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  *    object: $wgContLang
  36  *
  37  * NOT $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
  38  *
  39  * settings:
  40  *  $wgUseDynamicDates*, $wgInterwikiMagic*,
  41  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  42  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  43  *  $wgMaxArticleSize*
  44  *
  45  *  * only within ParserOptions
  46  * </pre>
  47  *
  48  * @ingroup Parser
  49  */
  50 class Parser {
  51         /**
  52          * Update this version number when the ParserOutput format
  53          * changes in an incompatible way, so the parser cache
  54          * can automatically discard old data.
  55          */
  56         const VERSION = '1.6.4';
  57
  58         /**
  59          * Update this version number when the output of serialiseHalfParsedText()
  60          * changes in an incompatible way
  61          */
  62         const HALF_PARSED_VERSION = 2;
  63
  64         # Flags for Parser::setFunctionHook
  65         # Also available as global constants from Defines.php
  66         const SFH_NO_HASH = 1;
  67         const SFH_OBJECT_ARGS = 2;
  68
  69         # Constants needed for external link processing
  70         # Everything except bracket, space, or control characters
  71         # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
  72         # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
  73         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
  74         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
  75                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
  76
  77         # State constants for the definition list colon extraction
  78         const COLON_STATE_TEXT = 0;
  79         const COLON_STATE_TAG = 1;
  80         const COLON_STATE_TAGSTART = 2;
  81         const COLON_STATE_CLOSETAG = 3;
  82         const COLON_STATE_TAGSLASH = 4;
  83         const COLON_STATE_COMMENT = 5;
  84         const COLON_STATE_COMMENTDASH = 6;
  85         const COLON_STATE_COMMENTDASHDASH = 7;
  86
  87         # Flags for preprocessToDom
  88         const PTD_FOR_INCLUSION = 1;
  89
  90         # Allowed values for $this->mOutputType
  91         # Parameter to startExternalParse().
  92         const OT_HTML = 1; # like parse()
  93         const OT_WIKI = 2; # like preSaveTransform()
  94         const OT_PREPROCESS = 3; # like preprocess()
  95         const OT_MSG = 3;
  96         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  97
  98         # Marker Suffix needs to be accessible staticly.
  99         const MARKER_SUFFIX = "-QINU\x7f";
 100
 101         # Persistent:
 102         var $mTagHooks = array();
 103         var $mTransparentTagHooks = array();
 104         var $mFunctionHooks = array();
 105         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
 106         var $mFunctionTagHooks = array();
 107         var $mStripList  = array();
 108         var $mDefaultStripList  = array();
 109         var $mVarCache = array();
 110         var $mImageParams = array();
 111         var $mImageParamsMagicArray = array();
 112         var $mMarkerIndex = 0;
 113         var $mFirstCall = true;
 114
 115         # Initialised by initialiseVariables()
 116
 117         /**
 118          * @var MagicWordArray
 119          */
 120         var $mVariables;
 121
 122         /**
 123          * @var MagicWordArray
 124          */
 125         var $mSubstWords;
 126         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 127
 128         # Cleared with clearState():
 129         /**
 130          * @var ParserOutput
 131          */
 132         var $mOutput;
 133         var $mAutonumber, $mDTopen;
 134
 135         /**
 136          * @var StripState
 137          */
 138         var $mStripState;
 139
 140         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 141         /**
 142          * @var LinkHolderArray
 143          */
 144         var $mLinkHolders;
 145
 146         var $mLinkID;
 147         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 148         var $mTplExpandCache; # empty-frame expansion cache
 149         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 150         var $mExpensiveFunctionCount; # number of expensive parser function calls
 151         var $mShowToc, $mForceTocPosition;
 152
 153         /**
 154          * @var User
 155          */
 156         var $mUser; # User object; only used when doing pre-save transform
 157
 158         # Temporary
 159         # These are variables reset at least once per parse regardless of $clearState
 160
 161         /**
 162          * @var ParserOptions
 163          */
 164         var $mOptions;
 165
 166         /**
 167          * @var Title
 168          */
 169         var $mTitle;        # Title context, used for self-link rendering and similar things
 170         var $mOutputType;   # Output type, one of the OT_xxx constants
 171         var $ot;            # Shortcut alias, see setOutputType()
 172         var $mRevisionObject; # The revision object of the specified revision ID
 173         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 174         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 175         var $mRevisionUser; # User to display in {{REVISIONUSER}} tag
 176         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 177
 178         /**
 179          * @var string
 180          */
 181         var $mUniqPrefix;
 182
 183         /**
 184          * Constructor
 185          *
 186          * @param $conf array
 187          */
 188         public function __construct( $conf = array() ) {
 189                 $this->mConf = $conf;
 190                 $this->mUrlProtocols = wfUrlProtocols();
 191                 $this->mExtLinkBracketedRegex = '/\[((' . wfUrlProtocols() . ')'.
