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 = array();
  93         var $mTransparentTagHooks = array();
  94         var $mFunctionHooks = array();
  95         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
  96         var $mFunctionTagHooks = array();
  97         var $mStripList  = array();
  98         var $mDefaultStripList  = array();
  99         var $mVarCache = array();
 100         var $mImageParams = array();
 101         var $mImageParamsMagicArray = array();
 102         var $mMarkerIndex = 0;
 103         var $mFirstCall = true;
 104         var $mVariables, $mSubstWords; # Initialised by initialiseVariables()
 105         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 106
 107
 108         # Cleared with clearState():
 109         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 110         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 111         var $mLinkHolders, $mLinkID;
 112         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 113         var $mTplExpandCache; # empty-frame expansion cache
 114         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 115         var $mExpensiveFunctionCount; # number of expensive parser function calls
 116         var $mUser; # User object; only used when doing pre-save transform
 117
 118         # Temporary
 119         # These are variables reset at least once per parse regardless of $clearState
 120         var $mOptions;      # ParserOptions object
 121         var $mTitle;        # Title context, used for self-link rendering and similar things
 122         var $mOutputType;   # Output type, one of the OT_xxx constants
 123         var $ot;            # Shortcut alias, see setOutputType()
 124         var $mRevisionObject; # The revision object of the specified revision ID
 125         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 126         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 127         var $mRevisionUser; # Userto display in {{REVISIONUSER}} tag
 128         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 129
 130         /**
 131          * Constructor
 132          *
 133          * @public
 134          */
 135         function __construct( $conf = array() ) {
 136                 $this->mConf = $conf;
 137                 $this->mUrlProtocols = wfUrlProtocols();
 138                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 139                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 140                 if ( isset( $conf['preprocessorClass'] ) ) {
 141                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 142                 } elseif ( extension_loaded( 'domxml' ) ) {
 143                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 144                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 145                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 146                 } elseif ( extension_loaded( 'dom' ) ) {
 147                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 148                 } else {
 149                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 150                 }
 151         }
 152
 153         /**
 154          * Reduce memory usage to reduce the impact of circular references
 155          */
 156         function __destruct() {
 157                 if ( isset( $this->mLinkHolders ) ) {
 158                         $this->mLinkHolders->__destruct();
 159                 }
 160                 foreach ( $this as $name => $value ) {
 161                         unset( $this->$name );
 162                 }
 163         }
 164
 165         /**
 166          * Do various kinds of initialisation on the first call of the parser
 167          */
 168         function firstCallInit() {
 169                 if ( !$this->mFirstCall ) {
 170                         return;
 171                 }
 172                 $this->mFirstCall = false;
 173
 174                 wfProfileIn( __METHOD__ );
 175
 176                 CoreParserFunctions::register( $this );
 177                 CoreTagHooks::register( $this );
 178                 $this->initialiseVariables();
 179
 180                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 181                 wfProfileOut( __METHOD__ );
 182         }
 183
 184         /**
 185          * Clear Parser state
 186          *
 187          * @private
 188          */
 189         function clearState() {
 190                 wfProfileIn( __METHOD__ );
 191                 if ( $this->mFirstCall ) {
 192                         $this->firstCallInit();
 193                 }
 194                 $this->mOutput = new ParserOutput;
 195                 $this->mAutonumber = 0;
 196                 $this->mLastSection = '';
 197                 $this->mDTopen = false;
 198                 $this->mIncludeCount = array();
 199                 $this->mStripState = new StripState;
 200                 $this->mArgStack = false;
 201                 $this->mInPre = false;
 202                 $this->mLinkHolders = new LinkHolderArray( $this );
 203                 $this->mLinkID = 0;
 204                 $this->mRevisionObject = $this->mRevisionTimestamp =
 205                         $this->mRevisionId = $this->mRevisionUser = null;
 206                 $this->mVarCache = array();
 207                 $this->mUser = null;
 208
 209                 /**
 210                  * Prefix for temporary replacement strings for the multipass parser.
 211                  * \x07 should never appear in input as it's disallowed in XML.
 212                  * Using it at the front also gives us a little extra robustness
 213                  * since it shouldn't match when butted up against identifier-like
 214                  * string constructs.
 215                  *
 216                  * Must not consist of all title characters, or else it will change
 217                  * the behaviour of <nowiki> in a link.
 218                  */
 219                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 220                 # Changed to \x7f to allow XML double-parsing -- TS
 221                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 222
 223
 224                 # Clear these on every parse, bug 4549
 225                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 226
 227                 $this->mShowToc = true;
 228                 $this->mForceTocPosition = false;
 229                 $this->mIncludeSizes = array(
 230                         'post-expand' => 0,
 231                         'arg' => 0,
 232                 );
 233                 $this->mPPNodeCount = 0;
 234                 $this->mDefaultSort = false;
 235                 $this->mHeadings = array();
 236                 $this->mDoubleUnderscores = array();
 237                 $this->mExpensiveFunctionCount = 0;
 238
 239                 # Fix cloning
 240                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 241                         $this->mPreprocessor = null;
 242                 }
 243
 244                 wfRunHooks( 'ParserClearState', array( &$this ) );
 245                 wfProfileOut( __METHOD__ );
 246         }
 247
 248         /**
 249          * Convert wikitext to HTML
 250          * Do not call this function recursively.
 251          *
 252          * @param $text String: text we want to parse
 253          * @param $title A title object
 254          * @param $options ParserOptions
 255          * @param $linestart boolean
 256          * @param $clearState boolean
 257          * @param $revid Int: number to pass in {{REVISIONID}}
 258          * @return ParserOutput a ParserOutput
 259          */
 260         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 261                 /**
 262                  * First pass--just handle <nowiki> sections, pass the rest off
 263                  * to internalParse() which does all the real work.
 264                  */
 265
 266                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 267                 $fname = __METHOD__.'-' . wfGetCaller();
 268                 wfProfileIn( __METHOD__ );
 269                 wfProfileIn( $fname );
 270
 271                 if ( $clearState ) {
 272                         $this->clearState();
 273                 }
 274
 275                 $options->resetUsage();
 276                 $this->mOptions = $options;
 277                 $this->setTitle( $title ); # Page title has to be set for the pre-processor
 278
 279                 $oldRevisionId = $this->mRevisionId;
 280                 $oldRevisionObject = $this->mRevisionObject;
 281                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 282                 $oldRevisionUser = $this->mRevisionUser;
 283                 if ( $revid !== null ) {
 284                         $this->mRevisionId = $revid;
 285                         $this->mRevisionObject = null;
 286                         $this->mRevisionTimestamp = null;
 287                         $this->mRevisionUser = null;
 288                 }
 289                 $this->setOutputType( self::OT_HTML );
 290                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 291                 # No more strip!
 292                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 293                 $text = $this->internalParse( $text );
 294
 295                 $text = $this->mStripState->unstripGeneral( $text );
 296
 297                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 298                 $fixtags = array(
 299                         # french spaces, last one Guillemet-left
 300                         # only if there is something before the space
 301                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 302                         # french spaces, Guillemet-right
 303                         '/(\\302\\253) /' => '\\1&#160;',
 304                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 305                 );
 306                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 307
 308                 $text = $this->doBlockLevels( $text, $linestart );
 309
 310                 $this->replaceLinkHolders( $text );
 311
 312                 /**
 313                  * The page doesn't get language converted if
 314                  * a) It's disabled
 315                  * b) Content isn't converted
 316                  * c) It's a conversion table
 317                  */
 318                 if ( !( $wgDisableLangConversion
 319                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 320                                 || $this->mTitle->isConversionTable() ) ) {
 321
 322                         # The position of the convert() call should not be changed. it
 323                         # assumes that the links are all replaced and the only thing left
 324                         # is the <nowiki> mark.
 325
 326                         $text = $wgContLang->convert( $text );
 327                 }
 328
 329                 /**
 330                  * A page get its title converted except:
 331                  * a) Language conversion is globally disabled
 332                  * b) Title convert is globally disabled
 333                  * c) The page is a redirect page
 334                  * d) User request with a "linkconvert" set to "no"
 335                  * e) A "nocontentconvert" magic word has been set
 336                  * f) A "notitleconvert" magic word has been set
 337                  * g) User sets "noconvertlink" in his/her preference
 338                  *
 339                  * Note that if a user tries to set a title in a conversion
 340                  * rule but content conversion was not done, then the parser
 341                  * won't pick it up.  This is probably expected behavior.
 342                  */
 343                 if ( !( $wgDisableLangConversion
 344                                 || $wgDisableTitleConversion
 345                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 346                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 347                                 || $this->mOutput->getDisplayTitle() !== false ) )
 348                 {
 349                         $convruletitle = $wgContLang->getConvRuleTitle();
 350                         if ( $convruletitle ) {
 351                                 $this->mOutput->setTitleText( $convruletitle );
 352                         } else {
 353                                 $titleText = $wgContLang->convertTitle( $title );
 354                                 $this->mOutput->setTitleText( $titleText );
 355                         }
 356                 }
 357
 358                 $text = $this->mStripState->unstripNoWiki( $text );
 359
 360                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 361
 362 //!JF Move to its own function
 363
 364                 $uniq_prefix = $this->mUniqPrefix;
 365                 $matches = array();
 366                 $elements = array_keys( $this->mTransparentTagHooks );
 367                 $text = $this->extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 368
 369                 foreach ( $matches as $marker => $data ) {
 370                         list( $element, $content, $params, $tag ) = $data;
 371                         $tagName = strtolower( $element );
 372                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 373                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 374                         } else {
 375                                 $output = $tag;
 376                         }
 377                         $this->mStripState->general->setPair( $marker, $output );
 378                 }
 379                 $text = $this->mStripState->unstripGeneral( $text );
 380
 381                 $text = Sanitizer::normalizeCharReferences( $text );
 382
 383                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 384                         $text = MWTidy::tidy( $text );
 385                 } else {
 386                         # attempt to sanitize at least some nesting problems
 387                         # (bug #2702 and quite a few others)
 388                         $tidyregs = array(
 389                                 # ''Something [http://www.cool.com cool''] -->
 390                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 391                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 392                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 393                                 # fix up an anchor inside another anchor, only
 394                                 # at least for a single single nested link (bug 3695)
 395                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 396                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 397                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 398                                 # contains a div, so fix it up here; replace
 399                                 # div with escaped text
 400                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 401                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 402                                 # remove empty italic or bold tag pairs, some
 403                                 # introduced by rules above
 404                                 '/<([bi])><\/\\1>/' => '',
 405                         );
 406
 407                         $text = preg_replace(
 408                                 array_keys( $tidyregs ),
 409                                 array_values( $tidyregs ),
 410                                 $text );
 411                 }
 412                 global $wgExpensiveParserFunctionLimit;
 413                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 414                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 415                 }
 416
 417                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 418
 419                 # Information on include size limits, for the benefit of users who try to skirt them
 420                 if ( $this->mOptions->getEnableLimitReport() ) {
 421                         $max = $this->mOptions->getMaxIncludeSize();
 422                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 423                         $limitReport =
 424                                 "NewPP limit report\n" .
 425                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 426                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 427                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 428                                 $PFreport;
 429                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 430                         $text .= "\n<!-- \n$limitReport-->\n";
 431                 }
 432                 $this->mOutput->setText( $text );
 433
 434                 $this->mRevisionId = $oldRevisionId;
 435                 $this->mRevisionObject = $oldRevisionObject;
 436                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 437                 $this->mRevisionUser = $oldRevisionUser;
 438                 wfProfileOut( $fname );
 439                 wfProfileOut( __METHOD__ );
 440
 441                 return $this->mOutput;
 442         }
 443
 444         /**
 445          * Recursive parser entry point that can be called from an extension tag
 446          * hook.
 447          *
 448          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 449          *
 450          * @param $text String: text extension wants to have parsed
 451          * @param $frame PPFrame: The frame to use for expanding any template variables
 452          */
 453         function recursiveTagParse( $text, $frame=false ) {
 454                 wfProfileIn( __METHOD__ );
 455                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 456                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 457                 $text = $this->internalParse( $text, false, $frame );
 458                 wfProfileOut( __METHOD__ );
 459                 return $text;
 460         }
 461
 462         /**
 463          * Expand templates and variables in the text, producing valid, static wikitext.
 464          * Also removes comments.
 465          */
 466         function preprocess( $text, $title, $options, $revid = null ) {
 467                 wfProfileIn( __METHOD__ );
 468                 $this->clearState();
 469                 $this->setOutputType( self::OT_PREPROCESS );
 470                 $options->resetUsage();
 471                 $this->mOptions = $options;
 472                 $this->setTitle( $title );
 473                 if ( $revid !== null ) {
 474                         $this->mRevisionId = $revid;
 475                 }
 476                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 477                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 478                 $text = $this->replaceVariables( $text );
 479                 $text = $this->mStripState->unstripBoth( $text );
 480                 wfProfileOut( __METHOD__ );
 481                 return $text;
 482         }
 483
 484         /**
 485          * Process the wikitext for the ?preload= feature. (bug 5210)
 486          *
 487          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 488          * comments, templates, arguments, tags hooks and parser functions are untouched.
 489          */
 490         public function getPreloadText( $text, $title, $options ) {
 491                 # Parser (re)initialisation
 492                 $this->clearState();
 493                 $this->setOutputType( self::OT_PLAIN );
 494                 $options->resetUsage();
 495                 $this->mOptions = $options;
 496                 $this->setTitle( $title );
 497
 498                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 499                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 500                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 501         }
 502
 503         /**
 504          * Get a random string
 505          *
 506          * @private
 507          * @static
 508          */
 509         static private function getRandomString() {
 510                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 511         }
 512
 513         /**
 514          * Set the current user.
 515          * Should only be used when doing pre-save transform.
 516          *
 517          * @param $user Mixed: User object or null (to reset)
 518          */
 519         function setUser( $user ) {
 520                 $this->mUser = $user;
 521         }
 522
 523         /**
 524          * Accessor for mUniqPrefix.
 525          *
 526          * @return String
 527          */
 528         public function uniqPrefix() {
 529                 if ( !isset( $this->mUniqPrefix ) ) {
 530                         # @todo Fixme: this is probably *horribly wrong*
 531                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 532                         # if this is called for a parser cache hit, the parser may not
 533                         # have ever been initialized in the first place.
 534                         # Not really sure what the heck is supposed to be going on here.
 535                         return '';
 536                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 537                 }
 538                 return $this->mUniqPrefix;
 539         }
 540
 541         /**
 542          * Set the context title
 543          */
 544         function setTitle( $t ) {
 545                 if ( !$t || $t instanceof FakeTitle ) {
 546                         $t = Title::newFromText( 'NO TITLE' );
 547                 }
 548
 549                 if ( strval( $t->getFragment() ) !== '' ) {
 550                         # Strip the fragment to avoid various odd effects
 551                         $this->mTitle = clone $t;
 552                         $this->mTitle->setFragment( '' );
 553                 } else {
 554                         $this->mTitle = $t;
 555                 }
 556         }
 557
 558         /**
 559          * Accessor for the Title object
 560          *
 561          * @return Title object
 562          */
 563         function getTitle() {
 564                 return $this->mTitle;
 565         }
 566
 567         /**
 568          * Accessor/mutator for the Title object
 569          *
 570          * @param $x New Title object or null to just get the current one
 571          * @return Title object
 572          */
 573         function Title( $x = null ) {
 574                 return wfSetVar( $this->mTitle, $x );
 575         }
 576
 577         /**
 578          * Set the output type
 579          *
 580          * @param $ot Integer: new value
 581          */
 582         function setOutputType( $ot ) {
 583                 $this->mOutputType = $ot;
 584                 # Shortcut alias
 585                 $this->ot = array(
 586                         'html' => $ot == self::OT_HTML,
 587                         'wiki' => $ot == self::OT_WIKI,
 588                         'pre' => $ot == self::OT_PREPROCESS,
 589                         'plain' => $ot == self::OT_PLAIN,
 590                 );
 591         }
 592
 593         /**
 594          * Accessor/mutator for the output type
 595          *
 596          * @param $x New value or null to just get the current one
 597          * @return Integer
 598          */
 599         function OutputType( $x = null ) {
 600                 return wfSetVar( $this->mOutputType, $x );
 601         }
 602
 603         /**
 604          * Get the ParserOutput object
 605          *
 606          * @return ParserOutput object
 607          */
 608         function getOutput() {
 609                 return $this->mOutput;
 610         }
 611
 612         /**
 613          * Get the ParserOptions object
 614          *
 615          * @return ParserOptions object
 616          */
 617         function getOptions() {
 618                 return $this->mOptions;
 619         }
 620
 621         /**
 622          * Accessor/mutator for the ParserOptions object
 623          *
 624          * @param $x New value or null to just get the current one
 625          * @return Current ParserOptions object
 626          */
 627         function Options( $x = null ) {
 628                 return wfSetVar( $this->mOptions, $x );
 629         }
 630
 631         function nextLinkID() {
 632                 return $this->mLinkID++;
 633         }
 634
 635         function getFunctionLang() {
 636                 global $wgLang, $wgContLang;
 637
 638                 $target = $this->mOptions->getTargetLanguage();
 639                 if ( $target !== null ) {
 640                         return $target;
 641                 } else {
 642                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 643                 }
 644         }
 645
 646         /**
 647          * Get a User object either from $this->mUser, if set, or from the
 648          * ParserOptions object otherwise
 649          *
 650          * @return User object
 651          */
 652         function getUser() {
 653                 if ( !is_null( $this->mUser ) ) {
 654                         return $this->mUser;
 655                 }
 656                 return $this->mOptions->getUser();
 657         }
 658
 659         /**
 660          * Get a preprocessor object
 661          *
 662          * @return Preprocessor instance
 663          */
 664         function getPreprocessor() {
 665                 if ( !isset( $this->mPreprocessor ) ) {
 666                         $class = $this->mPreprocessorClass;
 667                         $this->mPreprocessor = new $class( $this );
 668                 }
 669                 return $this->mPreprocessor;
 670         }
 671
 672         /**
 673          * Replaces all occurrences of HTML-style comments and the given tags
 674          * in the text with a random marker and returns the next text. The output
 675          * parameter $matches will be an associative array filled with data in
 676          * the form:
 677          *   'UNIQ-xxxxx' => array(
 678          *     'element',
 679          *     'tag content',
 680          *     array( 'param' => 'x' ),
 681          *     '<element param="x">tag content</element>' ) )
 682          *
 683          * @param $elements list of element names. Comments are always extracted.
 684          * @param $text Source text string.
 685          * @param $matches Out parameter, Array: extracted tags
 686          * @param $uniq_prefix
 687          * @return String: stripped text
 688          *
 689          * @static
 690          */
 691         public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 692                 static $n = 1;
 693                 $stripped = '';
 694                 $matches = array();
 695
 696                 $taglist = implode( '|', $elements );
 697                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 698
 699                 while ( $text != '' ) {
 700                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 701                         $stripped .= $p[0];
 702                         if ( count( $p ) < 5 ) {
 703                                 break;
 704                         }
 705                         if ( count( $p ) > 5 ) {
 706                                 # comment
 707                                 $element    = $p[4];
 708                                 $attributes = '';
 709                                 $close      = '';
 710                                 $inside     = $p[5];
 711                         } else {
 712                                 # tag
 713                                 $element    = $p[1];
 714                                 $attributes = $p[2];
 715                                 $close      = $p[3];
 716                                 $inside     = $p[4];
 717                         }
 718
 719                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 720                         $stripped .= $marker;
 721
 722                         if ( $close === '/>' ) {
 723                                 # Empty element tag, <tag />
 724                                 $content = null;
 725                                 $text = $inside;
 726                                 $tail = null;
 727                         } else {
 728                                 if ( $element === '!--' ) {
 729                                         $end = '/(-->)/';
 730                                 } else {
 731                                         $end = "/(<\\/$element\\s*>)/i";
 732                                 }
 733                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 734                                 $content = $q[0];
 735                                 if ( count( $q ) < 3 ) {
 736                                         # No end tag -- let it run out to the end of the text.
 737                                         $tail = '';
 738                                         $text = '';
 739                                 } else {
 740                                         $tail = $q[1];
 741                                         $text = $q[2];
 742                                 }
 743                         }
 744
 745                         $matches[$marker] = array( $element,
 746                                 $content,
 747                                 Sanitizer::decodeTagAttributes( $attributes ),
 748                                 "<$element$attributes$close$content$tail" );
 749                 }
 750                 return $stripped;
 751         }
 752
 753         /**
 754          * Get a list of strippable XML-like elements
 755          */
 756         function getStripList() {
 757                 return $this->mStripList;
 758         }
 759
 760         /**
 761          * @deprecated use replaceVariables
 762          */
 763         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 764                 return $text;
 765         }
 766
 767         /**
 768          * Restores pre, math, and other extensions removed by strip()
 769          *
 770          * always call unstripNoWiki() after this one
 771          * @private
 772          * @deprecated use $this->mStripState->unstrip()
 773          */
 774         function unstrip( $text, $state ) {
 775                 return $state->unstripGeneral( $text );
 776         }
 777
 778         /**
 779          * Always call this after unstrip() to preserve the order
 780          *
 781          * @private
 782          * @deprecated use $this->mStripState->unstrip()
 783          */
 784         function unstripNoWiki( $text, $state ) {
 785                 return $state->unstripNoWiki( $text );
 786         }
 787
 788         /**
 789          * @deprecated use $this->mStripState->unstripBoth()
 790          */
 791         function unstripForHTML( $text ) {
 792                 return $this->mStripState->unstripBoth( $text );
 793         }
 794
 795         /**
 796          * Add an item to the strip state
 797          * Returns the unique tag which must be inserted into the stripped text
 798          * The tag will be replaced with the original text in unstrip()
 799          *
 800          * @private
 801          */
 802         function insertStripItem( $text ) {
 803                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 804                 $this->mMarkerIndex++;
 805                 $this->mStripState->general->setPair( $rnd, $text );
 806                 return $rnd;
 807         }
 808
 809         /**
 810          * Interface with html tidy
 811          * @deprecated Use MWTidy::tidy()
 812          */
 813         public static function tidy( $text ) {
 814                 wfDeprecated( __METHOD__ );
 815                 return MWTidy::tidy( $text );
 816         }
 817
 818         /**
 819          * parse the wiki syntax used to render tables
 820          *
 821          * @private
 822          */
 823         function doTableStuff( $text ) {
 824                 wfProfileIn( __METHOD__ );
 825
 826                 $lines = StringUtils::explode( "\n", $text );
 827                 $out = '';
 828                 $td_history = array(); # Is currently a td tag open?
