
5329                 } while ( $text !== $oldText );
5330                 wfProfileOut( __METHOD__ );
5331                 return $text;
5332         }
5333
5334         function unstripBoth( $text ) {
5335                 wfProfileIn( __METHOD__ );
5336                 do {
5337                         $oldText = $text;
5338                         $text = $this->general->replace( $text );
5339                         $text = $this->nowiki->replace( $text );
5340                 } while ( $text !== $oldText );
5341                 wfProfileOut( __METHOD__ );
5342                 return $text;
5343         }
5344 }
5345
5346 /**
5347  * @todo document, briefly.
5348  * @ingroup Parser
5349  */
5350 class OnlyIncludeReplacer {
5351         var $output = '';
5352
5353         function replace( $matches ) {
5354                 if ( substr( $matches[1], -1 ) === "\n" ) {
5355                         $this->output .= substr( $matches[1], 0, -1 );
5356                 } else {
5357                         $this->output .= $matches[1];
5358                 }
5359         }
5360 }