 192                         self::EXT_LINK_URL_CLASS.'+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
 193                 if ( isset( $conf['preprocessorClass'] ) ) {
 194                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 195                 } elseif ( defined( 'MW_COMPILED' ) ) {
 196                         # Preprocessor_Hash is much faster than Preprocessor_DOM in compiled mode
 197                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 198                 } elseif ( extension_loaded( 'domxml' ) ) {
 199                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 200                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 201                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 202                 } elseif ( extension_loaded( 'dom' ) ) {
 203                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 204                 } else {
 205                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 206                 }
 207                 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
 208         }
 209
 210         /**
 211          * Reduce memory usage to reduce the impact of circular references
 212          */
 213         function __destruct() {
 214                 if ( isset( $this->mLinkHolders ) ) {
 215                         unset( $this->mLinkHolders );
 216                 }
 217                 foreach ( $this as $name => $value ) {
 218                         unset( $this->$name );
 219                 }
 220         }
 221
 222         /**
 223          * Do various kinds of initialisation on the first call of the parser
 224          */
 225         function firstCallInit() {
 226                 if ( !$this->mFirstCall ) {
 227                         return;
 228                 }
 229                 $this->mFirstCall = false;
 230
 231                 wfProfileIn( __METHOD__ );
 232
 233                 CoreParserFunctions::register( $this );
 234                 CoreTagHooks::register( $this );
 235                 $this->initialiseVariables();
 236
 237                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 238                 wfProfileOut( __METHOD__ );
 239         }
 240
 241         /**
 242          * Clear Parser state
 243          *
 244          * @private
 245          */
 246         function clearState() {
 247                 wfProfileIn( __METHOD__ );
 248                 if ( $this->mFirstCall ) {
 249                         $this->firstCallInit();
 250                 }
 251                 $this->mOutput = new ParserOutput;
 252                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 253                 $this->mAutonumber = 0;
 254                 $this->mLastSection = '';
 255                 $this->mDTopen = false;
 256                 $this->mIncludeCount = array();
 257                 $this->mArgStack = false;
 258                 $this->mInPre = false;
 259                 $this->mLinkHolders = new LinkHolderArray( $this );
 260                 $this->mLinkID = 0;
 261                 $this->mRevisionObject = $this->mRevisionTimestamp =
 262                         $this->mRevisionId = $this->mRevisionUser = null;
 263                 $this->mVarCache = array();
 264                 $this->mUser = null;
 265
 266                 /**
 267                  * Prefix for temporary replacement strings for the multipass parser.
 268                  * \x07 should never appear in input as it's disallowed in XML.
 269                  * Using it at the front also gives us a little extra robustness
 270                  * since it shouldn't match when butted up against identifier-like
 271                  * string constructs.
 272                  *
 273                  * Must not consist of all title characters, or else it will change
 274                  * the behaviour of <nowiki> in a link.
 275                  */
 276                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 277                 # Changed to \x7f to allow XML double-parsing -- TS
 278                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 279                 $this->mStripState = new StripState( $this->mUniqPrefix );
 280
 281
 282                 # Clear these on every parse, bug 4549
 283                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 284
 285                 $this->mShowToc = true;
 286                 $this->mForceTocPosition = false;
 287                 $this->mIncludeSizes = array(
 288                         'post-expand' => 0,
 289                         'arg' => 0,
 290                 );
 291                 $this->mPPNodeCount = 0;
 292                 $this->mDefaultSort = false;
 293                 $this->mHeadings = array();
 294                 $this->mDoubleUnderscores = array();
 295                 $this->mExpensiveFunctionCount = 0;
 296
 297                 # Fix cloning
 298                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 299                         $this->mPreprocessor = null;
 300                 }
 301
 302                 wfRunHooks( 'ParserClearState', array( &$this ) );
 303                 wfProfileOut( __METHOD__ );
 304         }
 305
 306         /**
 307          * Convert wikitext to HTML
 308          * Do not call this function recursively.