 829                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 830                 $tr_history = array(); # Is currently a tr tag open?
 831                 $tr_attributes = array(); # history of tr attributes
 832                 $has_opened_tr = array(); # Did this table open a <tr> element?
 833                 $indent_level = 0; # indent level of the table
 834
 835                 foreach ( $lines as $outLine ) {
 836                         $line = trim( $outLine );
 837
 838                         if ( $line === '' ) { # empty line, go to next line
 839                                 $out .= $outLine."\n";
 840                                 continue;
 841                         }
 842
 843                         $first_character = $line[0];
 844                         $matches = array();
 845
 846                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 847                                 # First check if we are starting a new table
 848                                 $indent_level = strlen( $matches[1] );
 849
 850                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 851                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 852
 853                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 854                                 array_push( $td_history , false );
 855                                 array_push( $last_tag_history , '' );
 856                                 array_push( $tr_history , false );
 857                                 array_push( $tr_attributes , '' );
 858                                 array_push( $has_opened_tr , false );
 859                         } elseif ( count( $td_history ) == 0 ) {
 860                                 # Don't do any of the following
 861                                 $out .= $outLine."\n";
 862                                 continue;
 863                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 864                                 # We are ending a table
 865                                 $line = '</table>' . substr( $line , 2 );
 866                                 $last_tag = array_pop( $last_tag_history );
 867
 868                                 if ( !array_pop( $has_opened_tr ) ) {
 869                                         $line = "<tr><td></td></tr>{$line}";
 870                                 }
 871
 872                                 if ( array_pop( $tr_history ) ) {
 873                                         $line = "</tr>{$line}";
 874                                 }
 875
 876                                 if ( array_pop( $td_history ) ) {
 877                                         $line = "</{$last_tag}>{$line}";
 878                                 }
 879                                 array_pop( $tr_attributes );
 880                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 881                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 882                                 # Now we have a table row
 883                                 $line = preg_replace( '#^\|-+#', '', $line );
 884
 885                                 # Whats after the tag is now only attributes
 886                                 $attributes = $this->mStripState->unstripBoth( $line );
 887                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 888                                 array_pop( $tr_attributes );
 889                                 array_push( $tr_attributes, $attributes );
 890
 891                                 $line = '';
 892                                 $last_tag = array_pop( $last_tag_history );
 893                                 array_pop( $has_opened_tr );
 894                                 array_push( $has_opened_tr , true );
 895
 896                                 if ( array_pop( $tr_history ) ) {
 897                                         $line = '</tr>';
 898                                 }
 899
 900                                 if ( array_pop( $td_history ) ) {
 901                                         $line = "</{$last_tag}>{$line}";
 902                                 }
 903
 904                                 $outLine = $line;
 905                                 array_push( $tr_history , false );
 906                                 array_push( $td_history , false );
 907                                 array_push( $last_tag_history , '' );
 908                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 909                                 # This might be cell elements, td, th or captions
 910                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 911                                         $first_character = '+';
 912                                         $line = substr( $line , 1 );
 913                                 }
 914
 915                                 $line = substr( $line , 1 );
 916
 917                                 if ( $first_character === '!' ) {
 918                                         $line = str_replace( '!!' , '||' , $line );
 919                                 }
 920
 921                                 # Split up multiple cells on the same line.
 922                                 # FIXME : This can result in improper nesting of tags processed
 923                                 # by earlier parser steps, but should avoid splitting up eg
 924                                 # attribute values containing literal "||".
 925                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 926
 927                                 $outLine = '';
 928
 929                                 # Loop through each table cell
 930                                 foreach ( $cells as $cell ) {
 931                                         $previous = '';
 932                                         if ( $first_character !== '+' ) {
 933                                                 $tr_after = array_pop( $tr_attributes );
 934                                                 if ( !array_pop( $tr_history ) ) {
 935                                                         $previous = "<tr{$tr_after}>\n";
 936                                                 }
 937                                                 array_push( $tr_history , true );
 938                                                 array_push( $tr_attributes , '' );
 939                                                 array_pop( $has_opened_tr );
 940                                                 array_push( $has_opened_tr , true );
 941                                         }
 942
 943                                         $last_tag = array_pop( $last_tag_history );
 944
 945                                         if ( array_pop( $td_history ) ) {
 946                                                 $previous = "</{$last_tag}>\n{$previous}";
 947                                         }
 948
 949                                         if ( $first_character === '|' ) {
 950                                                 $last_tag = 'td';
 951                                         } elseif ( $first_character === '!' ) {
 952                                                 $last_tag = 'th';
 953                                         } elseif ( $first_character === '+' ) {
 954                                                 $last_tag = 'caption';
 955                                         } else {
 956                                                 $last_tag = '';
 957                                         }
 958
 959                                         array_push( $last_tag_history , $last_tag );
 960
 961                                         # A cell could contain both parameters and data
 962                                         $cell_data = explode( '|' , $cell , 2 );
 963
 964                                         # Bug 553: Note that a '|' inside an invalid link should not
 965                                         # be mistaken as delimiting cell parameters
 966                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 967                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 968                                         } elseif ( count( $cell_data ) == 1 ) {
 969                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 970                                         } else {
 971                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 972                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 973                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 974                                         }
 975
 976                                         $outLine .= $cell;
 977                                         array_push( $td_history , true );
 978                                 }
 979                         }
 980                         $out .= $outLine . "\n";
 981                 }
 982
 983                 # Closing open td, tr && table
 984                 while ( count( $td_history ) > 0 ) {
 985                         if ( array_pop( $td_history ) ) {
 986                                 $out .= "</td>\n";
 987                         }
 988                         if ( array_pop( $tr_history ) ) {
 989                                 $out .= "</tr>\n";
 990                         }
 991                         if ( !array_pop( $has_opened_tr ) ) {
 992                                 $out .= "<tr><td></td></tr>\n" ;
 993                         }
 994
 995                         $out .= "</table>\n";
 996                 }
 997
 998                 # Remove trailing line-ending (b/c)
 999                 if ( substr( $out, -1 ) === "\n" ) {
1000                         $out = substr( $out, 0, -1 );
1001                 }
1002
1003                 # special case: don't return empty table
1004                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1005                         $out = '';
1006                 }
1007
1008                 wfProfileOut( __METHOD__ );
1009
1010                 return $out;
1011         }
1012
1013         /**
1014          * Helper function for parse() that transforms wiki markup into
1015          * HTML. Only called for $mOutputType == self::OT_HTML.
1016          *
1017          * @private
1018          */
1019         function internalParse( $text, $isMain = true, $frame=false ) {
1020                 wfProfileIn( __METHOD__ );
1021
1022                 $origText = $text;
1023
1024                 # Hook to suspend the parser in this state
1025                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1026                         wfProfileOut( __METHOD__ );
1027                         return $text ;
1028                 }
1029
1030                 # if $frame is provided, then use $frame for replacing any variables
1031                 if ( $frame ) {
1032                         # use frame depth to infer how include/noinclude tags should be handled
1033                         # depth=0 means this is the top-level document; otherwise it's an included document
1034                         if ( !$frame->depth ) {
1035                                 $flag = 0;
1036                         } else {
1037                                 $flag = Parser::PTD_FOR_INCLUSION;
1038                         }
1039                         $dom = $this->preprocessToDom( $text, $flag );
1040                         $text = $frame->expand( $dom );
1041                 } else {
1042                         # if $frame is not provided, then use old-style replaceVariables
1043                         $text = $this->replaceVariables( $text );
1044                 }
1045
1046                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1047                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1048
1049                 # Tables need to come after variable replacement for things to work
1050                 # properly; putting them before other transformations should keep
1051                 # exciting things like link expansions from showing up in surprising
1052                 # places.
1053                 $text = $this->doTableStuff( $text );
1054
1055                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1056
1057                 $text = $this->doDoubleUnderscore( $text );
1058
1059                 $text = $this->doHeadings( $text );
1060                 if ( $this->mOptions->getUseDynamicDates() ) {
1061                         $df = DateFormatter::getInstance();
1062                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1063                 }
1064                 $text = $this->replaceInternalLinks( $text );
1065                 $text = $this->doAllQuotes( $text );
1066                 $text = $this->replaceExternalLinks( $text );
1067
1068                 # replaceInternalLinks may sometimes leave behind
1069                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1070                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1071
1072                 $text = $this->doMagicLinks( $text );
1073                 $text = $this->formatHeadings( $text, $origText, $isMain );
1074
1075                 wfProfileOut( __METHOD__ );
1076                 return $text;
1077         }
1078
1079         /**
1080          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1081          * magic external links.
1082          *
1083          * DML
1084          * @private
1085          */
1086         function doMagicLinks( $text ) {
1087                 wfProfileIn( __METHOD__ );
1088                 $prots = $this->mUrlProtocols;
1089                 $urlChar = self::EXT_LINK_URL_CLASS;
1090                 $text = preg_replace_callback(
1091                         '!(?:                           # Start cases
1092                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1093                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1094                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1095                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1096                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1097                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1098                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1099                                     [0-9Xx]                 # check digit
1100                                     \b)
1101                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1102                 wfProfileOut( __METHOD__ );
1103                 return $text;
1104         }
1105
1106         function magicLinkCallback( $m ) {
1107                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1108                         # Skip anchor
1109                         return $m[0];
1110                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1111                         # Skip HTML element
1112                         return $m[0];
1113                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1114                         # Free external link
1115                         return $this->makeFreeExternalLink( $m[0] );
1116                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1117                         # RFC or PMID
1118                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1119                                 $keyword = 'RFC';
1120                                 $urlmsg = 'rfcurl';
1121                                 $CssClass = 'mw-magiclink-rfc';
1122                                 $id = $m[4];
1123                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1124                                 $keyword = 'PMID';
1125                                 $urlmsg = 'pubmedurl';
1126                                 $CssClass = 'mw-magiclink-pmid';
1127                                 $id = $m[4];
1128                         } else {
1129                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1130                                         substr( $m[0], 0, 20 ) . '"' );
1131                         }
1132                         $url = wfMsgForContent( $urlmsg, $id);
1133                         $sk = $this->mOptions->getSkin( $this->mTitle );
1134                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1135                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1136                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1137                         # ISBN
1138                         $isbn = $m[5];
1139                         $num = strtr( $isbn, array(
1140                                 '-' => '',
1141                                 ' ' => '',
1142                                 'x' => 'X',
1143                         ));
1144                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1145                         return'<a href="' .
1146                                 $titleObj->escapeLocalUrl() .
1147                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1148                 } else {
1149                         return $m[0];
1150                 }
1151         }
1152
1153         /**
1154          * Make a free external link, given a user-supplied URL
1155          * @return HTML
1156          * @private
1157          */
1158         function makeFreeExternalLink( $url ) {
1159                 global $wgContLang;
1160                 wfProfileIn( __METHOD__ );
1161
1162                 $sk = $this->mOptions->getSkin( $this->mTitle );
1163                 $trail = '';
1164
1165                 # The characters '<' and '>' (which were escaped by
1166                 # removeHTMLtags()) should not be included in
1167                 # URLs, per RFC 2396.
1168                 $m2 = array();
1169                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1170                         $trail = substr( $url, $m2[0][1] ) . $trail;
1171                         $url = substr( $url, 0, $m2[0][1] );
1172                 }
1173
1174                 # Move trailing punctuation to $trail
1175                 $sep = ',;\.:!?';
1176                 # If there is no left bracket, then consider right brackets fair game too
1177                 if ( strpos( $url, '(' ) === false ) {
1178                         $sep .= ')';
1179                 }
1180
1181                 $numSepChars = strspn( strrev( $url ), $sep );
1182                 if ( $numSepChars ) {
1183                         $trail = substr( $url, -$numSepChars ) . $trail;
1184                         $url = substr( $url, 0, -$numSepChars );
1185                 }
1186
1187                 $url = Sanitizer::cleanUrl( $url );
1188
1189                 # Is this an external image?
1190                 $text = $this->maybeMakeExternalImage( $url );
1191                 if ( $text === false ) {
1192                         # Not an image, make a link
1193                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1194                                 $this->getExternalLinkAttribs( $url ) );
1195                         # Register it in the output object...
1196                         # Replace unnecessary URL escape codes with their equivalent characters
1197                         $pasteurized = self::replaceUnusualEscapes( $url );
1198                         $this->mOutput->addExternalLink( $pasteurized );
1199                 }
1200                 wfProfileOut( __METHOD__ );
1201                 return $text . $trail;
1202         }
1203
1204
1205         /**
1206          * Parse headers and return html
1207          *
1208          * @private
1209          */
1210         function doHeadings( $text ) {
1211                 wfProfileIn( __METHOD__ );
1212                 for ( $i = 6; $i >= 1; --$i ) {
1213                         $h = str_repeat( '=', $i );
1214                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1215                           "<h$i>\\1</h$i>", $text );
1216                 }
1217                 wfProfileOut( __METHOD__ );
1218                 return $text;
1219         }
1220
1221         /**
1222          * Replace single quotes with HTML markup
1223          * @private
1224          * @return string the altered text
1225          */
1226         function doAllQuotes( $text ) {
1227                 wfProfileIn( __METHOD__ );
1228                 $outtext = '';
1229                 $lines = StringUtils::explode( "\n", $text );
1230                 foreach ( $lines as $line ) {
1231                         $outtext .= $this->doQuotes( $line ) . "\n";
1232                 }
1233                 $outtext = substr( $outtext, 0,-1 );
1234                 wfProfileOut( __METHOD__ );
1235                 return $outtext;
1236         }
1237
1238         /**
1239          * Helper function for doAllQuotes()
1240          */
1241         public function doQuotes( $text ) {
1242                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1243                 if ( count( $arr ) == 1 ) {
1244                         return $text;
1245                 } else {
1246                         # First, do some preliminary work. This may shift some apostrophes from
1247                         # being mark-up to being text. It also counts the number of occurrences
1248                         # of bold and italics mark-ups.
1249                         $numbold = 0;
1250                         $numitalics = 0;
1251                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1252                                 if ( ( $i % 2 ) == 1 ) {
1253                                         # If there are ever four apostrophes, assume the first is supposed to
1254                                         # be text, and the remaining three constitute mark-up for bold text.
1255                                         if ( strlen( $arr[$i] ) == 4 ) {
1256                                                 $arr[$i-1] .= "'";
1257                                                 $arr[$i] = "'''";
1258                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1259                                                 # If there are more than 5 apostrophes in a row, assume they're all
1260                                                 # text except for the last 5.
1261                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1262                                                 $arr[$i] = "'''''";
1263                                         }
1264                                         # Count the number of occurrences of bold and italics mark-ups.
1265                                         # We are not counting sequences of five apostrophes.
1266                                         if ( strlen( $arr[$i] ) == 2 ) {
1267                                                 $numitalics++;
1268                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1269                                                 $numbold++;
1270                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1271                                                 $numitalics++;
1272                                                 $numbold++;
1273                                         }
1274                                 }
1275                         }
1276
1277                         # If there is an odd number of both bold and italics, it is likely
1278                         # that one of the bold ones was meant to be an apostrophe followed
1279                         # by italics. Which one we cannot know for certain, but it is more
1280                         # likely to be one that has a single-letter word before it.
1281                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1282                                 $i = 0;
1283                                 $firstsingleletterword = -1;
1284                                 $firstmultiletterword = -1;
1285                                 $firstspace = -1;
1286                                 foreach ( $arr as $r ) {
1287                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1288                                                 $x1 = substr( $arr[$i-1], -1 );
1289                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1290                                                 if ( $x1 === ' ' ) {
1291                                                         if ( $firstspace == -1 ) {
1292                                                                 $firstspace = $i;
1293                                                         }
1294                                                 } elseif ( $x2 === ' ') {
1295                                                         if ( $firstsingleletterword == -1 ) {
1296                                                                 $firstsingleletterword = $i;
1297                                                         }
1298                                                 } else {
1299                                                         if ( $firstmultiletterword == -1 ) {
1300                                                                 $firstmultiletterword = $i;
1301                                                         }
1302                                                 }
1303                                         }
1304                                         $i++;
1305                                 }
1306
1307                                 # If there is a single-letter word, use it!
1308                                 if ( $firstsingleletterword > -1 ) {
1309                                         $arr[$firstsingleletterword] = "''";
1310                                         $arr[$firstsingleletterword-1] .= "'";
1311                                 } elseif ( $firstmultiletterword > -1 ) {
1312                                         # If not, but there's a multi-letter word, use that one.
1313                                         $arr[$firstmultiletterword] = "''";
1314                                         $arr[$firstmultiletterword-1] .= "'";
1315                                 } elseif ( $firstspace > -1 ) {
1316                                         # ... otherwise use the first one that has neither.
1317                                         # (notice that it is possible for all three to be -1 if, for example,
1318                                         # there is only one pentuple-apostrophe in the line)
1319                                         $arr[$firstspace] = "''";
1320                                         $arr[$firstspace-1] .= "'";
1321                                 }
1322                         }
1323
1324                         # Now let's actually convert our apostrophic mush to HTML!
1325                         $output = '';
1326                         $buffer = '';
1327                         $state = '';
1328                         $i = 0;
1329                         foreach ( $arr as $r ) {
1330                                 if ( ( $i % 2 ) == 0 ) {
1331                                         if ( $state === 'both' ) {
1332                                                 $buffer .= $r;
1333                                         } else {
1334                                                 $output .= $r;
1335                                         }
1336                                 } else {
1337                                         if ( strlen( $r ) == 2 ) {
1338                                                 if ( $state === 'i' ) {
1339                                                         $output .= '</i>'; $state = '';
1340                                                 } elseif ( $state === 'bi' ) {
1341                                                         $output .= '</i>'; $state = 'b';
1342                                                 } elseif ( $state === 'ib' ) {
1343                                                         $output .= '</b></i><b>'; $state = 'b';
1344                                                 } elseif ( $state === 'both' ) {
1345                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1346                                                 } else { # $state can be 'b' or ''
1347                                                         $output .= '<i>'; $state .= 'i';
1348                                                 }
1349                                         } elseif ( strlen( $r ) == 3 ) {
1350                                                 if ( $state === 'b' ) {
1351                                                         $output .= '</b>'; $state = '';
1352                                                 } elseif ( $state === 'bi' ) {
1353                                                         $output .= '</i></b><i>'; $state = 'i';
1354                                                 } elseif ( $state === 'ib' ) {
1355                                                         $output .= '</b>'; $state = 'i';
1356                                                 } elseif ( $state === 'both' ) {
1357                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1358                                                 } else { # $state can be 'i' or ''
1359                                                         $output .= '<b>'; $state .= 'b';
1360                                                 }
1361                                         } elseif ( strlen( $r ) == 5 ) {
1362                                                 if ( $state === 'b' ) {
1363                                                         $output .= '</b><i>'; $state = 'i';
1364                                                 } elseif ( $state === 'i' ) {
1365                                                         $output .= '</i><b>'; $state = 'b';
1366                                                 } elseif ( $state === 'bi' ) {
1367                                                         $output .= '</i></b>'; $state = '';
1368                                                 } elseif ( $state === 'ib' ) {
1369                                                         $output .= '</b></i>'; $state = '';
1370                                                 } elseif ( $state === 'both' ) {
1371                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1372                                                 } else { # ($state == '')
1373                                                         $buffer = ''; $state = 'both';
1374                                                 }
1375                                         }
1376                                 }
1377                                 $i++;
1378                         }
1379                         # Now close all remaining tags.  Notice that the order is important.
1380                         if ( $state === 'b' || $state === 'ib' ) {
1381                                 $output .= '</b>';
1382                         }
1383                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1384                                 $output .= '</i>';
1385                         }
1386                         if ( $state === 'bi' ) {
1387                                 $output .= '</b>';
1388                         }
1389                         # There might be lonely ''''', so make sure we have a buffer
1390                         if ( $state === 'both' && $buffer ) {
1391                                 $output .= '<b><i>'.$buffer.'</i></b>';
1392                         }
1393                         return $output;
1394                 }
1395         }
1396
1397         /**
1398          * Replace external links (REL)
1399          *
1400          * Note: this is all very hackish and the order of execution matters a lot.
