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