 309          *
 310          * @param $text String: text we want to parse
 311          * @param $title Title object
 312          * @param $options ParserOptions
 313          * @param $linestart boolean
 314          * @param $clearState boolean
 315          * @param $revid Int: number to pass in {{REVISIONID}}
 316          * @return ParserOutput a ParserOutput
 317          */
 318         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 319                 /**
 320                  * First pass--just handle <nowiki> sections, pass the rest off
 321                  * to internalParse() which does all the real work.
 322                  */
 323
 324                 global $wgUseTidy, $wgAlwaysUseTidy, $wgDisableLangConversion, $wgDisableTitleConversion;
 325                 $fname = __METHOD__.'-' . wfGetCaller();
 326                 wfProfileIn( __METHOD__ );
 327                 wfProfileIn( $fname );
 328
 329                 $this->startParse( $title, $options, self::OT_HTML, $clearState );
 330
 331                 $oldRevisionId = $this->mRevisionId;
 332                 $oldRevisionObject = $this->mRevisionObject;
 333                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 334                 $oldRevisionUser = $this->mRevisionUser;
 335                 if ( $revid !== null ) {
 336                         $this->mRevisionId = $revid;
 337                         $this->mRevisionObject = null;
 338                         $this->mRevisionTimestamp = null;
 339                         $this->mRevisionUser = null;
 340                 }
 341
 342                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 343                 # No more strip!
 344                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 345                 $text = $this->internalParse( $text );
 346
 347                 $text = $this->mStripState->unstripGeneral( $text );
 348
 349                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 350                 $fixtags = array(
 351                         # french spaces, last one Guillemet-left
 352                         # only if there is something before the space
 353                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
 354                         # french spaces, Guillemet-right
 355                         '/(\\302\\253) /' => '\\1&#160;',
 356                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 357                 );
 358                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 359
 360                 $text = $this->doBlockLevels( $text, $linestart );
 361
 362                 $this->replaceLinkHolders( $text );
 363
 364                 /**
 365                  * The input doesn't get language converted if
 366                  * a) It's disabled
 367                  * b) Content isn't converted
 368                  * c) It's a conversion table
 369                  * d) it is an interface message (which is in the user language)
 370                  */
 371                 if ( !( $wgDisableLangConversion
 372                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 373                                 || $this->mTitle->isConversionTable()
 374                                 || $this->mOptions->getInterfaceMessage() ) ) {
 375
 376                         # The position of the convert() call should not be changed. it
 377                         # assumes that the links are all replaced and the only thing left
 378                         # is the <nowiki> mark.
 379                         $text = $this->getFunctionLang()->convert( $text );
 380                 }
 381
 382                 /**
 383                  * A converted title will be provided in the output object if title and
 384                  * content conversion are enabled, the article text does not contain
 385                  * a conversion-suppressing double-underscore tag, and no
 386                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 387                  * automatic link conversion.
 388                  */
 389                 if ( !( $wgDisableLangConversion
 390                                 || $wgDisableTitleConversion
 391                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 392                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 393                                 || $this->mOutput->getDisplayTitle() !== false ) )
 394                 {
 395                         $convruletitle = $this->getFunctionLang()->getConvRuleTitle();
 396                         if ( $convruletitle ) {
 397                                 $this->mOutput->setTitleText( $convruletitle );
 398                         } else {
 399                                 $titleText = $this->getFunctionLang()->convertTitle( $title );
 400                                 $this->mOutput->setTitleText( $titleText );
 401                         }
 402                 }
 403
 404                 $text = $this->mStripState->unstripNoWiki( $text );
 405
 406                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 407
 408                 $text = $this->replaceTransparentTags( $text );
 409                 $text = $this->mStripState->unstripGeneral( $text );
 410
 411                 $text = Sanitizer::normalizeCharReferences( $text );
 412
 413                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 414                         $text = MWTidy::tidy( $text );
 415                 } else {
 416                         # attempt to sanitize at least some nesting problems
 417                         # (bug #2702 and quite a few others)
 418                         $tidyregs = array(
 419                                 # ''Something [http://www.cool.com cool''] -->
 420                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 421                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 422                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 423                                 # fix up an anchor inside another anchor, only
 424                                 # at least for a single single nested link (bug 3695)
 425                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 426                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 427                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 428                                 # contains a div, so fix it up here; replace
 429                                 # div with escaped text
 430                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 431                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 432                                 # remove empty italic or bold tag pairs, some
 433                                 # introduced by rules above
 434                                 '/<([bi])><\/\\1>/' => '',
 435                         );
 436
 437                         $text = preg_replace(
 438                                 array_keys( $tidyregs ),
 439                                 array_values( $tidyregs ),
 440                                 $text );
 441                 }
 442                 global $wgExpensiveParserFunctionLimit;
 443                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 444                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 445                 }
 446
 447                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 448
 449                 # Information on include size limits, for the benefit of users who try to skirt them
 450                 if ( $this->mOptions->getEnableLimitReport() ) {
 451                         $max = $this->mOptions->getMaxIncludeSize();
 452                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 453                         $limitReport =
 454                                 "NewPP limit report\n" .
