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() / cleanSigInSig()
  26  *     Cleans a signature before saving it to preferences
  27  * getSection()
  28  *     Return the content of a section from an article for section editing
  29  * replaceSection()
  30  *     Replaces a section by number inside an article
  31  * getPreloadText()
  32  *     Removes <noinclude> sections, and <includeonly> tags.
  33  *
  34  * Globals used:
  35  *    objects:   $wgLang, $wgContLang
  36  *
  37  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  38  *
  39  * settings:
  40  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  41  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  42  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  43  *  $wgMaxArticleSize*
  44  *
  45  *  * only within ParserOptions
  46  * </pre>
  47  *
  48  * @ingroup Parser
  49  */
  50 class Parser {
  51         /**
  52          * Update this version number when the ParserOutput format
  53          * changes in an incompatible way, so the parser cache
  54          * can automatically discard old data.
  55          */
  56         const VERSION = '1.6.4';
  57
  58         # Flags for Parser::setFunctionHook
  59         # Also available as global constants from Defines.php
  60         const SFH_NO_HASH = 1;
  61         const SFH_OBJECT_ARGS = 2;
  62
  63         # Constants needed for external link processing
  64         # Everything except bracket, space, or control characters
  65         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  66         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  67                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  68
  69         # State constants for the definition list colon extraction
  70         const COLON_STATE_TEXT = 0;
  71         const COLON_STATE_TAG = 1;
  72         const COLON_STATE_TAGSTART = 2;
  73         const COLON_STATE_CLOSETAG = 3;
  74         const COLON_STATE_TAGSLASH = 4;
  75         const COLON_STATE_COMMENT = 5;
  76         const COLON_STATE_COMMENTDASH = 6;
  77         const COLON_STATE_COMMENTDASHDASH = 7;
  78
  79         # Flags for preprocessToDom
  80         const PTD_FOR_INCLUSION = 1;
  81
  82         # Allowed values for $this->mOutputType
  83         # Parameter to startExternalParse().
  84         const OT_HTML = 1; # like parse()
  85         const OT_WIKI = 2; # like preSaveTransform()
  86         const OT_PREPROCESS = 3; # like preprocess()
  87         const OT_MSG = 3;
  88         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  89
  90         # Marker Suffix needs to be accessible staticly.
  91         const MARKER_SUFFIX = "-QINU\x7f";
  92
  93         # Persistent:
  94         var $mTagHooks = array();
  95         var $mTransparentTagHooks = array();
  96         var $mFunctionHooks = array();
  97         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
  98         var $mFunctionTagHooks = array();
  99         var $mStripList  = array();
 100         var $mDefaultStripList  = array();
 101         var $mVarCache = array();
 102         var $mImageParams = array();
 103         var $mImageParamsMagicArray = array();
 104         var $mMarkerIndex = 0;
 105         var $mFirstCall = true;
 106         var $mVariables, $mSubstWords; # Initialised by initialiseVariables()
 107         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 108
 109
 110         # Cleared with clearState():
 111         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 112         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 113         var $mLinkHolders, $mLinkID;
 114         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 115         var $mTplExpandCache; # empty-frame expansion cache
 116         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 117         var $mExpensiveFunctionCount; # number of expensive parser function calls
 118         var $mUser; # User object; only used when doing pre-save transform
 119
 120         # Temporary
 121         # These are variables reset at least once per parse regardless of $clearState
 122         var $mOptions;      # ParserOptions object
 123         var $mTitle;        # Title context, used for self-link rendering and similar things
 124         var $mOutputType;   # Output type, one of the OT_xxx constants
 125         var $ot;            # Shortcut alias, see setOutputType()
 126         var $mRevisionObject; # The revision object of the specified revision ID
 127         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 128         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 129         var $mRevisionUser; # Userto display in {{REVISIONUSER}} tag
 130         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 131
 132         /**
 133          * Constructor
 134          *
 135          * @public
 136          */
 137         function __construct( $conf = array() ) {
 138                 $this->mConf = $conf;
 139                 $this->mUrlProtocols = wfUrlProtocols();
 140                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 141                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 142                 if ( isset( $conf['preprocessorClass'] ) ) {
 143                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 144                 } elseif ( extension_loaded( 'domxml' ) ) {
 145                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 146                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 147                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 148                 } elseif ( extension_loaded( 'dom' ) ) {
 149                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 150                 } else {
 151                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 152                 }
 153         }
 154
 155         /**
 156          * Reduce memory usage to reduce the impact of circular references
 157          */
 158         function __destruct() {
 159                 if ( isset( $this->mLinkHolders ) ) {
 160                         $this->mLinkHolders->__destruct();
 161                 }
 162                 foreach ( $this as $name => $value ) {
 163                         unset( $this->$name );
 164                 }
 165         }
 166
 167         /**
 168          * Do various kinds of initialisation on the first call of the parser
 169          */
 170         function firstCallInit() {
 171                 if ( !$this->mFirstCall ) {
 172                         return;
 173                 }
 174                 $this->mFirstCall = false;
 175
 176                 wfProfileIn( __METHOD__ );
 177
 178                 CoreParserFunctions::register( $this );
 179                 CoreTagHooks::register( $this );
 180                 $this->initialiseVariables();
 181
 182                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 183                 wfProfileOut( __METHOD__ );
 184         }
 185
 186         /**
 187          * Clear Parser state
 188          *
 189          * @private
 190          */
 191         function clearState() {
 192                 wfProfileIn( __METHOD__ );
 193                 if ( $this->mFirstCall ) {
 194                         $this->firstCallInit();
 195                 }
 196                 $this->mOutput = new ParserOutput;
 197                 if( $this->mOptions ) {
 198                         $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 199                 }
 200                 $this->mAutonumber = 0;
 201                 $this->mLastSection = '';
 202                 $this->mDTopen = false;
 203                 $this->mIncludeCount = array();
 204                 $this->mStripState = new StripState;
 205                 $this->mArgStack = false;
 206                 $this->mInPre = false;
 207                 $this->mLinkHolders = new LinkHolderArray( $this );
 208                 $this->mLinkID = 0;
 209                 $this->mRevisionObject = $this->mRevisionTimestamp =
 210                         $this->mRevisionId = $this->mRevisionUser = null;
 211                 $this->mVarCache = array();
 212                 $this->mUser = null;
 213
 214                 /**
 215                  * Prefix for temporary replacement strings for the multipass parser.
 216                  * \x07 should never appear in input as it's disallowed in XML.
 217                  * Using it at the front also gives us a little extra robustness
 218                  * since it shouldn't match when butted up against identifier-like
 219                  * string constructs.
 220                  *
 221                  * Must not consist of all title characters, or else it will change
 222                  * the behaviour of <nowiki> in a link.
 223                  */
 224                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 225                 # Changed to \x7f to allow XML double-parsing -- TS
 226                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 227
 228
 229                 # Clear these on every parse, bug 4549
 230                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 231
 232                 $this->mShowToc = true;
 233                 $this->mForceTocPosition = false;
 234                 $this->mIncludeSizes = array(
 235                         'post-expand' => 0,
 236                         'arg' => 0,
 237                 );
 238                 $this->mPPNodeCount = 0;
 239                 $this->mDefaultSort = false;
 240                 $this->mHeadings = array();
 241                 $this->mDoubleUnderscores = array();
 242                 $this->mExpensiveFunctionCount = 0;
 243
 244                 # Fix cloning
 245                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 246                         $this->mPreprocessor = null;
 247                 }
 248
 249                 wfRunHooks( 'ParserClearState', array( &$this ) );
 250                 wfProfileOut( __METHOD__ );
 251         }
 252
 253         /**
 254          * Convert wikitext to HTML
 255          * Do not call this function recursively.
 256          *
 257          * @param $text String: text we want to parse
 258          * @param $title A title object
 259          * @param $options ParserOptions
 260          * @param $linestart boolean
 261          * @param $clearState boolean
 262          * @param $revid Int: number to pass in {{REVISIONID}}
 263          * @return ParserOutput a ParserOutput
 264          */
 265         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 266                 /**
 267                  * First pass--just handle <nowiki> sections, pass the rest off
 268                  * to internalParse() which does all the real work.
 269                  */
 270
 271                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 272                 $fname = __METHOD__.'-' . wfGetCaller();
 273                 wfProfileIn( __METHOD__ );
 274                 wfProfileIn( $fname );
 275
 276                 $this->startExternalParse( $title, $options, self::OT_HTML, $clearState );
 277
 278                 $oldRevisionId = $this->mRevisionId;
 279                 $oldRevisionObject = $this->mRevisionObject;
 280                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 281                 $oldRevisionUser = $this->mRevisionUser;
 282                 if ( $revid !== null ) {
 283                         $this->mRevisionId = $revid;
 284                         $this->mRevisionObject = null;
 285                         $this->mRevisionTimestamp = null;
 286                         $this->mRevisionUser = null;
 287                 }
 288
 289                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 290                 # No more strip!
 291                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 292                 $text = $this->internalParse( $text );
 293
 294                 $text = $this->mStripState->unstripGeneral( $text );
 295
 296                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 297                 $fixtags = array(
 298                         # french spaces, last one Guillemet-left
 299                         # only if there is something before the space
 300                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 301                         # french spaces, Guillemet-right
 302                         '/(\\302\\253) /' => '\\1&#160;',
 303                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 304                 );
 305                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 306
 307                 $text = $this->doBlockLevels( $text, $linestart );
 308
 309                 $this->replaceLinkHolders( $text );
 310
 311                 /**
 312                  * The page doesn't get language converted if
 313                  * a) It's disabled
 314                  * b) Content isn't converted
 315                  * c) It's a conversion table
 316                  */
 317                 if ( !( $wgDisableLangConversion
 318                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 319                                 || $this->mTitle->isConversionTable() ) ) {
 320
 321                         # The position of the convert() call should not be changed. it
 322                         # assumes that the links are all replaced and the only thing left
 323                         # is the <nowiki> mark.
 324
 325                         $text = $wgContLang->convert( $text );
 326                 }
 327
 328                 /**
 329                  * A converted title will be provided in the output object if title and
 330                  * content conversion are enabled, the article text does not contain
 331                  * a conversion-suppressing double-underscore tag, and no
 332                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 333                  * automatic link conversion.
 334                  */
 335                 if ( !( $wgDisableLangConversion
 336                                 || $wgDisableTitleConversion
 337                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 338                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 339                                 || $this->mOutput->getDisplayTitle() !== false ) )
 340                 {
 341                         $convruletitle = $wgContLang->getConvRuleTitle();
 342                         if ( $convruletitle ) {
 343                                 $this->mOutput->setTitleText( $convruletitle );
 344                         } else {
 345                                 $titleText = $wgContLang->convertTitle( $title );
 346                                 $this->mOutput->setTitleText( $titleText );
 347                         }
 348                 }
 349
 350                 $text = $this->mStripState->unstripNoWiki( $text );
 351
 352                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 353
 354 //!JF Move to its own function
 355
 356                 $uniq_prefix = $this->mUniqPrefix;
 357                 $matches = array();
 358                 $elements = array_keys( $this->mTransparentTagHooks );
 359                 $text = $this->extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 360
 361                 foreach ( $matches as $marker => $data ) {
 362                         list( $element, $content, $params, $tag ) = $data;
 363                         $tagName = strtolower( $element );
 364                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 365                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 366                         } else {
 367                                 $output = $tag;
 368                         }
 369                         $this->mStripState->general->setPair( $marker, $output );
 370                 }
 371                 $text = $this->mStripState->unstripGeneral( $text );
 372
 373                 $text = Sanitizer::normalizeCharReferences( $text );
 374
 375                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 376                         $text = MWTidy::tidy( $text );
 377                 } else {
 378                         # attempt to sanitize at least some nesting problems
 379                         # (bug #2702 and quite a few others)
 380                         $tidyregs = array(
 381                                 # ''Something [http://www.cool.com cool''] -->
 382                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 383                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 384                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 385                                 # fix up an anchor inside another anchor, only
 386                                 # at least for a single single nested link (bug 3695)
 387                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 388                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 389                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 390                                 # contains a div, so fix it up here; replace
 391                                 # div with escaped text
 392                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 393                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 394                                 # remove empty italic or bold tag pairs, some
 395                                 # introduced by rules above
 396                                 '/<([bi])><\/\\1>/' => '',
 397                         );
 398
 399                         $text = preg_replace(
 400                                 array_keys( $tidyregs ),
 401                                 array_values( $tidyregs ),
 402                                 $text );
 403                 }
 404                 global $wgExpensiveParserFunctionLimit;
 405                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 406                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 407                 }
 408
 409                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 410
 411                 # Information on include size limits, for the benefit of users who try to skirt them
 412                 if ( $this->mOptions->getEnableLimitReport() ) {
 413                         $max = $this->mOptions->getMaxIncludeSize();
 414                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 415                         $limitReport =
 416                                 "NewPP limit report\n" .
 417                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 418                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 419                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 420                                 $PFreport;
 421                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 422                         $text .= "\n<!-- \n$limitReport-->\n";
 423                 }
 424                 $this->mOutput->setText( $text );
 425
 426                 $this->mRevisionId = $oldRevisionId;
 427                 $this->mRevisionObject = $oldRevisionObject;
 428                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 429                 $this->mRevisionUser = $oldRevisionUser;
 430                 wfProfileOut( $fname );
 431                 wfProfileOut( __METHOD__ );
 432
 433                 return $this->mOutput;
 434         }
 435
 436         /**
 437          * Recursive parser entry point that can be called from an extension tag
 438          * hook.
 439          *
 440          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 441          *
 442          * @param $text String: text extension wants to have parsed
 443          * @param $frame PPFrame: The frame to use for expanding any template variables
 444          */
 445         function recursiveTagParse( $text, $frame=false ) {
 446                 wfProfileIn( __METHOD__ );
 447                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 448                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 449                 $text = $this->internalParse( $text, false, $frame );
 450                 wfProfileOut( __METHOD__ );
 451                 return $text;
 452         }
 453
 454         /**
 455          * Expand templates and variables in the text, producing valid, static wikitext.
 456          * Also removes comments.
 457          */
 458         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 459                 wfProfileIn( __METHOD__ );
 460                 $this->startExternalParse( $title, $options, self::OT_PREPROCESS, true );
 461                 if ( $revid !== null ) {
 462                         $this->mRevisionId = $revid;
 463                 }
 464                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 465                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 466                 $text = $this->replaceVariables( $text );
 467                 $text = $this->mStripState->unstripBoth( $text );
 468                 wfProfileOut( __METHOD__ );
 469                 return $text;
 470         }
 471
 472         /**
 473          * Process the wikitext for the ?preload= feature. (bug 5210)
 474          *
 475          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 476          * comments, templates, arguments, tags hooks and parser functions are untouched.
 477          */
 478         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 479                 # Parser (re)initialisation
 480                 $this->startExternalParse( $title, $options, self::OT_PLAIN, true );
 481
 482                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 483                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 484                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 485         }
 486
 487         /**
 488          * Get a random string
 489          *
 490          * @private
 491          * @static
 492          */
 493         static private function getRandomString() {
 494                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 495         }
 496
 497         /**
 498          * Set the current user.
 499          * Should only be used when doing pre-save transform.
 500          *
 501          * @param $user Mixed: User object or null (to reset)
 502          */
 503         function setUser( $user ) {
 504                 $this->mUser = $user;
 505         }
 506
 507         /**
 508          * Accessor for mUniqPrefix.
 509          *
 510          * @return String
 511          */
 512         public function uniqPrefix() {
 513                 if ( !isset( $this->mUniqPrefix ) ) {
 514                         # @todo Fixme: this is probably *horribly wrong*
 515                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 516                         # if this is called for a parser cache hit, the parser may not
 517                         # have ever been initialized in the first place.
 518                         # Not really sure what the heck is supposed to be going on here.
 519                         return '';
 520                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 521                 }
 522                 return $this->mUniqPrefix;
 523         }
 524
 525         /**
 526          * Set the context title
 527          */
 528         function setTitle( $t ) {
 529                 if ( !$t || $t instanceof FakeTitle ) {
 530                         $t = Title::newFromText( 'NO TITLE' );
 531                 }
 532
 533                 if ( strval( $t->getFragment() ) !== '' ) {
 534                         # Strip the fragment to avoid various odd effects
 535                         $this->mTitle = clone $t;
 536                         $this->mTitle->setFragment( '' );
 537                 } else {
 538                         $this->mTitle = $t;
 539                 }
 540         }
 541
 542         /**
 543          * Accessor for the Title object
 544          *
 545          * @return Title object
 546          */
 547         function getTitle() {
 548                 return $this->mTitle;
 549         }
 550
 551         /**
 552          * Accessor/mutator for the Title object
 553          *
 554          * @param $x New Title object or null to just get the current one
 555          * @return Title object
 556          */
 557         function Title( $x = null ) {
 558                 return wfSetVar( $this->mTitle, $x );
 559         }
 560
 561         /**
 562          * Set the output type
 563          *
 564          * @param $ot Integer: new value
 565          */
 566         function setOutputType( $ot ) {
 567                 $this->mOutputType = $ot;
 568                 # Shortcut alias
 569                 $this->ot = array(
 570                         'html' => $ot == self::OT_HTML,
 571                         'wiki' => $ot == self::OT_WIKI,
 572                         'pre' => $ot == self::OT_PREPROCESS,
 573                         'plain' => $ot == self::OT_PLAIN,
 574                 );
 575         }
 576
 577         /**
 578          * Accessor/mutator for the output type
 579          *
 580          * @param $x New value or null to just get the current one
 581          * @return Integer
 582          */
 583         function OutputType( $x = null ) {
 584                 return wfSetVar( $this->mOutputType, $x );
 585         }
 586
 587         /**
 588          * Get the ParserOutput object
 589          *
 590          * @return ParserOutput object
 591          */
 592         function getOutput() {
 593                 return $this->mOutput;
 594         }
 595
 596         /**
 597          * Get the ParserOptions object
 598          *
 599          * @return ParserOptions object
 600          */
 601         function getOptions() {
 602                 return $this->mOptions;
 603         }
 604
 605         /**
 606          * Accessor/mutator for the ParserOptions object
 607          *
 608          * @param $x New value or null to just get the current one
 609          * @return Current ParserOptions object
 610          */
 611         function Options( $x = null ) {
 612                 return wfSetVar( $this->mOptions, $x );
 613         }
 614
 615         function nextLinkID() {
 616                 return $this->mLinkID++;
 617         }
 618
 619         function getFunctionLang() {
 620                 global $wgLang, $wgContLang;
 621
 622                 $target = $this->mOptions->getTargetLanguage();
 623                 if ( $target !== null ) {
 624                         return $target;
 625                 } else {
 626                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 627                 }
 628         }
 629
 630         /**
 631          * Get a User object either from $this->mUser, if set, or from the
 632          * ParserOptions object otherwise
 633          *
 634          * @return User object
 635          */
 636         function getUser() {
 637                 if ( !is_null( $this->mUser ) ) {
 638                         return $this->mUser;
 639                 }
 640                 return $this->mOptions->getUser();
 641         }
 642
 643         /**
 644          * Get a preprocessor object
 645          *
 646          * @return Preprocessor instance
 647          */
 648         function getPreprocessor() {
 649                 if ( !isset( $this->mPreprocessor ) ) {
 650                         $class = $this->mPreprocessorClass;
 651                         $this->mPreprocessor = new $class( $this );
 652                 }
 653                 return $this->mPreprocessor;
 654         }
 655
 656         /**
 657          * Replaces all occurrences of HTML-style comments and the given tags
 658          * in the text with a random marker and returns the next text. The output
 659          * parameter $matches will be an associative array filled with data in
 660          * the form:
 661          *   'UNIQ-xxxxx' => array(
 662          *     'element',
 663          *     'tag content',
 664          *     array( 'param' => 'x' ),
 665          *     '<element param="x">tag content</element>' ) )
 666          *
 667          * @param $elements list of element names. Comments are always extracted.
 668          * @param $text Source text string.
 669          * @param $matches Out parameter, Array: extracted tags
 670          * @param $uniq_prefix
 671          * @return String: stripped text
 672          *
 673          * @static
 674          */
 675         public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 676                 static $n = 1;
 677                 $stripped = '';
 678                 $matches = array();
 679
 680                 $taglist = implode( '|', $elements );
 681                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 682
 683                 while ( $text != '' ) {
 684                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 685                         $stripped .= $p[0];
 686                         if ( count( $p ) < 5 ) {
 687                                 break;
 688                         }
 689                         if ( count( $p ) > 5 ) {
 690                                 # comment
 691                                 $element    = $p[4];
 692                                 $attributes = '';
 693                                 $close      = '';
 694                                 $inside     = $p[5];
 695                         } else {
 696                                 # tag
 697                                 $element    = $p[1];
 698                                 $attributes = $p[2];
 699                                 $close      = $p[3];
 700                                 $inside     = $p[4];
 701                         }
 702
 703                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 704                         $stripped .= $marker;
 705
 706                         if ( $close === '/>' ) {
 707                                 # Empty element tag, <tag />
 708                                 $content = null;
 709                                 $text = $inside;
 710                                 $tail = null;
 711                         } else {
 712                                 if ( $element === '!--' ) {
 713                                         $end = '/(-->)/';
 714                                 } else {
 715                                         $end = "/(<\\/$element\\s*>)/i";
 716                                 }
 717                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 718                                 $content = $q[0];
 719                                 if ( count( $q ) < 3 ) {
 720                                         # No end tag -- let it run out to the end of the text.