1401          * Make sure to run maintenance/parserTests.php if you change this code.
1402          *
1403          * @private
1404          */
1405         function replaceExternalLinks( $text ) {
1406                 global $wgContLang;
1407                 wfProfileIn( __METHOD__ );
1408
1409                 $sk = $this->mOptions->getSkin( $this->mTitle );
1410
1411                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1412                 $s = array_shift( $bits );
1413
1414                 $i = 0;
1415                 while ( $i<count( $bits ) ) {
1416                         $url = $bits[$i++];
1417                         $protocol = $bits[$i++];
1418                         $text = $bits[$i++];
1419                         $trail = $bits[$i++];
1420
1421                         # The characters '<' and '>' (which were escaped by
1422                         # removeHTMLtags()) should not be included in
1423                         # URLs, per RFC 2396.
1424                         $m2 = array();
1425                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1426                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1427                                 $url = substr( $url, 0, $m2[0][1] );
1428                         }
1429
1430                         # If the link text is an image URL, replace it with an <img> tag
1431                         # This happened by accident in the original parser, but some people used it extensively
1432                         $img = $this->maybeMakeExternalImage( $text );
1433                         if ( $img !== false ) {
1434                                 $text = $img;
1435                         }
1436
1437                         $dtrail = '';
1438
1439                         # Set linktype for CSS - if URL==text, link is essentially free
1440                         $linktype = ( $text === $url ) ? 'free' : 'text';
1441
1442                         # No link text, e.g. [http://domain.tld/some.link]
1443                         if ( $text == '' ) {
1444                                 # Autonumber if allowed. See bug #5918
1445                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1446                                         $langObj = $this->getFunctionLang();
1447                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1448                                         $linktype = 'autonumber';
1449                                 } else {
1450                                         # Otherwise just use the URL
1451                                         $text = htmlspecialchars( $url );
1452                                         $linktype = 'free';
1453                                 }
1454                         } else {
1455                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1456                                 # Check for trail
1457                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1458                         }
1459
1460                         $text = $wgContLang->markNoConversion( $text );
1461
1462                         $url = Sanitizer::cleanUrl( $url );
1463
1464                         # Use the encoded URL
1465                         # This means that users can paste URLs directly into the text
1466                         # Funny characters like ö aren't valid in URLs anyway
1467                         # This was changed in August 2004
1468                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1469                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1470
1471                         # Register link in the output object.
1472                         # Replace unnecessary URL escape codes with the referenced character
1473                         # This prevents spammers from hiding links from the filters
1474                         $pasteurized = self::replaceUnusualEscapes( $url );
1475                         $this->mOutput->addExternalLink( $pasteurized );
1476                 }
1477
1478                 wfProfileOut( __METHOD__ );
1479                 return $s;
1480         }
1481
1482         /**
1483          * Get an associative array of additional HTML attributes appropriate for a
1484          * particular external link.  This currently may include rel => nofollow
1485          * (depending on configuration, namespace, and the URL's domain) and/or a
1486          * target attribute (depending on configuration).
1487          *
1488          * @param $url String: optional URL, to extract the domain from for rel =>
1489          *   nofollow if appropriate
1490          * @return Array: associative array of HTML attributes
1491          */
1492         function getExternalLinkAttribs( $url = false ) {
1493                 $attribs = array();
1494                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1495                 $ns = $this->mTitle->getNamespace();
1496                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1497                         $attribs['rel'] = 'nofollow';
1498
1499                         global $wgNoFollowDomainExceptions;
1500                         if ( $wgNoFollowDomainExceptions ) {
1501                                 $bits = wfParseUrl( $url );
1502                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1503                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1504                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1505                                                         unset( $attribs['rel'] );
1506                                                         break;
1507                                                 }
1508                                         }
1509                                 }
1510                         }
1511                 }
1512                 if ( $this->mOptions->getExternalLinkTarget() ) {
1513                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1514                 }
1515                 return $attribs;
1516         }
1517
1518
1519         /**
1520          * Replace unusual URL escape codes with their equivalent characters
1521          *
1522          * @param $url String
1523          * @return String
1524          *
1525          * @todo  This can merge genuinely required bits in the path or query string,
1526          *        breaking legit URLs. A proper fix would treat the various parts of
1527          *        the URL differently; as a workaround, just use the output for
1528          *        statistical records, not for actual linking/output.
1529          */
1530         static function replaceUnusualEscapes( $url ) {
1531                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1532                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1533         }
1534
1535         /**
1536          * Callback function used in replaceUnusualEscapes().
1537          * Replaces unusual URL escape codes with their equivalent character
1538          */
1539         private static function replaceUnusualEscapesCallback( $matches ) {
1540                 $char = urldecode( $matches[0] );
1541                 $ord = ord( $char );
1542                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1543                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1544                         # No, shouldn't be escaped
1545                         return $char;
1546                 } else {
1547                         # Yes, leave it escaped
1548                         return $matches[0];
1549                 }
1550         }
1551
1552         /**
1553          * make an image if it's allowed, either through the global
1554          * option, through the exception, or through the on-wiki whitelist
1555          * @private
1556          */
1557         function maybeMakeExternalImage( $url ) {
1558                 $sk = $this->mOptions->getSkin( $this->mTitle );
1559                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1560                 $imagesexception = !empty( $imagesfrom );
1561                 $text = false;
1562                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1563                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1564                         $imagematch = false;
1565                         foreach ( $imagesfrom as $match ) {
1566                                 if ( strpos( $url, $match ) === 0 ) {
1567                                         $imagematch = true;
1568                                         break;
1569                                 }
1570                         }
1571                 } elseif ( $imagesexception ) {
1572                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1573                 } else {
1574                         $imagematch = false;
1575                 }
1576                 if ( $this->mOptions->getAllowExternalImages()
1577                      || ( $imagesexception && $imagematch ) ) {
1578                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1579                                 # Image found
1580                                 $text = $sk->makeExternalImage( $url );
1581                         }
1582                 }
1583                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1584                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1585                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1586                         foreach ( $whitelist as $entry ) {
1587                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1588                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1589                                         continue;
1590                                 }
1591                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1592                                         # Image matches a whitelist entry
1593                                         $text = $sk->makeExternalImage( $url );
1594                                         break;
1595                                 }
1596                         }
1597                 }
1598                 return $text;
1599         }
1600
1601         /**
1602          * Process [[ ]] wikilinks
1603          * @return String: processed text
1604          *
1605          * @private
1606          */
1607         function replaceInternalLinks( $s ) {
1608                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1609                 return $s;
1610         }
1611
1612         /**
1613          * Process [[ ]] wikilinks (RIL)
1614          * @return LinkHolderArray
1615          *
1616          * @private
1617          */
1618         function replaceInternalLinks2( &$s ) {
1619                 global $wgContLang;
1620
1621                 wfProfileIn( __METHOD__ );
1622
1623                 wfProfileIn( __METHOD__.'-setup' );
1624                 static $tc = FALSE, $e1, $e1_img;
1625                 # the % is needed to support urlencoded titles as well
1626                 if ( !$tc ) {
1627                         $tc = Title::legalChars() . '#%';
1628                         # Match a link having the form [[namespace:link|alternate]]trail
1629                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1630                         # Match cases where there is no "]]", which might still be images
1631                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1632                 }
1633
1634                 $sk = $this->mOptions->getSkin( $this->mTitle );
1635                 $holders = new LinkHolderArray( $this );
1636
1637                 # split the entire text string on occurences of [[
1638                 $a = StringUtils::explode( '[[', ' ' . $s );
1639                 # get the first element (all text up to first [[), and remove the space we added
1640                 $s = $a->current();
1641                 $a->next();
1642                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1643                 $s = substr( $s, 1 );
1644
1645                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1646                 $e2 = null;
1647                 if ( $useLinkPrefixExtension ) {
1648                         # Match the end of a line for a word that's not followed by whitespace,
1649                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1650                         $e2 = wfMsgForContent( 'linkprefix' );
1651                 }
1652
1653                 if ( is_null( $this->mTitle ) ) {
1654                         wfProfileOut( __METHOD__.'-setup' );
1655                         wfProfileOut( __METHOD__ );
1656                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1657                 }
1658                 $nottalk = !$this->mTitle->isTalkPage();
1659
1660                 if ( $useLinkPrefixExtension ) {
1661                         $m = array();
1662                         if ( preg_match( $e2, $s, $m ) ) {
1663                                 $first_prefix = $m[2];
1664                         } else {
1665                                 $first_prefix = false;
1666                         }
1667                 } else {
1668                         $prefix = '';
1669                 }
1670
1671                 if ( $wgContLang->hasVariants() ) {
1672                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1673                 } else {
1674                         $selflink = array( $this->mTitle->getPrefixedText() );
1675                 }
1676                 $useSubpages = $this->areSubpagesAllowed();
1677                 wfProfileOut( __METHOD__.'-setup' );
1678
1679                 # Loop for each link
1680                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1681                         # Check for excessive memory usage
1682                         if ( $holders->isBig() ) {
1683                                 # Too big
1684                                 # Do the existence check, replace the link holders and clear the array
1685                                 $holders->replace( $s );
1686                                 $holders->clear();
1687                         }
1688
1689                         if ( $useLinkPrefixExtension ) {
1690                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1691                                 if ( preg_match( $e2, $s, $m ) ) {
1692                                         $prefix = $m[2];
1693                                         $s = $m[1];
1694                                 } else {
1695                                         $prefix='';
1696                                 }
1697                                 # first link
1698                                 if ( $first_prefix ) {
1699                                         $prefix = $first_prefix;
1700                                         $first_prefix = false;
1701                                 }
1702                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1703                         }
1704
1705                         $might_be_img = false;
1706
1707                         wfProfileIn( __METHOD__."-e1" );
1708                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1709                                 $text = $m[2];
1710                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1711                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1712                                 # the real problem is with the $e1 regex
1713                                 # See bug 1300.
1714                                 #
1715                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1716                                 # and no image is in sight. See bug 2095.
1717                                 #
1718                                 if ( $text !== '' &&
1719                                         substr( $m[3], 0, 1 ) === ']' &&
1720                                         strpos( $text, '[' ) !== false
1721                                 )
1722                                 {
1723                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1724                                         $m[3] = substr( $m[3], 1 );
1725                                 }
1726                                 # fix up urlencoded title texts
1727                                 if ( strpos( $m[1], '%' ) !== false ) {
1728                                         # Should anchors '#' also be rejected?
1729                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode( $m[1] ) );
1730                                 }
1731                                 $trail = $m[3];
1732                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1733                                 $might_be_img = true;
1734                                 $text = $m[2];
1735                                 if ( strpos( $m[1], '%' ) !== false ) {
1736                                         $m[1] = urldecode( $m[1] );
1737                                 }
1738                                 $trail = "";
1739                         } else { # Invalid form; output directly
1740                                 $s .= $prefix . '[[' . $line ;
1741                                 wfProfileOut( __METHOD__."-e1" );
1742                                 continue;
1743                         }
1744                         wfProfileOut( __METHOD__."-e1" );
1745                         wfProfileIn( __METHOD__."-misc" );
1746
1747                         # Don't allow internal links to pages containing
1748                         # PROTO: where PROTO is a valid URL protocol; these
1749                         # should be external links.
1750                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1751                                 $s .= $prefix . '[[' . $line ;
1752                                 wfProfileOut( __METHOD__."-misc" );
1753                                 continue;
1754                         }
1755
1756                         # Make subpage if necessary
1757                         if ( $useSubpages ) {
1758                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1759                         } else {
1760                                 $link = $m[1];
1761                         }
1762
1763                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1764                         if ( !$noforce ) {
1765                                 # Strip off leading ':'
1766                                 $link = substr( $link, 1 );
1767                         }
1768
1769                         wfProfileOut( __METHOD__."-misc" );
1770                         wfProfileIn( __METHOD__."-title" );
1771                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1772                         if ( $nt === null ) {
1773                                 $s .= $prefix . '[[' . $line;
1774                                 wfProfileOut( __METHOD__."-title" );
1775                                 continue;
1776                         }
1777
1778                         $ns = $nt->getNamespace();
1779                         $iw = $nt->getInterWiki();
1780                         wfProfileOut( __METHOD__."-title" );
1781
1782                         if ( $might_be_img ) { # if this is actually an invalid link
1783                                 wfProfileIn( __METHOD__."-might_be_img" );
1784                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1785                                         $found = false;
1786                                         while ( true ) {
1787                                                 # look at the next 'line' to see if we can close it there
1788                                                 $a->next();
1789                                                 $next_line = $a->current();
1790                                                 if ( $next_line === false || $next_line === null ) {
1791                                                         break;
1792                                                 }
1793                                                 $m = explode( ']]', $next_line, 3 );
1794                                                 if ( count( $m ) == 3 ) {
1795                                                         # the first ]] closes the inner link, the second the image
1796                                                         $found = true;
1797                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1798                                                         $trail = $m[2];
1799                                                         break;
1800                                                 } elseif ( count( $m ) == 2 ) {
1801                                                         # if there's exactly one ]] that's fine, we'll keep looking
1802                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1803                                                 } else {
1804                                                         # if $next_line is invalid too, we need look no further
1805                                                         $text .= '[[' . $next_line;
1806                                                         break;
1807                                                 }
1808                                         }
1809                                         if ( !$found ) {
1810                                                 # we couldn't find the end of this imageLink, so output it raw
1811                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1812                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1813                                                 $s .= "{$prefix}[[$link|$text";
1814                                                 # note: no $trail, because without an end, there *is* no trail
1815                                                 wfProfileOut( __METHOD__."-might_be_img" );
1816                                                 continue;
1817                                         }
1818                                 } else { # it's not an image, so output it raw
1819                                         $s .= "{$prefix}[[$link|$text";
1820                                         # note: no $trail, because without an end, there *is* no trail
1821                                         wfProfileOut( __METHOD__."-might_be_img" );
1822                                         continue;
1823                                 }
1824                                 wfProfileOut( __METHOD__."-might_be_img" );
1825                         }
1826
1827                         $wasblank = ( $text  == '' );
1828                         if ( $wasblank ) {
1829                                 $text = $link;
1830                         } else {
1831                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1832                                 # [[Lista d''e paise d''o munno]] -> <a href="">Lista d''e paise d''o munno</a>
1833                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']] -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1834                                 $text = $this->doQuotes($text);
1835                         }
1836
1837                         # Link not escaped by : , create the various objects
1838                         if ( $noforce ) {
1839
1840                                 # Interwikis
1841                                 wfProfileIn( __METHOD__."-interwiki" );
1842                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1843                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1844                                         $s = rtrim( $s . $prefix );
1845                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1846                                         wfProfileOut( __METHOD__."-interwiki" );
1847                                         continue;
1848                                 }
1849                                 wfProfileOut( __METHOD__."-interwiki" );
1850
1851                                 if ( $ns == NS_FILE ) {
1852                                         wfProfileIn( __METHOD__."-image" );
1853                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1854                                                 if ( $wasblank ) {
1855                                                         # if no parameters were passed, $text
1856                                                         # becomes something like "File:Foo.png",
1857                                                         # which we don't want to pass on to the
1858                                                         # image generator
1859                                                         $text = '';
1860                                                 } else {
1861                                                         # recursively parse links inside the image caption
1862                                                         # actually, this will parse them in any other parameters, too,
1863                                                         # but it might be hard to fix that, and it doesn't matter ATM
1864                                                         $text = $this->replaceExternalLinks( $text );
1865                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1866                                                 }
1867                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1868                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1869                                         } else {
1870                                                 $s .= $prefix . $trail;
1871                                         }
1872                                         $this->mOutput->addImage( $nt->getDBkey() );
1873                                         wfProfileOut( __METHOD__."-image" );
1874                                         continue;
1875
1876                                 }
1877
1878                                 if ( $ns == NS_CATEGORY ) {
1879                                         wfProfileIn( __METHOD__."-category" );
1880                                         $s = rtrim( $s . "\n" ); # bug 87
1881
1882                                         if ( $wasblank ) {
1883                                                 $sortkey = $this->getDefaultSort();
1884                                         } else {
1885                                                 $sortkey = $text;
1886                                         }
1887                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1888                                         $sortkey = str_replace( "\n", '', $sortkey );
1889                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1890                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1891
1892                                         /**
1893                                          * Strip the whitespace Category links produce, see bug 87
1894                                          * @todo We might want to use trim($tmp, "\n") here.
1895                                          */
1896                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1897
1898                                         wfProfileOut( __METHOD__."-category" );
1899                                         continue;
1900                                 }
1901                         }
1902
1903                         # Self-link checking
1904                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1905                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1906                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1907                                         continue;
1908                                 }
1909                         }
1910
1911                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1912                         # FIXME: Should do batch file existence checks, see comment below
1913                         if ( $ns == NS_MEDIA ) {
1914                                 wfProfileIn( __METHOD__."-media" );
1915                                 # Give extensions a chance to select the file revision for us
1916                                 $skip = $time = false;
1917                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1918                                 if ( $skip ) {
1919                                         $link = $sk->link( $nt );
1920                                 } else {
1921                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1922                                 }
1923                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1924                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1925                                 $this->mOutput->addImage( $nt->getDBkey() );
1926                                 wfProfileOut( __METHOD__."-media" );
1927                                 continue;
1928                         }
1929
1930                         wfProfileIn( __METHOD__."-always_known" );
1931                         # Some titles, such as valid special pages or files in foreign repos, should
1932                         # be shown as bluelinks even though they're not included in the page table
1933                         #
1934                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1935                         # batch file existence checks for NS_FILE and NS_MEDIA
1936                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1937                                 $this->mOutput->addLink( $nt );
1938                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1939                         } else {
1940                                 # Links will be added to the output link list after checking
1941                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1942                         }
1943                         wfProfileOut( __METHOD__."-always_known" );
1944                 }
1945                 wfProfileOut( __METHOD__ );
1946                 return $holders;
1947         }
1948
1949         /**
1950          * Make a link placeholder. The text returned can be later resolved to a real link with
1951          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1952          * parsing of interwiki links, and secondly to allow all existence checks and
1953          * article length checks (for stub links) to be bundled into a single query.
1954          *
1955          * @deprecated
1956          */
1957         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1958                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1959         }
1960
1961         /**
1962          * Render a forced-blue link inline; protect against double expansion of
1963          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1964          * Since this little disaster has to split off the trail text to avoid
1965          * breaking URLs in the following text without breaking trails on the
1966          * wiki links, it's been made into a horrible function.
1967          *
1968          * @param $nt Title
1969          * @param $text String
1970          * @param $query String
1971          * @param $trail String
1972          * @param $prefix String
1973          * @return String: HTML-wikitext mix oh yuck
1974          */
1975         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1976                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1977                 $sk = $this->mOptions->getSkin( $this->mTitle );
1978                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1979                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1980                 return $this->armorLinks( $link ) . $trail;
1981         }
1982
1983         /**
1984          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1985          * going to go through further parsing steps before inline URL expansion.
1986          *
1987          * Not needed quite as much as it used to be since free links are a bit
1988          * more sensible these days. But bracketed links are still an issue.
1989          *
1990          * @param $text String: more-or-less HTML
1991          * @return String: less-or-more HTML with NOPARSE bits
1992          */
1993         function armorLinks( $text ) {
1994                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1995                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1996         }
1997
1998         /**
1999          * Return true if subpage links should be expanded on this page.
2000          * @return Boolean
2001          */
2002         function areSubpagesAllowed() {
2003                 # Some namespaces don't allow subpages
2004                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2005         }
2006
2007         /**
2008          * Handle link to subpage if necessary
2009          *
2010          * @param $target String: the source of the link
2011          * @param &$text String: the link text, modified as necessary
2012          * @return string the full name of the link
2013          * @private
2014          */
2015         function maybeDoSubpageLink( $target, &$text ) {
2016                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2017         }
2018
2019         /**#@+
2020          * Used by doBlockLevels()
2021          * @private
2022          */
2023         function closeParagraph() {
2024                 $result = '';
2025                 if ( $this->mLastSection != '' ) {
2026                         $result = '</' . $this->mLastSection  . ">\n";
2027                 }
2028                 $this->mInPre = false;
2029                 $this->mLastSection = '';
2030                 return $result;
2031         }
2032
2033         /**
2034          * getCommon() returns the length of the longest common substring
2035          * of both arguments, starting at the beginning of both.
2036          * @private
2037          */
2038         function getCommon( $st1, $st2 ) {
2039                 $fl = strlen( $st1 );
2040                 $shorter = strlen( $st2 );
2041                 if ( $fl < $shorter ) {
2042                         $shorter = $fl;
2043                 }
2044
2045                 for ( $i = 0; $i < $shorter; ++$i ) {
2046                         if ( $st1{$i} != $st2{$i} ) {
2047                                 break;
2048                         }
2049                 }
2050                 return $i;
2051         }
2052
2053         /**
2054          * These next three functions open, continue, and close the list
2055          * element appropriate to the prefix character passed into them.