 455                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 456                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 457                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 458                                 $PFreport;
 459                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 460                         $text .= "\n<!-- \n$limitReport-->\n";
 461                 }
 462                 $this->mOutput->setText( $text );
 463
 464                 $this->mRevisionId = $oldRevisionId;
 465                 $this->mRevisionObject = $oldRevisionObject;
 466                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 467                 $this->mRevisionUser = $oldRevisionUser;
 468                 wfProfileOut( $fname );
 469                 wfProfileOut( __METHOD__ );
 470
 471                 return $this->mOutput;
 472         }
 473
 474         /**
 475          * Recursive parser entry point that can be called from an extension tag
 476          * hook.
 477          *
 478          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 479          *
 480          * @param $text String: text extension wants to have parsed
 481          * @param $frame PPFrame: The frame to use for expanding any template variables
 482          *
 483          * @return string
 484          */
 485         function recursiveTagParse( $text, $frame=false ) {
 486                 wfProfileIn( __METHOD__ );
 487                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 488                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 489                 $text = $this->internalParse( $text, false, $frame );
 490                 wfProfileOut( __METHOD__ );
 491                 return $text;
 492         }
 493
 494         /**
 495          * Expand templates and variables in the text, producing valid, static wikitext.
 496          * Also removes comments.
 497          */
 498         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 499                 wfProfileIn( __METHOD__ );
 500                 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
 501                 if ( $revid !== null ) {
 502                         $this->mRevisionId = $revid;
 503                 }
 504                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 505                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 506                 $text = $this->replaceVariables( $text );
 507                 $text = $this->mStripState->unstripBoth( $text );
 508                 wfProfileOut( __METHOD__ );
 509                 return $text;
 510         }
 511
 512         /**
 513          * Recursive parser entry point that can be called from an extension tag
 514          * hook.
 515          *
 516          * @param $text String: text to be expanded
 517          * @param $frame PPFrame: The frame to use for expanding any template variables
 518          * @return String
 519          */
 520         public function recursivePreprocess( $text, $frame = false ) {
 521                 wfProfileIn( __METHOD__ );
 522                 $text = $this->replaceVariables( $text, $frame );
 523                 $text = $this->mStripState->unstripBoth( $text );
 524                 wfProfileOut( __METHOD__ );
 525                 return $text;
 526         }
 527
 528         /**
 529          * Process the wikitext for the ?preload= feature. (bug 5210)
 530          *
 531          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 532          * comments, templates, arguments, tags hooks and parser functions are untouched.
 533          *
 534          * @param $text String
 535          * @param $title Title
 536          * @param $options ParserOptions
 537          * @return String
 538          */
 539         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 540                 # Parser (re)initialisation
 541                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 542
 543                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 544                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 545                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 546                 $text = $this->mStripState->unstripBoth( $text );
 547                 return $text;
 548         }
 549
 550         /**
 551          * Get a random string
 552          *
 553          * @return string
 554          */
 555         static public function getRandomString() {
 556                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 557         }
 558
 559         /**
 560          * Set the current user.
 561          * Should only be used when doing pre-save transform.
 562          *
 563          * @param $user Mixed: User object or null (to reset)
 564          */
 565         function setUser( $user ) {
 566                 $this->mUser = $user;
 567         }
 568
 569         /**
 570          * Accessor for mUniqPrefix.
 571          *
 572          * @return String
 573          */
 574         public function uniqPrefix() {
 575                 if ( !isset( $this->mUniqPrefix ) ) {
 576                         # @todo FIXME: This is probably *horribly wrong*
 577                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 578                         # if this is called for a parser cache hit, the parser may not
 579                         # have ever been initialized in the first place.