 721                                         $tail = '';
 722                                         $text = '';
 723                                 } else {
 724                                         $tail = $q[1];
 725                                         $text = $q[2];
 726                                 }
 727                         }
 728
 729                         $matches[$marker] = array( $element,
 730                                 $content,
 731                                 Sanitizer::decodeTagAttributes( $attributes ),
 732                                 "<$element$attributes$close$content$tail" );
 733                 }
 734                 return $stripped;
 735         }
 736
 737         /**
 738          * Get a list of strippable XML-like elements
 739          */
 740         function getStripList() {
 741                 return $this->mStripList;
 742         }
 743
 744         /**
 745          * @deprecated use replaceVariables
 746          */
 747         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 748                 return $text;
 749         }
 750
 751         /**
 752          * Restores pre, math, and other extensions removed by strip()
 753          *
 754          * always call unstripNoWiki() after this one
 755          * @private
 756          * @deprecated use $this->mStripState->unstrip()
 757          */
 758         function unstrip( $text, $state ) {
 759                 return $state->unstripGeneral( $text );
 760         }
 761
 762         /**
 763          * Always call this after unstrip() to preserve the order
 764          *
 765          * @private
 766          * @deprecated use $this->mStripState->unstrip()
 767          */
 768         function unstripNoWiki( $text, $state ) {
 769                 return $state->unstripNoWiki( $text );
 770         }
 771
 772         /**
 773          * @deprecated use $this->mStripState->unstripBoth()
 774          */
 775         function unstripForHTML( $text ) {
 776                 return $this->mStripState->unstripBoth( $text );
 777         }
 778
 779         /**
 780          * Add an item to the strip state
 781          * Returns the unique tag which must be inserted into the stripped text
 782          * The tag will be replaced with the original text in unstrip()
 783          *
 784          * @private
 785          */
 786         function insertStripItem( $text ) {
 787                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 788                 $this->mMarkerIndex++;
 789                 $this->mStripState->general->setPair( $rnd, $text );
 790                 return $rnd;
 791         }
 792
 793         /**
 794          * Interface with html tidy
 795          * @deprecated Use MWTidy::tidy()
 796          */
 797         public static function tidy( $text ) {
 798                 wfDeprecated( __METHOD__ );
 799                 return MWTidy::tidy( $text );
 800         }
 801
 802         /**
 803          * parse the wiki syntax used to render tables
 804          *
 805          * @private
 806          */
 807         function doTableStuff( $text ) {
 808                 wfProfileIn( __METHOD__ );
 809
 810                 $lines = StringUtils::explode( "\n", $text );
 811                 $out = '';
 812                 $td_history = array(); # Is currently a td tag open?
 813                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 814                 $tr_history = array(); # Is currently a tr tag open?
 815                 $tr_attributes = array(); # history of tr attributes
 816                 $has_opened_tr = array(); # Did this table open a <tr> element?
 817                 $indent_level = 0; # indent level of the table
 818
 819                 foreach ( $lines as $outLine ) {
 820                         $line = trim( $outLine );
 821
 822                         if ( $line === '' ) { # empty line, go to next line
 823                                 $out .= $outLine."\n";
 824                                 continue;
 825                         }
 826
 827                         $first_character = $line[0];
 828                         $matches = array();
 829
 830                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 831                                 # First check if we are starting a new table
 832                                 $indent_level = strlen( $matches[1] );
 833
 834                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 835                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 836
 837                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 838                                 array_push( $td_history , false );
 839                                 array_push( $last_tag_history , '' );
 840                                 array_push( $tr_history , false );
 841                                 array_push( $tr_attributes , '' );
 842                                 array_push( $has_opened_tr , false );
 843                         } elseif ( count( $td_history ) == 0 ) {
 844                                 # Don't do any of the following
 845                                 $out .= $outLine."\n";
 846                                 continue;
 847                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 848                                 # We are ending a table
 849                                 $line = '</table>' . substr( $line , 2 );
 850                                 $last_tag = array_pop( $last_tag_history );
 851
 852                                 if ( !array_pop( $has_opened_tr ) ) {
 853                                         $line = "<tr><td></td></tr>{$line}";
 854                                 }
 855
 856                                 if ( array_pop( $tr_history ) ) {
 857                                         $line = "</tr>{$line}";
 858                                 }
 859
 860                                 if ( array_pop( $td_history ) ) {
 861                                         $line = "</{$last_tag}>{$line}";
 862                                 }
 863                                 array_pop( $tr_attributes );
 864                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 865                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 866                                 # Now we have a table row
 867                                 $line = preg_replace( '#^\|-+#', '', $line );
 868
 869                                 # Whats after the tag is now only attributes
 870                                 $attributes = $this->mStripState->unstripBoth( $line );
 871                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 872                                 array_pop( $tr_attributes );
 873                                 array_push( $tr_attributes, $attributes );
 874
 875                                 $line = '';
 876                                 $last_tag = array_pop( $last_tag_history );
 877                                 array_pop( $has_opened_tr );
 878                                 array_push( $has_opened_tr , true );
 879
 880                                 if ( array_pop( $tr_history ) ) {
 881                                         $line = '</tr>';
 882                                 }
 883
 884                                 if ( array_pop( $td_history ) ) {
 885                                         $line = "</{$last_tag}>{$line}";
 886                                 }
 887
 888                                 $outLine = $line;
 889                                 array_push( $tr_history , false );
 890                                 array_push( $td_history , false );
 891                                 array_push( $last_tag_history , '' );
 892                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 893                                 # This might be cell elements, td, th or captions
 894                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 895                                         $first_character = '+';
 896                                         $line = substr( $line , 1 );
 897                                 }
 898
 899                                 $line = substr( $line , 1 );
 900
 901                                 if ( $first_character === '!' ) {
 902                                         $line = str_replace( '!!' , '||' , $line );
 903                                 }
 904
 905                                 # Split up multiple cells on the same line.
 906                                 # FIXME : This can result in improper nesting of tags processed
 907                                 # by earlier parser steps, but should avoid splitting up eg
 908                                 # attribute values containing literal "||".
 909                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 910
 911                                 $outLine = '';
 912
 913                                 # Loop through each table cell
 914                                 foreach ( $cells as $cell ) {
 915                                         $previous = '';
 916                                         if ( $first_character !== '+' ) {
 917                                                 $tr_after = array_pop( $tr_attributes );
 918                                                 if ( !array_pop( $tr_history ) ) {
 919                                                         $previous = "<tr{$tr_after}>\n";
 920                                                 }
 921                                                 array_push( $tr_history , true );
 922                                                 array_push( $tr_attributes , '' );
 923                                                 array_pop( $has_opened_tr );
 924                                                 array_push( $has_opened_tr , true );
 925                                         }
 926
 927                                         $last_tag = array_pop( $last_tag_history );
 928
 929                                         if ( array_pop( $td_history ) ) {
 930                                                 $previous = "</{$last_tag}>\n{$previous}";
 931                                         }
 932
 933                                         if ( $first_character === '|' ) {
 934                                                 $last_tag = 'td';
 935                                         } elseif ( $first_character === '!' ) {
 936                                                 $last_tag = 'th';
 937                                         } elseif ( $first_character === '+' ) {
 938                                                 $last_tag = 'caption';
 939                                         } else {
 940                                                 $last_tag = '';
 941                                         }
 942
 943                                         array_push( $last_tag_history , $last_tag );
 944
 945                                         # A cell could contain both parameters and data
 946                                         $cell_data = explode( '|' , $cell , 2 );
 947
 948                                         # Bug 553: Note that a '|' inside an invalid link should not
 949                                         # be mistaken as delimiting cell parameters
 950                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 951                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 952                                         } elseif ( count( $cell_data ) == 1 ) {
 953                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 954                                         } else {
 955                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 956                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 957                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 958                                         }
 959
 960                                         $outLine .= $cell;
 961                                         array_push( $td_history , true );
 962                                 }
 963                         }
 964                         $out .= $outLine . "\n";
 965                 }
 966
 967                 # Closing open td, tr && table
 968                 while ( count( $td_history ) > 0 ) {
 969                         if ( array_pop( $td_history ) ) {
 970                                 $out .= "</td>\n";
 971                         }
 972                         if ( array_pop( $tr_history ) ) {
 973                                 $out .= "</tr>\n";
 974                         }
 975                         if ( !array_pop( $has_opened_tr ) ) {
 976                                 $out .= "<tr><td></td></tr>\n" ;
 977                         }
 978
 979                         $out .= "</table>\n";
 980                 }
 981
 982                 # Remove trailing line-ending (b/c)
 983                 if ( substr( $out, -1 ) === "\n" ) {
 984                         $out = substr( $out, 0, -1 );
 985                 }
 986
 987                 # special case: don't return empty table
 988                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 989                         $out = '';
 990                 }
 991
 992                 wfProfileOut( __METHOD__ );
 993
 994                 return $out;
 995         }
 996
 997         /**
 998          * Helper function for parse() that transforms wiki markup into
 999          * HTML. Only called for $mOutputType == self::OT_HTML.
1000          *
1001          * @private
1002          */
1003         function internalParse( $text, $isMain = true, $frame=false ) {
1004                 wfProfileIn( __METHOD__ );
1005
1006                 $origText = $text;
1007
1008                 # Hook to suspend the parser in this state
1009                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1010                         wfProfileOut( __METHOD__ );
1011                         return $text ;
1012                 }
1013
1014                 # if $frame is provided, then use $frame for replacing any variables
1015                 if ( $frame ) {
1016                         # use frame depth to infer how include/noinclude tags should be handled
1017                         # depth=0 means this is the top-level document; otherwise it's an included document
1018                         if ( !$frame->depth ) {
1019                                 $flag = 0;
1020                         } else {
1021                                 $flag = Parser::PTD_FOR_INCLUSION;
1022                         }
1023                         $dom = $this->preprocessToDom( $text, $flag );
1024                         $text = $frame->expand( $dom );
1025                 } else {
1026                         # if $frame is not provided, then use old-style replaceVariables
1027                         $text = $this->replaceVariables( $text );
1028                 }
1029
1030                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1031                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1032
1033                 # Tables need to come after variable replacement for things to work
1034                 # properly; putting them before other transformations should keep
1035                 # exciting things like link expansions from showing up in surprising
1036                 # places.
1037                 $text = $this->doTableStuff( $text );
1038
1039                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1040
1041                 $text = $this->doDoubleUnderscore( $text );
1042
1043                 $text = $this->doHeadings( $text );
1044                 if ( $this->mOptions->getUseDynamicDates() ) {
1045                         $df = DateFormatter::getInstance();
1046                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1047                 }
1048                 $text = $this->replaceInternalLinks( $text );
1049                 $text = $this->doAllQuotes( $text );
1050                 $text = $this->replaceExternalLinks( $text );
1051
1052                 # replaceInternalLinks may sometimes leave behind
1053                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1054                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1055
1056                 $text = $this->doMagicLinks( $text );
1057                 $text = $this->formatHeadings( $text, $origText, $isMain );
1058
1059                 wfProfileOut( __METHOD__ );
1060                 return $text;
1061         }
1062
1063         /**
1064          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1065          * magic external links.
1066          *
1067          * DML
1068          * @private
1069          */
1070         function doMagicLinks( $text ) {
1071                 wfProfileIn( __METHOD__ );
1072                 $prots = $this->mUrlProtocols;
1073                 $urlChar = self::EXT_LINK_URL_CLASS;
1074                 $text = preg_replace_callback(
1075                         '!(?:                           # Start cases
1076                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1077                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1078                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1079                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1080                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1081                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1082                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1083                                         [0-9Xx]                 # check digit
1084                                         \b)
1085                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1086                 wfProfileOut( __METHOD__ );
1087                 return $text;
1088         }
1089
1090         function magicLinkCallback( $m ) {
1091                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1092                         # Skip anchor
1093                         return $m[0];
1094                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1095                         # Skip HTML element
1096                         return $m[0];
1097                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1098                         # Free external link
1099                         return $this->makeFreeExternalLink( $m[0] );
1100                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1101                         # RFC or PMID
1102                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1103                                 $keyword = 'RFC';
1104                                 $urlmsg = 'rfcurl';
1105                                 $CssClass = 'mw-magiclink-rfc';
1106                                 $id = $m[4];
1107                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1108                                 $keyword = 'PMID';
1109                                 $urlmsg = 'pubmedurl';
1110                                 $CssClass = 'mw-magiclink-pmid';
1111                                 $id = $m[4];
1112                         } else {
1113                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1114                                         substr( $m[0], 0, 20 ) . '"' );
1115                         }
1116                         $url = wfMsgForContent( $urlmsg, $id );
1117                         $sk = $this->mOptions->getSkin( $this->mTitle );
1118                         return $sk->makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1119                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1120                         # ISBN
1121                         $isbn = $m[5];
1122                         $num = strtr( $isbn, array(
1123                                 '-' => '',
1124                                 ' ' => '',
1125                                 'x' => 'X',
1126                         ));
1127                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1128                         return'<a href="' .
1129                                 $titleObj->escapeLocalUrl() .
1130                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1131                 } else {
1132                         return $m[0];
1133                 }
1134         }
1135
1136         /**
1137          * Make a free external link, given a user-supplied URL
1138          * @return HTML
1139          * @private
1140          */
1141         function makeFreeExternalLink( $url ) {
1142                 global $wgContLang;
1143                 wfProfileIn( __METHOD__ );
1144
1145                 $sk = $this->mOptions->getSkin( $this->mTitle );
1146                 $trail = '';
1147
1148                 # The characters '<' and '>' (which were escaped by
1149                 # removeHTMLtags()) should not be included in
1150                 # URLs, per RFC 2396.
1151                 $m2 = array();
1152                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1153                         $trail = substr( $url, $m2[0][1] ) . $trail;
1154                         $url = substr( $url, 0, $m2[0][1] );
1155                 }
1156
1157                 # Move trailing punctuation to $trail
1158                 $sep = ',;\.:!?';
1159                 # If there is no left bracket, then consider right brackets fair game too
1160                 if ( strpos( $url, '(' ) === false ) {
1161                         $sep .= ')';
1162                 }
1163
1164                 $numSepChars = strspn( strrev( $url ), $sep );
1165                 if ( $numSepChars ) {
1166                         $trail = substr( $url, -$numSepChars ) . $trail;
1167                         $url = substr( $url, 0, -$numSepChars );
1168                 }
1169
1170                 $url = Sanitizer::cleanUrl( $url );
1171
1172                 # Is this an external image?
1173                 $text = $this->maybeMakeExternalImage( $url );
1174                 if ( $text === false ) {
1175                         # Not an image, make a link
1176                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1177                                 $this->getExternalLinkAttribs( $url ) );
1178                         # Register it in the output object...
1179                         # Replace unnecessary URL escape codes with their equivalent characters
1180                         $pasteurized = self::replaceUnusualEscapes( $url );
1181                         $this->mOutput->addExternalLink( $pasteurized );
1182                 }
1183                 wfProfileOut( __METHOD__ );
1184                 return $text . $trail;
1185         }
1186
1187
1188         /**
1189          * Parse headers and return html
1190          *
1191          * @private
1192          */
1193         function doHeadings( $text ) {
1194                 wfProfileIn( __METHOD__ );
1195                 for ( $i = 6; $i >= 1; --$i ) {
1196                         $h = str_repeat( '=', $i );
1197                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1198                           "<h$i>\\1</h$i>", $text );
1199                 }
1200                 wfProfileOut( __METHOD__ );
1201                 return $text;
1202         }
1203
1204         /**
1205          * Replace single quotes with HTML markup
1206          * @private
1207          * @return string the altered text
1208          */
1209         function doAllQuotes( $text ) {
1210                 wfProfileIn( __METHOD__ );
1211                 $outtext = '';
1212                 $lines = StringUtils::explode( "\n", $text );
1213                 foreach ( $lines as $line ) {
1214                         $outtext .= $this->doQuotes( $line ) . "\n";
1215                 }
1216                 $outtext = substr( $outtext, 0,-1 );
1217                 wfProfileOut( __METHOD__ );
1218                 return $outtext;
1219         }
1220
1221         /**
1222          * Helper function for doAllQuotes()
1223          */
1224         public function doQuotes( $text ) {
1225                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1226                 if ( count( $arr ) == 1 ) {
1227                         return $text;
1228                 } else {
1229                         # First, do some preliminary work. This may shift some apostrophes from
1230                         # being mark-up to being text. It also counts the number of occurrences
1231                         # of bold and italics mark-ups.
1232                         $numbold = 0;
1233                         $numitalics = 0;
1234                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1235                                 if ( ( $i % 2 ) == 1 ) {
1236                                         # If there are ever four apostrophes, assume the first is supposed to
1237                                         # be text, and the remaining three constitute mark-up for bold text.
1238                                         if ( strlen( $arr[$i] ) == 4 ) {
1239                                                 $arr[$i-1] .= "'";
1240                                                 $arr[$i] = "'''";
1241                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1242                                                 # If there are more than 5 apostrophes in a row, assume they're all
1243                                                 # text except for the last 5.
1244                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1245                                                 $arr[$i] = "'''''";
1246                                         }
1247                                         # Count the number of occurrences of bold and italics mark-ups.
1248                                         # We are not counting sequences of five apostrophes.
1249                                         if ( strlen( $arr[$i] ) == 2 ) {
1250                                                 $numitalics++;
1251                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1252                                                 $numbold++;
1253                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1254                                                 $numitalics++;
1255                                                 $numbold++;
1256                                         }
1257                                 }
1258                         }
1259
1260                         # If there is an odd number of both bold and italics, it is likely
1261                         # that one of the bold ones was meant to be an apostrophe followed
1262                         # by italics. Which one we cannot know for certain, but it is more
1263                         # likely to be one that has a single-letter word before it.
1264                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1265                                 $i = 0;
1266                                 $firstsingleletterword = -1;
1267                                 $firstmultiletterword = -1;
1268                                 $firstspace = -1;
1269                                 foreach ( $arr as $r ) {
1270                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1271                                                 $x1 = substr( $arr[$i-1], -1 );
1272                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1273                                                 if ( $x1 === ' ' ) {
1274                                                         if ( $firstspace == -1 ) {
1275                                                                 $firstspace = $i;
1276                                                         }
1277                                                 } elseif ( $x2 === ' ') {
1278                                                         if ( $firstsingleletterword == -1 ) {
1279                                                                 $firstsingleletterword = $i;
1280                                                         }
1281                                                 } else {
1282                                                         if ( $firstmultiletterword == -1 ) {
1283                                                                 $firstmultiletterword = $i;
1284                                                         }
1285                                                 }
1286                                         }
1287                                         $i++;
1288                                 }
1289
1290                                 # If there is a single-letter word, use it!
1291                                 if ( $firstsingleletterword > -1 ) {
1292                                         $arr[$firstsingleletterword] = "''";
1293                                         $arr[$firstsingleletterword-1] .= "'";
1294                                 } elseif ( $firstmultiletterword > -1 ) {
1295                                         # If not, but there's a multi-letter word, use that one.
1296                                         $arr[$firstmultiletterword] = "''";
1297                                         $arr[$firstmultiletterword-1] .= "'";
1298                                 } elseif ( $firstspace > -1 ) {
1299                                         # ... otherwise use the first one that has neither.
1300                                         # (notice that it is possible for all three to be -1 if, for example,
1301                                         # there is only one pentuple-apostrophe in the line)
1302                                         $arr[$firstspace] = "''";
1303                                         $arr[$firstspace-1] .= "'";
1304                                 }
1305                         }
1306
1307                         # Now let's actually convert our apostrophic mush to HTML!
1308                         $output = '';
1309                         $buffer = '';
1310                         $state = '';
1311                         $i = 0;
1312                         foreach ( $arr as $r ) {
1313                                 if ( ( $i % 2 ) == 0 ) {
1314                                         if ( $state === 'both' ) {
1315                                                 $buffer .= $r;
1316                                         } else {
1317                                                 $output .= $r;
1318                                         }
1319                                 } else {
1320                                         if ( strlen( $r ) == 2 ) {
1321                                                 if ( $state === 'i' ) {
1322                                                         $output .= '</i>'; $state = '';
1323                                                 } elseif ( $state === 'bi' ) {
1324                                                         $output .= '</i>'; $state = 'b';
1325                                                 } elseif ( $state === 'ib' ) {
1326                                                         $output .= '</b></i><b>'; $state = 'b';
1327                                                 } elseif ( $state === 'both' ) {
1328                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1329                                                 } else { # $state can be 'b' or ''
1330                                                         $output .= '<i>'; $state .= 'i';
1331                                                 }
1332                                         } elseif ( strlen( $r ) == 3 ) {
1333                                                 if ( $state === 'b' ) {
1334                                                         $output .= '</b>'; $state = '';
1335                                                 } elseif ( $state === 'bi' ) {
1336                                                         $output .= '</i></b><i>'; $state = 'i';
1337                                                 } elseif ( $state === 'ib' ) {
1338                                                         $output .= '</b>'; $state = 'i';
1339                                                 } elseif ( $state === 'both' ) {
1340                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1341                                                 } else { # $state can be 'i' or ''
1342                                                         $output .= '<b>'; $state .= 'b';
1343                                                 }
1344                                         } elseif ( strlen( $r ) == 5 ) {
1345                                                 if ( $state === 'b' ) {
1346                                                         $output .= '</b><i>'; $state = 'i';
1347                                                 } elseif ( $state === 'i' ) {
1348                                                         $output .= '</i><b>'; $state = 'b';
1349                                                 } elseif ( $state === 'bi' ) {
1350                                                         $output .= '</i></b>'; $state = '';
1351                                                 } elseif ( $state === 'ib' ) {
1352                                                         $output .= '</b></i>'; $state = '';
1353                                                 } elseif ( $state === 'both' ) {
1354                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1355                                                 } else { # ($state == '')
1356                                                         $buffer = ''; $state = 'both';
1357                                                 }
1358                                         }
1359                                 }
1360                                 $i++;
1361                         }
1362                         # Now close all remaining tags.  Notice that the order is important.