2056          * @private
2057          */
2058         function openList( $char ) {
2059                 $result = $this->closeParagraph();
2060
2061                 if ( '*' === $char ) {
2062                         $result .= '<ul><li>';
2063                 } elseif ( '#' === $char ) {
2064                         $result .= '<ol><li>';
2065                 } elseif ( ':' === $char ) {
2066                         $result .= '<dl><dd>';
2067                 } elseif ( ';' === $char ) {
2068                         $result .= '<dl><dt>';
2069                         $this->mDTopen = true;
2070                 } else {
2071                         $result = '<!-- ERR 1 -->';
2072                 }
2073
2074                 return $result;
2075         }
2076
2077         /**
2078          * TODO: document
2079          * @param $char String
2080          * @private
2081          */
2082         function nextItem( $char ) {
2083                 if ( '*' === $char || '#' === $char ) {
2084                         return '</li><li>';
2085                 } elseif ( ':' === $char || ';' === $char ) {
2086                         $close = '</dd>';
2087                         if ( $this->mDTopen ) {
2088                                 $close = '</dt>';
2089                         }
2090                         if ( ';' === $char ) {
2091                                 $this->mDTopen = true;
2092                                 return $close . '<dt>';
2093                         } else {
2094                                 $this->mDTopen = false;
2095                                 return $close . '<dd>';
2096                         }
2097                 }
2098                 return '<!-- ERR 2 -->';
2099         }
2100
2101         /**
2102          * TODO: document
2103          * @param $char String
2104          * @private
2105          */
2106         function closeList( $char ) {
2107                 if ( '*' === $char ) {
2108                         $text = '</li></ul>';
2109                 } elseif ( '#' === $char ) {
2110                         $text = '</li></ol>';
2111                 } elseif ( ':' === $char ) {
2112                         if ( $this->mDTopen ) {
2113                                 $this->mDTopen = false;
2114                                 $text = '</dt></dl>';
2115                         } else {
2116                                 $text = '</dd></dl>';
2117                         }
2118                 } else {
2119                         return '<!-- ERR 3 -->';
2120                 }
2121                 return $text."\n";
2122         }
2123         /**#@-*/
2124
2125         /**
2126          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2127          *
2128          * @param $text String
2129          * @param $linestart Boolean: whether or not this is at the start of a line.
2130          * @private
2131          * @return string the lists rendered as HTML
2132          */
2133         function doBlockLevels( $text, $linestart ) {
2134                 wfProfileIn( __METHOD__ );
2135
2136                 # Parsing through the text line by line.  The main thing
2137                 # happening here is handling of block-level elements p, pre,
2138                 # and making lists from lines starting with * # : etc.
2139                 #
2140                 $textLines = StringUtils::explode( "\n", $text );
2141
2142                 $lastPrefix = $output = '';
2143                 $this->mDTopen = $inBlockElem = false;
2144                 $prefixLength = 0;
2145                 $paragraphStack = false;
2146
2147                 foreach ( $textLines as $oLine ) {
2148                         # Fix up $linestart
2149                         if ( !$linestart ) {
2150                                 $output .= $oLine;
2151                                 $linestart = true;
2152                                 continue;
2153                         }
2154                         # * = ul
2155                         # # = ol
2156                         # ; = dt
2157                         # : = dd
2158
2159                         $lastPrefixLength = strlen( $lastPrefix );
2160                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2161                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2162                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2163                         if ( !$this->mInPre ) {
2164                                 # Multiple prefixes may abut each other for nested lists.
2165                                 $prefixLength = strspn( $oLine, '*#:;' );
2166                                 $prefix = substr( $oLine, 0, $prefixLength );
2167
2168                                 # eh?
2169                                 # ; and : are both from definition-lists, so they're equivalent
2170                                 #  for the purposes of determining whether or not we need to open/close
2171                                 #  elements.
2172                                 $prefix2 = str_replace( ';', ':', $prefix );
2173                                 $t = substr( $oLine, $prefixLength );
2174                                 $this->mInPre = (bool)$preOpenMatch;
2175                         } else {
2176                                 # Don't interpret any other prefixes in preformatted text
2177                                 $prefixLength = 0;
2178                                 $prefix = $prefix2 = '';
2179                                 $t = $oLine;
2180                         }
2181
2182                         # List generation
2183                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2184                                 # Same as the last item, so no need to deal with nesting or opening stuff
2185                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2186                                 $paragraphStack = false;
2187
2188                                 if ( substr( $prefix, -1 ) === ';') {
2189                                         # The one nasty exception: definition lists work like this:
2190                                         # ; title : definition text
2191                                         # So we check for : in the remainder text to split up the
2192                                         # title and definition, without b0rking links.
2193                                         $term = $t2 = '';
2194                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2195                                                 $t = $t2;
2196                                                 $output .= $term . $this->nextItem( ':' );
2197                                         }
2198                                 }
2199                         } elseif ( $prefixLength || $lastPrefixLength ) {
2200                                 # We need to open or close prefixes, or both.
2201
2202                                 # Either open or close a level...
2203                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2204                                 $paragraphStack = false;
2205
2206                                 # Close all the prefixes which aren't shared.
2207                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2208                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2209                                         --$lastPrefixLength;
2210                                 }
2211
2212                                 # Continue the current prefix if appropriate.
2213                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2214                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2215                                 }
2216
2217                                 # Open prefixes where appropriate.
2218                                 while ( $prefixLength > $commonPrefixLength ) {
2219                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2220                                         $output .= $this->openList( $char );
2221
2222                                         if ( ';' === $char ) {
2223                                                 # FIXME: This is dupe of code above
2224                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2225                                                         $t = $t2;
2226                                                         $output .= $term . $this->nextItem( ':' );
2227                                                 }
2228                                         }
2229                                         ++$commonPrefixLength;
2230                                 }
2231                                 $lastPrefix = $prefix2;
2232                         }
2233
2234                         # If we have no prefixes, go to paragraph mode.
2235                         if ( 0 == $prefixLength ) {
2236                                 wfProfileIn( __METHOD__."-paragraph" );
2237                                 # No prefix (not in list)--go to paragraph mode
2238                                 # XXX: use a stack for nestable elements like span, table and div
2239                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2240                                 $closematch = preg_match(
2241                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2242                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2243                                 if ( $openmatch or $closematch ) {
2244                                         $paragraphStack = false;
2245                                         # TODO bug 5718: paragraph closed
2246                                         $output .= $this->closeParagraph();
2247                                         if ( $preOpenMatch and !$preCloseMatch ) {
2248                                                 $this->mInPre = true;
2249                                         }
2250                                         $inBlockElem = !$closematch;
2251                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2252                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2253                                                 # pre
2254                                                 if ( $this->mLastSection !== 'pre' ) {
2255                                                         $paragraphStack = false;
2256                                                         $output .= $this->closeParagraph().'<pre>';
2257                                                         $this->mLastSection = 'pre';
2258                                                 }
2259                                                 $t = substr( $t, 1 );
2260                                         } else {
2261                                                 # paragraph
2262                                                 if ( trim( $t ) === '' ) {
2263                                                         if ( $paragraphStack ) {
2264                                                                 $output .= $paragraphStack.'<br />';
2265                                                                 $paragraphStack = false;
2266                                                                 $this->mLastSection = 'p';
2267                                                         } else {
2268                                                                 if ( $this->mLastSection !== 'p' ) {
2269                                                                         $output .= $this->closeParagraph();
2270                                                                         $this->mLastSection = '';
2271                                                                         $paragraphStack = '<p>';
2272                                                                 } else {
2273                                                                         $paragraphStack = '</p><p>';
2274                                                                 }
2275                                                         }
2276                                                 } else {
2277                                                         if ( $paragraphStack ) {
2278                                                                 $output .= $paragraphStack;
2279                                                                 $paragraphStack = false;
2280                                                                 $this->mLastSection = 'p';
2281                                                         } elseif ( $this->mLastSection !== 'p' ) {
2282                                                                 $output .= $this->closeParagraph().'<p>';
2283                                                                 $this->mLastSection = 'p';
2284                                                         }
2285                                                 }
2286                                         }
2287                                 }
2288                                 wfProfileOut( __METHOD__."-paragraph" );
2289                         }
2290                         # somewhere above we forget to get out of pre block (bug 785)
2291                         if ( $preCloseMatch && $this->mInPre ) {
2292                                 $this->mInPre = false;
2293                         }
2294                         if ( $paragraphStack === false ) {
2295                                 $output .= $t."\n";
2296                         }
2297                 }
2298                 while ( $prefixLength ) {
2299                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2300                         --$prefixLength;
2301                 }
2302                 if ( $this->mLastSection != '' ) {
2303                         $output .= '</' . $this->mLastSection . '>';
2304                         $this->mLastSection = '';
2305                 }
2306
2307                 wfProfileOut( __METHOD__ );
2308                 return $output;
2309         }
2310
2311         /**
2312          * Split up a string on ':', ignoring any occurences inside tags
2313          * to prevent illegal overlapping.
2314          *
2315          * @param $str String: the string to split
2316          * @param &$before String: set to everything before the ':'
2317          * @param &$after String: set to everything after the ':'
2318          * return String: the position of the ':', or false if none found
2319          */
2320         function findColonNoLinks( $str, &$before, &$after ) {
2321                 wfProfileIn( __METHOD__ );
2322
2323                 $pos = strpos( $str, ':' );
2324                 if ( $pos === false ) {
2325                         # Nothing to find!
2326                         wfProfileOut( __METHOD__ );
2327                         return false;
2328                 }
2329
2330                 $lt = strpos( $str, '<' );
2331                 if ( $lt === false || $lt > $pos ) {
2332                         # Easy; no tag nesting to worry about
2333                         $before = substr( $str, 0, $pos );
2334                         $after = substr( $str, $pos+1 );
2335                         wfProfileOut( __METHOD__ );
2336                         return $pos;
2337                 }
2338
2339                 # Ugly state machine to walk through avoiding tags.
2340                 $state = self::COLON_STATE_TEXT;
2341                 $stack = 0;
2342                 $len = strlen( $str );
2343                 for( $i = 0; $i < $len; $i++ ) {
2344                         $c = $str{$i};
2345
2346                         switch( $state ) {
2347                         # (Using the number is a performance hack for common cases)
2348                         case 0: # self::COLON_STATE_TEXT:
2349                                 switch( $c ) {
2350                                 case "<":
2351                                         # Could be either a <start> tag or an </end> tag
2352                                         $state = self::COLON_STATE_TAGSTART;
2353                                         break;
2354                                 case ":":
2355                                         if ( $stack == 0 ) {
2356                                                 # We found it!
2357                                                 $before = substr( $str, 0, $i );
2358                                                 $after = substr( $str, $i + 1 );
2359                                                 wfProfileOut( __METHOD__ );
2360                                                 return $i;
2361                                         }
2362                                         # Embedded in a tag; don't break it.
2363                                         break;
2364                                 default:
2365                                         # Skip ahead looking for something interesting
2366                                         $colon = strpos( $str, ':', $i );
2367                                         if ( $colon === false ) {
2368                                                 # Nothing else interesting
2369                                                 wfProfileOut( __METHOD__ );
2370                                                 return false;
2371                                         }
2372                                         $lt = strpos( $str, '<', $i );
2373                                         if ( $stack === 0 ) {
2374                                                 if ( $lt === false || $colon < $lt ) {
2375                                                         # We found it!
2376                                                         $before = substr( $str, 0, $colon );
2377                                                         $after = substr( $str, $colon + 1 );
2378                                                         wfProfileOut( __METHOD__ );
2379                                                         return $i;
2380                                                 }
2381                                         }
2382                                         if ( $lt === false ) {
2383                                                 # Nothing else interesting to find; abort!
2384                                                 # We're nested, but there's no close tags left. Abort!
2385                                                 break 2;
2386                                         }
2387                                         # Skip ahead to next tag start
2388                                         $i = $lt;
2389                                         $state = self::COLON_STATE_TAGSTART;
2390                                 }
2391                                 break;
2392                         case 1: # self::COLON_STATE_TAG:
2393                                 # In a <tag>
2394                                 switch( $c ) {
2395                                 case ">":
2396                                         $stack++;
2397                                         $state = self::COLON_STATE_TEXT;
2398                                         break;
2399                                 case "/":
2400                                         # Slash may be followed by >?
2401                                         $state = self::COLON_STATE_TAGSLASH;
2402                                         break;
2403                                 default:
2404                                         # ignore
2405                                 }
2406                                 break;
2407                         case 2: # self::COLON_STATE_TAGSTART:
2408                                 switch( $c ) {
2409                                 case "/":
2410                                         $state = self::COLON_STATE_CLOSETAG;
2411                                         break;
2412                                 case "!":
2413                                         $state = self::COLON_STATE_COMMENT;
2414                                         break;
2415                                 case ">":
2416                                         # Illegal early close? This shouldn't happen D:
2417                                         $state = self::COLON_STATE_TEXT;
2418                                         break;
2419                                 default:
2420                                         $state = self::COLON_STATE_TAG;
2421                                 }
2422                                 break;
2423                         case 3: # self::COLON_STATE_CLOSETAG:
2424                                 # In a </tag>
2425                                 if ( $c === ">" ) {
2426                                         $stack--;
2427                                         if ( $stack < 0 ) {
2428                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2429                                                 wfProfileOut( __METHOD__ );
2430                                                 return false;
2431                                         }
2432                                         $state = self::COLON_STATE_TEXT;
2433                                 }
2434                                 break;
2435                         case self::COLON_STATE_TAGSLASH:
2436                                 if ( $c === ">" ) {
2437                                         # Yes, a self-closed tag <blah/>
2438                                         $state = self::COLON_STATE_TEXT;
2439                                 } else {
2440                                         # Probably we're jumping the gun, and this is an attribute
2441                                         $state = self::COLON_STATE_TAG;
2442                                 }
2443                                 break;
2444                         case 5: # self::COLON_STATE_COMMENT:
2445                                 if ( $c === "-" ) {
2446                                         $state = self::COLON_STATE_COMMENTDASH;
2447                                 }
2448                                 break;
2449                         case self::COLON_STATE_COMMENTDASH:
2450                                 if ( $c === "-" ) {
2451                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2452                                 } else {
2453                                         $state = self::COLON_STATE_COMMENT;
2454                                 }
2455                                 break;
2456                         case self::COLON_STATE_COMMENTDASHDASH:
2457                                 if ( $c === ">" ) {
2458                                         $state = self::COLON_STATE_TEXT;
2459                                 } else {
2460                                         $state = self::COLON_STATE_COMMENT;
2461                                 }
2462                                 break;
2463                         default:
2464                                 throw new MWException( "State machine error in " . __METHOD__ );
2465                         }
2466                 }
2467                 if ( $stack > 0 ) {
2468                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2469                         return false;
2470                 }
2471                 wfProfileOut( __METHOD__ );
2472                 return false;
2473         }
2474
2475         /**
2476          * Return value of a magic variable (like PAGENAME)
2477          *
2478          * @private
2479          */
2480         function getVariableValue( $index, $frame=false ) {
2481                 global $wgContLang, $wgSitename, $wgServer;
2482                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2483
2484                 /**
2485                  * Some of these require message or data lookups and can be
2486                  * expensive to check many times.
2487                  */
2488                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2489                         if ( isset( $this->mVarCache[$index] ) ) {
2490                                 return $this->mVarCache[$index];
2491                         }
2492                 }
2493
2494                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2495                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2496
2497                 # Use the time zone
2498                 global $wgLocaltimezone;
2499                 if ( isset( $wgLocaltimezone ) ) {
2500                         $oldtz = date_default_timezone_get();
2501                         date_default_timezone_set( $wgLocaltimezone );
2502                 }
2503
2504                 $localTimestamp = date( 'YmdHis', $ts );
2505                 $localMonth = date( 'm', $ts );
2506                 $localMonth1 = date( 'n', $ts );
2507                 $localMonthName = date( 'n', $ts );
2508                 $localDay = date( 'j', $ts );
2509                 $localDay2 = date( 'd', $ts );
2510                 $localDayOfWeek = date( 'w', $ts );
2511                 $localWeek = date( 'W', $ts );
2512                 $localYear = date( 'Y', $ts );
2513                 $localHour = date( 'H', $ts );
2514                 if ( isset( $wgLocaltimezone ) ) {
2515                         date_default_timezone_set( $oldtz );
2516                 }
2517
2518                 switch ( $index ) {
2519                         case 'currentmonth':
2520                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2521                                 break;
2522                         case 'currentmonth1':
2523                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2524                                 break;
2525                         case 'currentmonthname':
2526                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2527                                 break;
2528                         case 'currentmonthnamegen':
2529                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2530                                 break;
2531                         case 'currentmonthabbrev':
2532                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2533                                 break;
2534                         case 'currentday':
2535                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2536                                 break;
2537                         case 'currentday2':
2538                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2539                                 break;
2540                         case 'localmonth':
2541                                 $value = $wgContLang->formatNum( $localMonth );
2542                                 break;
2543                         case 'localmonth1':
2544                                 $value = $wgContLang->formatNum( $localMonth1 );
2545                                 break;
2546                         case 'localmonthname':
2547                                 $value = $wgContLang->getMonthName( $localMonthName );
2548                                 break;
2549                         case 'localmonthnamegen':
2550                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2551                                 break;
2552                         case 'localmonthabbrev':
2553                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2554                                 break;
2555                         case 'localday':
2556                                 $value = $wgContLang->formatNum( $localDay );
2557                                 break;
2558                         case 'localday2':
2559                                 $value = $wgContLang->formatNum( $localDay2 );
2560                                 break;
2561                         case 'pagename':
2562                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2563                                 break;
2564                         case 'pagenamee':
2565                                 $value = $this->mTitle->getPartialURL();
2566                                 break;
2567                         case 'fullpagename':
2568                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2569                                 break;
2570                         case 'fullpagenamee':
2571                                 $value = $this->mTitle->getPrefixedURL();
2572                                 break;
2573                         case 'subpagename':
2574                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2575                                 break;
2576                         case 'subpagenamee':
2577                                 $value = $this->mTitle->getSubpageUrlForm();
2578                                 break;
2579                         case 'basepagename':
2580                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2581                                 break;
2582                         case 'basepagenamee':
2583                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2584                                 break;
2585                         case 'talkpagename':
2586                                 if ( $this->mTitle->canTalk() ) {
2587                                         $talkPage = $this->mTitle->getTalkPage();
2588                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2589                                 } else {
2590                                         $value = '';
2591                                 }
2592                                 break;
2593                         case 'talkpagenamee':
2594                                 if ( $this->mTitle->canTalk() ) {
2595                                         $talkPage = $this->mTitle->getTalkPage();
2596                                         $value = $talkPage->getPrefixedUrl();
2597                                 } else {
2598                                         $value = '';
2599                                 }
2600                                 break;
2601                         case 'subjectpagename':
2602                                 $subjPage = $this->mTitle->getSubjectPage();
2603                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2604                                 break;
2605                         case 'subjectpagenamee':
2606                                 $subjPage = $this->mTitle->getSubjectPage();
2607                                 $value = $subjPage->getPrefixedUrl();
2608                                 break;
2609                         case 'revisionid':
2610                                 # Let the edit saving system know we should parse the page
2611                                 # *after* a revision ID has been assigned.
2612                                 $this->mOutput->setFlag( 'vary-revision' );
2613                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2614                                 $value = $this->mRevisionId;
2615                                 break;
2616                         case 'revisionday':
2617                                 # Let the edit saving system know we should parse the page
2618                                 # *after* a revision ID has been assigned. This is for null edits.
2619                                 $this->mOutput->setFlag( 'vary-revision' );
2620                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2621                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2622                                 break;
2623                         case 'revisionday2':
2624                                 # Let the edit saving system know we should parse the page
2625                                 # *after* a revision ID has been assigned. This is for null edits.
2626                                 $this->mOutput->setFlag( 'vary-revision' );
2627                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2628                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2629                                 break;
2630                         case 'revisionmonth':
2631                                 # Let the edit saving system know we should parse the page
2632                                 # *after* a revision ID has been assigned. This is for null edits.
2633                                 $this->mOutput->setFlag( 'vary-revision' );
2634                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2635                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2636                                 break;
2637                         case 'revisionmonth1':
2638                                 # Let the edit saving system know we should parse the page
2639                                 # *after* a revision ID has been assigned. This is for null edits.
2640                                 $this->mOutput->setFlag( 'vary-revision' );
2641                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2642                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2643                                 break;
2644                         case 'revisionyear':
2645                                 # Let the edit saving system know we should parse the page
2646                                 # *after* a revision ID has been assigned. This is for null edits.
2647                                 $this->mOutput->setFlag( 'vary-revision' );
2648                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2649                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2650                                 break;
2651                         case 'revisiontimestamp':
2652                                 # Let the edit saving system know we should parse the page
2653                                 # *after* a revision ID has been assigned. This is for null edits.
2654                                 $this->mOutput->setFlag( 'vary-revision' );
2655                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2656                                 $value = $this->getRevisionTimestamp();
2657                                 break;
2658                         case 'revisionuser':
2659                                 # Let the edit saving system know we should parse the page
2660                                 # *after* a revision ID has been assigned. This is for null edits.