 580                         # Not really sure what the heck is supposed to be going on here.
 581                         return '';
 582                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 583                 }
 584                 return $this->mUniqPrefix;
 585         }
 586
 587         /**
 588          * Set the context title
 589          *
 590          * @param $t Title
 591          */
 592         function setTitle( $t ) {
 593                 if ( !$t || $t instanceof FakeTitle ) {
 594                         $t = Title::newFromText( 'NO TITLE' );
 595                 }
 596
 597                 if ( strval( $t->getFragment() ) !== '' ) {
 598                         # Strip the fragment to avoid various odd effects
 599                         $this->mTitle = clone $t;
 600                         $this->mTitle->setFragment( '' );
 601                 } else {
 602                         $this->mTitle = $t;
 603                 }
 604         }
 605
 606         /**
 607          * Accessor for the Title object
 608          *
 609          * @return Title object
 610          */
 611         function getTitle() {
 612                 return $this->mTitle;
 613         }
 614
 615         /**
 616          * Accessor/mutator for the Title object
 617          *
 618          * @param $x New Title object or null to just get the current one
 619          * @return Title object
 620          */
 621         function Title( $x = null ) {
 622                 return wfSetVar( $this->mTitle, $x );
 623         }
 624
 625         /**
 626          * Set the output type
 627          *
 628          * @param $ot Integer: new value
 629          */
 630         function setOutputType( $ot ) {
 631                 $this->mOutputType = $ot;
 632                 # Shortcut alias
 633                 $this->ot = array(
 634                         'html' => $ot == self::OT_HTML,
 635                         'wiki' => $ot == self::OT_WIKI,
 636                         'pre' => $ot == self::OT_PREPROCESS,
 637                         'plain' => $ot == self::OT_PLAIN,
 638                 );
 639         }
 640
 641         /**
 642          * Accessor/mutator for the output type
 643          *
 644          * @param $x New value or null to just get the current one
 645          * @return Integer
 646          */
 647         function OutputType( $x = null ) {
 648                 return wfSetVar( $this->mOutputType, $x );
 649         }
 650
 651         /**
 652          * Get the ParserOutput object
 653          *
 654          * @return ParserOutput object
 655          */
 656         function getOutput() {
 657                 return $this->mOutput;
 658         }
 659
 660         /**
 661          * Get the ParserOptions object
 662          *
 663          * @return ParserOptions object
 664          */
 665         function getOptions() {
 666                 return $this->mOptions;
 667         }
 668
 669         /**
 670          * Accessor/mutator for the ParserOptions object
 671          *
 672          * @param $x New value or null to just get the current one
 673          * @return Current ParserOptions object
 674          */
 675         function Options( $x = null ) {
 676                 return wfSetVar( $this->mOptions, $x );
 677         }
 678
 679         /**
 680          * @return int
 681          */
 682         function nextLinkID() {
 683                 return $this->mLinkID++;
 684         }
 685
 686         /**
 687          * @param $id int
 688          */
 689         function setLinkID( $id ) {
 690                 $this->mLinkID = $id;
 691         }
 692
 693         /**
 694          * @return Language
 695          */
 696         function getFunctionLang() {
 697                 $target = $this->mOptions->getTargetLanguage();
 698                 if ( $target !== null ) {
 699                         return $target;
 700                 } elseif( $this->mOptions->getInterfaceMessage() ) {
 701                         return $this->mOptions->getUserLangObj();
 702                 } elseif( is_null( $this->mTitle ) ) {
 703                         throw new MWException( __METHOD__.': $this->mTitle is null' );
 704                 }
 705                 return $this->mTitle->getPageLanguage();
 706         }
 707
 708         /**
 709          * Get a User object either from $this->mUser, if set, or from the
 710          * ParserOptions object otherwise
 711          *
 712          * @return User object
 713          */
 714         function getUser() {
 715                 if ( !is_null( $this->mUser ) ) {
 716                         return $this->mUser;
 717                 }
 718                 return $this->mOptions->getUser();
 719         }
 720
 721         /**
 722          * Get a preprocessor object
 723          *
 724          * @return Preprocessor instance
 725          */
 726         function getPreprocessor() {
 727                 if ( !isset( $this->mPreprocessor ) ) {
 728                         $class = $this->mPreprocessorClass;
 729                         $this->mPreprocessor = new $class( $this );
 730                 }
 731                 return $this->mPreprocessor;
 732         }
 733
 734         /**
 735          * Replaces all occurrences of HTML-style comments and the given tags
 736          * in the text with a random marker and returns the next text. The output
 737          * parameter $matches will be an associative array filled with data in
 738          * the form:
 739          *   'UNIQ-xxxxx' => array(
 740          *     'element',
 741          *     'tag content',
 742          *     array( 'param' => 'x' ),
 743          *     '<element param="x">tag content</element>' ) )
 744          *
 745          * @param $elements array list of element names. Comments are always extracted.