1363                         if ( $state === 'b' || $state === 'ib' ) {
1364                                 $output .= '</b>';
1365                         }
1366                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1367                                 $output .= '</i>';
1368                         }
1369                         if ( $state === 'bi' ) {
1370                                 $output .= '</b>';
1371                         }
1372                         # There might be lonely ''''', so make sure we have a buffer
1373                         if ( $state === 'both' && $buffer ) {
1374                                 $output .= '<b><i>'.$buffer.'</i></b>';
1375                         }
1376                         return $output;
1377                 }
1378         }
1379
1380         /**
1381          * Replace external links (REL)
1382          *
1383          * Note: this is all very hackish and the order of execution matters a lot.
1384          * Make sure to run maintenance/parserTests.php if you change this code.
1385          *
1386          * @private
1387          */
1388         function replaceExternalLinks( $text ) {
1389                 global $wgContLang;
1390                 wfProfileIn( __METHOD__ );
1391
1392                 $sk = $this->mOptions->getSkin( $this->mTitle );
1393
1394                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1395                 $s = array_shift( $bits );
1396
1397                 $i = 0;
1398                 while ( $i<count( $bits ) ) {
1399                         $url = $bits[$i++];
1400                         $protocol = $bits[$i++];
1401                         $text = $bits[$i++];
1402                         $trail = $bits[$i++];
1403
1404                         # The characters '<' and '>' (which were escaped by
1405                         # removeHTMLtags()) should not be included in
1406                         # URLs, per RFC 2396.
1407                         $m2 = array();
1408                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1409                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1410                                 $url = substr( $url, 0, $m2[0][1] );
1411                         }
1412
1413                         # If the link text is an image URL, replace it with an <img> tag
1414                         # This happened by accident in the original parser, but some people used it extensively
1415                         $img = $this->maybeMakeExternalImage( $text );
1416                         if ( $img !== false ) {
1417                                 $text = $img;
1418                         }
1419
1420                         $dtrail = '';
1421
1422                         # Set linktype for CSS - if URL==text, link is essentially free
1423                         $linktype = ( $text === $url ) ? 'free' : 'text';
1424
1425                         # No link text, e.g. [http://domain.tld/some.link]
1426                         if ( $text == '' ) {
1427                                 # Autonumber if allowed. See bug #5918
1428                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1429                                         $langObj = $this->getFunctionLang();
1430                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1431                                         $linktype = 'autonumber';
1432                                 } else {
1433                                         # Otherwise just use the URL
1434                                         $text = htmlspecialchars( $url );
1435                                         $linktype = 'free';
1436                                 }
1437                         } else {
1438                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1439                                 # Check for trail
1440                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1441                         }
1442
1443                         $text = $wgContLang->markNoConversion( $text );
1444
1445                         $url = Sanitizer::cleanUrl( $url );
1446
1447                         # Use the encoded URL
1448                         # This means that users can paste URLs directly into the text
1449                         # Funny characters like ö aren't valid in URLs anyway
1450                         # This was changed in August 2004
1451                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1452                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1453
1454                         # Register link in the output object.
1455                         # Replace unnecessary URL escape codes with the referenced character
1456                         # This prevents spammers from hiding links from the filters
1457                         $pasteurized = self::replaceUnusualEscapes( $url );
1458                         $this->mOutput->addExternalLink( $pasteurized );
1459                 }
1460
1461                 wfProfileOut( __METHOD__ );
1462                 return $s;
1463         }
1464
1465         /**
1466          * Get an associative array of additional HTML attributes appropriate for a
1467          * particular external link.  This currently may include rel => nofollow
1468          * (depending on configuration, namespace, and the URL's domain) and/or a
1469          * target attribute (depending on configuration).
1470          *
1471          * @param $url String: optional URL, to extract the domain from for rel =>
1472          *   nofollow if appropriate
1473          * @return Array: associative array of HTML attributes
1474          */
1475         function getExternalLinkAttribs( $url = false ) {
1476                 $attribs = array();
1477                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1478                 $ns = $this->mTitle->getNamespace();
1479                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1480                         $attribs['rel'] = 'nofollow';
1481
1482                         global $wgNoFollowDomainExceptions;
1483                         if ( $wgNoFollowDomainExceptions ) {
1484                                 $bits = wfParseUrl( $url );
1485                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1486                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1487                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1488                                                         unset( $attribs['rel'] );
1489                                                         break;
1490                                                 }
1491                                         }
1492                                 }
1493                         }
1494                 }
1495                 if ( $this->mOptions->getExternalLinkTarget() ) {
1496                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1497                 }
1498                 return $attribs;
1499         }
1500
1501
1502         /**
1503          * Replace unusual URL escape codes with their equivalent characters
1504          *
1505          * @param $url String
1506          * @return String
1507          *
1508          * @todo  This can merge genuinely required bits in the path or query string,
1509          *        breaking legit URLs. A proper fix would treat the various parts of
1510          *        the URL differently; as a workaround, just use the output for
1511          *        statistical records, not for actual linking/output.
1512          */
1513         static function replaceUnusualEscapes( $url ) {
1514                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1515                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1516         }
1517
1518         /**
1519          * Callback function used in replaceUnusualEscapes().
1520          * Replaces unusual URL escape codes with their equivalent character
1521          */
1522         private static function replaceUnusualEscapesCallback( $matches ) {
1523                 $char = urldecode( $matches[0] );
1524                 $ord = ord( $char );
1525                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1526                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1527                         # No, shouldn't be escaped
1528                         return $char;
1529                 } else {
1530                         # Yes, leave it escaped
1531                         return $matches[0];
1532                 }
1533         }
1534
1535         /**
1536          * make an image if it's allowed, either through the global
1537          * option, through the exception, or through the on-wiki whitelist
1538          * @private
1539          */
1540         function maybeMakeExternalImage( $url ) {
1541                 $sk = $this->mOptions->getSkin( $this->mTitle );
1542                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1543                 $imagesexception = !empty( $imagesfrom );
1544                 $text = false;
1545                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1546                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1547                         $imagematch = false;
1548                         foreach ( $imagesfrom as $match ) {
1549                                 if ( strpos( $url, $match ) === 0 ) {
1550                                         $imagematch = true;
1551                                         break;
1552                                 }
1553                         }
1554                 } elseif ( $imagesexception ) {
1555                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1556                 } else {
1557                         $imagematch = false;
1558                 }
1559                 if ( $this->mOptions->getAllowExternalImages()
1560                          || ( $imagesexception && $imagematch ) ) {
1561                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1562                                 # Image found
1563                                 $text = $sk->makeExternalImage( $url );
1564                         }
1565                 }
1566                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1567                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1568                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1569                         foreach ( $whitelist as $entry ) {
1570                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1571                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1572                                         continue;
1573                                 }
1574                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1575                                         # Image matches a whitelist entry
1576                                         $text = $sk->makeExternalImage( $url );
1577                                         break;
1578                                 }
1579                         }
1580                 }
1581                 return $text;
1582         }
1583
1584         /**
1585          * Process [[ ]] wikilinks
1586          * @return String: processed text
1587          *
1588          * @private
1589          */
1590         function replaceInternalLinks( $s ) {
1591                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1592                 return $s;
1593         }
1594
1595         /**
1596          * Process [[ ]] wikilinks (RIL)
1597          * @return LinkHolderArray
1598          *
1599          * @private
1600          */
1601         function replaceInternalLinks2( &$s ) {
1602                 global $wgContLang;
1603
1604                 wfProfileIn( __METHOD__ );
1605
1606                 wfProfileIn( __METHOD__.'-setup' );
1607                 static $tc = FALSE, $e1, $e1_img;
1608                 # the % is needed to support urlencoded titles as well
1609                 if ( !$tc ) {
1610                         $tc = Title::legalChars() . '#%';
1611                         # Match a link having the form [[namespace:link|alternate]]trail
1612                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1613                         # Match cases where there is no "]]", which might still be images
1614                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1615                 }
1616
1617                 $sk = $this->mOptions->getSkin( $this->mTitle );
1618                 $holders = new LinkHolderArray( $this );
1619
1620                 # split the entire text string on occurences of [[
1621                 $a = StringUtils::explode( '[[', ' ' . $s );
1622                 # get the first element (all text up to first [[), and remove the space we added
1623                 $s = $a->current();
1624                 $a->next();
1625                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1626                 $s = substr( $s, 1 );
1627
1628                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1629                 $e2 = null;
1630                 if ( $useLinkPrefixExtension ) {
1631                         # Match the end of a line for a word that's not followed by whitespace,
1632                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1633                         $e2 = wfMsgForContent( 'linkprefix' );
1634                 }
1635
1636                 if ( is_null( $this->mTitle ) ) {
1637                         wfProfileOut( __METHOD__.'-setup' );
1638                         wfProfileOut( __METHOD__ );
1639                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1640                 }
1641                 $nottalk = !$this->mTitle->isTalkPage();
1642
1643                 if ( $useLinkPrefixExtension ) {
1644                         $m = array();
1645                         if ( preg_match( $e2, $s, $m ) ) {
1646                                 $first_prefix = $m[2];
1647                         } else {
1648                                 $first_prefix = false;
1649                         }
1650                 } else {
1651                         $prefix = '';
1652                 }
1653
1654                 if ( $wgContLang->hasVariants() ) {
1655                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1656                 } else {
1657                         $selflink = array( $this->mTitle->getPrefixedText() );
1658                 }
1659                 $useSubpages = $this->areSubpagesAllowed();
1660                 wfProfileOut( __METHOD__.'-setup' );
1661
1662                 # Loop for each link
1663                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1664                         # Check for excessive memory usage
1665                         if ( $holders->isBig() ) {
1666                                 # Too big
1667                                 # Do the existence check, replace the link holders and clear the array
1668                                 $holders->replace( $s );
1669                                 $holders->clear();
1670                         }
1671
1672                         if ( $useLinkPrefixExtension ) {
1673                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1674                                 if ( preg_match( $e2, $s, $m ) ) {
1675                                         $prefix = $m[2];
1676                                         $s = $m[1];
1677                                 } else {
1678                                         $prefix='';
1679                                 }
1680                                 # first link
1681                                 if ( $first_prefix ) {
1682                                         $prefix = $first_prefix;
1683                                         $first_prefix = false;
1684                                 }
1685                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1686                         }
1687
1688                         $might_be_img = false;
1689
1690                         wfProfileIn( __METHOD__."-e1" );
1691                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1692                                 $text = $m[2];
1693                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1694                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1695                                 # the real problem is with the $e1 regex
1696                                 # See bug 1300.
1697                                 #
1698                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1699                                 # and no image is in sight. See bug 2095.
1700                                 #
1701                                 if ( $text !== '' &&
1702                                         substr( $m[3], 0, 1 ) === ']' &&
1703                                         strpos( $text, '[' ) !== false
1704                                 )
1705                                 {
1706                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1707                                         $m[3] = substr( $m[3], 1 );
1708                                 }
1709                                 # fix up urlencoded title texts
1710                                 if ( strpos( $m[1], '%' ) !== false ) {
1711                                         # Should anchors '#' also be rejected?
1712                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1713                                 }
1714                                 $trail = $m[3];
1715                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1716                                 $might_be_img = true;
1717                                 $text = $m[2];
1718                                 if ( strpos( $m[1], '%' ) !== false ) {
1719                                         $m[1] = rawurldecode( $m[1] );
1720                                 }
1721                                 $trail = "";
1722                         } else { # Invalid form; output directly
1723                                 $s .= $prefix . '[[' . $line ;
1724                                 wfProfileOut( __METHOD__."-e1" );
1725                                 continue;
1726                         }
1727                         wfProfileOut( __METHOD__."-e1" );
1728                         wfProfileIn( __METHOD__."-misc" );
1729
1730                         # Don't allow internal links to pages containing
1731                         # PROTO: where PROTO is a valid URL protocol; these
1732                         # should be external links.
1733                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1734                                 $s .= $prefix . '[[' . $line ;
1735                                 wfProfileOut( __METHOD__."-misc" );
1736                                 continue;
1737                         }
1738
1739                         # Make subpage if necessary
1740                         if ( $useSubpages ) {
1741                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1742                         } else {
1743                                 $link = $m[1];
1744                         }
1745
1746                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1747                         if ( !$noforce ) {
1748                                 # Strip off leading ':'
1749                                 $link = substr( $link, 1 );
1750                         }
1751
1752                         wfProfileOut( __METHOD__."-misc" );
1753                         wfProfileIn( __METHOD__."-title" );
1754                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1755                         if ( $nt === null ) {
1756                                 $s .= $prefix . '[[' . $line;
1757                                 wfProfileOut( __METHOD__."-title" );
1758                                 continue;
1759                         }
1760
1761                         $ns = $nt->getNamespace();
1762                         $iw = $nt->getInterWiki();
1763                         wfProfileOut( __METHOD__."-title" );
1764
1765                         if ( $might_be_img ) { # if this is actually an invalid link
1766                                 wfProfileIn( __METHOD__."-might_be_img" );
1767                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1768                                         $found = false;
1769                                         while ( true ) {
1770                                                 # look at the next 'line' to see if we can close it there
1771                                                 $a->next();
1772                                                 $next_line = $a->current();
1773                                                 if ( $next_line === false || $next_line === null ) {
1774                                                         break;
1775                                                 }
1776                                                 $m = explode( ']]', $next_line, 3 );
1777                                                 if ( count( $m ) == 3 ) {
1778                                                         # the first ]] closes the inner link, the second the image
1779                                                         $found = true;
1780                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1781                                                         $trail = $m[2];
1782                                                         break;
1783                                                 } elseif ( count( $m ) == 2 ) {
1784                                                         # if there's exactly one ]] that's fine, we'll keep looking
1785                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1786                                                 } else {
1787                                                         # if $next_line is invalid too, we need look no further
1788                                                         $text .= '[[' . $next_line;
1789                                                         break;
1790                                                 }
1791                                         }
1792                                         if ( !$found ) {
1793                                                 # we couldn't find the end of this imageLink, so output it raw
1794                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1795                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1796                                                 $s .= "{$prefix}[[$link|$text";
1797                                                 # note: no $trail, because without an end, there *is* no trail
1798                                                 wfProfileOut( __METHOD__."-might_be_img" );
1799                                                 continue;
1800                                         }
1801                                 } else { # it's not an image, so output it raw
1802                                         $s .= "{$prefix}[[$link|$text";
1803                                         # note: no $trail, because without an end, there *is* no trail
1804                                         wfProfileOut( __METHOD__."-might_be_img" );
1805                                         continue;
1806                                 }
1807                                 wfProfileOut( __METHOD__."-might_be_img" );
1808                         }
1809
1810                         $wasblank = ( $text  == '' );
1811                         if ( $wasblank ) {
1812                                 $text = $link;
1813                         } else {
1814                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1815                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1816                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1817                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1818                                 $text = $this->doQuotes( $text );
1819                         }
1820
1821                         # Link not escaped by : , create the various objects
1822                         if ( $noforce ) {
1823
1824                                 # Interwikis
1825                                 wfProfileIn( __METHOD__."-interwiki" );
1826                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1827                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1828                                         $s = rtrim( $s . $prefix );
1829                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1830                                         wfProfileOut( __METHOD__."-interwiki" );
1831                                         continue;
1832                                 }
1833                                 wfProfileOut( __METHOD__."-interwiki" );
1834
1835                                 if ( $ns == NS_FILE ) {
1836                                         wfProfileIn( __METHOD__."-image" );
1837                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1838                                                 if ( $wasblank ) {
1839                                                         # if no parameters were passed, $text
1840                                                         # becomes something like "File:Foo.png",
1841                                                         # which we don't want to pass on to the
1842                                                         # image generator
1843                                                         $text = '';
1844                                                 } else {
1845                                                         # recursively parse links inside the image caption
1846                                                         # actually, this will parse them in any other parameters, too,
1847                                                         # but it might be hard to fix that, and it doesn't matter ATM
1848                                                         $text = $this->replaceExternalLinks( $text );
1849                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1850                                                 }
1851                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1852                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1853                                         } else {
1854                                                 $s .= $prefix . $trail;
1855                                         }
1856                                         $this->mOutput->addImage( $nt->getDBkey() );
1857                                         wfProfileOut( __METHOD__."-image" );
1858                                         continue;
1859
1860                                 }
1861
1862                                 if ( $ns == NS_CATEGORY ) {
1863                                         wfProfileIn( __METHOD__."-category" );
1864                                         $s = rtrim( $s . "\n" ); # bug 87
1865
1866                                         if ( $wasblank ) {
1867                                                 $sortkey = $this->getDefaultSort();
1868                                         } else {
1869                                                 $sortkey = $text;
1870                                         }
1871                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1872                                         $sortkey = str_replace( "\n", '', $sortkey );
1873                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1874                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1875
1876                                         /**
1877                                          * Strip the whitespace Category links produce, see bug 87
1878                                          * @todo We might want to use trim($tmp, "\n") here.
1879                                          */
1880                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1881
1882                                         wfProfileOut( __METHOD__."-category" );
1883                                         continue;
1884                                 }
1885                         }
1886
1887                         # Self-link checking
1888                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1889                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1890                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1891                                         continue;
1892                                 }
1893                         }
1894
1895                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1896                         # FIXME: Should do batch file existence checks, see comment below
1897                         if ( $ns == NS_MEDIA ) {
1898                                 wfProfileIn( __METHOD__."-media" );
1899                                 # Give extensions a chance to select the file revision for us
1900                                 $skip = $time = false;
1901                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1902                                 if ( $skip ) {
1903                                         $link = $sk->link( $nt );
1904                                 } else {
1905                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1906                                 }
1907                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1908                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1909                                 $this->mOutput->addImage( $nt->getDBkey() );
1910                                 wfProfileOut( __METHOD__."-media" );
1911                                 continue;
1912                         }
1913
1914                         wfProfileIn( __METHOD__."-always_known" );
1915                         # Some titles, such as valid special pages or files in foreign repos, should
1916                         # be shown as bluelinks even though they're not included in the page table
1917                         #
1918                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1919                         # batch file existence checks for NS_FILE and NS_MEDIA
1920                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1921                                 $this->mOutput->addLink( $nt );
1922                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1923                         } else {
1924                                 # Links will be added to the output link list after checking
1925                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1926                         }
1927                         wfProfileOut( __METHOD__."-always_known" );
1928                 }
1929                 wfProfileOut( __METHOD__ );
1930                 return $holders;
1931         }
1932
1933         /**
1934          * Make a link placeholder. The text returned can be later resolved to a real link with
1935          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1936          * parsing of interwiki links, and secondly to allow all existence checks and
1937          * article length checks (for stub links) to be bundled into a single query.
1938          *
1939          * @deprecated
1940          */
1941         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1942                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1943         }
1944
1945         /**
1946          * Render a forced-blue link inline; protect against double expansion of
1947          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1948          * Since this little disaster has to split off the trail text to avoid
1949          * breaking URLs in the following text without breaking trails on the
1950          * wiki links, it's been made into a horrible function.
1951          *
1952          * @param $nt Title
1953          * @param $text String
1954          * @param $query String
1955          * @param $trail String
1956          * @param $prefix String
1957          * @return String: HTML-wikitext mix oh yuck
1958          */
1959         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1960                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1961                 $sk = $this->mOptions->getSkin( $this->mTitle );
1962                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1963                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1964                 return $this->armorLinks( $link ) . $trail;
1965         }
1966
1967         /**
1968          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1969          * going to go through further parsing steps before inline URL expansion.
1970          *
1971          * Not needed quite as much as it used to be since free links are a bit
1972          * more sensible these days. But bracketed links are still an issue.
1973          *
1974          * @param $text String: more-or-less HTML
1975          * @return String: less-or-more HTML with NOPARSE bits
1976          */
1977         function armorLinks( $text ) {
1978                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1979                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1980         }
1981
1982         /**
1983          * Return true if subpage links should be expanded on this page.
1984          * @return Boolean
1985          */
1986         function areSubpagesAllowed() {
1987                 # Some namespaces don't allow subpages
1988                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1989         }
1990
1991         /**
1992          * Handle link to subpage if necessary
1993          *
1994          * @param $target String: the source of the link
1995          * @param &$text String: the link text, modified as necessary
1996          * @return string the full name of the link
1997          * @private
1998          */
1999         function maybeDoSubpageLink( $target, &$text ) {
2000                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2001         }
2002
2003         /**#@+
2004          * Used by doBlockLevels()
2005          * @private
2006          */
2007         function closeParagraph() {
2008                 $result = '';
2009                 if ( $this->mLastSection != '' ) {
2010                         $result = '</' . $this->mLastSection  . ">\n";
2011                 }
2012                 $this->mInPre = false;
2013                 $this->mLastSection = '';
2014                 return $result;
2015         }
2016
2017         /**
2018          * getCommon() returns the length of the longest common substring
2019          * of both arguments, starting at the beginning of both.