2661                                 $this->mOutput->setFlag( 'vary-revision' );
2662                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2663                                 $value = $this->getRevisionUser();
2664                                 break;
2665                         case 'namespace':
2666                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2667                                 break;
2668                         case 'namespacee':
2669                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2670                                 break;
2671                         case 'talkspace':
2672                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2673                                 break;
2674                         case 'talkspacee':
2675                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2676                                 break;
2677                         case 'subjectspace':
2678                                 $value = $this->mTitle->getSubjectNsText();
2679                                 break;
2680                         case 'subjectspacee':
2681                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2682                                 break;
2683                         case 'currentdayname':
2684                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2685                                 break;
2686                         case 'currentyear':
2687                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2688                                 break;
2689                         case 'currenttime':
2690                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2691                                 break;
2692                         case 'currenthour':
2693                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2694                                 break;
2695                         case 'currentweek':
2696                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2697                                 # int to remove the padding
2698                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2699                                 break;
2700                         case 'currentdow':
2701                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2702                                 break;
2703                         case 'localdayname':
2704                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2705                                 break;
2706                         case 'localyear':
2707                                 $value = $wgContLang->formatNum( $localYear, true );
2708                                 break;
2709                         case 'localtime':
2710                                 $value = $wgContLang->time( $localTimestamp, false, false );
2711                                 break;
2712                         case 'localhour':
2713                                 $value = $wgContLang->formatNum( $localHour, true );
2714                                 break;
2715                         case 'localweek':
2716                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2717                                 # int to remove the padding
2718                                 $value = $wgContLang->formatNum( (int)$localWeek );
2719                                 break;
2720                         case 'localdow':
2721                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2722                                 break;
2723                         case 'numberofarticles':
2724                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2725                                 break;
2726                         case 'numberoffiles':
2727                                 $value = $wgContLang->formatNum( SiteStats::images() );
2728                                 break;
2729                         case 'numberofusers':
2730                                 $value = $wgContLang->formatNum( SiteStats::users() );
2731                                 break;
2732                         case 'numberofactiveusers':
2733                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2734                                 break;
2735                         case 'numberofpages':
2736                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2737                                 break;
2738                         case 'numberofadmins':
2739                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2740                                 break;
2741                         case 'numberofedits':
2742                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2743                                 break;
2744                         case 'numberofviews':
2745                                 $value = $wgContLang->formatNum( SiteStats::views() );
2746                                 break;
2747                         case 'currenttimestamp':
2748                                 $value = wfTimestamp( TS_MW, $ts );
2749                                 break;
2750                         case 'localtimestamp':
2751                                 $value = $localTimestamp;
2752                                 break;
2753                         case 'currentversion':
2754                                 $value = SpecialVersion::getVersion();
2755                                 break;
2756                         case 'articlepath':
2757                                 return $wgArticlePath;
2758                         case 'sitename':
2759                                 return $wgSitename;
2760                         case 'server':
2761                                 return $wgServer;
2762                         case 'servername':
2763                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2764                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2765                                 wfRestoreWarnings();
2766                                 return $serverName ? $serverName : $wgServer;
2767                         case 'scriptpath':
2768                                 return $wgScriptPath;
2769                         case 'stylepath':
2770                                 return $wgStylePath;
2771                         case 'directionmark':
2772                                 return $wgContLang->getDirMark();
2773                         case 'contentlanguage':
2774                                 global $wgLanguageCode;
2775                                 return $wgLanguageCode;
2776                         default:
2777                                 $ret = null;
2778                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2779                                         return $ret;
2780                                 } else {
2781                                         return null;
2782                                 }
2783                 }
2784
2785                 if ( $index )
2786                         $this->mVarCache[$index] = $value;
2787
2788                 return $value;
2789         }
2790
2791         /**
2792          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2793          *
2794          * @private
2795          */
2796         function initialiseVariables() {
2797                 wfProfileIn( __METHOD__ );
2798                 $variableIDs = MagicWord::getVariableIDs();
2799                 $substIDs = MagicWord::getSubstIDs();
2800
2801                 $this->mVariables = new MagicWordArray( $variableIDs );
2802                 $this->mSubstWords = new MagicWordArray( $substIDs );
2803                 wfProfileOut( __METHOD__ );
2804         }
2805
2806         /**
2807          * Preprocess some wikitext and return the document tree.
2808          * This is the ghost of replace_variables().
2809          *
2810          * @param $text String: The text to parse
2811          * @param $flags Integer: bitwise combination of:
2812          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2813          *                                     included. Default is to assume a direct page view.
2814          *
2815          * The generated DOM tree must depend only on the input text and the flags.
2816          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2817          *
2818          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2819          * change in the DOM tree for a given text, must be passed through the section identifier
2820          * in the section edit link and thus back to extractSections().
2821          *
2822          * The output of this function is currently only cached in process memory, but a persistent
2823          * cache may be implemented at a later date which takes further advantage of these strict
2824          * dependency requirements.
2825          *
2826          * @private
2827          */
2828         function preprocessToDom( $text, $flags = 0 ) {
2829                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2830                 return $dom;
2831         }
2832
2833         /**
2834          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2835          */
2836         public static function splitWhitespace( $s ) {
2837                 $ltrimmed = ltrim( $s );
2838                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2839                 $trimmed = rtrim( $ltrimmed );
2840                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2841                 if ( $diff > 0 ) {
2842                         $w2 = substr( $ltrimmed, -$diff );
2843                 } else {
2844                         $w2 = '';
2845                 }
2846                 return array( $w1, $trimmed, $w2 );
2847         }
2848
2849         /**
2850          * Replace magic variables, templates, and template arguments
2851          * with the appropriate text. Templates are substituted recursively,
2852          * taking care to avoid infinite loops.
2853          *
2854          * Note that the substitution depends on value of $mOutputType:
2855          *  self::OT_WIKI: only {{subst:}} templates
2856          *  self::OT_PREPROCESS: templates but not extension tags
2857          *  self::OT_HTML: all templates and extension tags
2858          *
2859          * @param $text String: the text to transform
2860          * @param $frame PPFrame Object describing the arguments passed to the template.
2861          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2862          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2863          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2864          * @private
2865          */
2866         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2867                 # Is there any text? Also, Prevent too big inclusions!
2868                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2869                         return $text;
2870                 }
2871                 wfProfileIn( __METHOD__ );
2872
2873                 if ( $frame === false ) {
2874                         $frame = $this->getPreprocessor()->newFrame();
2875                 } elseif ( !( $frame instanceof PPFrame ) ) {
2876                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2877                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2878                 }
2879
2880                 $dom = $this->preprocessToDom( $text );
2881                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2882                 $text = $frame->expand( $dom, $flags );
2883
2884                 wfProfileOut( __METHOD__ );
2885                 return $text;
2886         }
2887
2888         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2889         static function createAssocArgs( $args ) {
2890                 $assocArgs = array();
2891                 $index = 1;
2892                 foreach ( $args as $arg ) {
2893                         $eqpos = strpos( $arg, '=' );
2894                         if ( $eqpos === false ) {
2895                                 $assocArgs[$index++] = $arg;
2896                         } else {
2897                                 $name = trim( substr( $arg, 0, $eqpos ) );
2898                                 $value = trim( substr( $arg, $eqpos+1 ) );
2899                                 if ( $value === false ) {
2900                                         $value = '';
2901                                 }
2902                                 if ( $name !== false ) {
2903                                         $assocArgs[$name] = $value;
2904                                 }
2905                         }
2906                 }
2907
2908                 return $assocArgs;
2909         }
2910
2911         /**
2912          * Warn the user when a parser limitation is reached
2913          * Will warn at most once the user per limitation type
2914          *
2915          * @param $limitationType String: should be one of:
2916          *   'expensive-parserfunction' (corresponding messages:
2917          *       'expensive-parserfunction-warning',
2918          *       'expensive-parserfunction-category')
2919          *   'post-expand-template-argument' (corresponding messages:
2920          *       'post-expand-template-argument-warning',
2921          *       'post-expand-template-argument-category')
2922          *   'post-expand-template-inclusion' (corresponding messages:
2923          *       'post-expand-template-inclusion-warning',
2924          *       'post-expand-template-inclusion-category')
2925          * @param $current Current value
2926          * @param $max Maximum allowed, when an explicit limit has been
2927          *       exceeded, provide the values (optional)
2928          */
2929         function limitationWarn( $limitationType, $current=null, $max=null) {
2930                 # does no harm if $current and $max are present but are unnecessary for the message
2931                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2932                 $this->mOutput->addWarning( $warning );
2933                 $this->addTrackingCategory( "$limitationType-category" );
2934         }
2935
2936         /**
2937          * Return the text of a template, after recursively
2938          * replacing any variables or templates within the template.
2939          *
2940          * @param $piece Array: the parts of the template
2941          *  $piece['title']: the title, i.e. the part before the |
2942          *  $piece['parts']: the parameter array
2943          *  $piece['lineStart']: whether the brace was at the start of a line
2944          * @param $frame PPFrame The current frame, contains template arguments
2945          * @return String: the text of the template
2946          * @private
2947          */
2948         function braceSubstitution( $piece, $frame ) {
2949                 global $wgContLang, $wgNonincludableNamespaces;
2950                 wfProfileIn( __METHOD__ );
2951                 wfProfileIn( __METHOD__.'-setup' );
2952
2953                 # Flags
2954                 $found = false;             # $text has been filled
2955                 $nowiki = false;            # wiki markup in $text should be escaped
2956                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2957                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2958                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2959                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2960
2961                 # Title object, where $text came from
2962                 $title = null;
2963
2964                 # $part1 is the bit before the first |, and must contain only title characters.
2965                 # Various prefixes will be stripped from it later.
2966                 $titleWithSpaces = $frame->expand( $piece['title'] );
2967                 $part1 = trim( $titleWithSpaces );
2968                 $titleText = false;
2969
2970                 # Original title text preserved for various purposes
2971                 $originalTitle = $part1;
2972
2973                 # $args is a list of argument nodes, starting from index 0, not including $part1
2974                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2975                 wfProfileOut( __METHOD__.'-setup' );
2976
2977                 # SUBST
2978                 wfProfileIn( __METHOD__.'-modifiers' );
2979                 if ( !$found ) {
2980
2981                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2982
2983                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2984                         # Decide whether to expand template or keep wikitext as-is.
2985                         if ( $this->ot['wiki'] ) {
2986                                 if ( $substMatch === false ) {
2987                                         $literal = true;  # literal when in PST with no prefix
2988                                 } else {
2989                                         $literal = false; # expand when in PST with subst: or safesubst:
2990                                 }
2991                         } else {
2992                                 if ( $substMatch == 'subst' ) {
2993                                         $literal = true;  # literal when not in PST with plain subst:
2994                                 } else {
2995                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2996                                 }
2997                         }
2998                         if ( $literal ) {
2999                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3000                                 $isLocalObj = true;
3001                                 $found = true;
3002                         }
3003                 }
3004
3005                 # Variables
3006                 if ( !$found && $args->getLength() == 0 ) {
3007                         $id = $this->mVariables->matchStartToEnd( $part1 );
3008                         if ( $id !== false ) {
3009                                 $text = $this->getVariableValue( $id, $frame );
3010                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3011                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3012                                 }
3013                                 $found = true;
3014                         }
3015                 }
3016
3017                 # MSG, MSGNW and RAW
3018                 if ( !$found ) {
3019                         # Check for MSGNW:
3020                         $mwMsgnw = MagicWord::get( 'msgnw' );
3021                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3022                                 $nowiki = true;
3023                         } else {
3024                                 # Remove obsolete MSG:
3025                                 $mwMsg = MagicWord::get( 'msg' );
3026                                 $mwMsg->matchStartAndRemove( $part1 );
3027                         }
3028
3029                         # Check for RAW:
3030                         $mwRaw = MagicWord::get( 'raw' );
3031                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3032                                 $forceRawInterwiki = true;
3033                         }
3034                 }
3035                 wfProfileOut( __METHOD__.'-modifiers' );
3036
3037                 # Parser functions
3038                 if ( !$found ) {
3039                         wfProfileIn( __METHOD__ . '-pfunc' );
3040
3041                         $colonPos = strpos( $part1, ':' );
3042                         if ( $colonPos !== false ) {
3043                                 # Case sensitive functions
3044                                 $function = substr( $part1, 0, $colonPos );
3045                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3046                                         $function = $this->mFunctionSynonyms[1][$function];
3047                                 } else {
3048                                         # Case insensitive functions
3049                                         $function = $wgContLang->lc( $function );
3050                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3051                                                 $function = $this->mFunctionSynonyms[0][$function];
3052                                         } else {
3053                                                 $function = false;
3054                                         }
3055                                 }
3056                                 if ( $function ) {
3057                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3058                                         $initialArgs = array( &$this );
3059                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3060                                         if ( $flags & SFH_OBJECT_ARGS ) {
3061                                                 # Add a frame parameter, and pass the arguments as an array
3062                                                 $allArgs = $initialArgs;
3063                                                 $allArgs[] = $frame;
3064                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3065                                                         $funcArgs[] = $args->item( $i );
3066                                                 }
3067                                                 $allArgs[] = $funcArgs;
3068                                         } else {
3069                                                 # Convert arguments to plain text
3070                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3071                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3072                                                 }
3073                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3074                                         }
3075
3076                                         # Workaround for PHP bug 35229 and similar
3077                                         if ( !is_callable( $callback ) ) {
3078                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3079                                                 wfProfileOut( __METHOD__ );
3080                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3081                                         }
3082                                         $result = call_user_func_array( $callback, $allArgs );
3083                                         $found = true;
3084                                         $noparse = true;
3085                                         $preprocessFlags = 0;
3086
3087                                         if ( is_array( $result ) ) {
3088                                                 if ( isset( $result[0] ) ) {
3089                                                         $text = $result[0];
3090                                                         unset( $result[0] );
3091                                                 }
3092
3093                                                 # Extract flags into the local scope
3094                                                 # This allows callers to set flags such as nowiki, found, etc.
3095                                                 extract( $result );
3096                                         } else {
3097                                                 $text = $result;
3098                                         }
3099                                         if ( !$noparse ) {
3100                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3101                                                 $isChildObj = true;
3102                                         }
3103                                 }
3104                         }
3105                         wfProfileOut( __METHOD__ . '-pfunc' );
3106                 }
3107
3108                 # Finish mangling title and then check for loops.
3109                 # Set $title to a Title object and $titleText to the PDBK
3110                 if ( !$found ) {
3111                         $ns = NS_TEMPLATE;
3112                         # Split the title into page and subpage
3113                         $subpage = '';
3114                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3115                         if ( $subpage !== '' ) {
3116                                 $ns = $this->mTitle->getNamespace();
3117                         }
3118                         $title = Title::newFromText( $part1, $ns );
3119                         if ( $title ) {
3120                                 $titleText = $title->getPrefixedText();
3121                                 # Check for language variants if the template is not found
3122                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3123                                         $wgContLang->findVariantLink( $part1, $title, true );
3124                                 }
3125                                 # Do recursion depth check
3126                                 $limit = $this->mOptions->getMaxTemplateDepth();
3127                                 if ( $frame->depth >= $limit ) {
3128                                         $found = true;
3129                                         $text = '<span class="error">'
3130                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3131                                                 . '</span>';
3132                                 }
3133                         }
3134                 }
3135
3136                 # Load from database
3137                 if ( !$found && $title ) {
3138                         wfProfileIn( __METHOD__ . '-loadtpl' );
3139                         if ( !$title->isExternal() ) {
3140                                 if ( $title->getNamespace() == NS_SPECIAL
3141                                         && $this->mOptions->getAllowSpecialInclusion()
3142                                         && $this->ot['html'] )
3143                                 {
3144                                         $text = SpecialPage::capturePath( $title );
3145                                         if ( is_string( $text ) ) {
3146                                                 $found = true;
3147                                                 $isHTML = true;
3148                                                 $this->disableCache();
3149                                         }
3150                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3151                                         $found = false; # access denied
3152                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3153                                 } else {
3154                                         list( $text, $title ) = $this->getTemplateDom( $title );
3155                                         if ( $text !== false ) {
3156                                                 $found = true;
3157                                                 $isChildObj = true;
3158                                         }
3159                                 }
3160
3161                                 # If the title is valid but undisplayable, make a link to it
3162                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3163                                         $text = "[[:$titleText]]";
3164                                         $found = true;
3165                                 }
3166                         } elseif ( $title->isTrans() ) {
3167                                 # Interwiki transclusion
3168                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3169                                         $text = $this->interwikiTransclude( $title, 'render' );
3170                                         $isHTML = true;
3171                                 } else {
3172                                         $text = $this->interwikiTransclude( $title, 'raw' );
3173                                         # Preprocess it like a template
3174                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3175                                         $isChildObj = true;
3176                                 }
3177                                 $found = true;
3178                         }
3179
3180                         # Do infinite loop check
3181                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3182                         if ( !$frame->loopCheck( $title ) ) {
3183                                 $found = true;
3184                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3185                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3186                         }
3187                         wfProfileOut( __METHOD__ . '-loadtpl' );
3188                 }
3189
3190                 # If we haven't found text to substitute by now, we're done
3191                 # Recover the source wikitext and return it
3192                 if ( !$found ) {
3193                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3194                         wfProfileOut( __METHOD__ );
3195                         return array( 'object' => $text );
3196                 }
3197
3198                 # Expand DOM-style return values in a child frame
3199                 if ( $isChildObj ) {
3200                         # Clean up argument array
3201                         $newFrame = $frame->newChild( $args, $title );
3202
3203                         if ( $nowiki ) {
3204                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3205                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3206                                 # Expansion is eligible for the empty-frame cache
3207                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3208                                         $text = $this->mTplExpandCache[$titleText];
3209                                 } else {
3210                                         $text = $newFrame->expand( $text );
3211                                         $this->mTplExpandCache[$titleText] = $text;
3212                                 }
3213                         } else {
3214                                 # Uncached expansion
3215                                 $text = $newFrame->expand( $text );
3216                         }
3217                 }
3218                 if ( $isLocalObj && $nowiki ) {
3219                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3220                         $isLocalObj = false;
3221                 }
3222
3223                 # Replace raw HTML by a placeholder
3224                 # Add a blank line preceding, to prevent it from mucking up
3225                 # immediately preceding headings
3226                 if ( $isHTML ) {
3227                         $text = "\n\n" . $this->insertStripItem( $text );
3228                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3229                         # Escape nowiki-style return values
3230                         $text = wfEscapeWikiText( $text );
3231                 } elseif ( is_string( $text )
3232                         && !$piece['lineStart']
3233                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3234                 {
3235                         # Bug 529: if the template begins with a table or block-level
3236                         # element, it should be treated as beginning a new line.
3237                         # This behaviour is somewhat controversial.
3238                         $text = "\n" . $text;
3239                 }
3240
3241                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3242                         # Error, oversize inclusion
3243                         if ( $titleText !== false ) {
3244                                 # Make a working, properly escaped link if possible (bug 23588)
3245                                 $text = "[[:$titleText]]";
3246                         } else {
3247                                 # This will probably not be a working link, but at least it may
3248                                 # provide some hint of where the problem is
3249                                 preg_replace( '/^:/', '', $originalTitle );
3250                                 $text = "[[:$originalTitle]]";
3251                         }
3252                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3253                         $this->limitationWarn( 'post-expand-template-inclusion' );
3254                 }
3255
3256                 if ( $isLocalObj ) {
3257                         $ret = array( 'object' => $text );
3258                 } else {
3259                         $ret = array( 'text' => $text );
3260                 }
3261
3262                 wfProfileOut( __METHOD__ );
3263                 return $ret;
3264         }
3265
3266         /**
3267          * Get the semi-parsed DOM representation of a template with a given title,
3268          * and its redirect destination title. Cached.
3269          */
3270         function getTemplateDom( $title ) {
3271                 $cacheTitle = $title;
3272                 $titleText = $title->getPrefixedDBkey();
3273
3274                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3275                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3276                         $title = Title::makeTitle( $ns, $dbk );
3277                         $titleText = $title->getPrefixedDBkey();
3278                 }
3279                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3280                         return array( $this->mTplDomCache[$titleText], $title );
3281                 }
3282
3283                 # Cache miss, go to the database
3284                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3285
3286                 if ( $text === false ) {
3287                         $this->mTplDomCache[$titleText] = false;
3288                         return array( false, $title );
3289                 }
3290
3291                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3292                 $this->mTplDomCache[ $titleText ] = $dom;
3293
3294                 if ( !$title->equals( $cacheTitle ) ) {
3295                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3296                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3297                 }
3298
3299                 return array( $dom, $title );
3300         }
3301
3302         /**
3303          * Fetch the unparsed text of a template and register a reference to it.
3304          */
3305         function fetchTemplateAndTitle( $title ) {
3306                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3307                 $stuff = call_user_func( $templateCb, $title, $this );
3308                 $text = $stuff['text'];
3309                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3310                 if ( isset( $stuff['deps'] ) ) {
3311                         foreach ( $stuff['deps'] as $dep ) {
3312                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3313                         }
3314                 }
3315                 return array( $text, $finalTitle );
3316         }
3317
3318         function fetchTemplate( $title ) {
3319                 $rv = $this->fetchTemplateAndTitle( $title );
3320                 return $rv[0];
3321         }
3322
3323         /**
3324          * Static function to get a template
3325          * Can be overridden via ParserOptions::setTemplateCallback().
3326          */
3327         static function statelessFetchTemplate( $title, $parser=false ) {
3328                 $text = $skip = false;
3329                 $finalTitle = $title;
3330                 $deps = array();
3331
3332                 # Loop to fetch the article, with up to 1 redirect
3333                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3334                         # Give extensions a chance to select the revision instead
3335                         $id = false; # Assume current
3336                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3337
3338                         if ( $skip ) {
3339                                 $text = false;
3340                                 $deps[] = array(
3341                                         'title' => $title,
3342                                         'page_id' => $title->getArticleID(),
3343                                         'rev_id' => null );
3344                                 break;
3345                         }
3346                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3347                         $rev_id = $rev ? $rev->getId() : 0;
3348                         # If there is no current revision, there is no page
3349                         if ( $id === false && !$rev ) {
3350                                 $linkCache = LinkCache::singleton();
3351                                 $linkCache->addBadLinkObj( $title );
3352                         }
3353
3354                         $deps[] = array(
3355                                 'title' => $title,
3356                                 'page_id' => $title->getArticleID(),
3357                                 'rev_id' => $rev_id );
3358
3359                         if ( $rev ) {
3360                                 $text = $rev->getText();
3361                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3362                                 global $wgContLang;
3363                                 $message = $wgContLang->lcfirst( $title->getText() );
3364                                 $text = wfMsgForContentNoTrans( $message );
3365                                 if ( wfEmptyMsg( $message, $text ) ) {
3366                                         $text = false;
3367                                         break;
3368                                 }
3369                         } else {
3370                                 break;
3371                         }
3372                         if ( $text === false ) {
3373                                 break;
3374                         }
3375                         # Redirect?
3376                         $finalTitle = $title;
3377                         $title = Title::newFromRedirect( $text );
3378                 }
3379                 return array(
3380                         'text' => $text,
3381                         'finalTitle' => $finalTitle,
3382                         'deps' => $deps );
3383         }
3384
3385         /**
3386          * Transclude an interwiki link.
3387          */
3388         function interwikiTransclude( $title, $action ) {
3389                 global $wgEnableScaryTranscluding;
3390
3391                 if ( !$wgEnableScaryTranscluding ) {
3392                         return wfMsgForContent('scarytranscludedisabled');
3393                 }
3394
3395                 $url = $title->getFullUrl( "action=$action" );
3396
3397                 if ( strlen( $url ) > 255 ) {
3398                         return wfMsgForContent( 'scarytranscludetoolong' );
3399                 }
3400                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3401         }
3402
3403         function fetchScaryTemplateMaybeFromCache( $url ) {
3404                 global $wgTranscludeCacheExpiry;
3405                 $dbr = wfGetDB( DB_SLAVE );
3406                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3407                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3408                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3409                 if ( $obj ) {
3410                         return $obj->tc_contents;
3411                 }
3412
3413                 $text = Http::get( $url );
3414                 if ( !$text ) {
3415                         return wfMsgForContent( 'scarytranscludefailed', $url );
3416                 }
3417
3418                 $dbw = wfGetDB( DB_MASTER );
3419                 $dbw->replace( 'transcache', array('tc_url'), array(
3420                         'tc_url' => $url,
3421                         'tc_time' => $dbw->timestamp( time() ),
3422                         'tc_contents' => $text)
3423                 );
3424                 return $text;
3425         }
3426
3427
3428         /**
3429          * Triple brace replacement -- used for template arguments
3430          * @private
3431          */
3432         function argSubstitution( $piece, $frame ) {
3433                 wfProfileIn( __METHOD__ );
3434
3435                 $error = false;
3436                 $parts = $piece['parts'];
3437                 $nameWithSpaces = $frame->expand( $piece['title'] );
3438                 $argName = trim( $nameWithSpaces );
3439                 $object = false;
3440                 $text = $frame->getArgument( $argName );
3441                 if (  $text === false && $parts->getLength() > 0
3442                   && (
3443                     $this->ot['html']
3444                     || $this->ot['pre']
3445                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3446                   )
3447                 ) {
3448                         # No match in frame, use the supplied default
3449                         $object = $parts->item( 0 )->getChildren();
3450                 }
3451                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3452                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3453                         $this->limitationWarn( 'post-expand-template-argument' );
3454                 }
3455
3456                 if ( $text === false && $object === false ) {
3457                         # No match anywhere
3458                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3459                 }
3460                 if ( $error !== false ) {
3461                         $text .= $error;
3462                 }
3463                 if ( $object !== false ) {
3464                         $ret = array( 'object' => $object );
3465                 } else {
3466                         $ret = array( 'text' => $text );
3467                 }
3468
3469                 wfProfileOut( __METHOD__ );
3470                 return $ret;
3471         }
3472
3473         /**
3474          * Return the text to be used for a given extension tag.
3475          * This is the ghost of strip().
3476          *
3477          * @param $params Associative array of parameters:
3478          *     name       PPNode for the tag name
3479          *     attr       PPNode for unparsed text where tag attributes are thought to be
3480          *     attributes Optional associative array of parsed attributes
3481          *     inner      Contents of extension element
3482          *     noClose    Original text did not have a close tag
3483          * @param $frame PPFrame
3484          */
3485         function extensionSubstitution( $params, $frame ) {
3486                 $name = $frame->expand( $params['name'] );
3487                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3488                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3489                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3490
3491                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3492                         ( $this->ot['html'] || $this->ot['pre'] );
3493                 if ( $isFunctionTag ) {
3494                         $markerType = 'none';
3495                 } else {
3496                         $markerType = 'general';
3497                 }
3498                 if ( $this->ot['html'] || $isFunctionTag ) {
3499                         $name = strtolower( $name );
3500                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3501                         if ( isset( $params['attributes'] ) ) {
3502                                 $attributes = $attributes + $params['attributes'];
3503                         }
3504
3505                         if ( isset( $this->mTagHooks[$name] ) ) {
3506                                 # Workaround for PHP bug 35229 and similar
3507                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3508                                         throw new MWException( "Tag hook for $name is not callable\n" );
3509                                 }
3510                                 $output = call_user_func_array( $this->mTagHooks[$name],
3511                                         array( $content, $attributes, $this, $frame ) );
3512                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3513                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3514                                 if ( !is_callable( $callback ) ) {
3515                                         throw new MWException( "Tag hook for $name is not callable\n" );
3516                                 }
3517
3518                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3519                         } else {
3520                                 $output = '<span class="error">Invalid tag extension name: ' .
3521                                         htmlspecialchars( $name ) . '</span>';
3522                         }
3523
3524                         if ( is_array( $output ) ) {
3525                                 # Extract flags to local scope (to override $markerType)
3526                                 $flags = $output;
3527                                 $output = $flags[0];
3528                                 unset( $flags[0] );
3529                                 extract( $flags );
3530                         }
3531                 } else {
3532                         if ( is_null( $attrText ) ) {
3533                                 $attrText = '';
3534                         }
3535                         if ( isset( $params['attributes'] ) ) {
3536                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3537                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3538                                                 htmlspecialchars( $attrValue ) . '"';
3539                                 }
3540                         }
3541                         if ( $content === null ) {
3542                                 $output = "<$name$attrText/>";
3543                         } else {
3544                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3545                                 $output = "<$name$attrText>$content$close";
3546                         }
3547                 }
3548
3549                 if ( $markerType === 'none' ) {
3550                         return $output;
3551                 } elseif ( $markerType === 'nowiki' ) {
3552                         $this->mStripState->nowiki->setPair( $marker, $output );
3553                 } elseif ( $markerType === 'general' ) {
3554                         $this->mStripState->general->setPair( $marker, $output );
3555                 } else {
3556                         throw new MWException( __METHOD__.': invalid marker type' );
3557                 }
3558                 return $marker;
3559         }
3560
3561         /**
3562          * Increment an include size counter
3563          *
3564          * @param $type String: the type of expansion
3565          * @param $size Integer: the size of the text
3566          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3567          */
3568         function incrementIncludeSize( $type, $size ) {
3569                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3570                         return false;
3571                 } else {
3572                         $this->mIncludeSizes[$type] += $size;
3573                         return true;
3574                 }
3575         }
3576
3577         /**
3578          * Increment the expensive function count
3579          *
3580          * @return Boolean: false if the limit has been exceeded
3581          */
3582         function incrementExpensiveFunctionCount() {
3583                 global $wgExpensiveParserFunctionLimit;
3584                 $this->mExpensiveFunctionCount++;
3585                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3586                         return true;
3587                 }
3588                 return false;
3589         }
3590
3591         /**
3592          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3593          * Fills $this->mDoubleUnderscores, returns the modified text
3594          */
3595         function doDoubleUnderscore( $text ) {
3596                 wfProfileIn( __METHOD__ );
3597
3598                 # The position of __TOC__ needs to be recorded
3599                 $mw = MagicWord::get( 'toc' );
3600                 if ( $mw->match( $text ) ) {
3601                         $this->mShowToc = true;
3602                         $this->mForceTocPosition = true;
3603
3604                         # Set a placeholder. At the end we'll fill it in with the TOC.
3605                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3606
3607                         # Only keep the first one.
3608                         $text = $mw->replace( '', $text );
3609                 }
3610
3611                 # Now match and remove the rest of them
3612                 $mwa = MagicWord::getDoubleUnderscoreArray();
3613                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3614
3615                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3616                         $this->mOutput->mNoGallery = true;
3617                 }
3618                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3619                         $this->mShowToc = false;
3620                 }
3621                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3622                         $this->addTrackingCategory( 'hidden-category-category' );
3623                 }
3624                 # (bug 8068) Allow control over whether robots index a page.
3625                 #
3626                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3627                 # is not desirable, the last one on the page should win.
3628                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3629                         $this->mOutput->setIndexPolicy( 'noindex' );
3630                         $this->addTrackingCategory( 'noindex-category' );
3631                 }
3632                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3633                         $this->mOutput->setIndexPolicy( 'index' );
3634                         $this->addTrackingCategory( 'index-category' );
3635                 }
3636
3637                 # Cache all double underscores in the database
3638                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3639                         $this->mOutput->setProperty( $key, '' );
3640                 }
3641
3642                 wfProfileOut( __METHOD__ );
3643                 return $text;
3644         }
3645
3646         /**
3647          * Add a tracking category, getting the title from a system message,
3648          * or print a debug message if the title is invalid.
3649          *
3650          * @param $msg String: message key
3651          * @return Boolean: whether the addition was successful
3652          */
3653         protected function addTrackingCategory( $msg ) {
3654                 $cat = wfMsgForContent( $msg );
3655
3656                 # Allow tracking categories to be disabled by setting them to "-"
3657                 if ( $cat === '-' ) {
3658                         return false;
3659                 }
3660
3661                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3662                 if ( $containerCategory ) {
3663                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3664                         return true;
3665                 } else {
3666                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3667                         return false;
3668                 }
3669         }
3670
3671         /**
3672          * This function accomplishes several tasks:
3673          * 1) Auto-number headings if that option is enabled
3674          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3675          * 3) Add a Table of contents on the top for users who have enabled the option
3676          * 4) Auto-anchor headings
3677          *
3678          * It loops through all headlines, collects the necessary data, then splits up the
3679          * string and re-inserts the newly formatted headlines.
3680          *
3681          * @param $text String
3682          * @param $origText String: original, untouched wikitext
3683          * @param $isMain Boolean
3684          * @private
3685          */
3686         function formatHeadings( $text, $origText, $isMain=true ) {
3687                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3688
3689                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3690
3691                 # Inhibit editsection links if requested in the page
3692                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3693                         $showEditLink = 0;
3694                 } else {
3695                         $showEditLink = $this->mOptions->getEditSection();
3696                 }
3697
3698                 # Get all headlines for numbering them and adding funky stuff like [edit]
3699                 # links - this is for later, but we need the number of headlines right now
3700                 $matches = array();
3701                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3702
3703                 # if there are fewer than 4 headlines in the article, do not show TOC
3704                 # unless it's been explicitly enabled.
3705                 $enoughToc = $this->mShowToc &&
3706                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3707
3708                 # Allow user to stipulate that a page should have a "new section"
3709                 # link added via __NEWSECTIONLINK__
3710                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3711                         $this->mOutput->setNewSection( true );
3712                 }
3713
3714                 # Allow user to remove the "new section"
3715                 # link via __NONEWSECTIONLINK__
3716                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3717                         $this->mOutput->hideNewSection( true );
3718                 }
3719
3720                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3721                 # override above conditions and always show TOC above first header
3722                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3723                         $this->mShowToc = true;
3724                         $enoughToc = true;
3725                 }
3726
3727                 # We need this to perform operations on the HTML
3728                 $sk = $this->mOptions->getSkin( $this->mTitle );
3729
3730                 # headline counter
3731                 $headlineCount = 0;
3732                 $numVisible = 0;
3733
3734                 # Ugh .. the TOC should have neat indentation levels which can be
3735                 # passed to the skin functions. These are determined here
3736                 $toc = '';
3737                 $full = '';
3738                 $head = array();
3739                 $sublevelCount = array();
3740                 $levelCount = array();
3741                 $level = 0;
3742                 $prevlevel = 0;
3743                 $toclevel = 0;
3744                 $prevtoclevel = 0;
3745                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3746                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3747                 $oldType = $this->mOutputType;
3748                 $this->setOutputType( self::OT_WIKI );
3749                 $frame = $this->getPreprocessor()->newFrame();
3750                 $root = $this->preprocessToDom( $origText );
3751                 $node = $root->getFirstChild();
3752                 $byteOffset = 0;
3753                 $tocraw = array();
3754                 $refers = array();
3755
3756                 foreach ( $matches[3] as $headline ) {
3757                         $isTemplate = false;
3758                         $titleText = false;
3759                         $sectionIndex = false;
3760                         $numbering = '';
3761                         $markerMatches = array();
3762                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3763                                 $serial = $markerMatches[1];
3764                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3765                                 $isTemplate = ( $titleText != $baseTitleText );
3766                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3767                         }
3768
3769                         if ( $toclevel ) {
3770                                 $prevlevel = $level;
3771                         }
3772                         $level = $matches[1][$headlineCount];
3773
3774                         if ( $level > $prevlevel ) {
3775                                 # Increase TOC level
3776                                 $toclevel++;
3777                                 $sublevelCount[$toclevel] = 0;
3778                                 if ( $toclevel<$wgMaxTocLevel ) {
3779                                         $prevtoclevel = $toclevel;
3780                                         $toc .= $sk->tocIndent();
3781                                         $numVisible++;
3782                                 }
3783                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3784                                 # Decrease TOC level, find level to jump to
3785
3786                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3787                                         if ( $levelCount[$i] == $level ) {
3788                                                 # Found last matching level
3789                                                 $toclevel = $i;
3790                                                 break;
3791                                         } elseif ( $levelCount[$i] < $level ) {
3792                                                 # Found first matching level below current level
3793                                                 $toclevel = $i + 1;
3794                                                 break;
3795                                         }
3796                                 }
3797                                 if ( $i == 0 ) {
3798                                         $toclevel = 1;
3799                                 }
3800                                 if ( $toclevel<$wgMaxTocLevel ) {
3801                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3802                                                 # Unindent only if the previous toc level was shown :p
3803                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3804                                                 $prevtoclevel = $toclevel;
3805                                         } else {
3806                                                 $toc .= $sk->tocLineEnd();
3807                                         }
3808                                 }
3809                         } else {
3810                                 # No change in level, end TOC line
3811                                 if ( $toclevel<$wgMaxTocLevel ) {
3812                                         $toc .= $sk->tocLineEnd();
3813                                 }
3814                         }
3815
3816                         $levelCount[$toclevel] = $level;
3817
3818                         # count number of headlines for each level
3819                         @$sublevelCount[$toclevel]++;
3820                         $dot = 0;
3821                         for( $i = 1; $i <= $toclevel; $i++ ) {
3822                                 if ( !empty( $sublevelCount[$i] ) ) {
3823                                         if ( $dot ) {
3824                                                 $numbering .= '.';
3825                                         }
3826                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3827                                         $dot = 1;
3828                                 }
3829                         }
3830
3831                         # The safe header is a version of the header text safe to use for links
3832                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3833                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3834
3835                         # Remove link placeholders by the link text.
3836                         #     <!--LINK number-->
3837                         # turns into
3838                         #     link text with suffix
3839                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3840
3841                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3842                         $tocline = preg_replace(
3843                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3844                                 array( '',                          '<$1>' ),
3845                                 $safeHeadline
3846                         );
3847                         $tocline = trim( $tocline );
3848
3849                         # For the anchor, strip out HTML-y stuff period
3850                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3851                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
3852
3853                         # Save headline for section edit hint before it's escaped
3854                         $headlineHint = $safeHeadline;
3855
3856                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3857                                 # For reverse compatibility, provide an id that's
3858                                 # HTML4-compatible, like we used to.
3859                                 #
3860                                 # It may be worth noting, academically, that it's possible for
3861                                 # the legacy anchor to conflict with a non-legacy headline
3862                                 # anchor on the page.  In this case likely the "correct" thing
3863                                 # would be to either drop the legacy anchors or make sure
3864                                 # they're numbered first.  However, this would require people
3865                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3866                                 # manually, so let's not bother worrying about it.
3867                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3868                                         array( 'noninitial', 'legacy' ) );
3869                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3870
3871                                 if ( $legacyHeadline == $safeHeadline ) {
3872                                         # No reason to have both (in fact, we can't)
3873                                         $legacyHeadline = false;
3874                                 }
3875                         } else {
3876                                 $legacyHeadline = false;
3877                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3878                                         'noninitial' );
3879                         }
3880
3881                         # HTML names must be case-insensitively unique (bug 10721).
3882                         # This does not apply to Unicode characters per
3883                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
3884                         # FIXME: We may be changing them depending on the current locale.
3885                         $arrayKey = strtolower( $safeHeadline );
3886                         if ( $legacyHeadline === false ) {
3887                                 $legacyArrayKey = false;
3888                         } else {
3889                                 $legacyArrayKey = strtolower( $legacyHeadline );
3890                         }
3891
3892                         # count how many in assoc. array so we can track dupes in anchors
3893                         if ( isset( $refers[$arrayKey] ) ) {
3894                                 $refers[$arrayKey]++;
3895                         } else {
3896                                 $refers[$arrayKey] = 1;
3897                         }
3898                         if ( isset( $refers[$legacyArrayKey] ) ) {
3899                                 $refers[$legacyArrayKey]++;
3900                         } else {
3901                                 $refers[$legacyArrayKey] = 1;
3902                         }
3903
3904                         # Don't number the heading if it is the only one (looks silly)
3905                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3906                                 # the two are different if the line contains a link
3907                                 $headline = $numbering . ' ' . $headline;
3908                         }
3909
3910                         # Create the anchor for linking from the TOC to the section
3911                         $anchor = $safeHeadline;
3912                         $legacyAnchor = $legacyHeadline;
3913                         if ( $refers[$arrayKey] > 1 ) {
3914                                 $anchor .= '_' . $refers[$arrayKey];
3915                         }
3916                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3917                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3918                         }
3919                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3920                                 $toc .= $sk->tocLine( $anchor, $tocline,
3921                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3922                         }
3923
3924                         # Add the section to the section tree
3925                         # Find the DOM node for this header
3926                         while ( $node && !$isTemplate ) {
3927                                 if ( $node->getName() === 'h' ) {
3928                                         $bits = $node->splitHeading();
3929                                         if ( $bits['i'] == $sectionIndex ) {
3930                                                 break;
3931                                         }
3932                                 }
3933                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3934                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3935                                 $node = $node->getNextSibling();
3936                         }
3937                         $tocraw[] = array(
3938                                 'toclevel' => $toclevel,
3939                                 'level' => $level,
3940                                 'line' => $tocline,
3941                                 'number' => $numbering,
3942                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3943                                 'fromtitle' => $titleText,
3944                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3945                                 'anchor' => $anchor,
3946                         );
3947
3948                         # give headline the correct <h#> tag
3949                         if ( $showEditLink && $sectionIndex !== false ) {
3950                                 if ( $isTemplate ) {
3951                                         # Put a T flag in the section identifier, to indicate to extractSections()
3952                                         # that sections inside <includeonly> should be counted.
3953                                         $editlink = $sk->doEditSectionLink( Title::newFromText( $titleText ), "T-$sectionIndex", null, $this->mOptions->getUserLang() );
3954                                 } else {
3955                                         $editlink = $sk->doEditSectionLink( $this->mTitle, $sectionIndex, $headlineHint, $this->mOptions->getUserLang() );
3956                                 }
3957                         } else {
3958                                 $editlink = '';
3959                         }
3960                         $head[$headlineCount] = $sk->makeHeadline( $level,
3961                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3962                                 $editlink, $legacyAnchor );
3963
3964                         $headlineCount++;
3965                 }
3966
3967                 $this->setOutputType( $oldType );
3968
3969                 # Never ever show TOC if no headers
3970                 if ( $numVisible < 1 ) {
3971                         $enoughToc = false;
3972                 }
3973
3974                 if ( $enoughToc ) {
3975                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3976                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3977                         }
3978                         $toc = $sk->tocList( $toc );
3979                         $this->mOutput->setTOCHTML( $toc );
3980                 }
3981
3982                 if ( $isMain ) {
3983                         $this->mOutput->setSections( $tocraw );
3984                 }
3985
3986                 # split up and insert constructed headlines
3987
3988                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3989                 $i = 0;
3990
3991                 foreach ( $blocks as $block ) {
3992                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3993                                 # This is the [edit] link that appears for the top block of text when
3994                                 # section editing is enabled
3995
3996                                 # Disabled because it broke block formatting
3997                                 # For example, a bullet point in the top line
3998                                 # $full .= $sk->editSectionLink(0);
3999                         }
4000                         $full .= $block;
4001                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4002                                 # Top anchor now in skin
4003                                 $full = $full.$toc;
4004                         }
4005
4006                         if ( !empty( $head[$i] ) ) {
4007                                 $full .= $head[$i];
4008                         }
4009                         $i++;
4010                 }
4011                 if ( $this->mForceTocPosition ) {
4012                         return str_replace( '<!--MWTOC-->', $toc, $full );
4013                 } else {
4014                         return $full;
4015                 }
4016         }
4017
4018         /**
4019          * Transform wiki markup when saving a page by doing \r\n -> \n
4020          * conversion, substitting signatures, {{subst:}} templates, etc.