2020          * @private
2021          */
2022         function getCommon( $st1, $st2 ) {
2023                 $fl = strlen( $st1 );
2024                 $shorter = strlen( $st2 );
2025                 if ( $fl < $shorter ) {
2026                         $shorter = $fl;
2027                 }
2028
2029                 for ( $i = 0; $i < $shorter; ++$i ) {
2030                         if ( $st1{$i} != $st2{$i} ) {
2031                                 break;
2032                         }
2033                 }
2034                 return $i;
2035         }
2036
2037         /**
2038          * These next three functions open, continue, and close the list
2039          * element appropriate to the prefix character passed into them.
2040          * @private
2041          */
2042         function openList( $char ) {
2043                 $result = $this->closeParagraph();
2044
2045                 if ( '*' === $char ) {
2046                         $result .= '<ul><li>';
2047                 } elseif ( '#' === $char ) {
2048                         $result .= '<ol><li>';
2049                 } elseif ( ':' === $char ) {
2050                         $result .= '<dl><dd>';
2051                 } elseif ( ';' === $char ) {
2052                         $result .= '<dl><dt>';
2053                         $this->mDTopen = true;
2054                 } else {
2055                         $result = '<!-- ERR 1 -->';
2056                 }
2057
2058                 return $result;
2059         }
2060
2061         /**
2062          * TODO: document
2063          * @param $char String
2064          * @private
2065          */
2066         function nextItem( $char ) {
2067                 if ( '*' === $char || '#' === $char ) {
2068                         return '</li><li>';
2069                 } elseif ( ':' === $char || ';' === $char ) {
2070                         $close = '</dd>';
2071                         if ( $this->mDTopen ) {
2072                                 $close = '</dt>';
2073                         }
2074                         if ( ';' === $char ) {
2075                                 $this->mDTopen = true;
2076                                 return $close . '<dt>';
2077                         } else {
2078                                 $this->mDTopen = false;
2079                                 return $close . '<dd>';
2080                         }
2081                 }
2082                 return '<!-- ERR 2 -->';
2083         }
2084
2085         /**
2086          * TODO: document
2087          * @param $char String
2088          * @private
2089          */
2090         function closeList( $char ) {
2091                 if ( '*' === $char ) {
2092                         $text = '</li></ul>';
2093                 } elseif ( '#' === $char ) {
2094                         $text = '</li></ol>';
2095                 } elseif ( ':' === $char ) {
2096                         if ( $this->mDTopen ) {
2097                                 $this->mDTopen = false;
2098                                 $text = '</dt></dl>';
2099                         } else {
2100                                 $text = '</dd></dl>';
2101                         }
2102                 } else {
2103                         return '<!-- ERR 3 -->';
2104                 }
2105                 return $text."\n";
2106         }
2107         /**#@-*/
2108
2109         /**
2110          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2111          *
2112          * @param $text String
2113          * @param $linestart Boolean: whether or not this is at the start of a line.
2114          * @private
2115          * @return string the lists rendered as HTML
2116          */
2117         function doBlockLevels( $text, $linestart ) {
2118                 wfProfileIn( __METHOD__ );
2119
2120                 # Parsing through the text line by line.  The main thing
2121                 # happening here is handling of block-level elements p, pre,
2122                 # and making lists from lines starting with * # : etc.
2123                 #
2124                 $textLines = StringUtils::explode( "\n", $text );
2125
2126                 $lastPrefix = $output = '';
2127                 $this->mDTopen = $inBlockElem = false;
2128                 $prefixLength = 0;
2129                 $paragraphStack = false;
2130
2131                 foreach ( $textLines as $oLine ) {
2132                         # Fix up $linestart
2133                         if ( !$linestart ) {
2134                                 $output .= $oLine;
2135                                 $linestart = true;
2136                                 continue;
2137                         }
2138                         # * = ul
2139                         # # = ol
2140                         # ; = dt
2141                         # : = dd
2142
2143                         $lastPrefixLength = strlen( $lastPrefix );
2144                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2145                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2146                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2147                         if ( !$this->mInPre ) {
2148                                 # Multiple prefixes may abut each other for nested lists.
2149                                 $prefixLength = strspn( $oLine, '*#:;' );
2150                                 $prefix = substr( $oLine, 0, $prefixLength );
2151
2152                                 # eh?
2153                                 # ; and : are both from definition-lists, so they're equivalent
2154                                 #  for the purposes of determining whether or not we need to open/close
2155                                 #  elements.
2156                                 $prefix2 = str_replace( ';', ':', $prefix );
2157                                 $t = substr( $oLine, $prefixLength );
2158                                 $this->mInPre = (bool)$preOpenMatch;
2159                         } else {
2160                                 # Don't interpret any other prefixes in preformatted text
2161                                 $prefixLength = 0;
2162                                 $prefix = $prefix2 = '';
2163                                 $t = $oLine;
2164                         }
2165
2166                         # List generation
2167                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2168                                 # Same as the last item, so no need to deal with nesting or opening stuff
2169                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2170                                 $paragraphStack = false;
2171
2172                                 if ( substr( $prefix, -1 ) === ';') {
2173                                         # The one nasty exception: definition lists work like this:
2174                                         # ; title : definition text
2175                                         # So we check for : in the remainder text to split up the
2176                                         # title and definition, without b0rking links.
2177                                         $term = $t2 = '';
2178                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2179                                                 $t = $t2;
2180                                                 $output .= $term . $this->nextItem( ':' );
2181                                         }
2182                                 }
2183                         } elseif ( $prefixLength || $lastPrefixLength ) {
2184                                 # We need to open or close prefixes, or both.
2185
2186                                 # Either open or close a level...
2187                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2188                                 $paragraphStack = false;
2189
2190                                 # Close all the prefixes which aren't shared.
2191                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2192                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2193                                         --$lastPrefixLength;
2194                                 }
2195
2196                                 # Continue the current prefix if appropriate.
2197                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2198                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2199                                 }
2200
2201                                 # Open prefixes where appropriate.
2202                                 while ( $prefixLength > $commonPrefixLength ) {
2203                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2204                                         $output .= $this->openList( $char );
2205
2206                                         if ( ';' === $char ) {
2207                                                 # FIXME: This is dupe of code above
2208                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2209                                                         $t = $t2;
2210                                                         $output .= $term . $this->nextItem( ':' );
2211                                                 }
2212                                         }
2213                                         ++$commonPrefixLength;
2214                                 }
2215                                 $lastPrefix = $prefix2;
2216                         }
2217
2218                         # If we have no prefixes, go to paragraph mode.
2219                         if ( 0 == $prefixLength ) {
2220                                 wfProfileIn( __METHOD__."-paragraph" );
2221                                 # No prefix (not in list)--go to paragraph mode
2222                                 # XXX: use a stack for nestable elements like span, table and div
2223                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2224                                 $closematch = preg_match(
2225                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2226                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2227                                 if ( $openmatch or $closematch ) {
2228                                         $paragraphStack = false;
2229                                         # TODO bug 5718: paragraph closed
2230                                         $output .= $this->closeParagraph();
2231                                         if ( $preOpenMatch and !$preCloseMatch ) {
2232                                                 $this->mInPre = true;
2233                                         }
2234                                         $inBlockElem = !$closematch;
2235                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2236                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2237                                                 # pre
2238                                                 if ( $this->mLastSection !== 'pre' ) {
2239                                                         $paragraphStack = false;
2240                                                         $output .= $this->closeParagraph().'<pre>';
2241                                                         $this->mLastSection = 'pre';
2242                                                 }
2243                                                 $t = substr( $t, 1 );
2244                                         } else {
2245                                                 # paragraph
2246                                                 if ( trim( $t ) === '' ) {
2247                                                         if ( $paragraphStack ) {
2248                                                                 $output .= $paragraphStack.'<br />';
2249                                                                 $paragraphStack = false;
2250                                                                 $this->mLastSection = 'p';
2251                                                         } else {
2252                                                                 if ( $this->mLastSection !== 'p' ) {
2253                                                                         $output .= $this->closeParagraph();
2254                                                                         $this->mLastSection = '';
2255                                                                         $paragraphStack = '<p>';
2256                                                                 } else {
2257                                                                         $paragraphStack = '</p><p>';
2258                                                                 }
2259                                                         }
2260                                                 } else {
2261                                                         if ( $paragraphStack ) {
2262                                                                 $output .= $paragraphStack;
2263                                                                 $paragraphStack = false;
2264                                                                 $this->mLastSection = 'p';
2265                                                         } elseif ( $this->mLastSection !== 'p' ) {
2266                                                                 $output .= $this->closeParagraph().'<p>';
2267                                                                 $this->mLastSection = 'p';
2268                                                         }
2269                                                 }
2270                                         }
2271                                 }
2272                                 wfProfileOut( __METHOD__."-paragraph" );
2273                         }
2274                         # somewhere above we forget to get out of pre block (bug 785)
2275                         if ( $preCloseMatch && $this->mInPre ) {
2276                                 $this->mInPre = false;
2277                         }
2278                         if ( $paragraphStack === false ) {
2279                                 $output .= $t."\n";
2280                         }
2281                 }
2282                 while ( $prefixLength ) {
2283                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2284                         --$prefixLength;
2285                 }
2286                 if ( $this->mLastSection != '' ) {
2287                         $output .= '</' . $this->mLastSection . '>';
2288                         $this->mLastSection = '';
2289                 }
2290
2291                 wfProfileOut( __METHOD__ );
2292                 return $output;
2293         }
2294
2295         /**
2296          * Split up a string on ':', ignoring any occurences inside tags
2297          * to prevent illegal overlapping.
2298          *
2299          * @param $str String: the string to split
2300          * @param &$before String: set to everything before the ':'
2301          * @param &$after String: set to everything after the ':'
2302          * return String: the position of the ':', or false if none found
2303          */
2304         function findColonNoLinks( $str, &$before, &$after ) {
2305                 wfProfileIn( __METHOD__ );
2306
2307                 $pos = strpos( $str, ':' );
2308                 if ( $pos === false ) {
2309                         # Nothing to find!
2310                         wfProfileOut( __METHOD__ );
2311                         return false;
2312                 }
2313
2314                 $lt = strpos( $str, '<' );
2315                 if ( $lt === false || $lt > $pos ) {
2316                         # Easy; no tag nesting to worry about
2317                         $before = substr( $str, 0, $pos );
2318                         $after = substr( $str, $pos+1 );
2319                         wfProfileOut( __METHOD__ );
2320                         return $pos;
2321                 }
2322
2323                 # Ugly state machine to walk through avoiding tags.
2324                 $state = self::COLON_STATE_TEXT;
2325                 $stack = 0;
2326                 $len = strlen( $str );
2327                 for( $i = 0; $i < $len; $i++ ) {
2328                         $c = $str{$i};
2329
2330                         switch( $state ) {
2331                         # (Using the number is a performance hack for common cases)
2332                         case 0: # self::COLON_STATE_TEXT:
2333                                 switch( $c ) {
2334                                 case "<":
2335                                         # Could be either a <start> tag or an </end> tag
2336                                         $state = self::COLON_STATE_TAGSTART;
2337                                         break;
2338                                 case ":":
2339                                         if ( $stack == 0 ) {
2340                                                 # We found it!
2341                                                 $before = substr( $str, 0, $i );
2342                                                 $after = substr( $str, $i + 1 );
2343                                                 wfProfileOut( __METHOD__ );
2344                                                 return $i;
2345                                         }
2346                                         # Embedded in a tag; don't break it.
2347                                         break;
2348                                 default:
2349                                         # Skip ahead looking for something interesting
2350                                         $colon = strpos( $str, ':', $i );
2351                                         if ( $colon === false ) {
2352                                                 # Nothing else interesting
2353                                                 wfProfileOut( __METHOD__ );
2354                                                 return false;
2355                                         }
2356                                         $lt = strpos( $str, '<', $i );
2357                                         if ( $stack === 0 ) {
2358                                                 if ( $lt === false || $colon < $lt ) {
2359                                                         # We found it!
2360                                                         $before = substr( $str, 0, $colon );
2361                                                         $after = substr( $str, $colon + 1 );
2362                                                         wfProfileOut( __METHOD__ );
2363                                                         return $i;
2364                                                 }
2365                                         }
2366                                         if ( $lt === false ) {
2367                                                 # Nothing else interesting to find; abort!
2368                                                 # We're nested, but there's no close tags left. Abort!
2369                                                 break 2;
2370                                         }
2371                                         # Skip ahead to next tag start
2372                                         $i = $lt;
2373                                         $state = self::COLON_STATE_TAGSTART;
2374                                 }
2375                                 break;
2376                         case 1: # self::COLON_STATE_TAG:
2377                                 # In a <tag>
2378                                 switch( $c ) {
2379                                 case ">":
2380                                         $stack++;
2381                                         $state = self::COLON_STATE_TEXT;
2382                                         break;
2383                                 case "/":
2384                                         # Slash may be followed by >?
2385                                         $state = self::COLON_STATE_TAGSLASH;
2386                                         break;
2387                                 default:
2388                                         # ignore
2389                                 }
2390                                 break;
2391                         case 2: # self::COLON_STATE_TAGSTART:
2392                                 switch( $c ) {
2393                                 case "/":
2394                                         $state = self::COLON_STATE_CLOSETAG;
2395                                         break;
2396                                 case "!":
2397                                         $state = self::COLON_STATE_COMMENT;
2398                                         break;
2399                                 case ">":
2400                                         # Illegal early close? This shouldn't happen D:
2401                                         $state = self::COLON_STATE_TEXT;
2402                                         break;
2403                                 default:
2404                                         $state = self::COLON_STATE_TAG;
2405                                 }
2406                                 break;
2407                         case 3: # self::COLON_STATE_CLOSETAG:
2408                                 # In a </tag>
2409                                 if ( $c === ">" ) {
2410                                         $stack--;
2411                                         if ( $stack < 0 ) {
2412                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2413                                                 wfProfileOut( __METHOD__ );
2414                                                 return false;
2415                                         }
2416                                         $state = self::COLON_STATE_TEXT;
2417                                 }
2418                                 break;
2419                         case self::COLON_STATE_TAGSLASH:
2420                                 if ( $c === ">" ) {
2421                                         # Yes, a self-closed tag <blah/>
2422                                         $state = self::COLON_STATE_TEXT;
2423                                 } else {
2424                                         # Probably we're jumping the gun, and this is an attribute
2425                                         $state = self::COLON_STATE_TAG;
2426                                 }
2427                                 break;
2428                         case 5: # self::COLON_STATE_COMMENT:
2429                                 if ( $c === "-" ) {
2430                                         $state = self::COLON_STATE_COMMENTDASH;
2431                                 }
2432                                 break;
2433                         case self::COLON_STATE_COMMENTDASH:
2434                                 if ( $c === "-" ) {
2435                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2436                                 } else {
2437                                         $state = self::COLON_STATE_COMMENT;
2438                                 }
2439                                 break;
2440                         case self::COLON_STATE_COMMENTDASHDASH:
2441                                 if ( $c === ">" ) {
2442                                         $state = self::COLON_STATE_TEXT;
2443                                 } else {
2444                                         $state = self::COLON_STATE_COMMENT;
2445                                 }
2446                                 break;
2447                         default:
2448                                 throw new MWException( "State machine error in " . __METHOD__ );
2449                         }
2450                 }
2451                 if ( $stack > 0 ) {
2452                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2453                         return false;
2454                 }
2455                 wfProfileOut( __METHOD__ );
2456                 return false;
2457         }
2458
2459         /**
2460          * Return value of a magic variable (like PAGENAME)
2461          *
2462          * @private
2463          */
2464         function getVariableValue( $index, $frame=false ) {
2465                 global $wgContLang, $wgSitename, $wgServer;
2466                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2467
2468                 /**
2469                  * Some of these require message or data lookups and can be
2470                  * expensive to check many times.
2471                  */
2472                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2473                         if ( isset( $this->mVarCache[$index] ) ) {
2474                                 return $this->mVarCache[$index];
2475                         }
2476                 }
2477
2478                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2479                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2480
2481                 # Use the time zone
2482                 global $wgLocaltimezone;
2483                 if ( isset( $wgLocaltimezone ) ) {
2484                         $oldtz = date_default_timezone_get();
2485                         date_default_timezone_set( $wgLocaltimezone );
2486                 }
2487
2488                 $localTimestamp = date( 'YmdHis', $ts );
2489                 $localMonth = date( 'm', $ts );
2490                 $localMonth1 = date( 'n', $ts );
2491                 $localMonthName = date( 'n', $ts );
2492                 $localDay = date( 'j', $ts );
2493                 $localDay2 = date( 'd', $ts );
2494                 $localDayOfWeek = date( 'w', $ts );
2495                 $localWeek = date( 'W', $ts );
2496                 $localYear = date( 'Y', $ts );
2497                 $localHour = date( 'H', $ts );
2498                 if ( isset( $wgLocaltimezone ) ) {
2499                         date_default_timezone_set( $oldtz );
2500                 }
2501
2502                 switch ( $index ) {
2503                         case 'currentmonth':
2504                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2505                                 break;
2506                         case 'currentmonth1':
2507                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2508                                 break;
2509                         case 'currentmonthname':
2510                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2511                                 break;
2512                         case 'currentmonthnamegen':
2513                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2514                                 break;
2515                         case 'currentmonthabbrev':
2516                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2517                                 break;
2518                         case 'currentday':
2519                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2520                                 break;
2521                         case 'currentday2':
2522                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2523                                 break;
2524                         case 'localmonth':
2525                                 $value = $wgContLang->formatNum( $localMonth );
2526                                 break;
2527                         case 'localmonth1':
2528                                 $value = $wgContLang->formatNum( $localMonth1 );
2529                                 break;
2530                         case 'localmonthname':
2531                                 $value = $wgContLang->getMonthName( $localMonthName );
2532                                 break;
2533                         case 'localmonthnamegen':
2534                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2535                                 break;
2536                         case 'localmonthabbrev':
2537                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2538                                 break;
2539                         case 'localday':
2540                                 $value = $wgContLang->formatNum( $localDay );
2541                                 break;
2542                         case 'localday2':
2543                                 $value = $wgContLang->formatNum( $localDay2 );
2544                                 break;
2545                         case 'pagename':
2546                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2547                                 break;
2548                         case 'pagenamee':
2549                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2550                                 break;
2551                         case 'fullpagename':
2552                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2553                                 break;
2554                         case 'fullpagenamee':
2555                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2556                                 break;
2557                         case 'subpagename':
2558                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2559                                 break;
2560                         case 'subpagenamee':
2561                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2562                                 break;
2563                         case 'basepagename':
2564                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2565                                 break;
2566                         case 'basepagenamee':
2567                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2568                                 break;
2569                         case 'talkpagename':
2570                                 if ( $this->mTitle->canTalk() ) {
2571                                         $talkPage = $this->mTitle->getTalkPage();
2572                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2573                                 } else {
2574                                         $value = '';
2575                                 }
2576                                 break;
2577                         case 'talkpagenamee':
2578                                 if ( $this->mTitle->canTalk() ) {
2579                                         $talkPage = $this->mTitle->getTalkPage();
2580                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2581                                 } else {
2582                                         $value = '';
2583                                 }
2584                                 break;
2585                         case 'subjectpagename':
2586                                 $subjPage = $this->mTitle->getSubjectPage();
2587                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2588                                 break;
2589                         case 'subjectpagenamee':
2590                                 $subjPage = $this->mTitle->getSubjectPage();
2591                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2592                                 break;
2593                         case 'revisionid':
2594                                 # Let the edit saving system know we should parse the page
2595                                 # *after* a revision ID has been assigned.
2596                                 $this->mOutput->setFlag( 'vary-revision' );
2597                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2598                                 $value = $this->mRevisionId;
2599                                 break;
2600                         case 'revisionday':
2601                                 # Let the edit saving system know we should parse the page
2602                                 # *after* a revision ID has been assigned. This is for null edits.
2603                                 $this->mOutput->setFlag( 'vary-revision' );
2604                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2605                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2606                                 break;
2607                         case 'revisionday2':
2608                                 # Let the edit saving system know we should parse the page
2609                                 # *after* a revision ID has been assigned. This is for null edits.
2610                                 $this->mOutput->setFlag( 'vary-revision' );
2611                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2612                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2613                                 break;
2614                         case 'revisionmonth':
2615                                 # Let the edit saving system know we should parse the page
2616                                 # *after* a revision ID has been assigned. This is for null edits.
2617                                 $this->mOutput->setFlag( 'vary-revision' );
2618                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2619                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2620                                 break;
2621                         case 'revisionmonth1':
2622                                 # Let the edit saving system know we should parse the page
2623                                 # *after* a revision ID has been assigned. This is for null edits.
2624                                 $this->mOutput->setFlag( 'vary-revision' );
2625                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2626                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2627                                 break;
2628                         case 'revisionyear':
2629                                 # Let the edit saving system know we should parse the page
2630                                 # *after* a revision ID has been assigned. This is for null edits.
2631                                 $this->mOutput->setFlag( 'vary-revision' );
2632                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2633                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2634                                 break;
2635                         case 'revisiontimestamp':
2636                                 # Let the edit saving system know we should parse the page
2637                                 # *after* a revision ID has been assigned. This is for null edits.
2638                                 $this->mOutput->setFlag( 'vary-revision' );
2639                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2640                                 $value = $this->getRevisionTimestamp();
2641                                 break;
2642                         case 'revisionuser':
2643                                 # Let the edit saving system know we should parse the page
2644                                 # *after* a revision ID has been assigned. This is for null edits.