4021          *
4022          * @param $text String: the text to transform
4023          * @param $title Title: the Title object for the current article
4024          * @param $user User: the User object describing the current user
4025          * @param $options ParserOptions: parsing options
4026          * @param $clearState Boolean: whether to clear the parser state first
4027          * @return String: the altered wiki markup
4028          */
4029         public function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
4030                 $options->resetUsage();
4031                 $this->mOptions = $options;
4032                 $this->setTitle( $title );
4033                 $this->setUser( $user );
4034                 $this->setOutputType( self::OT_WIKI );
4035
4036                 if ( $clearState ) {
4037                         $this->clearState();
4038                 }
4039
4040                 $pairs = array(
4041                         "\r\n" => "\n",
4042                 );
4043                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4044                 $text = $this->pstPass2( $text, $user );
4045                 $text = $this->mStripState->unstripBoth( $text );
4046
4047                 $this->setUser( null ); #Reset
4048
4049                 return $text;
4050         }
4051
4052         /**
4053          * Pre-save transform helper function
4054          * @private
4055          */
4056         function pstPass2( $text, $user ) {
4057                 global $wgContLang, $wgLocaltimezone;
4058
4059                 # Note: This is the timestamp saved as hardcoded wikitext to
4060                 # the database, we use $wgContLang here in order to give
4061                 # everyone the same signature and use the default one rather
4062                 # than the one selected in each user's preferences.
4063                 # (see also bug 12815)
4064                 $ts = $this->mOptions->getTimestamp();
4065                 if ( isset( $wgLocaltimezone ) ) {
4066                         $tz = $wgLocaltimezone;
4067                 } else {
4068                         $tz = date_default_timezone_get();
4069                 }
4070
4071                 $unixts = wfTimestamp( TS_UNIX, $ts );
4072                 $oldtz = date_default_timezone_get();
4073                 date_default_timezone_set( $tz );
4074                 $ts = date( 'YmdHis', $unixts );
4075                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4076
4077                 # Allow translation of timezones through wiki. date() can return
4078                 # whatever crap the system uses, localised or not, so we cannot
4079                 # ship premade translations.
4080                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4081                 $value = wfMsgForContent( $key );
4082                 if ( !wfEmptyMsg( $key, $value ) ) {
4083                         $tzMsg = $value;
4084                 }
4085
4086                 date_default_timezone_set( $oldtz );
4087
4088                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4089
4090                 # Variable replacement
4091                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4092                 $text = $this->replaceVariables( $text );
4093
4094                 # Signatures
4095                 $sigText = $this->getUserSig( $user );
4096                 $text = strtr( $text, array(
4097                         '~~~~~' => $d,
4098                         '~~~~' => "$sigText $d",
4099                         '~~~' => $sigText
4100                 ) );
4101
4102                 # Context links: [[|name]] and [[name (context)|]]
4103                 global $wgLegalTitleChars;
4104                 $tc = "[$wgLegalTitleChars]";
4105                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4106
4107                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4108                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4109                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4110                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4111
4112                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4113                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4114                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4115                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4116
4117                 $t = $this->mTitle->getText();
4118                 $m = array();
4119                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4120                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4121                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4122                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4123                 } else {
4124                         # if there's no context, don't bother duplicating the title
4125                         $text = preg_replace( $p2, '[[\\1]]', $text );
4126                 }
4127
4128                 # Trim trailing whitespace
4129                 $text = rtrim( $text );
4130
4131                 return $text;
4132         }
4133
4134         /**
4135          * Fetch the user's signature text, if any, and normalize to
4136          * validated, ready-to-insert wikitext.
4137          * If you have pre-fetched the nickname or the fancySig option, you can
4138          * specify them here to save a database query.
4139          *
4140          * @param $user User
4141          * @param $nickname String: nickname to use or false to use user's default nickname
4142          * @param $fancySig Boolean: whether the nicknname is the complete signature
4143          *                  or null to use default value
4144          * @return string
4145          */
4146         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4147                 global $wgMaxSigChars;
4148
4149                 $username = $user->getName();
4150
4151                 # If not given, retrieve from the user object.
4152                 if ( $nickname === false )
4153                         $nickname = $user->getOption( 'nickname' );
4154
4155                 if ( is_null( $fancySig ) ) {
4156                         $fancySig = $user->getBoolOption( 'fancysig' );
4157                 }
4158
4159                 $nickname = $nickname == null ? $username : $nickname;
4160
4161                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4162                         $nickname = $username;
4163                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4164                 } elseif ( $fancySig !== false ) {
4165                         # Sig. might contain markup; validate this
4166                         if ( $this->validateSig( $nickname ) !== false ) {
4167                                 # Validated; clean up (if needed) and return it
4168                                 return $this->cleanSig( $nickname, true );
4169                         } else {
4170                                 # Failed to validate; fall back to the default
4171                                 $nickname = $username;
4172                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4173                         }
4174                 }
4175
4176                 # Make sure nickname doesnt get a sig in a sig
4177                 $nickname = $this->cleanSigInSig( $nickname );
4178
4179                 # If we're still here, make it a link to the user page
4180                 $userText = wfEscapeWikiText( $username );
4181                 $nickText = wfEscapeWikiText( $nickname );
4182                 if ( $user->isAnon() )  {
4183                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4184                 } else {
4185                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4186                 }
4187         }
4188
4189         /**
4190          * Check that the user's signature contains no bad XML
4191          *
4192          * @param $text String
4193          * @return mixed An expanded string, or false if invalid.
4194          */
4195         function validateSig( $text ) {
4196                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4197         }
4198
4199         /**
4200          * Clean up signature text
4201          *
4202          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4203          * 2) Substitute all transclusions
4204          *
4205          * @param $text String
4206          * @param $parsing Whether we're cleaning (preferences save) or parsing
4207          * @return String: signature text
4208          */
4209         function cleanSig( $text, $parsing = false ) {
4210                 if ( !$parsing ) {
4211                         global $wgTitle;
4212                         $this->clearState();
4213                         $this->setTitle( $wgTitle );
4214                         $this->mOptions = new ParserOptions;
4215                         $this->setOutputType = self::OT_PREPROCESS;
4216                 }
4217
4218                 # Option to disable this feature
4219                 if ( !$this->mOptions->getCleanSignatures() ) {
4220                         return $text;
4221                 }
4222
4223                 # FIXME: regex doesn't respect extension tags or nowiki
4224                 #  => Move this logic to braceSubstitution()
4225                 $substWord = MagicWord::get( 'subst' );
4226                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4227                 $substText = '{{' . $substWord->getSynonym( 0 );
4228
4229                 $text = preg_replace( $substRegex, $substText, $text );
4230                 $text = $this->cleanSigInSig( $text );
4231                 $dom = $this->preprocessToDom( $text );
4232                 $frame = $this->getPreprocessor()->newFrame();
4233                 $text = $frame->expand( $dom );
4234
4235                 if ( !$parsing ) {
4236                         $text = $this->mStripState->unstripBoth( $text );
4237                 }
4238
4239                 return $text;
4240         }
4241
4242         /**
4243          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4244          *
4245          * @param $text String
4246          * @return String: signature text with /~{3,5}/ removed
4247          */
4248         function cleanSigInSig( $text ) {
4249                 $text = preg_replace( '/~{3,5}/', '', $text );
4250                 return $text;
4251         }
4252
4253         /**
4254          * Set up some variables which are usually set up in parse()
4255          * so that an external function can call some class members with confidence
4256          */
4257         public function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4258                 $this->setTitle( $title );
4259                 $options->resetUsage();
4260                 $this->mOptions = $options;
4261                 $this->setOutputType( $outputType );
4262                 if ( $clearState ) {
4263                         $this->clearState();
4264                 }
4265         }
4266
4267         /**
4268          * Wrapper for preprocess()
4269          *
4270          * @param $text String: the text to preprocess
4271          * @param $options ParserOptions: options
4272          * @return String
4273          */
4274         public function transformMsg( $text, $options ) {
4275                 global $wgTitle;
4276                 static $executing = false;
4277
4278                 # Guard against infinite recursion
4279                 if ( $executing ) {
4280                         return $text;
4281                 }
4282                 $executing = true;
4283
4284                 wfProfileIn( __METHOD__ );
4285                 $text = $this->preprocess( $text, $wgTitle, $options );
4286
4287                 $executing = false;
4288                 wfProfileOut( __METHOD__ );
4289                 return $text;
4290         }
4291
4292         /**
4293          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4294          * The callback should have the following form:
4295          *    function myParserHook( $text, $params, $parser ) { ... }
4296          *
4297          * Transform and return $text. Use $parser for any required context, e.g. use
4298          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4299          *
4300          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4301          * @param $callback Mixed: the callback function (and object) to use for the tag
4302          * @return The old value of the mTagHooks array associated with the hook
4303          */
4304         public function setHook( $tag, $callback ) {
4305                 $tag = strtolower( $tag );
4306                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4307                 $this->mTagHooks[$tag] = $callback;
4308                 if ( !in_array( $tag, $this->mStripList ) ) {
4309                         $this->mStripList[] = $tag;
4310                 }
4311
4312                 return $oldVal;
4313         }
4314
4315         function setTransparentTagHook( $tag, $callback ) {
4316                 $tag = strtolower( $tag );
4317                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4318                 $this->mTransparentTagHooks[$tag] = $callback;
4319
4320                 return $oldVal;
4321         }
4322
4323         /**
4324          * Remove all tag hooks
4325          */
4326         function clearTagHooks() {
4327                 $this->mTagHooks = array();
4328                 $this->mStripList = $this->mDefaultStripList;
4329         }
4330
4331         /**
4332          * Create a function, e.g. {{sum:1|2|3}}
4333          * The callback function should have the form:
4334          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4335          *
4336          * Or with SFH_OBJECT_ARGS:
4337          *    function myParserFunction( $parser, $frame, $args ) { ... }
4338          *
4339          * The callback may either return the text result of the function, or an array with the text
4340          * in element 0, and a number of flags in the other elements. The names of the flags are
4341          * specified in the keys. Valid flags are:
4342          *   found                     The text returned is valid, stop processing the template. This
4343          *                             is on by default.
4344          *   nowiki                    Wiki markup in the return value should be escaped
4345          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4346          *
4347          * @param $id String: The magic word ID
4348          * @param $callback Mixed: the callback function (and object) to use
4349          * @param $flags Integer: a combination of the following flags:
4350          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4351          *
4352          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4353          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4354          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4355          *     the arguments, and to control the way they are expanded.
4356          *
4357          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4358          *     arguments, for instance:
4359          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4360          *
4361          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4362          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4363          *     working if/when this is changed.
4364          *
4365          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4366          *     expansion.
4367          *
4368          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4369          *     about the methods available in PPFrame and PPNode.
4370          *
4371          * @return The old callback function for this name, if any
4372          */
4373         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4374                 global $wgContLang;
4375
4376                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4377                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4378
4379                 # Add to function cache
4380                 $mw = MagicWord::get( $id );
4381                 if ( !$mw )
4382                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4383
4384                 $synonyms = $mw->getSynonyms();
4385                 $sensitive = intval( $mw->isCaseSensitive() );
4386
4387                 foreach ( $synonyms as $syn ) {
4388                         # Case
4389                         if ( !$sensitive ) {
4390                                 $syn = $wgContLang->lc( $syn );
4391                         }
4392                         # Add leading hash
4393                         if ( !( $flags & SFH_NO_HASH ) ) {
4394                                 $syn = '#' . $syn;
4395                         }
4396                         # Remove trailing colon
4397                         if ( substr( $syn, -1, 1 ) === ':' ) {
4398                                 $syn = substr( $syn, 0, -1 );
4399                         }
4400                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4401                 }
4402                 return $oldVal;
4403         }
4404
4405         /**
4406          * Get all registered function hook identifiers
4407          *
4408          * @return Array
4409          */
4410         function getFunctionHooks() {
4411                 return array_keys( $this->mFunctionHooks );
4412         }
4413
4414         /**
4415          * Create a tag function, e.g. <test>some stuff</test>.
4416          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4417          * Unlike parser functions, their content is not preprocessed.
4418          */
4419         function setFunctionTagHook( $tag, $callback, $flags ) {
4420                 $tag = strtolower( $tag );
4421                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4422                         $this->mFunctionTagHooks[$tag] : null;
4423                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4424
4425                 if ( !in_array( $tag, $this->mStripList ) ) {
4426                         $this->mStripList[] = $tag;
4427                 }
4428
4429                 return $old;
4430         }
4431
4432         /**
4433          * FIXME: update documentation. makeLinkObj() is deprecated.
4434          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4435          * Placeholders created in Skin::makeLinkObj()
4436          * Returns an array of link CSS classes, indexed by PDBK.
4437          */
4438         function replaceLinkHolders( &$text, $options = 0 ) {
4439                 return $this->mLinkHolders->replace( $text );
4440         }
4441
4442         /**
4443          * Replace <!--LINK--> link placeholders with plain text of links
4444          * (not HTML-formatted).
4445          *
4446          * @param $text String
4447          * @return String
4448          */
4449         function replaceLinkHoldersText( $text ) {
4450                 return $this->mLinkHolders->replaceText( $text );
4451         }
4452
4453         /**
4454          * Renders an image gallery from a text with one line per image.
4455          * text labels may be given by using |-style alternative text. E.g.
4456          *   Image:one.jpg|The number "1"
4457          *   Image:tree.jpg|A tree
4458          * given as text will return the HTML of a gallery with two images,
4459          * labeled 'The number "1"' and
4460          * 'A tree'.
4461          */
4462         function renderImageGallery( $text, $params ) {
4463                 $ig = new ImageGallery();
4464                 $ig->setContextTitle( $this->mTitle );
4465                 $ig->setShowBytes( false );
4466                 $ig->setShowFilename( false );
4467                 $ig->setParser( $this );
4468                 $ig->setHideBadImages();
4469                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4470                 $ig->useSkin( $this->mOptions->getSkin( $this->mTitle ) );
4471                 $ig->mRevisionId = $this->mRevisionId;
4472
4473                 if ( isset( $params['showfilename'] ) ) {
4474                         $ig->setShowFilename( true );
4475                 } else {
4476                         $ig->setShowFilename( false );
4477                 }
4478                 if ( isset( $params['caption'] ) ) {
4479                         $caption = $params['caption'];
4480                         $caption = htmlspecialchars( $caption );
4481                         $caption = $this->replaceInternalLinks( $caption );
4482                         $ig->setCaptionHtml( $caption );
4483                 }
4484                 if ( isset( $params['perrow'] ) ) {
4485                         $ig->setPerRow( $params['perrow'] );
4486                 }
4487                 if ( isset( $params['widths'] ) ) {
4488                         $ig->setWidths( $params['widths'] );
4489                 }
4490                 if ( isset( $params['heights'] ) ) {
4491                         $ig->setHeights( $params['heights'] );
4492                 }
4493
4494                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4495
4496                 $lines = StringUtils::explode( "\n", $text );
4497                 foreach ( $lines as $line ) {
4498                         # match lines like these:
4499                         # Image:someimage.jpg|This is some image
4500                         $matches = array();
4501                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4502                         # Skip empty lines
4503                         if ( count( $matches ) == 0 ) {
4504                                 continue;
4505                         }
4506
4507                         if ( strpos( $matches[0], '%' ) !== false ) {
4508                                 $matches[1] = urldecode( $matches[1] );
4509                         }
4510                         $tp = Title::newFromText( $matches[1] );
4511                         $nt =& $tp;
4512                         if ( is_null( $nt ) ) {
4513                                 # Bogus title. Ignore these so we don't bomb out later.
4514                                 continue;
4515                         }
4516                         if ( isset( $matches[3] ) ) {
4517                                 $label = $matches[3];
4518                         } else {
4519                                 $label = '';
4520                         }
4521
4522                         $html = $this->recursiveTagParse( trim( $label ) );
4523
4524                         $ig->add( $nt, $html );
4525
4526                         # Only add real images (bug #5586)
4527                         if ( $nt->getNamespace() == NS_FILE ) {
4528                                 $this->mOutput->addImage( $nt->getDBkey() );
4529                         }
4530                 }
4531                 return $ig->toHTML();
4532         }
4533
4534         function getImageParams( $handler ) {
4535                 if ( $handler ) {
4536                         $handlerClass = get_class( $handler );
4537                 } else {
4538                         $handlerClass = '';
4539                 }
4540                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4541                         # Initialise static lists
4542                         static $internalParamNames = array(
4543                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4544                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4545                                         'bottom', 'text-bottom' ),
4546                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4547                                         'upright', 'border', 'link', 'alt' ),
4548                         );
4549                         static $internalParamMap;
4550                         if ( !$internalParamMap ) {
4551                                 $internalParamMap = array();
4552                                 foreach ( $internalParamNames as $type => $names ) {
4553                                         foreach ( $names as $name ) {
4554                                                 $magicName = str_replace( '-', '_', "img_$name" );
4555                                                 $internalParamMap[$magicName] = array( $type, $name );
4556                                         }
4557                                 }
4558                         }
4559
4560                         # Add handler params
4561                         $paramMap = $internalParamMap;
4562                         if ( $handler ) {
4563                                 $handlerParamMap = $handler->getParamMap();
4564                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4565                                         $paramMap[$magic] = array( 'handler', $paramName );
4566                                 }
4567                         }
4568                         $this->mImageParams[$handlerClass] = $paramMap;
4569                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4570                 }
4571                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4572         }
4573
4574         /**
4575          * Parse image options text and use it to make an image
4576          *
4577          * @param $title Title
4578          * @param $options String
4579          * @param $holders LinkHolderArray
4580          */
4581         function makeImage( $title, $options, $holders = false ) {
4582                 # Check if the options text is of the form "options|alt text"
4583                 # Options are:
4584                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4585                 #  * left       no resizing, just left align. label is used for alt= only
4586                 #  * right      same, but right aligned
4587                 #  * none       same, but not aligned
4588                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4589                 #  * center     center the image
4590                 #  * frame      Keep original image size, no magnify-button.
4591                 #  * framed     Same as "frame"
4592                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4593                 #  * upright    reduce width for upright images, rounded to full __0 px
4594                 #  * border     draw a 1px border around the image
4595                 #  * alt        Text for HTML alt attribute (defaults to empty)
4596                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4597                 # vertical-align values (no % or length right now):
4598                 #  * baseline
4599                 #  * sub
4600                 #  * super
4601                 #  * top
4602                 #  * text-top
4603                 #  * middle
4604                 #  * bottom
4605                 #  * text-bottom
4606
4607                 $parts = StringUtils::explode( "|", $options );
4608                 $sk = $this->mOptions->getSkin( $this->mTitle );
4609
4610                 # Give extensions a chance to select the file revision for us
4611                 $skip = $time = $descQuery = false;
4612                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4613
4614                 if ( $skip ) {
4615                         return $sk->link( $title );
4616                 }
4617
4618                 # Get the file
4619                 $file = wfFindFile( $title, array( 'time' => $time ) );
4620                 # Get parameter map
4621                 $handler = $file ? $file->getHandler() : false;
4622
4623                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4624
4625                 # Process the input parameters
4626                 $caption = '';
4627                 $params = array( 'frame' => array(), 'handler' => array(),
4628                         'horizAlign' => array(), 'vertAlign' => array() );
4629                 foreach ( $parts as $part ) {
4630                         $part = trim( $part );
4631                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4632                         $validated = false;
4633                         if ( isset( $paramMap[$magicName] ) ) {
4634                                 list( $type, $paramName ) = $paramMap[$magicName];
4635
4636                                 # Special case; width and height come in one variable together
4637                                 if ( $type === 'handler' && $paramName === 'width' ) {
4638                                         $m = array();
4639                                         # (bug 13500) In both cases (width/height and width only),
4640                                         # permit trailing "px" for backward compatibility.
4641                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4642                                                 $width = intval( $m[1] );
4643                                                 $height = intval( $m[2] );
4644                                                 if ( $handler->validateParam( 'width', $width ) ) {
4645                                                         $params[$type]['width'] = $width;
4646                                                         $validated = true;
4647                                                 }
4648                                                 if ( $handler->validateParam( 'height', $height ) ) {
4649                                                         $params[$type]['height'] = $height;
4650                                                         $validated = true;
4651                                                 }
4652                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4653                                                 $width = intval( $value );
4654                                                 if ( $handler->validateParam( 'width', $width ) ) {
4655                                                         $params[$type]['width'] = $width;
4656                                                         $validated = true;
4657                                                 }
4658                                         } # else no validation -- bug 13436
4659                                 } else {
4660                                         if ( $type === 'handler' ) {
4661                                                 # Validate handler parameter
4662                                                 $validated = $handler->validateParam( $paramName, $value );
4663                                         } else {
4664                                                 # Validate internal parameters
4665                                                 switch( $paramName ) {
4666                                                 case 'manualthumb':
4667                                                 case 'alt':
4668                                                         # @todo Fixme: possibly check validity here for
4669                                                         # manualthumb? downstream behavior seems odd with
4670                                                         # missing manual thumbs.