2645                                 $this->mOutput->setFlag( 'vary-revision' );
2646                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2647                                 $value = $this->getRevisionUser();
2648                                 break;
2649                         case 'namespace':
2650                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2651                                 break;
2652                         case 'namespacee':
2653                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2654                                 break;
2655                         case 'talkspace':
2656                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2657                                 break;
2658                         case 'talkspacee':
2659                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2660                                 break;
2661                         case 'subjectspace':
2662                                 $value = $this->mTitle->getSubjectNsText();
2663                                 break;
2664                         case 'subjectspacee':
2665                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2666                                 break;
2667                         case 'currentdayname':
2668                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2669                                 break;
2670                         case 'currentyear':
2671                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2672                                 break;
2673                         case 'currenttime':
2674                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2675                                 break;
2676                         case 'currenthour':
2677                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2678                                 break;
2679                         case 'currentweek':
2680                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2681                                 # int to remove the padding
2682                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2683                                 break;
2684                         case 'currentdow':
2685                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2686                                 break;
2687                         case 'localdayname':
2688                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2689                                 break;
2690                         case 'localyear':
2691                                 $value = $wgContLang->formatNum( $localYear, true );
2692                                 break;
2693                         case 'localtime':
2694                                 $value = $wgContLang->time( $localTimestamp, false, false );
2695                                 break;
2696                         case 'localhour':
2697                                 $value = $wgContLang->formatNum( $localHour, true );
2698                                 break;
2699                         case 'localweek':
2700                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2701                                 # int to remove the padding
2702                                 $value = $wgContLang->formatNum( (int)$localWeek );
2703                                 break;
2704                         case 'localdow':
2705                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2706                                 break;
2707                         case 'numberofarticles':
2708                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2709                                 break;
2710                         case 'numberoffiles':
2711                                 $value = $wgContLang->formatNum( SiteStats::images() );
2712                                 break;
2713                         case 'numberofusers':
2714                                 $value = $wgContLang->formatNum( SiteStats::users() );
2715                                 break;
2716                         case 'numberofactiveusers':
2717                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2718                                 break;
2719                         case 'numberofpages':
2720                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2721                                 break;
2722                         case 'numberofadmins':
2723                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2724                                 break;
2725                         case 'numberofedits':
2726                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2727                                 break;
2728                         case 'numberofviews':
2729                                 $value = $wgContLang->formatNum( SiteStats::views() );
2730                                 break;
2731                         case 'currenttimestamp':
2732                                 $value = wfTimestamp( TS_MW, $ts );
2733                                 break;
2734                         case 'localtimestamp':
2735                                 $value = $localTimestamp;
2736                                 break;
2737                         case 'currentversion':
2738                                 $value = SpecialVersion::getVersion();
2739                                 break;
2740                         case 'articlepath':
2741                                 return $wgArticlePath;
2742                         case 'sitename':
2743                                 return $wgSitename;
2744                         case 'server':
2745                                 return $wgServer;
2746                         case 'servername':
2747                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2748                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2749                                 wfRestoreWarnings();
2750                                 return $serverName ? $serverName : $wgServer;
2751                         case 'scriptpath':
2752                                 return $wgScriptPath;
2753                         case 'stylepath':
2754                                 return $wgStylePath;
2755                         case 'directionmark':
2756                                 return $wgContLang->getDirMark();
2757                         case 'contentlanguage':
2758                                 global $wgLanguageCode;
2759                                 return $wgLanguageCode;
2760                         default:
2761                                 $ret = null;
2762                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2763                                         return $ret;
2764                                 } else {
2765                                         return null;
2766                                 }
2767                 }
2768
2769                 if ( $index )
2770                         $this->mVarCache[$index] = $value;
2771
2772                 return $value;
2773         }
2774
2775         /**
2776          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2777          *
2778          * @private
2779          */
2780         function initialiseVariables() {
2781                 wfProfileIn( __METHOD__ );
2782                 $variableIDs = MagicWord::getVariableIDs();
2783                 $substIDs = MagicWord::getSubstIDs();
2784
2785                 $this->mVariables = new MagicWordArray( $variableIDs );
2786                 $this->mSubstWords = new MagicWordArray( $substIDs );
2787                 wfProfileOut( __METHOD__ );
2788         }
2789
2790         /**
2791          * Preprocess some wikitext and return the document tree.
2792          * This is the ghost of replace_variables().
2793          *
2794          * @param $text String: The text to parse
2795          * @param $flags Integer: bitwise combination of:
2796          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2797          *                                     included. Default is to assume a direct page view.
2798          *
2799          * The generated DOM tree must depend only on the input text and the flags.
2800          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2801          *
2802          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2803          * change in the DOM tree for a given text, must be passed through the section identifier
2804          * in the section edit link and thus back to extractSections().
2805          *
2806          * The output of this function is currently only cached in process memory, but a persistent
2807          * cache may be implemented at a later date which takes further advantage of these strict
2808          * dependency requirements.
2809          *
2810          * @private
2811          */
2812         function preprocessToDom( $text, $flags = 0 ) {
2813                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2814                 return $dom;
2815         }
2816
2817         /**
2818          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2819          */
2820         public static function splitWhitespace( $s ) {
2821                 $ltrimmed = ltrim( $s );
2822                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2823                 $trimmed = rtrim( $ltrimmed );
2824                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2825                 if ( $diff > 0 ) {
2826                         $w2 = substr( $ltrimmed, -$diff );
2827                 } else {
2828                         $w2 = '';
2829                 }
2830                 return array( $w1, $trimmed, $w2 );
2831         }
2832
2833         /**
2834          * Replace magic variables, templates, and template arguments
2835          * with the appropriate text. Templates are substituted recursively,
2836          * taking care to avoid infinite loops.
2837          *
2838          * Note that the substitution depends on value of $mOutputType:
2839          *  self::OT_WIKI: only {{subst:}} templates
2840          *  self::OT_PREPROCESS: templates but not extension tags
2841          *  self::OT_HTML: all templates and extension tags
2842          *
2843          * @param $text String: the text to transform
2844          * @param $frame PPFrame Object describing the arguments passed to the template.
2845          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2846          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2847          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2848          * @private
2849          */
2850         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2851                 # Is there any text? Also, Prevent too big inclusions!
2852                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2853                         return $text;
2854                 }
2855                 wfProfileIn( __METHOD__ );
2856
2857                 if ( $frame === false ) {
2858                         $frame = $this->getPreprocessor()->newFrame();
2859                 } elseif ( !( $frame instanceof PPFrame ) ) {
2860                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2861                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2862                 }
2863
2864                 $dom = $this->preprocessToDom( $text );
2865                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2866                 $text = $frame->expand( $dom, $flags );
2867
2868                 wfProfileOut( __METHOD__ );
2869                 return $text;
2870         }
2871
2872         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2873         static function createAssocArgs( $args ) {
2874                 $assocArgs = array();
2875                 $index = 1;
2876                 foreach ( $args as $arg ) {
2877                         $eqpos = strpos( $arg, '=' );
2878                         if ( $eqpos === false ) {
2879                                 $assocArgs[$index++] = $arg;
2880                         } else {
2881                                 $name = trim( substr( $arg, 0, $eqpos ) );
2882                                 $value = trim( substr( $arg, $eqpos+1 ) );
2883                                 if ( $value === false ) {
2884                                         $value = '';
2885                                 }
2886                                 if ( $name !== false ) {
2887                                         $assocArgs[$name] = $value;
2888                                 }
2889                         }
2890                 }
2891
2892                 return $assocArgs;
2893         }
2894
2895         /**
2896          * Warn the user when a parser limitation is reached
2897          * Will warn at most once the user per limitation type
2898          *
2899          * @param $limitationType String: should be one of:
2900          *   'expensive-parserfunction' (corresponding messages:
2901          *       'expensive-parserfunction-warning',
2902          *       'expensive-parserfunction-category')
2903          *   'post-expand-template-argument' (corresponding messages:
2904          *       'post-expand-template-argument-warning',
2905          *       'post-expand-template-argument-category')
2906          *   'post-expand-template-inclusion' (corresponding messages:
2907          *       'post-expand-template-inclusion-warning',
2908          *       'post-expand-template-inclusion-category')
2909          * @param $current Current value
2910          * @param $max Maximum allowed, when an explicit limit has been
2911          *       exceeded, provide the values (optional)
2912          */
2913         function limitationWarn( $limitationType, $current=null, $max=null) {
2914                 # does no harm if $current and $max are present but are unnecessary for the message
2915                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2916                 $this->mOutput->addWarning( $warning );
2917                 $this->addTrackingCategory( "$limitationType-category" );
2918         }
2919
2920         /**
2921          * Return the text of a template, after recursively
2922          * replacing any variables or templates within the template.
2923          *
2924          * @param $piece Array: the parts of the template
2925          *  $piece['title']: the title, i.e. the part before the |
2926          *  $piece['parts']: the parameter array
2927          *  $piece['lineStart']: whether the brace was at the start of a line
2928          * @param $frame PPFrame The current frame, contains template arguments
2929          * @return String: the text of the template
2930          * @private
2931          */
2932         function braceSubstitution( $piece, $frame ) {
2933                 global $wgContLang, $wgNonincludableNamespaces;
2934                 wfProfileIn( __METHOD__ );
2935                 wfProfileIn( __METHOD__.'-setup' );
2936
2937                 # Flags
2938                 $found = false;             # $text has been filled
2939                 $nowiki = false;            # wiki markup in $text should be escaped
2940                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2941                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2942                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2943                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2944
2945                 # Title object, where $text came from
2946                 $title = null;
2947
2948                 # $part1 is the bit before the first |, and must contain only title characters.
2949                 # Various prefixes will be stripped from it later.
2950                 $titleWithSpaces = $frame->expand( $piece['title'] );
2951                 $part1 = trim( $titleWithSpaces );
2952                 $titleText = false;
2953
2954                 # Original title text preserved for various purposes
2955                 $originalTitle = $part1;
2956
2957                 # $args is a list of argument nodes, starting from index 0, not including $part1
2958                 # *** FIXME if piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
2959                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2960                 wfProfileOut( __METHOD__.'-setup' );
2961
2962                 # SUBST
2963                 wfProfileIn( __METHOD__.'-modifiers' );
2964                 if ( !$found ) {
2965
2966                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2967
2968                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2969                         # Decide whether to expand template or keep wikitext as-is.
2970                         if ( $this->ot['wiki'] ) {
2971                                 if ( $substMatch === false ) {
2972                                         $literal = true;  # literal when in PST with no prefix
2973                                 } else {
2974                                         $literal = false; # expand when in PST with subst: or safesubst:
2975                                 }
2976                         } else {
2977                                 if ( $substMatch == 'subst' ) {
2978                                         $literal = true;  # literal when not in PST with plain subst:
2979                                 } else {
2980                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2981                                 }
2982                         }
2983                         if ( $literal ) {
2984                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2985                                 $isLocalObj = true;
2986                                 $found = true;
2987                         }
2988                 }
2989
2990                 # Variables
2991                 if ( !$found && $args->getLength() == 0 ) {
2992                         $id = $this->mVariables->matchStartToEnd( $part1 );
2993                         if ( $id !== false ) {
2994                                 $text = $this->getVariableValue( $id, $frame );
2995                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
2996                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
2997                                 }
2998                                 $found = true;
2999                         }
3000                 }
3001
3002                 # MSG, MSGNW and RAW
3003                 if ( !$found ) {
3004                         # Check for MSGNW:
3005                         $mwMsgnw = MagicWord::get( 'msgnw' );
3006                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3007                                 $nowiki = true;
3008                         } else {
3009                                 # Remove obsolete MSG:
3010                                 $mwMsg = MagicWord::get( 'msg' );
3011                                 $mwMsg->matchStartAndRemove( $part1 );
3012                         }
3013
3014                         # Check for RAW:
3015                         $mwRaw = MagicWord::get( 'raw' );
3016                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3017                                 $forceRawInterwiki = true;
3018                         }
3019                 }
3020                 wfProfileOut( __METHOD__.'-modifiers' );
3021
3022                 # Parser functions
3023                 if ( !$found ) {
3024                         wfProfileIn( __METHOD__ . '-pfunc' );
3025
3026                         $colonPos = strpos( $part1, ':' );
3027                         if ( $colonPos !== false ) {
3028                                 # Case sensitive functions
3029                                 $function = substr( $part1, 0, $colonPos );
3030                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3031                                         $function = $this->mFunctionSynonyms[1][$function];
3032                                 } else {
3033                                         # Case insensitive functions
3034                                         $function = $wgContLang->lc( $function );
3035                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3036                                                 $function = $this->mFunctionSynonyms[0][$function];
3037                                         } else {
3038                                                 $function = false;
3039                                         }
3040                                 }
3041                                 if ( $function ) {
3042                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3043                                         $initialArgs = array( &$this );
3044                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3045                                         if ( $flags & SFH_OBJECT_ARGS ) {
3046                                                 # Add a frame parameter, and pass the arguments as an array
3047                                                 $allArgs = $initialArgs;
3048                                                 $allArgs[] = $frame;
3049                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3050                                                         $funcArgs[] = $args->item( $i );
3051                                                 }
3052                                                 $allArgs[] = $funcArgs;
3053                                         } else {
3054                                                 # Convert arguments to plain text
3055                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3056                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3057                                                 }
3058                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3059                                         }
3060
3061                                         # Workaround for PHP bug 35229 and similar
3062                                         if ( !is_callable( $callback ) ) {
3063                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3064                                                 wfProfileOut( __METHOD__ );
3065                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3066                                         }
3067                                         $result = call_user_func_array( $callback, $allArgs );
3068                                         $found = true;
3069                                         $noparse = true;
3070                                         $preprocessFlags = 0;
3071
3072                                         if ( is_array( $result ) ) {
3073                                                 if ( isset( $result[0] ) ) {
3074                                                         $text = $result[0];
3075                                                         unset( $result[0] );
3076                                                 }
3077
3078                                                 # Extract flags into the local scope
3079                                                 # This allows callers to set flags such as nowiki, found, etc.
3080                                                 extract( $result );
3081                                         } else {
3082                                                 $text = $result;
3083                                         }
3084                                         if ( !$noparse ) {
3085                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3086                                                 $isChildObj = true;
3087                                         }
3088                                 }
3089                         }
3090                         wfProfileOut( __METHOD__ . '-pfunc' );
3091                 }
3092
3093                 # Finish mangling title and then check for loops.
3094                 # Set $title to a Title object and $titleText to the PDBK
3095                 if ( !$found ) {
3096                         $ns = NS_TEMPLATE;
3097                         # Split the title into page and subpage
3098                         $subpage = '';
3099                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3100                         if ( $subpage !== '' ) {
3101                                 $ns = $this->mTitle->getNamespace();
3102                         }
3103                         $title = Title::newFromText( $part1, $ns );
3104                         if ( $title ) {
3105                                 $titleText = $title->getPrefixedText();
3106                                 # Check for language variants if the template is not found
3107                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3108                                         $wgContLang->findVariantLink( $part1, $title, true );
3109                                 }
3110                                 # Do recursion depth check
3111                                 $limit = $this->mOptions->getMaxTemplateDepth();
3112                                 if ( $frame->depth >= $limit ) {
3113                                         $found = true;
3114                                         $text = '<span class="error">'
3115                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3116                                                 . '</span>';
3117                                 }
3118                         }
3119                 }
3120
3121                 # Load from database
3122                 if ( !$found && $title ) {
3123                         wfProfileIn( __METHOD__ . '-loadtpl' );
3124                         if ( !$title->isExternal() ) {
3125                                 if ( $title->getNamespace() == NS_SPECIAL
3126                                         && $this->mOptions->getAllowSpecialInclusion()
3127                                         && $this->ot['html'] )
3128                                 {
3129                                         $text = SpecialPage::capturePath( $title );
3130                                         if ( is_string( $text ) ) {
3131                                                 $found = true;
3132                                                 $isHTML = true;
3133                                                 $this->disableCache();
3134                                         }
3135                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3136                                         $found = false; # access denied
3137                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3138                                 } else {
3139                                         list( $text, $title ) = $this->getTemplateDom( $title );
3140                                         if ( $text !== false ) {
3141                                                 $found = true;
3142                                                 $isChildObj = true;
3143                                         }
3144                                 }
3145
3146                                 # If the title is valid but undisplayable, make a link to it
3147                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3148                                         $text = "[[:$titleText]]";
3149                                         $found = true;
3150                                 }
3151                         } elseif ( $title->isTrans() ) {
3152                                 # Interwiki transclusion
3153                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3154                                         $text = $this->interwikiTransclude( $title, 'render' );
3155                                         $isHTML = true;
3156                                 } else {
3157                                         $text = $this->interwikiTransclude( $title, 'raw' );
3158                                         # Preprocess it like a template
3159                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3160                                         $isChildObj = true;
3161                                 }
3162                                 $found = true;
3163                         }
3164
3165                         # Do infinite loop check
3166                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3167                         if ( !$frame->loopCheck( $title ) ) {
3168                                 $found = true;
3169                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3170                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3171                         }
3172                         wfProfileOut( __METHOD__ . '-loadtpl' );
3173                 }
3174
3175                 # If we haven't found text to substitute by now, we're done
3176                 # Recover the source wikitext and return it
3177                 if ( !$found ) {
3178                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3179                         wfProfileOut( __METHOD__ );
3180                         return array( 'object' => $text );
3181                 }
3182
3183                 # Expand DOM-style return values in a child frame
3184                 if ( $isChildObj ) {
3185                         # Clean up argument array
3186                         $newFrame = $frame->newChild( $args, $title );
3187
3188                         if ( $nowiki ) {
3189                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3190                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3191                                 # Expansion is eligible for the empty-frame cache
3192                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3193                                         $text = $this->mTplExpandCache[$titleText];
3194                                 } else {
3195                                         $text = $newFrame->expand( $text );
3196                                         $this->mTplExpandCache[$titleText] = $text;
3197                                 }
3198                         } else {
3199                                 # Uncached expansion
3200                                 $text = $newFrame->expand( $text );
3201                         }
3202                 }
3203                 if ( $isLocalObj && $nowiki ) {
3204                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3205                         $isLocalObj = false;
3206                 }
3207
3208                 # Replace raw HTML by a placeholder
3209                 # Add a blank line preceding, to prevent it from mucking up
3210                 # immediately preceding headings
3211                 if ( $isHTML ) {
3212                         $text = "\n\n" . $this->insertStripItem( $text );
3213                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3214                         # Escape nowiki-style return values
3215                         $text = wfEscapeWikiText( $text );
3216                 } elseif ( is_string( $text )
3217                         && !$piece['lineStart']
3218                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3219                 {
3220                         # Bug 529: if the template begins with a table or block-level
3221                         # element, it should be treated as beginning a new line.
3222                         # This behaviour is somewhat controversial.
3223                         $text = "\n" . $text;
3224                 }
3225
3226                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3227                         # Error, oversize inclusion
3228                         if ( $titleText !== false ) {
3229                                 # Make a working, properly escaped link if possible (bug 23588)
3230                                 $text = "[[:$titleText]]";
3231                         } else {
3232                                 # This will probably not be a working link, but at least it may
3233                                 # provide some hint of where the problem is
3234                                 preg_replace( '/^:/', '', $originalTitle );
3235                                 $text = "[[:$originalTitle]]";
3236                         }
3237                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3238                         $this->limitationWarn( 'post-expand-template-inclusion' );
3239                 }
3240
3241                 if ( $isLocalObj ) {
3242                         $ret = array( 'object' => $text );
3243                 } else {
3244                         $ret = array( 'text' => $text );
3245                 }
3246
3247                 wfProfileOut( __METHOD__ );
3248                 return $ret;
3249         }
3250
3251         /**
3252          * Get the semi-parsed DOM representation of a template with a given title,
3253          * and its redirect destination title. Cached.
3254          */
3255         function getTemplateDom( $title ) {
3256                 $cacheTitle = $title;
3257                 $titleText = $title->getPrefixedDBkey();
3258
3259                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3260                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3261                         $title = Title::makeTitle( $ns, $dbk );
3262                         $titleText = $title->getPrefixedDBkey();
3263                 }
3264                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3265                         return array( $this->mTplDomCache[$titleText], $title );
3266                 }
3267
3268                 # Cache miss, go to the database
3269                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3270
3271                 if ( $text === false ) {
3272                         $this->mTplDomCache[$titleText] = false;
3273                         return array( false, $title );
3274                 }
3275
3276                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3277                 $this->mTplDomCache[ $titleText ] = $dom;
3278
3279                 if ( !$title->equals( $cacheTitle ) ) {
3280                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3281                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3282                 }
3283
3284                 return array( $dom, $title );
3285         }
3286
3287         /**
3288          * Fetch the unparsed text of a template and register a reference to it.
3289          */
3290         function fetchTemplateAndTitle( $title ) {
3291                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3292                 $stuff = call_user_func( $templateCb, $title, $this );
3293                 $text = $stuff['text'];
3294                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3295                 if ( isset( $stuff['deps'] ) ) {
3296                         foreach ( $stuff['deps'] as $dep ) {
3297                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3298                         }
3299                 }
3300                 return array( $text, $finalTitle );
3301         }
3302
3303         function fetchTemplate( $title ) {
3304                 $rv = $this->fetchTemplateAndTitle( $title );
3305                 return $rv[0];
3306         }
3307
3308         /**
3309          * Static function to get a template
3310          * Can be overridden via ParserOptions::setTemplateCallback().
3311          */
3312         static function statelessFetchTemplate( $title, $parser=false ) {
3313                 $text = $skip = false;
3314                 $finalTitle = $title;
3315                 $deps = array();
3316
3317                 # Loop to fetch the article, with up to 1 redirect
3318                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3319                         # Give extensions a chance to select the revision instead
3320                         $id = false; # Assume current
3321                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3322
3323                         if ( $skip ) {
3324                                 $text = false;
3325                                 $deps[] = array(
3326                                         'title' => $title,
3327                                         'page_id' => $title->getArticleID(),
3328                                         'rev_id' => null );
3329                                 break;
3330                         }
3331                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3332                         $rev_id = $rev ? $rev->getId() : 0;
3333                         # If there is no current revision, there is no page
3334                         if ( $id === false && !$rev ) {
3335                                 $linkCache = LinkCache::singleton();
3336                                 $linkCache->addBadLinkObj( $title );
3337                         }
3338
3339                         $deps[] = array(
3340                                 'title' => $title,
3341                                 'page_id' => $title->getArticleID(),
3342                                 'rev_id' => $rev_id );
3343
3344                         if ( $rev ) {
3345                                 $text = $rev->getText();
3346                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3347                                 global $wgContLang;
3348                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3349                                 if ( !$message->exists() ) {
3350                                         $text = false;
3351                                         break;
3352                                 }
3353                                 $text = $message->plain();
3354                         } else {
3355                                 break;
3356                         }
3357                         if ( $text === false ) {
3358                                 break;
3359                         }
3360                         # Redirect?
3361                         $finalTitle = $title;
3362                         $title = Title::newFromRedirect( $text );
3363                 }
3364                 return array(
3365                         'text' => $text,
3366                         'finalTitle' => $finalTitle,
3367                         'deps' => $deps );
3368         }
3369
3370         /**
3371          * Transclude an interwiki link.
3372          */
3373         function interwikiTransclude( $title, $action ) {
3374                 global $wgEnableScaryTranscluding;
3375
3376                 if ( !$wgEnableScaryTranscluding ) {
3377                         return wfMsgForContent('scarytranscludedisabled');
3378                 }
3379
3380                 $url = $title->getFullUrl( "action=$action" );
3381
3382                 if ( strlen( $url ) > 255 ) {
3383                         return wfMsgForContent( 'scarytranscludetoolong' );
3384                 }
3385                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3386         }
3387
3388         function fetchScaryTemplateMaybeFromCache( $url ) {
3389                 global $wgTranscludeCacheExpiry;
3390                 $dbr = wfGetDB( DB_SLAVE );
3391                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3392                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3393                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3394                 if ( $obj ) {
3395                         return $obj->tc_contents;
3396                 }
3397
3398                 $text = Http::get( $url );
3399                 if ( !$text ) {
3400                         return wfMsgForContent( 'scarytranscludefailed', $url );
3401                 }
3402
3403                 $dbw = wfGetDB( DB_MASTER );
3404                 $dbw->replace( 'transcache', array('tc_url'), array(
3405                         'tc_url' => $url,
3406                         'tc_time' => $dbw->timestamp( time() ),
3407                         'tc_contents' => $text)
3408                 );
3409                 return $text;
3410         }
3411
3412
3413         /**
3414          * Triple brace replacement -- used for template arguments
3415          * @private
3416          */
3417         function argSubstitution( $piece, $frame ) {
3418                 wfProfileIn( __METHOD__ );
3419
3420                 $error = false;
3421                 $parts = $piece['parts'];
3422                 $nameWithSpaces = $frame->expand( $piece['title'] );
3423                 $argName = trim( $nameWithSpaces );
3424                 $object = false;
3425                 $text = $frame->getArgument( $argName );
3426                 if (  $text === false && $parts->getLength() > 0
3427                   && (
3428                         $this->ot['html']
3429                         || $this->ot['pre']
3430                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3431                   )
3432                 ) {
3433                         # No match in frame, use the supplied default
3434                         $object = $parts->item( 0 )->getChildren();
3435                 }
3436                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3437                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3438                         $this->limitationWarn( 'post-expand-template-argument' );
3439                 }
3440
3441                 if ( $text === false && $object === false ) {
3442                         # No match anywhere
3443                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3444                 }
3445                 if ( $error !== false ) {
3446                         $text .= $error;
3447                 }
3448                 if ( $object !== false ) {
3449                         $ret = array( 'object' => $object );
3450                 } else {
3451                         $ret = array( 'text' => $text );
3452                 }
3453
3454                 wfProfileOut( __METHOD__ );
3455                 return $ret;
3456         }
3457
3458         /**
3459          * Return the text to be used for a given extension tag.
3460          * This is the ghost of strip().
3461          *
3462          * @param $params Associative array of parameters:
3463          *     name       PPNode for the tag name
3464          *     attr       PPNode for unparsed text where tag attributes are thought to be
3465          *     attributes Optional associative array of parsed attributes
3466          *     inner      Contents of extension element
3467          *     noClose    Original text did not have a close tag
3468          * @param $frame PPFrame
3469          */
3470         function extensionSubstitution( $params, $frame ) {
3471                 $name = $frame->expand( $params['name'] );
3472                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3473                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3474                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3475
3476                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3477                         ( $this->ot['html'] || $this->ot['pre'] );
3478                 if ( $isFunctionTag ) {
3479                         $markerType = 'none';
3480                 } else {
3481                         $markerType = 'general';
3482                 }
3483                 if ( $this->ot['html'] || $isFunctionTag ) {
3484                         $name = strtolower( $name );
3485                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3486                         if ( isset( $params['attributes'] ) ) {
3487                                 $attributes = $attributes + $params['attributes'];
3488                         }
3489
3490                         if ( isset( $this->mTagHooks[$name] ) ) {
3491                                 # Workaround for PHP bug 35229 and similar
3492                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3493                                         throw new MWException( "Tag hook for $name is not callable\n" );
3494                                 }
3495                                 $output = call_user_func_array( $this->mTagHooks[$name],
3496                                         array( $content, $attributes, $this, $frame ) );
3497                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3498                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3499                                 if ( !is_callable( $callback ) ) {
3500                                         throw new MWException( "Tag hook for $name is not callable\n" );
3501                                 }
3502
3503                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3504                         } else {
3505                                 $output = '<span class="error">Invalid tag extension name: ' .
3506                                         htmlspecialchars( $name ) . '</span>';
3507                         }
3508
3509                         if ( is_array( $output ) ) {
3510                                 # Extract flags to local scope (to override $markerType)
3511                                 $flags = $output;
3512                                 $output = $flags[0];
3513                                 unset( $flags[0] );
3514                                 extract( $flags );
3515                         }
3516                 } else {
3517                         if ( is_null( $attrText ) ) {
3518                                 $attrText = '';
3519                         }
3520                         if ( isset( $params['attributes'] ) ) {
3521                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3522                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3523                                                 htmlspecialchars( $attrValue ) . '"';
3524                                 }
3525                         }
3526                         if ( $content === null ) {
3527                                 $output = "<$name$attrText/>";
3528                         } else {
3529                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3530                                 $output = "<$name$attrText>$content$close";
3531                         }
3532                 }
3533
3534                 if ( $markerType === 'none' ) {
3535                         return $output;
3536                 } elseif ( $markerType === 'nowiki' ) {
3537                         $this->mStripState->nowiki->setPair( $marker, $output );
3538                 } elseif ( $markerType === 'general' ) {
3539                         $this->mStripState->general->setPair( $marker, $output );
3540                 } else {
3541                         throw new MWException( __METHOD__.': invalid marker type' );
3542                 }
3543                 return $marker;
3544         }
3545
3546         /**
3547          * Increment an include size counter
3548          *
3549          * @param $type String: the type of expansion
3550          * @param $size Integer: the size of the text
3551          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3552          */
3553         function incrementIncludeSize( $type, $size ) {
3554                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3555                         return false;
3556                 } else {
3557                         $this->mIncludeSizes[$type] += $size;
3558                         return true;
3559                 }
3560         }
3561
3562         /**
3563          * Increment the expensive function count
3564          *
3565          * @return Boolean: false if the limit has been exceeded
3566          */
3567         function incrementExpensiveFunctionCount() {
3568                 global $wgExpensiveParserFunctionLimit;
3569                 $this->mExpensiveFunctionCount++;
3570                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3571                         return true;
3572                 }
3573                 return false;
3574         }
3575
3576         /**
3577          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3578          * Fills $this->mDoubleUnderscores, returns the modified text
3579          */
3580         function doDoubleUnderscore( $text ) {
3581                 wfProfileIn( __METHOD__ );
3582
3583                 # The position of __TOC__ needs to be recorded
3584                 $mw = MagicWord::get( 'toc' );
3585                 if ( $mw->match( $text ) ) {
3586                         $this->mShowToc = true;
3587                         $this->mForceTocPosition = true;
3588
3589                         # Set a placeholder. At the end we'll fill it in with the TOC.
3590                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3591
3592                         # Only keep the first one.
3593                         $text = $mw->replace( '', $text );
3594                 }
3595
3596                 # Now match and remove the rest of them
3597                 $mwa = MagicWord::getDoubleUnderscoreArray();
3598                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3599
3600                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3601                         $this->mOutput->mNoGallery = true;
3602                 }
3603                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3604                         $this->mShowToc = false;
3605                 }
3606                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3607                         $this->addTrackingCategory( 'hidden-category-category' );
3608                 }
3609                 # (bug 8068) Allow control over whether robots index a page.
3610                 #
3611                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3612                 # is not desirable, the last one on the page should win.
3613                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3614                         $this->mOutput->setIndexPolicy( 'noindex' );
3615                         $this->addTrackingCategory( 'noindex-category' );
3616                 }
3617                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3618                         $this->mOutput->setIndexPolicy( 'index' );
3619                         $this->addTrackingCategory( 'index-category' );
3620                 }
3621
3622                 # Cache all double underscores in the database
3623                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3624                         $this->mOutput->setProperty( $key, '' );
3625                 }
3626
3627                 wfProfileOut( __METHOD__ );
3628                 return $text;
3629         }
3630
3631         /**
3632          * Add a tracking category, getting the title from a system message,
3633          * or print a debug message if the title is invalid.
3634          *
3635          * @param $msg String: message key
3636          * @return Boolean: whether the addition was successful
3637          */
3638         protected function addTrackingCategory( $msg ) {
3639                 $cat = wfMsgForContent( $msg );
3640
3641                 # Allow tracking categories to be disabled by setting them to "-"
3642                 if ( $cat === '-' ) {
3643                         return false;
3644                 }
3645
3646                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3647                 if ( $containerCategory ) {
3648                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3649                         return true;
3650                 } else {
3651                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3652                         return false;
3653                 }
3654         }
3655
3656         /**
3657          * This function accomplishes several tasks:
3658          * 1) Auto-number headings if that option is enabled
3659          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3660          * 3) Add a Table of contents on the top for users who have enabled the option
3661          * 4) Auto-anchor headings
3662          *
3663          * It loops through all headlines, collects the necessary data, then splits up the
3664          * string and re-inserts the newly formatted headlines.
3665          *
3666          * @param $text String
3667          * @param $origText String: original, untouched wikitext
3668          * @param $isMain Boolean
3669          * @private
3670          */
3671         function formatHeadings( $text, $origText, $isMain=true ) {
3672                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3673
3674                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3675
3676                 # Inhibit editsection links if requested in the page
3677                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3678                         $showEditLink = 0;
3679                 } else {
3680                         $showEditLink = $this->mOptions->getEditSection();
3681                 }
3682                 if ( $showEditLink ) {
3683                         $this->mOutput->setEditSectionTokens( true );
3684                 }
3685
3686                 # Get all headlines for numbering them and adding funky stuff like [edit]
3687                 # links - this is for later, but we need the number of headlines right now
3688                 $matches = array();
3689                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3690
3691                 # if there are fewer than 4 headlines in the article, do not show TOC
3692                 # unless it's been explicitly enabled.
3693                 $enoughToc = $this->mShowToc &&
3694                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3695
3696                 # Allow user to stipulate that a page should have a "new section"
3697                 # link added via __NEWSECTIONLINK__
3698                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3699                         $this->mOutput->setNewSection( true );
3700                 }
3701
3702                 # Allow user to remove the "new section"
3703                 # link via __NONEWSECTIONLINK__
3704                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3705                         $this->mOutput->hideNewSection( true );
3706                 }
3707
3708                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3709                 # override above conditions and always show TOC above first header
3710                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3711                         $this->mShowToc = true;
3712                         $enoughToc = true;
3713                 }
3714
3715                 # We need this to perform operations on the HTML
3716                 $sk = $this->mOptions->getSkin( $this->mTitle );
3717
3718                 # headline counter
3719                 $headlineCount = 0;
3720                 $numVisible = 0;
3721
3722                 # Ugh .. the TOC should have neat indentation levels which can be
3723                 # passed to the skin functions. These are determined here
3724                 $toc = '';
3725                 $full = '';
3726                 $head = array();
3727                 $sublevelCount = array();
3728                 $levelCount = array();
3729                 $level = 0;
3730                 $prevlevel = 0;
3731                 $toclevel = 0;
3732                 $prevtoclevel = 0;
3733                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3734                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3735                 $oldType = $this->mOutputType;
3736                 $this->setOutputType( self::OT_WIKI );
3737                 $frame = $this->getPreprocessor()->newFrame();
3738                 $root = $this->preprocessToDom( $origText );
3739                 $node = $root->getFirstChild();
3740                 $byteOffset = 0;
3741                 $tocraw = array();
3742                 $refers = array();
3743
3744                 foreach ( $matches[3] as $headline ) {
3745                         $isTemplate = false;
3746                         $titleText = false;
3747                         $sectionIndex = false;
3748                         $numbering = '';
3749                         $markerMatches = array();
3750                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3751                                 $serial = $markerMatches[1];
3752                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3753                                 $isTemplate = ( $titleText != $baseTitleText );
3754                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3755                         }
3756
3757                         if ( $toclevel ) {
3758                                 $prevlevel = $level;
3759                         }
3760                         $level = $matches[1][$headlineCount];
3761
3762                         if ( $level > $prevlevel ) {
3763                                 # Increase TOC level
3764                                 $toclevel++;
3765                                 $sublevelCount[$toclevel] = 0;
3766                                 if ( $toclevel<$wgMaxTocLevel ) {
3767                                         $prevtoclevel = $toclevel;
3768                                         $toc .= $sk->tocIndent();
3769                                         $numVisible++;
3770                                 }
3771                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3772                                 # Decrease TOC level, find level to jump to
3773
3774                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3775                                         if ( $levelCount[$i] == $level ) {
3776                                                 # Found last matching level
3777                                                 $toclevel = $i;
3778                                                 break;
3779                                         } elseif ( $levelCount[$i] < $level ) {
3780                                                 # Found first matching level below current level
3781                                                 $toclevel = $i + 1;
3782                                                 break;
3783                                         }
3784                                 }
3785                                 if ( $i == 0 ) {
3786                                         $toclevel = 1;
3787                                 }
3788                                 if ( $toclevel<$wgMaxTocLevel ) {
3789                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3790                                                 # Unindent only if the previous toc level was shown :p
3791                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3792                                                 $prevtoclevel = $toclevel;
3793                                         } else {
3794                                                 $toc .= $sk->tocLineEnd();
3795                                         }
3796                                 }
3797                         } else {
3798                                 # No change in level, end TOC line
3799                                 if ( $toclevel<$wgMaxTocLevel ) {
3800                                         $toc .= $sk->tocLineEnd();
3801                                 }
3802                         }
3803
3804                         $levelCount[$toclevel] = $level;
3805
3806                         # count number of headlines for each level
3807                         @$sublevelCount[$toclevel]++;
3808                         $dot = 0;
3809                         for( $i = 1; $i <= $toclevel; $i++ ) {
3810                                 if ( !empty( $sublevelCount[$i] ) ) {
3811                                         if ( $dot ) {
3812                                                 $numbering .= '.';
3813                                         }
3814                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3815                                         $dot = 1;
3816                                 }
3817                         }
3818
3819                         # The safe header is a version of the header text safe to use for links
3820                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3821                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3822
3823                         # Remove link placeholders by the link text.
3824                         #     <!--LINK number-->
3825                         # turns into
3826                         #     link text with suffix
3827                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3828
3829                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3830                         $tocline = preg_replace(
3831                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3832                                 array( '',                          '<$1>' ),
3833                                 $safeHeadline
3834                         );
3835                         $tocline = trim( $tocline );
3836
3837                         # For the anchor, strip out HTML-y stuff period
3838                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3839                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
3840
3841                         # Save headline for section edit hint before it's escaped
3842                         $headlineHint = $safeHeadline;
3843
3844                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3845                                 # For reverse compatibility, provide an id that's
3846                                 # HTML4-compatible, like we used to.
3847                                 #
3848                                 # It may be worth noting, academically, that it's possible for
3849                                 # the legacy anchor to conflict with a non-legacy headline
3850                                 # anchor on the page.  In this case likely the "correct" thing
3851                                 # would be to either drop the legacy anchors or make sure
3852                                 # they're numbered first.  However, this would require people
3853                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3854                                 # manually, so let's not bother worrying about it.
3855                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3856                                         array( 'noninitial', 'legacy' ) );
3857                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3858
3859                                 if ( $legacyHeadline == $safeHeadline ) {
3860                                         # No reason to have both (in fact, we can't)
3861                                         $legacyHeadline = false;
3862                                 }
3863                         } else {
3864                                 $legacyHeadline = false;
3865                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3866                                         'noninitial' );
3867                         }
3868
3869                         # HTML names must be case-insensitively unique (bug 10721).
3870                         # This does not apply to Unicode characters per
3871                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
3872                         # FIXME: We may be changing them depending on the current locale.
3873                         $arrayKey = strtolower( $safeHeadline );
3874                         if ( $legacyHeadline === false ) {
3875                                 $legacyArrayKey = false;
3876                         } else {
3877                                 $legacyArrayKey = strtolower( $legacyHeadline );
3878                         }
3879
3880                         # count how many in assoc. array so we can track dupes in anchors
3881                         if ( isset( $refers[$arrayKey] ) ) {
3882                                 $refers[$arrayKey]++;
3883                         } else {
3884                                 $refers[$arrayKey] = 1;
3885                         }
3886                         if ( isset( $refers[$legacyArrayKey] ) ) {
3887                                 $refers[$legacyArrayKey]++;
3888                         } else {
3889                                 $refers[$legacyArrayKey] = 1;
3890                         }
3891
3892                         # Don't number the heading if it is the only one (looks silly)
3893                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3894                                 # the two are different if the line contains a link
3895                                 $headline = $numbering . ' ' . $headline;
3896                         }
3897
3898                         # Create the anchor for linking from the TOC to the section
3899                         $anchor = $safeHeadline;
3900                         $legacyAnchor = $legacyHeadline;
3901                         if ( $refers[$arrayKey] > 1 ) {
3902                                 $anchor .= '_' . $refers[$arrayKey];
3903                         }
3904                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3905                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3906                         }
3907                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3908                                 $toc .= $sk->tocLine( $anchor, $tocline,
3909                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3910                         }
3911
3912                         # Add the section to the section tree
3913                         # Find the DOM node for this header
3914                         while ( $node && !$isTemplate ) {
3915                                 if ( $node->getName() === 'h' ) {
3916                                         $bits = $node->splitHeading();
3917                                         if ( $bits['i'] == $sectionIndex ) {
3918                                                 break;
3919                                         }
3920                                 }
3921                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3922                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3923                                 $node = $node->getNextSibling();
3924                         }
3925                         $tocraw[] = array(
3926                                 'toclevel' => $toclevel,
3927                                 'level' => $level,
3928                                 'line' => $tocline,
3929                                 'number' => $numbering,
3930                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3931                                 'fromtitle' => $titleText,
3932                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3933                                 'anchor' => $anchor,
3934                         );
3935
3936                         # give headline the correct <h#> tag
3937                         if ( $showEditLink && $sectionIndex !== false ) {
3938                                 // Output edit section links as markers with styles that can be customized by skins
3939                                 if ( $isTemplate ) {
3940                                         # Put a T flag in the section identifier, to indicate to extractSections()
3941                                         # that sections inside <includeonly> should be counted.
3942                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
3943                                 } else {
3944                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
3945                                 }
3946                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
3947                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
3948                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
3949                                 // so we don't have to worry about a user trying to input one of these markers directly.
3950                                 // We use a page and section attribute to stop the language converter from converting these important bits
3951                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
3952                                 // be able to convert that piece of data.
3953                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
3954                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
3955                                 if ( isset($editlinkArgs[2]) ) {
3956                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
3957                                 } else {
3958                                         $editlink .= '/>';
3959                                 }
3960                         } else {
3961                                 $editlink = '';
3962                         }
3963                         $head[$headlineCount] = $sk->makeHeadline( $level,
3964                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3965                                 $editlink, $legacyAnchor );
3966
3967                         $headlineCount++;
3968                 }
3969
3970                 $this->setOutputType( $oldType );
3971
3972                 # Never ever show TOC if no headers
3973                 if ( $numVisible < 1 ) {
3974                         $enoughToc = false;
3975                 }
3976
3977                 if ( $enoughToc ) {
3978                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3979                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3980                         }
3981                         $toc = $sk->tocList( $toc );
3982                         $this->mOutput->setTOCHTML( $toc );
3983                 }
3984
3985                 if ( $isMain ) {
3986                         $this->mOutput->setSections( $tocraw );
3987                 }
3988
3989                 # split up and insert constructed headlines
3990
3991                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3992                 $i = 0;
3993
3994                 foreach ( $blocks as $block ) {
3995                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3996                                 # This is the [edit] link that appears for the top block of text when
3997                                 # section editing is enabled
3998
3999                                 # Disabled because it broke block formatting
4000                                 # For example, a bullet point in the top line
4001                                 # $full .= $sk->editSectionLink(0);
4002                         }
4003                         $full .= $block;
4004                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4005                                 # Top anchor now in skin
4006                                 $full = $full.$toc;
4007                         }
4008
4009                         if ( !empty( $head[$i] ) ) {
4010                                 $full .= $head[$i];
4011                         }
4012                         $i++;
4013                 }
4014                 if ( $this->mForceTocPosition ) {
4015                         return str_replace( '<!--MWTOC-->', $toc, $full );
4016                 } else {
4017                         return $full;
4018                 }
4019         }
4020
4021         /**
4022          * Transform wiki markup when saving a page by doing \r\n -> \n
4023          * conversion, substitting signatures, {{subst:}} templates, etc.