4671                                                         $validated = true;
4672                                                         $value = $this->stripAltText( $value, $holders );
4673                                                         break;
4674                                                 case 'link':
4675                                                         $chars = self::EXT_LINK_URL_CLASS;
4676                                                         $prots = $this->mUrlProtocols;
4677                                                         if ( $value === '' ) {
4678                                                                 $paramName = 'no-link';
4679                                                                 $value = true;
4680                                                                 $validated = true;
4681                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4682                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4683                                                                         $paramName = 'link-url';
4684                                                                         $this->mOutput->addExternalLink( $value );
4685                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4686                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4687                                                                         }
4688                                                                         $validated = true;
4689                                                                 }
4690                                                         } else {
4691                                                                 $linkTitle = Title::newFromText( $value );
4692                                                                 if ( $linkTitle ) {
4693                                                                         $paramName = 'link-title';
4694                                                                         $value = $linkTitle;
4695                                                                         $this->mOutput->addLink( $linkTitle );
4696                                                                         $validated = true;
4697                                                                 }
4698                                                         }
4699                                                         break;
4700                                                 default:
4701                                                         # Most other things appear to be empty or numeric...
4702                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4703                                                 }
4704                                         }
4705
4706                                         if ( $validated ) {
4707                                                 $params[$type][$paramName] = $value;
4708                                         }
4709                                 }
4710                         }
4711                         if ( !$validated ) {
4712                                 $caption = $part;
4713                         }
4714                 }
4715
4716                 # Process alignment parameters
4717                 if ( $params['horizAlign'] ) {
4718                         $params['frame']['align'] = key( $params['horizAlign'] );
4719                 }
4720                 if ( $params['vertAlign'] ) {
4721                         $params['frame']['valign'] = key( $params['vertAlign'] );
4722                 }
4723
4724                 $params['frame']['caption'] = $caption;
4725
4726                 # Will the image be presented in a frame, with the caption below?
4727                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4728                                  isset( $params['frame']['framed'] ) ||
4729                                  isset( $params['frame']['thumbnail'] ) ||
4730                                  isset( $params['frame']['manualthumb'] );
4731
4732                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4733                 # came to also set the caption, ordinary text after the image -- which
4734                 # makes no sense, because that just repeats the text multiple times in
4735                 # screen readers.  It *also* came to set the title attribute.
4736                 #
4737                 # Now that we have an alt attribute, we should not set the alt text to
4738                 # equal the caption: that's worse than useless, it just repeats the
4739                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4740                 # use the unnamed parameter for alt text as well, just for the time be-
4741                 # ing, if the unnamed param is set and the alt param is not.
4742                 #
4743                 # For the future, we need to figure out if we want to tweak this more,
4744                 # e.g., introducing a title= parameter for the title; ignoring the un-
4745                 # named parameter entirely for images without a caption; adding an ex-
4746                 # plicit caption= parameter and preserving the old magic unnamed para-
4747                 # meter for BC; ...
4748                 if ( $imageIsFramed ) { # Framed image
4749                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4750                                 # No caption or alt text, add the filename as the alt text so
4751                                 # that screen readers at least get some description of the image
4752                                 $params['frame']['alt'] = $title->getText();
4753                         }
4754                         # Do not set $params['frame']['title'] because tooltips don't make sense
4755                         # for framed images
4756                 } else { # Inline image
4757                         if ( !isset( $params['frame']['alt'] ) ) {
4758                                 # No alt text, use the "caption" for the alt text
4759                                 if ( $caption !== '') {
4760                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4761                                 } else {
4762                                         # No caption, fall back to using the filename for the
4763                                         # alt text
4764                                         $params['frame']['alt'] = $title->getText();
4765                                 }
4766                         }
4767                         # Use the "caption" for the tooltip text
4768                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4769                 }
4770
4771                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4772
4773                 # Linker does the rest
4774                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery, $this->mOptions->getThumbSize() );
4775
4776                 # Give the handler a chance to modify the parser object
4777                 if ( $handler ) {
4778                         $handler->parserTransformHook( $this, $file );
4779                 }
4780
4781                 return $ret;
4782         }
4783
4784         protected function stripAltText( $caption, $holders ) {
4785                 # Strip bad stuff out of the title (tooltip).  We can't just use
4786                 # replaceLinkHoldersText() here, because if this function is called
4787                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4788                 if ( $holders ) {
4789                         $tooltip = $holders->replaceText( $caption );
4790                 } else {
4791                         $tooltip = $this->replaceLinkHoldersText( $caption );
4792                 }
4793
4794                 # make sure there are no placeholders in thumbnail attributes
4795                 # that are later expanded to html- so expand them now and
4796                 # remove the tags
4797                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4798                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4799
4800                 return $tooltip;
4801         }
4802
4803         /**
4804          * Set a flag in the output object indicating that the content is dynamic and
4805          * shouldn't be cached.
4806          */
4807         function disableCache() {
4808                 wfDebug( "Parser output marked as uncacheable.\n" );
4809                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
4810                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
4811         }
4812
4813         /**
4814          * Callback from the Sanitizer for expanding items found in HTML attribute
4815          * values, so they can be safely tested and escaped.
4816          *
4817          * @param $text String
4818          * @param $frame PPFrame
4819          * @return String
4820          * @private
4821          */
4822         function attributeStripCallback( &$text, $frame = false ) {
4823                 $text = $this->replaceVariables( $text, $frame );
4824                 $text = $this->mStripState->unstripBoth( $text );
4825                 return $text;
4826         }
4827
4828         /**
4829          * Accessor
4830          */
4831         function getTags() {
4832                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4833         }
4834
4835         /**
4836          * Break wikitext input into sections, and either pull or replace
4837          * some particular section's text.
4838          *
4839          * External callers should use the getSection and replaceSection methods.
4840          *
4841          * @param $text String: Page wikitext
4842          * @param $section String: a section identifier string of the form:
4843          *   <flag1> - <flag2> - ... - <section number>
4844          *
4845          * Currently the only recognised flag is "T", which means the target section number
4846          * was derived during a template inclusion parse, in other words this is a template
4847          * section edit link. If no flags are given, it was an ordinary section edit link.
4848          * This flag is required to avoid a section numbering mismatch when a section is
4849          * enclosed by <includeonly> (bug 6563).
4850          *
4851          * The section number 0 pulls the text before the first heading; other numbers will
4852          * pull the given section along with its lower-level subsections. If the section is
4853          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4854          *
4855          * @param $mode String: one of "get" or "replace"
4856          * @param $newText String: replacement text for section data.
4857          * @return String: for "get", the extracted section text.
4858          *                 for "replace", the whole page with the section replaced.
4859          */
4860         private function extractSections( $text, $section, $mode, $newText='' ) {
4861                 global $wgTitle;
4862                 $this->clearState();
4863                 $this->setTitle( $wgTitle ); # not generally used but removes an ugly failure mode
4864                 $this->mOptions = new ParserOptions;
4865                 $this->setOutputType( self::OT_PLAIN );
4866                 $outText = '';
4867                 $frame = $this->getPreprocessor()->newFrame();
4868
4869                 # Process section extraction flags
4870                 $flags = 0;
4871                 $sectionParts = explode( '-', $section );
4872                 $sectionIndex = array_pop( $sectionParts );
4873                 foreach ( $sectionParts as $part ) {
4874                         if ( $part === 'T' ) {
4875                                 $flags |= self::PTD_FOR_INCLUSION;
4876                         }
4877                 }
4878                 # Preprocess the text
4879                 $root = $this->preprocessToDom( $text, $flags );
4880
4881                 # <h> nodes indicate section breaks
4882                 # They can only occur at the top level, so we can find them by iterating the root's children
4883                 $node = $root->getFirstChild();
4884
4885                 # Find the target section
4886                 if ( $sectionIndex == 0 ) {
4887                         # Section zero doesn't nest, level=big
4888                         $targetLevel = 1000;
4889                 } else {
4890                         while ( $node ) {
4891                                 if ( $node->getName() === 'h' ) {
4892                                         $bits = $node->splitHeading();
4893                                         if ( $bits['i'] == $sectionIndex ) {
4894                                                 $targetLevel = $bits['level'];
4895                                                 break;
4896                                         }
4897                                 }
4898                                 if ( $mode === 'replace' ) {
4899                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4900                                 }
4901                                 $node = $node->getNextSibling();
4902                         }
4903                 }
4904
4905                 if ( !$node ) {
4906                         # Not found
4907                         if ( $mode === 'get' ) {
4908                                 return $newText;
4909                         } else {
4910                                 return $text;
4911                         }
4912                 }
4913
4914                 # Find the end of the section, including nested sections
4915                 do {
4916                         if ( $node->getName() === 'h' ) {
4917                                 $bits = $node->splitHeading();
4918                                 $curLevel = $bits['level'];
4919                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4920                                         break;
4921                                 }
4922                         }
4923                         if ( $mode === 'get' ) {
4924                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4925                         }
4926                         $node = $node->getNextSibling();
4927                 } while ( $node );
4928
4929                 # Write out the remainder (in replace mode only)
4930                 if ( $mode === 'replace' ) {
4931                         # Output the replacement text
4932                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4933                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4934                         # Only add trailing whitespace if there is newText
4935                         if ( $newText != "" ) {
4936                                 $outText .= $newText . "\n\n";
4937                         }
4938
4939                         while ( $node ) {
4940                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4941                                 $node = $node->getNextSibling();
4942                         }
4943                 }
4944
4945                 if ( is_string( $outText ) ) {
4946                         # Re-insert stripped tags
4947                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4948                 }
4949
4950                 return $outText;
4951         }
4952
4953         /**
4954          * This function returns the text of a section, specified by a number ($section).
4955          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4956          * the first section before any such heading (section 0).
4957          *
4958          * If a section contains subsections, these are also returned.
4959          *
4960          * @param $text String: text to look in
4961          * @param $section String: section identifier
4962          * @param $deftext String: default to return if section is not found
4963          * @return string text of the requested section
4964          */
4965         public function getSection( $text, $section, $deftext='' ) {
4966                 return $this->extractSections( $text, $section, "get", $deftext );
4967         }
4968
4969         public function replaceSection( $oldtext, $section, $text ) {
4970                 return $this->extractSections( $oldtext, $section, "replace", $text );
4971         }
4972
4973         /**
4974          * Get the ID of the revision we are parsing
4975          *
4976          * @return Mixed: integer or null
4977          */
4978         function getRevisionId() {
4979                 return $this->mRevisionId;
4980         }
4981
4982         /**
4983          * Get the revision object for $this->mRevisionId
4984          *
4985          * @return either a Revision object or null
4986          */
4987         protected function getRevisionObject() {
4988                 if ( !is_null( $this->mRevisionObject ) ) {
4989                         return $this->mRevisionObject;
4990                 }
4991                 if ( is_null( $this->mRevisionId ) ) {
4992                         return null;
4993                 }
4994
4995                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
4996                 return $this->mRevisionObject;
4997         }
4998
4999         /**
5000          * Get the timestamp associated with the current revision, adjusted for
5001          * the default server-local timestamp
5002          */
5003         function getRevisionTimestamp() {
5004                 if ( is_null( $this->mRevisionTimestamp ) ) {
5005                         wfProfileIn( __METHOD__ );
5006
5007                         $revObject = $this->getRevisionObject();
5008                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5009
5010                         if( $timestamp !== false ) {
5011                                 global $wgContLang;
5012
5013                                 # The cryptic '' timezone parameter tells to use the site-default
5014                                 # timezone offset instead of the user settings.
5015                                 #
5016                                 # Since this value will be saved into the parser cache, served
5017                                 # to other users, and potentially even used inside links and such,
5018                                 # it needs to be consistent for all visitors.
5019                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5020                         }
5021
5022                         wfProfileOut( __METHOD__ );
5023                 }
5024                 return $this->mRevisionTimestamp;
5025         }
5026
5027         /**
5028          * Get the name of the user that edited the last revision
5029          *
5030          * @return String: user name
5031          */
5032         function getRevisionUser() {
5033                 if( is_null( $this->mRevisionUser ) ) {
5034                         $revObject = $this->getRevisionObject();
5035
5036                         # if this template is subst: the revision id will be blank,
5037                         # so just use the current user's name
5038                         if( $revObject ) {
5039                                 $this->mRevisionUser = $revObject->getUserText();
5040                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5041                                 $this->mRevisionUser = $this->getUser()->getName();
5042                         }
5043                 }
5044                 return $this->mRevisionUser;
5045         }
5046
5047         /**
5048          * Mutator for $mDefaultSort
5049          *
5050          * @param $sort New value
5051          */
5052         public function setDefaultSort( $sort ) {
5053                 $this->mDefaultSort = $sort;
5054                 $this->mOutput->setProperty( 'defaultsort', $sort );
5055         }
5056
5057         /**
5058          * Accessor for $mDefaultSort
5059          * Will use the title/prefixed title if none is set
5060          *
5061          * @return string
5062          */
5063         public function getDefaultSort() {
5064                 if ( $this->mDefaultSort !== false ) {
5065                         return $this->mDefaultSort;
5066                 } else {
5067                         return $this->mTitle->getCategorySortkey();
5068                 }
5069         }
5070
5071         /**
5072          * Accessor for $mDefaultSort
5073          * Unlike getDefaultSort(), will return false if none is set
5074          *
5075          * @return string or false
5076          */
5077         public function getCustomDefaultSort() {
5078                 return $this->mDefaultSort;
5079         }
5080
5081         /**
5082          * Try to guess the section anchor name based on a wikitext fragment
5083          * presumably extracted from a heading, for example "Header" from
5084          * "== Header ==".
5085          */
5086         public function guessSectionNameFromWikiText( $text ) {
5087                 # Strip out wikitext links(they break the anchor)
5088                 $text = $this->stripSectionName( $text );
5089                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5090                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5091         }
5092
5093         /**
5094          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5095          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5096          * as something other than UTF-8 (apparently?), resulting in breakage.
5097          *
5098          * @param $text String: The section name
5099          * @return string An anchor
5100          */
5101         public function guessLegacySectionNameFromWikiText( $text ) {
5102                 # Strip out wikitext links(they break the anchor)
5103                 $text = $this->stripSectionName( $text );
5104                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5105                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5106         }
5107
5108         /**
5109          * Strips a text string of wikitext for use in a section anchor
5110          *
5111          * Accepts a text string and then removes all wikitext from the
5112          * string and leaves only the resultant text (i.e. the result of
5113          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5114          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5115          * to create valid section anchors by mimicing the output of the
5116          * parser when headings are parsed.
5117          *
5118          * @param $text String: text string to be stripped of wikitext
5119          * for use in a Section anchor
5120          * @return Filtered text string
5121          */
5122         public function stripSectionName( $text ) {
5123                 # Strip internal link markup
5124                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5125                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5126
5127                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5128                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5129                 # on how many empty links there are on the page - need to figure that out.
5130                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5131
5132                 # Parse wikitext quotes (italics & bold)
5133                 $text = $this->doQuotes( $text );
5134
5135                 # Strip HTML tags
5136                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5137                 return $text;
5138         }
5139
5140         /**
5141          * strip/replaceVariables/unstrip for preprocessor regression testing
5142          */
5143         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5144                 $this->clearState();
5145                 if ( !$title instanceof Title ) {
5146                         $title = Title::newFromText( $title );
5147                 }
5148                 $this->mTitle = $title;
5149                 $options->resetUsage();
5150                 $this->mOptions = $options;
5151                 $this->setOutputType( $outputType );
5152                 $text = $this->replaceVariables( $text );
5153                 $text = $this->mStripState->unstripBoth( $text );
5154                 $text = Sanitizer::removeHTMLtags( $text );
5155                 return $text;
5156         }
5157
5158         function testPst( $text, $title, $options ) {
5159                 global $wgUser;
5160                 if ( !$title instanceof Title ) {
5161                         $title = Title::newFromText( $title );
5162                 }
5163                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5164         }
5165
5166         function testPreprocess( $text, $title, $options ) {
5167                 if ( !$title instanceof Title ) {
5168                         $title = Title::newFromText( $title );
5169                 }
5170                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5171         }
5172
5173         function markerSkipCallback( $s, $callback ) {
5174                 $i = 0;
5175                 $out = '';
5176                 while ( $i < strlen( $s ) ) {
5177                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5178                         if ( $markerStart === false ) {
5179                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5180                                 break;
5181                         } else {
5182                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5183                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5184                                 if ( $markerEnd === false ) {
5185                                         $out .= substr( $s, $markerStart );
5186                                         break;
5187                                 } else {
5188                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5189                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5190                                         $i = $markerEnd;
5191                                 }
5192                         }
5193                 }
5194                 return $out;
5195         }
5196
5197         function serialiseHalfParsedText( $text ) {
5198                 $data = array();
5199                 $data['text'] = $text;
5200
5201                 # First, find all strip markers, and store their
5202                 #  data in an array.
5203                 $stripState = new StripState;
5204                 $pos = 0;
5205                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5206                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5207                 {
5208                         $end_pos += strlen( self::MARKER_SUFFIX );
5209                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5210
5211                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5212                                 $replaceArray = $stripState->general;
5213                                 $stripText = $this->mStripState->general->data[$marker];
5214                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5215                                 $replaceArray = $stripState->nowiki;
5216                                 $stripText = $this->mStripState->nowiki->data[$marker];
5217                         } else {
5218                                 throw new MWException( "Hanging strip marker: '$marker'." );
5219                         }
5220
5221                         $replaceArray->setPair( $marker, $stripText );
5222                         $pos = $end_pos;
5223                 }
5224                 $data['stripstate'] = $stripState;
5225
5226                 # Now, find all of our links, and store THEIR
5227                 #  data in an array! :)
5228                 $links = array( 'internal' => array(), 'interwiki' => array() );
5229                 $pos = 0;
5230
5231                 # Internal links
5232                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5233                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5234
5235                         $ns = trim( $ns );
5236                         if ( empty( $links['internal'][$ns] ) ) {
5237                                 $links['internal'][$ns] = array();
5238                         }
5239
5240                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5241                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5242                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5243                 }
5244
5245                 $pos = 0;
5246
5247                 # Interwiki links
5248                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5249                         $data = substr( $text, $start_pos );
5250                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5251                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5252                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5253                 }
5254
5255                 $data['linkholder'] = $links;
5256
5257                 return $data;
5258         }
5259
5260         /**
5261          * TODO: document
5262          * @param $data Array
5263          * @param $intPrefix String unique identifying prefix
5264          * @return String
5265          */
5266         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5267                 if ( !$intPrefix ) {
5268                         $intPrefix = self::getRandomString();
5269                 }
5270
5271                 # First, extract the strip state.
5272                 $stripState = $data['stripstate'];
5273                 $this->mStripState->general->merge( $stripState->general );
5274                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5275
5276                 # Now, extract the text, and renumber links
5277                 $text = $data['text'];
5278                 $links = $data['linkholder'];
5279
5280                 # Internal...
5281                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5282                         foreach ( $nsLinks as $key => $entry ) {
5283                                 $newKey = $intPrefix . '-' . $key;
5284                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5285
5286                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5287                         }
5288                 }
5289
5290                 # Interwiki...
5291                 foreach ( $links['interwiki'] as $key => $entry ) {
5292                         $newKey = "$intPrefix-$key";
5293                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5294
5295                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5296                 }
5297
5298                 # Should be good to go.
5299                 return $text;
5300         }
5301 }
5302
5303 /**
5304  * @todo document, briefly.
5305  * @ingroup Parser
5306  */
5307 class StripState {
5308         var $general, $nowiki;
5309
5310         function __construct() {
5311                 $this->general = new ReplacementArray;
5312                 $this->nowiki = new ReplacementArray;
5313         }
5314
5315         function unstripGeneral( $text ) {
5316                 wfProfileIn( __METHOD__ );
5317                 do {
5318                         $oldText = $text;
5319                         $text = $this->general->replace( $text );
5320                 } while ( $text !== $oldText );
5321                 wfProfileOut( __METHOD__ );
5322                 return $text;
5323         }
5324
5325         function unstripNoWiki( $text ) {
5326                 wfProfileIn( __METHOD__ );
5327                 do {
5328                         $oldText = $text;
5329                         $text = $this->nowiki->replace( $text );
5330                 } while ( $text !== $oldText );
5331                 wfProfileOut( __METHOD__ );
5332                 return $text;
5333         }
5334
5335         function unstripBoth( $text ) {
5336                 wfProfileIn( __METHOD__ );
5337                 do {
5338                         $oldText = $text;
5339                         $text = $this->general->replace( $text );
5340                         $text = $this->nowiki->replace( $text );
5341                 } while ( $text !== $oldText );
5342                 wfProfileOut( __METHOD__ );
5343                 return $text;
5344         }
5345 }
5346
5347 /**
5348  * @todo document, briefly.
5349  * @ingroup Parser
5350  */
5351 class OnlyIncludeReplacer {
5352         var $output = '';
5353
5354         function replace( $matches ) {
5355                 if ( substr( $matches[1], -1 ) === "\n" ) {
5356                         $this->output .= substr( $matches[1], 0, -1 );
5357                 } else {
5358                         $this->output .= $matches[1];
5359                 }
5360         }
5361 }