4024          *
4025          * @param $text String: the text to transform
4026          * @param $title Title: the Title object for the current article
4027          * @param $user User: the User object describing the current user
4028          * @param $options ParserOptions: parsing options
4029          * @param $clearState Boolean: whether to clear the parser state first
4030          * @return String: the altered wiki markup
4031          */
4032         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4033                 $this->startExternalParse( $title, $options, self::OT_WIKI, $clearState );
4034                 $this->setUser( $user );
4035
4036                 $pairs = array(
4037                         "\r\n" => "\n",
4038                 );
4039                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4040                 if( $options->getPreSaveTransform() ) {
4041                         $text = $this->pstPass2( $text, $user );
4042                 }
4043                 $text = $this->mStripState->unstripBoth( $text );
4044
4045                 $this->setUser( null ); #Reset
4046
4047                 return $text;
4048         }
4049
4050         /**
4051          * Pre-save transform helper function
4052          * @private
4053          */
4054         function pstPass2( $text, $user ) {
4055                 global $wgContLang, $wgLocaltimezone;
4056
4057                 # Note: This is the timestamp saved as hardcoded wikitext to
4058                 # the database, we use $wgContLang here in order to give
4059                 # everyone the same signature and use the default one rather
4060                 # than the one selected in each user's preferences.
4061                 # (see also bug 12815)
4062                 $ts = $this->mOptions->getTimestamp();
4063                 if ( isset( $wgLocaltimezone ) ) {
4064                         $tz = $wgLocaltimezone;
4065                 } else {
4066                         $tz = date_default_timezone_get();
4067                 }
4068
4069                 $unixts = wfTimestamp( TS_UNIX, $ts );
4070                 $oldtz = date_default_timezone_get();
4071                 date_default_timezone_set( $tz );
4072                 $ts = date( 'YmdHis', $unixts );
4073                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4074
4075                 # Allow translation of timezones through wiki. date() can return
4076                 # whatever crap the system uses, localised or not, so we cannot
4077                 # ship premade translations.
4078                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4079                 $msg = wfMessage( $key )->inContentLanguage();
4080                 if ( $msg->exists() ) {
4081                         $tzMsg = $msg->text();
4082                 }
4083
4084                 date_default_timezone_set( $oldtz );
4085
4086                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4087
4088                 # Variable replacement
4089                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4090                 $text = $this->replaceVariables( $text );
4091
4092                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4093                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4094
4095                 # Signatures
4096                 $sigText = $this->getUserSig( $user );
4097                 $text = strtr( $text, array(
4098                         '~~~~~' => $d,
4099                         '~~~~' => "$sigText $d",
4100                         '~~~' => $sigText
4101                 ) );
4102
4103                 # Context links: [[|name]] and [[name (context)|]]
4104                 global $wgLegalTitleChars;
4105                 $tc = "[$wgLegalTitleChars]";
4106                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4107
4108                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4109                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4110                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4111                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4112
4113                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4114                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4115                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4116                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4117
4118                 $t = $this->mTitle->getText();
4119                 $m = array();
4120                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4121                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4122                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4123                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4124                 } else {
4125                         # if there's no context, don't bother duplicating the title
4126                         $text = preg_replace( $p2, '[[\\1]]', $text );
4127                 }
4128
4129                 # Trim trailing whitespace
4130                 $text = rtrim( $text );
4131
4132                 return $text;
4133         }
4134
4135         /**
4136          * Fetch the user's signature text, if any, and normalize to
4137          * validated, ready-to-insert wikitext.
4138          * If you have pre-fetched the nickname or the fancySig option, you can
4139          * specify them here to save a database query.
4140          * Do not reuse this parser instance after calling getUserSig(),
4141          * as it may have changed if it's the $wgParser.
4142          *
4143          * @param $user User
4144          * @param $nickname String: nickname to use or false to use user's default nickname
4145          * @param $fancySig Boolean: whether the nicknname is the complete signature
4146          *                  or null to use default value
4147          * @return string
4148          */
4149         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4150                 global $wgMaxSigChars;
4151
4152                 $username = $user->getName();
4153
4154                 # If not given, retrieve from the user object.
4155                 if ( $nickname === false )
4156                         $nickname = $user->getOption( 'nickname' );
4157
4158                 if ( is_null( $fancySig ) ) {
4159                         $fancySig = $user->getBoolOption( 'fancysig' );
4160                 }
4161
4162                 $nickname = $nickname == null ? $username : $nickname;
4163
4164                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4165                         $nickname = $username;
4166                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4167                 } elseif ( $fancySig !== false ) {
4168                         # Sig. might contain markup; validate this
4169                         if ( $this->validateSig( $nickname ) !== false ) {
4170                                 # Validated; clean up (if needed) and return it
4171                                 return $this->cleanSig( $nickname, true );
4172                         } else {
4173                                 # Failed to validate; fall back to the default
4174                                 $nickname = $username;
4175                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4176                         }
4177                 }
4178
4179                 # Make sure nickname doesnt get a sig in a sig
4180                 $nickname = $this->cleanSigInSig( $nickname );
4181
4182                 # If we're still here, make it a link to the user page
4183                 $userText = wfEscapeWikiText( $username );
4184                 $nickText = wfEscapeWikiText( $nickname );
4185                 if ( $user->isAnon() )  {
4186                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4187                 } else {
4188                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4189                 }
4190         }
4191
4192         /**
4193          * Check that the user's signature contains no bad XML
4194          *
4195          * @param $text String
4196          * @return mixed An expanded string, or false if invalid.
4197          */
4198         function validateSig( $text ) {
4199                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4200         }
4201
4202         /**
4203          * Clean up signature text
4204          *
4205          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4206          * 2) Substitute all transclusions
4207          *
4208          * @param $text String
4209          * @param $parsing Whether we're cleaning (preferences save) or parsing
4210          * @return String: signature text
4211          */
4212         function cleanSig( $text, $parsing = false ) {
4213                 if ( !$parsing ) {
4214                         global $wgTitle;
4215                         $this->mOptions = new ParserOptions;
4216                         $this->clearState();
4217                         $this->setTitle( $wgTitle );
4218                         $this->setOutputType = self::OT_PREPROCESS;
4219                 }
4220
4221                 # Option to disable this feature
4222                 if ( !$this->mOptions->getCleanSignatures() ) {
4223                         return $text;
4224                 }
4225
4226                 # FIXME: regex doesn't respect extension tags or nowiki
4227                 #  => Move this logic to braceSubstitution()
4228                 $substWord = MagicWord::get( 'subst' );
4229                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4230                 $substText = '{{' . $substWord->getSynonym( 0 );
4231
4232                 $text = preg_replace( $substRegex, $substText, $text );
4233                 $text = $this->cleanSigInSig( $text );
4234                 $dom = $this->preprocessToDom( $text );
4235                 $frame = $this->getPreprocessor()->newFrame();
4236                 $text = $frame->expand( $dom );
4237
4238                 if ( !$parsing ) {
4239                         $text = $this->mStripState->unstripBoth( $text );
4240                 }
4241
4242                 return $text;
4243         }
4244
4245         /**
4246          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4247          *
4248          * @param $text String
4249          * @return String: signature text with /~{3,5}/ removed
4250          */
4251         function cleanSigInSig( $text ) {
4252                 $text = preg_replace( '/~{3,5}/', '', $text );
4253                 return $text;
4254         }
4255
4256         /**
4257          * Set up some variables which are usually set up in parse()
4258          * so that an external function can call some class members with confidence
4259          */
4260         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4261                 $this->setTitle( $title );
4262                 $this->mOptions = $options;
4263                 $this->setOutputType( $outputType );
4264                 if ( $clearState ) {
4265                         $this->clearState();
4266                 }
4267         }
4268
4269         /**
4270          * Wrapper for preprocess()
4271          *
4272          * @param $text String: the text to preprocess
4273          * @param $options ParserOptions: options
4274          * @return String
4275          */
4276         public function transformMsg( $text, $options ) {
4277                 global $wgTitle;
4278                 static $executing = false;
4279
4280                 # Guard against infinite recursion
4281                 if ( $executing ) {
4282                         return $text;
4283                 }
4284                 $executing = true;
4285
4286                 wfProfileIn( __METHOD__ );
4287                 $title = $wgTitle;
4288                 if ( !$title ) {
4289                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4290                         # Create a ghost title in such case
4291                         $title = Title::newFromText( 'Dwimmerlaik' );
4292                 }
4293                 $text = $this->preprocess( $text, $title, $options );
4294
4295                 $executing = false;
4296                 wfProfileOut( __METHOD__ );
4297                 return $text;
4298         }
4299
4300         /**
4301          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4302          * The callback should have the following form:
4303          *    function myParserHook( $text, $params, $parser ) { ... }
4304          *
4305          * Transform and return $text. Use $parser for any required context, e.g. use
4306          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4307          *
4308          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4309          * @param $callback Mixed: the callback function (and object) to use for the tag
4310          * @return The old value of the mTagHooks array associated with the hook
4311          */
4312         public function setHook( $tag, $callback ) {
4313                 $tag = strtolower( $tag );
4314                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4315                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4316                 $this->mTagHooks[$tag] = $callback;
4317                 if ( !in_array( $tag, $this->mStripList ) ) {
4318                         $this->mStripList[] = $tag;
4319                 }
4320
4321                 return $oldVal;
4322         }
4323
4324         function setTransparentTagHook( $tag, $callback ) {
4325                 $tag = strtolower( $tag );
4326                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4327                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4328                 $this->mTransparentTagHooks[$tag] = $callback;
4329
4330                 return $oldVal;
4331         }
4332
4333         /**
4334          * Remove all tag hooks
4335          */
4336         function clearTagHooks() {
4337                 $this->mTagHooks = array();
4338                 $this->mStripList = $this->mDefaultStripList;
4339         }
4340
4341         /**
4342          * Create a function, e.g. {{sum:1|2|3}}
4343          * The callback function should have the form:
4344          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4345          *
4346          * Or with SFH_OBJECT_ARGS:
4347          *    function myParserFunction( $parser, $frame, $args ) { ... }
4348          *
4349          * The callback may either return the text result of the function, or an array with the text
4350          * in element 0, and a number of flags in the other elements. The names of the flags are
4351          * specified in the keys. Valid flags are:
4352          *   found                     The text returned is valid, stop processing the template. This
4353          *                             is on by default.
4354          *   nowiki                    Wiki markup in the return value should be escaped
4355          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4356          *
4357          * @param $id String: The magic word ID
4358          * @param $callback Mixed: the callback function (and object) to use
4359          * @param $flags Integer: a combination of the following flags:
4360          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4361          *
4362          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4363          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4364          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4365          *     the arguments, and to control the way they are expanded.
4366          *
4367          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4368          *     arguments, for instance:
4369          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4370          *
4371          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4372          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4373          *     working if/when this is changed.
4374          *
4375          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4376          *     expansion.
4377          *
4378          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4379          *     about the methods available in PPFrame and PPNode.
4380          *
4381          * @return The old callback function for this name, if any
4382          */
4383         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4384                 global $wgContLang;
4385
4386                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4387                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4388
4389                 # Add to function cache
4390                 $mw = MagicWord::get( $id );
4391                 if ( !$mw )
4392                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4393
4394                 $synonyms = $mw->getSynonyms();
4395                 $sensitive = intval( $mw->isCaseSensitive() );
4396
4397                 foreach ( $synonyms as $syn ) {
4398                         # Case
4399                         if ( !$sensitive ) {
4400                                 $syn = $wgContLang->lc( $syn );
4401                         }
4402                         # Add leading hash
4403                         if ( !( $flags & SFH_NO_HASH ) ) {
4404                                 $syn = '#' . $syn;
4405                         }
4406                         # Remove trailing colon
4407                         if ( substr( $syn, -1, 1 ) === ':' ) {
4408                                 $syn = substr( $syn, 0, -1 );
4409                         }
4410                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4411                 }
4412                 return $oldVal;
4413         }
4414
4415         /**
4416          * Get all registered function hook identifiers
4417          *
4418          * @return Array
4419          */
4420         function getFunctionHooks() {
4421                 return array_keys( $this->mFunctionHooks );
4422         }
4423
4424         /**
4425          * Create a tag function, e.g. <test>some stuff</test>.
4426          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4427          * Unlike parser functions, their content is not preprocessed.
4428          */
4429         function setFunctionTagHook( $tag, $callback, $flags ) {
4430                 $tag = strtolower( $tag );
4431                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4432                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4433                         $this->mFunctionTagHooks[$tag] : null;
4434                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4435
4436                 if ( !in_array( $tag, $this->mStripList ) ) {
4437                         $this->mStripList[] = $tag;
4438                 }
4439
4440                 return $old;
4441         }
4442
4443         /**
4444          * FIXME: update documentation. makeLinkObj() is deprecated.
4445          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4446          * Placeholders created in Skin::makeLinkObj()
4447          * Returns an array of link CSS classes, indexed by PDBK.
4448          */
4449         function replaceLinkHolders( &$text, $options = 0 ) {
4450                 return $this->mLinkHolders->replace( $text );
4451         }
4452
4453         /**
4454          * Replace <!--LINK--> link placeholders with plain text of links
4455          * (not HTML-formatted).
4456          *
4457          * @param $text String
4458          * @return String
4459          */
4460         function replaceLinkHoldersText( $text ) {
4461                 return $this->mLinkHolders->replaceText( $text );
4462         }
4463
4464         /**
4465          * Renders an image gallery from a text with one line per image.
4466          * text labels may be given by using |-style alternative text. E.g.
4467          *   Image:one.jpg|The number "1"
4468          *   Image:tree.jpg|A tree
4469          * given as text will return the HTML of a gallery with two images,
4470          * labeled 'The number "1"' and
4471          * 'A tree'.
4472          */
4473         function renderImageGallery( $text, $params ) {
4474                 $ig = new ImageGallery();
4475                 $ig->setContextTitle( $this->mTitle );
4476                 $ig->setShowBytes( false );
4477                 $ig->setShowFilename( false );
4478                 $ig->setParser( $this );
4479                 $ig->setHideBadImages();
4480                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4481                 $ig->useSkin( $this->mOptions->getSkin( $this->mTitle ) );
4482                 $ig->mRevisionId = $this->mRevisionId;
4483
4484                 if ( isset( $params['showfilename'] ) ) {
4485                         $ig->setShowFilename( true );
4486                 } else {
4487                         $ig->setShowFilename( false );
4488                 }
4489                 if ( isset( $params['caption'] ) ) {
4490                         $caption = $params['caption'];
4491                         $caption = htmlspecialchars( $caption );
4492                         $caption = $this->replaceInternalLinks( $caption );
4493                         $ig->setCaptionHtml( $caption );
4494                 }
4495                 if ( isset( $params['perrow'] ) ) {
4496                         $ig->setPerRow( $params['perrow'] );
4497                 }
4498                 if ( isset( $params['widths'] ) ) {
4499                         $ig->setWidths( $params['widths'] );
4500                 }
4501                 if ( isset( $params['heights'] ) ) {
4502                         $ig->setHeights( $params['heights'] );
4503                 }
4504
4505                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4506
4507                 $lines = StringUtils::explode( "\n", $text );
4508                 foreach ( $lines as $line ) {
4509                         # match lines like these:
4510                         # Image:someimage.jpg|This is some image
4511                         $matches = array();
4512                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4513                         # Skip empty lines
4514                         if ( count( $matches ) == 0 ) {
4515                                 continue;
4516                         }
4517
4518                         if ( strpos( $matches[0], '%' ) !== false ) {
4519                                 $matches[1] = rawurldecode( $matches[1] );
4520                         }
4521                         $tp = Title::newFromText( $matches[1], NS_FILE );
4522                         $nt =& $tp;
4523                         if ( is_null( $nt ) ) {
4524                                 # Bogus title. Ignore these so we don't bomb out later.
4525                                 continue;
4526                         }
4527                         if ( isset( $matches[3] ) ) {
4528                                 $label = $matches[3];
4529                         } else {
4530                                 $label = '';
4531                         }
4532
4533                         $html = $this->recursiveTagParse( trim( $label ) );
4534
4535                         $ig->add( $nt, $html );
4536
4537                         # Only add real images (bug #5586)
4538                         if ( $nt->getNamespace() == NS_FILE ) {
4539                                 $this->mOutput->addImage( $nt->getDBkey() );
4540                         }
4541                 }
4542                 return $ig->toHTML();
4543         }
4544
4545         function getImageParams( $handler ) {
4546                 if ( $handler ) {
4547                         $handlerClass = get_class( $handler );
4548                 } else {
4549                         $handlerClass = '';
4550                 }
4551                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4552                         # Initialise static lists
4553                         static $internalParamNames = array(
4554                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4555                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4556                                         'bottom', 'text-bottom' ),
4557                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4558                                         'upright', 'border', 'link', 'alt' ),
4559                         );
4560                         static $internalParamMap;
4561                         if ( !$internalParamMap ) {
4562                                 $internalParamMap = array();
4563                                 foreach ( $internalParamNames as $type => $names ) {
4564                                         foreach ( $names as $name ) {
4565                                                 $magicName = str_replace( '-', '_', "img_$name" );
4566                                                 $internalParamMap[$magicName] = array( $type, $name );
4567                                         }
4568                                 }
4569                         }
4570
4571                         # Add handler params
4572                         $paramMap = $internalParamMap;
4573                         if ( $handler ) {
4574                                 $handlerParamMap = $handler->getParamMap();
4575                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4576                                         $paramMap[$magic] = array( 'handler', $paramName );
4577                                 }
4578                         }
4579                         $this->mImageParams[$handlerClass] = $paramMap;
4580                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4581                 }
4582                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4583         }
4584
4585         /**
4586          * Parse image options text and use it to make an image
4587          *
4588          * @param $title Title
4589          * @param $options String
4590          * @param $holders LinkHolderArray
4591          */
4592         function makeImage( $title, $options, $holders = false ) {
4593                 # Check if the options text is of the form "options|alt text"
4594                 # Options are:
4595                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4596                 #  * left       no resizing, just left align. label is used for alt= only
4597                 #  * right      same, but right aligned
4598                 #  * none       same, but not aligned
4599                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4600                 #  * center     center the image
4601                 #  * frame      Keep original image size, no magnify-button.
4602                 #  * framed     Same as "frame"
4603                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4604                 #  * upright    reduce width for upright images, rounded to full __0 px
4605                 #  * border     draw a 1px border around the image
4606                 #  * alt        Text for HTML alt attribute (defaults to empty)
4607                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4608                 # vertical-align values (no % or length right now):
4609                 #  * baseline
4610                 #  * sub
4611                 #  * super
4612                 #  * top
4613                 #  * text-top
4614                 #  * middle
4615                 #  * bottom
4616                 #  * text-bottom
4617
4618                 $parts = StringUtils::explode( "|", $options );
4619                 $sk = $this->mOptions->getSkin( $this->mTitle );
4620
4621                 # Give extensions a chance to select the file revision for us
4622                 $skip = $time = $descQuery = false;
4623                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4624
4625                 if ( $skip ) {
4626                         return $sk->link( $title );
4627                 }
4628
4629                 # Get the file
4630                 $file = wfFindFile( $title, array( 'time' => $time ) );
4631                 # Get parameter map
4632                 $handler = $file ? $file->getHandler() : false;
4633
4634                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4635
4636                 # Process the input parameters
4637                 $caption = '';
4638                 $params = array( 'frame' => array(), 'handler' => array(),
4639                         'horizAlign' => array(), 'vertAlign' => array() );
4640                 foreach ( $parts as $part ) {
4641                         $part = trim( $part );
4642                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4643                         $validated = false;
4644                         if ( isset( $paramMap[$magicName] ) ) {
4645                                 list( $type, $paramName ) = $paramMap[$magicName];
4646
4647                                 # Special case; width and height come in one variable together
4648                                 if ( $type === 'handler' && $paramName === 'width' ) {
4649                                         $m = array();
4650                                         # (bug 13500) In both cases (width/height and width only),
4651                                         # permit trailing "px" for backward compatibility.
4652                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4653                                                 $width = intval( $m[1] );
4654                                                 $height = intval( $m[2] );
4655                                                 if ( $handler->validateParam( 'width', $width ) ) {
4656                                                         $params[$type]['width'] = $width;
4657                                                         $validated = true;
4658                                                 }
4659                                                 if ( $handler->validateParam( 'height', $height ) ) {
4660                                                         $params[$type]['height'] = $height;
4661                                                         $validated = true;
4662                                                 }
4663                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4664                                                 $width = intval( $value );
4665                                                 if ( $handler->validateParam( 'width', $width ) ) {
4666                                                         $params[$type]['width'] = $width;
4667                                                         $validated = true;
4668                                                 }
4669                                         } # else no validation -- bug 13436
4670                                 } else {
4671                                         if ( $type === 'handler' ) {
4672                                                 # Validate handler parameter
4673                                                 $validated = $handler->validateParam( $paramName, $value );
4674                                         } else {
4675                                                 # Validate internal parameters
4676                                                 switch( $paramName ) {
4677                                                 case 'manualthumb':
4678                                                 case 'alt':
4679                                                         # @todo Fixme: possibly check validity here for
4680                                                         # manualthumb? downstream behavior seems odd with
4681                                                         # missing manual thumbs.
4682                                                         $validated = true;
4683                                                         $value = $this->stripAltText( $value, $holders );
4684                                                         break;
4685                                                 case 'link':
4686                                                         $chars = self::EXT_LINK_URL_CLASS;
4687                                                         $prots = $this->mUrlProtocols;
4688                                                         if ( $value === '' ) {
4689                                                                 $paramName = 'no-link';
4690                                                                 $value = true;
4691                                                                 $validated = true;
4692                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4693                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4694                                                                         $paramName = 'link-url';
4695                                                                         $this->mOutput->addExternalLink( $value );
4696                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4697                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4698                                                                         }
4699                                                                         $validated = true;
4700                                                                 }
4701                                                         } else {
4702                                                                 $linkTitle = Title::newFromText( $value );
4703                                                                 if ( $linkTitle ) {
4704                                                                         $paramName = 'link-title';
4705                                                                         $value = $linkTitle;
4706                                                                         $this->mOutput->addLink( $linkTitle );
4707                                                                         $validated = true;
4708                                                                 }
4709                                                         }
4710                                                         break;
4711                                                 default:
4712                                                         # Most other things appear to be empty or numeric...
4713                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4714                                                 }
4715                                         }
4716
4717                                         if ( $validated ) {
4718                                                 $params[$type][$paramName] = $value;
4719                                         }
4720                                 }
4721                         }
4722                         if ( !$validated ) {
4723                                 $caption = $part;
4724                         }
4725                 }
4726
4727                 # Process alignment parameters
4728                 if ( $params['horizAlign'] ) {
4729                         $params['frame']['align'] = key( $params['horizAlign'] );
4730                 }
4731                 if ( $params['vertAlign'] ) {
4732                         $params['frame']['valign'] = key( $params['vertAlign'] );
4733                 }
4734
4735                 $params['frame']['caption'] = $caption;
4736
4737                 # Will the image be presented in a frame, with the caption below?
4738                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4739                                                  isset( $params['frame']['framed'] ) ||
4740                                                  isset( $params['frame']['thumbnail'] ) ||
4741                                                  isset( $params['frame']['manualthumb'] );
4742
4743                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4744                 # came to also set the caption, ordinary text after the image -- which
4745                 # makes no sense, because that just repeats the text multiple times in
4746                 # screen readers.  It *also* came to set the title attribute.
4747                 #
4748                 # Now that we have an alt attribute, we should not set the alt text to
4749                 # equal the caption: that's worse than useless, it just repeats the
4750                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4751                 # use the unnamed parameter for alt text as well, just for the time be-
4752                 # ing, if the unnamed param is set and the alt param is not.
4753                 #
4754                 # For the future, we need to figure out if we want to tweak this more,
4755                 # e.g., introducing a title= parameter for the title; ignoring the un-
4756                 # named parameter entirely for images without a caption; adding an ex-
4757                 # plicit caption= parameter and preserving the old magic unnamed para-
4758                 # meter for BC; ...
4759                 if ( $imageIsFramed ) { # Framed image
4760                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4761                                 # No caption or alt text, add the filename as the alt text so
4762                                 # that screen readers at least get some description of the image
4763                                 $params['frame']['alt'] = $title->getText();
4764                         }
4765                         # Do not set $params['frame']['title'] because tooltips don't make sense
4766                         # for framed images
4767                 } else { # Inline image
4768                         if ( !isset( $params['frame']['alt'] ) ) {
4769                                 # No alt text, use the "caption" for the alt text
4770                                 if ( $caption !== '') {
4771                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4772                                 } else {
4773                                         # No caption, fall back to using the filename for the
4774                                         # alt text
4775                                         $params['frame']['alt'] = $title->getText();
4776                                 }
4777                         }
4778                         # Use the "caption" for the tooltip text
4779                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4780                 }
4781
4782                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4783
4784                 # Linker does the rest
4785                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery, $this->mOptions->getThumbSize() );
4786
4787                 # Give the handler a chance to modify the parser object
4788                 if ( $handler ) {
4789                         $handler->parserTransformHook( $this, $file );
4790                 }
4791
4792                 return $ret;
4793         }
4794
4795         protected function stripAltText( $caption, $holders ) {
4796                 # Strip bad stuff out of the title (tooltip).  We can't just use
4797                 # replaceLinkHoldersText() here, because if this function is called
4798                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4799                 if ( $holders ) {
4800                         $tooltip = $holders->replaceText( $caption );
4801                 } else {
4802                         $tooltip = $this->replaceLinkHoldersText( $caption );
4803                 }
4804
4805                 # make sure there are no placeholders in thumbnail attributes
4806                 # that are later expanded to html- so expand them now and
4807                 # remove the tags
4808                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4809                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4810
4811                 return $tooltip;
4812         }
4813
4814         /**
4815          * Set a flag in the output object indicating that the content is dynamic and
4816          * shouldn't be cached.
4817          */
4818         function disableCache() {
4819                 wfDebug( "Parser output marked as uncacheable.\n" );
4820                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
4821                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
4822         }
4823
4824         /**
4825          * Callback from the Sanitizer for expanding items found in HTML attribute
4826          * values, so they can be safely tested and escaped.
4827          *
4828          * @param $text String
4829          * @param $frame PPFrame
4830          * @return String
4831          * @private
4832          */
4833         function attributeStripCallback( &$text, $frame = false ) {
4834                 $text = $this->replaceVariables( $text, $frame );
4835                 $text = $this->mStripState->unstripBoth( $text );
4836                 return $text;
4837         }
4838
4839         /**
4840          * Accessor
4841          */
4842         function getTags() {
4843                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4844         }
4845
4846         /**
4847          * Break wikitext input into sections, and either pull or replace
4848          * some particular section's text.
4849          *
4850          * External callers should use the getSection and replaceSection methods.
4851          *
4852          * @param $text String: Page wikitext
4853          * @param $section String: a section identifier string of the form:
4854          *   <flag1> - <flag2> - ... - <section number>
4855          *
4856          * Currently the only recognised flag is "T", which means the target section number
4857          * was derived during a template inclusion parse, in other words this is a template
4858          * section edit link. If no flags are given, it was an ordinary section edit link.
4859          * This flag is required to avoid a section numbering mismatch when a section is
4860          * enclosed by <includeonly> (bug 6563).
4861          *
4862          * The section number 0 pulls the text before the first heading; other numbers will
4863          * pull the given section along with its lower-level subsections. If the section is
4864          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4865          *
4866          * @param $mode String: one of "get" or "replace"
4867          * @param $newText String: replacement text for section data.
4868          * @return String: for "get", the extracted section text.
4869          *                 for "replace", the whole page with the section replaced.
4870          */
4871         private function extractSections( $text, $section, $mode, $newText='' ) {
4872                 global $wgTitle; # not generally used but removes an ugly failure mode
4873                 $this->startExternalParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
4874                 $outText = '';
4875                 $frame = $this->getPreprocessor()->newFrame();
4876
4877                 # Process section extraction flags
4878                 $flags = 0;
4879                 $sectionParts = explode( '-', $section );
4880                 $sectionIndex = array_pop( $sectionParts );
4881                 foreach ( $sectionParts as $part ) {
4882                         if ( $part === 'T' ) {
4883                                 $flags |= self::PTD_FOR_INCLUSION;
4884                         }
4885                 }
4886                 # Preprocess the text
4887                 $root = $this->preprocessToDom( $text, $flags );
4888
4889                 # <h> nodes indicate section breaks
4890                 # They can only occur at the top level, so we can find them by iterating the root's children
4891                 $node = $root->getFirstChild();
4892
4893                 # Find the target section
4894                 if ( $sectionIndex == 0 ) {
4895                         # Section zero doesn't nest, level=big
4896                         $targetLevel = 1000;
4897                 } else {
4898                         while ( $node ) {
4899                                 if ( $node->getName() === 'h' ) {
4900                                         $bits = $node->splitHeading();
4901                                         if ( $bits['i'] == $sectionIndex ) {
4902                                                 $targetLevel = $bits['level'];
4903                                                 break;
4904                                         }
4905                                 }
4906                                 if ( $mode === 'replace' ) {
4907                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4908                                 }
4909                                 $node = $node->getNextSibling();
4910                         }
4911                 }
4912
4913                 if ( !$node ) {
4914                         # Not found
4915                         if ( $mode === 'get' ) {
4916                                 return $newText;
4917                         } else {
4918                                 return $text;
4919                         }
4920                 }
4921
4922                 # Find the end of the section, including nested sections
4923                 do {
4924                         if ( $node->getName() === 'h' ) {
4925                                 $bits = $node->splitHeading();
4926                                 $curLevel = $bits['level'];
4927                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4928                                         break;
4929                                 }
4930                         }
4931                         if ( $mode === 'get' ) {
4932                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4933                         }
4934                         $node = $node->getNextSibling();
4935                 } while ( $node );
4936
4937                 # Write out the remainder (in replace mode only)
4938                 if ( $mode === 'replace' ) {
4939                         # Output the replacement text
4940                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4941                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4942                         # Only add trailing whitespace if there is newText
4943                         if ( $newText != "" ) {
4944                                 $outText .= $newText . "\n\n";
4945                         }
4946
4947                         while ( $node ) {
4948                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4949                                 $node = $node->getNextSibling();
4950                         }
4951                 }
4952
4953                 if ( is_string( $outText ) ) {
4954                         # Re-insert stripped tags
4955                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4956                 }
4957
4958                 return $outText;
4959         }
4960
4961         /**
4962          * This function returns the text of a section, specified by a number ($section).
4963          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4964          * the first section before any such heading (section 0).
4965          *
4966          * If a section contains subsections, these are also returned.
4967          *
4968          * @param $text String: text to look in
4969          * @param $section String: section identifier
4970          * @param $deftext String: default to return if section is not found
4971          * @return string text of the requested section
4972          */
4973         public function getSection( $text, $section, $deftext='' ) {
4974                 return $this->extractSections( $text, $section, "get", $deftext );
4975         }
4976
4977         /**
4978          * This function returns $oldtext after the content of the section
4979          * specified by $section has been replaced with $text.
4980          *
4981          * @param $text String: former text of the article
4982          * @param $section Numeric: section identifier
4983          * @param $text String: replacing text
4984          * #return String: modified text
4985          */
4986         public function replaceSection( $oldtext, $section, $text ) {
4987                 return $this->extractSections( $oldtext, $section, "replace", $text );
4988         }
4989
4990         /**
4991          * Get the ID of the revision we are parsing
4992          *
4993          * @return Mixed: integer or null
4994          */
4995         function getRevisionId() {
4996                 return $this->mRevisionId;
4997         }
4998
4999         /**
5000          * Get the revision object for $this->mRevisionId
5001          *
5002          * @return either a Revision object or null
5003          */
5004         protected function getRevisionObject() {
5005                 if ( !is_null( $this->mRevisionObject ) ) {
5006                         return $this->mRevisionObject;
5007                 }
5008                 if ( is_null( $this->mRevisionId ) ) {
5009                         return null;
5010                 }
5011
5012                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5013                 return $this->mRevisionObject;
5014         }
5015
5016         /**
5017          * Get the timestamp associated with the current revision, adjusted for
5018          * the default server-local timestamp
5019          */
5020         function getRevisionTimestamp() {
5021                 if ( is_null( $this->mRevisionTimestamp ) ) {
5022                         wfProfileIn( __METHOD__ );
5023
5024                         $revObject = $this->getRevisionObject();
5025                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5026
5027                         if( $timestamp !== false ) {
5028                                 global $wgContLang;
5029
5030                                 # The cryptic '' timezone parameter tells to use the site-default
5031                                 # timezone offset instead of the user settings.
5032                                 #
5033                                 # Since this value will be saved into the parser cache, served
5034                                 # to other users, and potentially even used inside links and such,
5035                                 # it needs to be consistent for all visitors.
5036                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5037                         }
5038
5039                         wfProfileOut( __METHOD__ );
5040                 }
5041                 return $this->mRevisionTimestamp;
5042         }
5043
5044         /**
5045          * Get the name of the user that edited the last revision
5046          *
5047          * @return String: user name
5048          */
5049         function getRevisionUser() {
5050                 if( is_null( $this->mRevisionUser ) ) {
5051                         $revObject = $this->getRevisionObject();
5052
5053                         # if this template is subst: the revision id will be blank,
5054                         # so just use the current user's name
5055                         if( $revObject ) {
5056                                 $this->mRevisionUser = $revObject->getUserText();
5057                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5058                                 $this->mRevisionUser = $this->getUser()->getName();
5059                         }
5060                 }
5061                 return $this->mRevisionUser;
5062         }
5063
5064         /**
5065          * Mutator for $mDefaultSort
5066          *
5067          * @param $sort New value
5068          */
5069         public function setDefaultSort( $sort ) {
5070                 $this->mDefaultSort = $sort;
5071                 $this->mOutput->setProperty( 'defaultsort', $sort );
5072         }
5073
5074         /**
5075          * Accessor for $mDefaultSort
5076          * Will use the empty string if none is set.
5077          *
5078          * This value is treated as a prefix, so the
5079          * empty string is equivalent to sorting by
5080          * page name.
5081          *
5082          * @return string
5083          */
5084         public function getDefaultSort() {
5085                 if ( $this->mDefaultSort !== false ) {
5086                         return $this->mDefaultSort;
5087                 } else {
5088                         return '';
5089                 }
5090         }
5091
5092         /**
5093          * Accessor for $mDefaultSort
5094          * Unlike getDefaultSort(), will return false if none is set
5095          *
5096          * @return string or false
5097          */
5098         public function getCustomDefaultSort() {
5099                 return $this->mDefaultSort;
5100         }
5101
5102         /**
5103          * Try to guess the section anchor name based on a wikitext fragment
5104          * presumably extracted from a heading, for example "Header" from
5105          * "== Header ==".
5106          */
5107         public function guessSectionNameFromWikiText( $text ) {
5108                 # Strip out wikitext links(they break the anchor)
5109                 $text = $this->stripSectionName( $text );
5110                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5111                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5112         }
5113
5114         /**
5115          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5116          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5117          * as something other than UTF-8 (apparently?), resulting in breakage.
5118          *
5119          * @param $text String: The section name
5120          * @return string An anchor
5121          */
5122         public function guessLegacySectionNameFromWikiText( $text ) {
5123                 # Strip out wikitext links(they break the anchor)
5124                 $text = $this->stripSectionName( $text );
5125                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5126                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5127         }
5128
5129         /**
5130          * Strips a text string of wikitext for use in a section anchor
5131          *
5132          * Accepts a text string and then removes all wikitext from the
5133          * string and leaves only the resultant text (i.e. the result of
5134          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5135          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5136          * to create valid section anchors by mimicing the output of the
5137          * parser when headings are parsed.
5138          *
5139          * @param $text String: text string to be stripped of wikitext
5140          * for use in a Section anchor
5141          * @return Filtered text string
5142          */
5143         public function stripSectionName( $text ) {
5144                 # Strip internal link markup
5145                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5146                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5147
5148                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5149                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5150                 # on how many empty links there are on the page - need to figure that out.
5151                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5152
5153                 # Parse wikitext quotes (italics & bold)
5154                 $text = $this->doQuotes( $text );
5155
5156                 # Strip HTML tags
5157                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5158                 return $text;
5159         }
5160
5161         /**
5162          * strip/replaceVariables/unstrip for preprocessor regression testing
5163          */
5164         function testSrvus( $text, $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5165                 if ( !$title instanceof Title ) {
5166                         $title = Title::newFromText( $title );
5167                 }
5168                 $this->startExternalParse( $title, $options, $outputType, true );
5169
5170                 $text = $this->replaceVariables( $text );
5171                 $text = $this->mStripState->unstripBoth( $text );
5172                 $text = Sanitizer::removeHTMLtags( $text );
5173                 return $text;
5174         }
5175
5176         function testPst( $text, $title, $options ) {
5177                 global $wgUser;
5178                 if ( !$title instanceof Title ) {
5179                         $title = Title::newFromText( $title );
5180                 }
5181                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5182         }
5183
5184         function testPreprocess( $text, $title, $options ) {
5185                 if ( !$title instanceof Title ) {
5186                         $title = Title::newFromText( $title );
5187                 }
5188                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5189         }
5190
5191         function markerSkipCallback( $s, $callback ) {
5192                 $i = 0;
5193                 $out = '';
5194                 while ( $i < strlen( $s ) ) {
5195                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5196                         if ( $markerStart === false ) {
5197                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5198                                 break;
5199                         } else {
5200                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5201                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5202                                 if ( $markerEnd === false ) {
5203                                         $out .= substr( $s, $markerStart );
5204                                         break;
5205                                 } else {
5206                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5207                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5208                                         $i = $markerEnd;
5209                                 }
5210                         }
5211                 }
5212                 return $out;
5213         }
5214
5215         function serialiseHalfParsedText( $text ) {
5216                 $data = array();
5217                 $data['text'] = $text;
5218
5219                 # First, find all strip markers, and store their
5220                 #  data in an array.
5221                 $stripState = new StripState;
5222                 $pos = 0;
5223                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5224                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5225                 {
5226                         $end_pos += strlen( self::MARKER_SUFFIX );
5227                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5228
5229                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5230                                 $replaceArray = $stripState->general;
5231                                 $stripText = $this->mStripState->general->data[$marker];
5232                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5233                                 $replaceArray = $stripState->nowiki;
5234                                 $stripText = $this->mStripState->nowiki->data[$marker];
5235                         } else {
5236                                 throw new MWException( "Hanging strip marker: '$marker'." );
5237                         }
5238
5239                         $replaceArray->setPair( $marker, $stripText );
5240                         $pos = $end_pos;
5241                 }
5242                 $data['stripstate'] = $stripState;
5243
5244                 # Now, find all of our links, and store THEIR
5245                 #  data in an array! :)
5246                 $links = array( 'internal' => array(), 'interwiki' => array() );
5247                 $pos = 0;
5248
5249                 # Internal links
5250                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5251                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5252
5253                         $ns = trim( $ns );
5254                         if ( empty( $links['internal'][$ns] ) ) {
5255                                 $links['internal'][$ns] = array();
5256                         }
5257
5258                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5259                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5260                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5261                 }
5262
5263                 $pos = 0;
5264
5265                 # Interwiki links
5266                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5267                         $data = substr( $text, $start_pos );
5268                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5269                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5270                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5271                 }
5272
5273                 $data['linkholder'] = $links;
5274
5275                 return $data;
5276         }
5277
5278         /**
5279          * TODO: document
5280          * @param $data Array
5281          * @param $intPrefix String unique identifying prefix
5282          * @return String
5283          */
5284         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5285                 if ( !$intPrefix ) {
5286                         $intPrefix = self::getRandomString();
5287                 }
5288
5289                 # First, extract the strip state.
5290                 $stripState = $data['stripstate'];
5291                 $this->mStripState->general->merge( $stripState->general );
5292                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5293
5294                 # Now, extract the text, and renumber links
5295                 $text = $data['text'];
5296                 $links = $data['linkholder'];
5297
5298                 # Internal...
5299                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5300                         foreach ( $nsLinks as $key => $entry ) {
5301                                 $newKey = $intPrefix . '-' . $key;
5302                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5303
5304                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5305                         }
5306                 }
5307
5308                 # Interwiki...
5309                 foreach ( $links['interwiki'] as $key => $entry ) {
5310                         $newKey = "$intPrefix-$key";
5311                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5312
5313                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5314                 }
5315
5316                 # Should be good to go.
5317                 return $text;
5318         }
5319 }
5320
5321 /**
5322  * @todo document, briefly.
5323  * @ingroup Parser
5324  */
5325 class StripState {
5326         var $general, $nowiki;
5327
5328         function __construct() {
5329                 $this->general = new ReplacementArray;
5330                 $this->nowiki = new ReplacementArray;
5331         }
5332
5333         function unstripGeneral( $text ) {
5334                 wfProfileIn( __METHOD__ );
5335                 do {
5336                         $oldText = $text;
5337                         $text = $this->general->replace( $text );
5338                 } while ( $text !== $oldText );
5339                 wfProfileOut( __METHOD__ );
5340                 return $text;
5341         }
5342
5343         function unstripNoWiki( $text ) {
5344                 wfProfileIn( __METHOD__ );
5345                 do {
5346                         $oldText = $text;
5347                         $text = $this->nowiki->replace( $text );
5348                 } while ( $text !== $oldText );
5349                 wfProfileOut( __METHOD__ );
5350                 return $text;
5351         }
5352
5353         function unstripBoth( $text ) {
5354                 wfProfileIn( __METHOD__ );
5355                 do {
5356                         $oldText = $text;
5357                         $text = $this->general->replace( $text );
5358                         $text = $this->nowiki->replace( $text );
5359                 } while ( $text !== $oldText );
5360                 wfProfileOut( __METHOD__ );
5361                 return $text;
5362         }
5363 }
5364
5365 /**
5366  * @todo document, briefly.
5367  * @ingroup Parser
5368  */
5369 class OnlyIncludeReplacer {
5370         var $output = '';
5371
5372         function replace( $matches ) {
5373                 if ( substr( $matches[1], -1 ) === "\n" ) {
5374                         $this->output .= substr( $matches[1], 0, -1 );
5375                 } else {
5376                         $this->output .= $matches[1];
5377                 }
5378         }
5379 }