 746          * @param $text string Source text string.
 747          * @param $matches array Out parameter, Array: extracted tags
 748          * @param $uniq_prefix string
 749          * @return String: stripped text
 750          */
 751         public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 752                 static $n = 1;
 753                 $stripped = '';
 754                 $matches = array();
 755
 756                 $taglist = implode( '|', $elements );
 757                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 758
 759                 while ( $text != '' ) {
 760                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 761                         $stripped .= $p[0];
 762                         if ( count( $p ) < 5 ) {
 763                                 break;
 764                         }
 765                         if ( count( $p ) > 5 ) {
 766                                 # comment
 767                                 $element    = $p[4];
 768                                 $attributes = '';
 769                                 $close      = '';
 770                                 $inside     = $p[5];
 771                         } else {
 772                                 # tag
 773                                 $element    = $p[1];
 774                                 $attributes = $p[2];
 775                                 $close      = $p[3];
 776                                 $inside     = $p[4];
 777                         }
 778
 779                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 780                         $stripped .= $marker;
 781
 782                         if ( $close === '/>' ) {
 783                                 # Empty element tag, <tag />
 784                                 $content = null;
 785                                 $text = $inside;
 786                                 $tail = null;
 787                         } else {
 788                                 if ( $element === '!--' ) {
 789                                         $end = '/(-->)/';
 790                                 } else {
 791                                         $end = "/(<\\/$element\\s*>)/i";
 792                                 }
 793                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 794                                 $content = $q[0];
 795                                 if ( count( $q ) < 3 ) {
 796                                         # No end tag -- let it run out to the end of the text.
 797                                         $tail = '';
 798                                         $text = '';
 799                                 } else {
 800                                         $tail = $q[1];
 801                                         $text = $q[2];
 802                                 }
 803                         }
 804
 805                         $matches[$marker] = array( $element,
 806                                 $content,
 807                                 Sanitizer::decodeTagAttributes( $attributes ),
 808                                 "<$element$attributes$close$content$tail" );
 809                 }
 810                 return $stripped;
 811         }
 812
 813         /**
 814          * Get a list of strippable XML-like elements
 815          *
 816          * @return array
 817          */
 818         function getStripList() {
 819                 return $this->mStripList;
 820         }
 821
 822         /**
 823          * Add an item to the strip state
 824          * Returns the unique tag which must be inserted into the stripped text
 825          * The tag will be replaced with the original text in unstrip()
 826          *
 827          * @param $text string
 828          *
 829          * @return string
 830          */
 831         function insertStripItem( $text ) {
 832                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 833                 $this->mMarkerIndex++;
 834                 $this->mStripState->addGeneral( $rnd, $text );
 835                 return $rnd;
 836         }
 837
 838         /**
 839          * parse the wiki syntax used to render tables
 840          *
 841          * @private
 842          */
 843         function doTableStuff( $text ) {
 844                 wfProfileIn( __METHOD__ );
 845
 846                 $lines = StringUtils::explode( "\n", $text );
 847                 $out = '';
 848                 $td_history = array(); # Is currently a td tag open?
 849                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 850                 $tr_history = array(); # Is currently a tr tag open?
 851                 $tr_attributes = array(); # history of tr attributes
 852                 $has_opened_tr = array(); # Did this table open a <tr> element?
 853                 $indent_level = 0; # indent level of the table
 854
 855                 foreach ( $lines as $outLine ) {
 856                         $line = trim( $outLine );
 857
 858                         if ( $line === '' ) { # empty line, go to next line
 859                                 $out .= $outLine."\n";
 860                                 continue;
 861                         }
 862
 863                         $first_character = $line[0];
 864                         $matches = array();
 865
 866                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 867                                 # First check if we are starting a new table
 868                                 $indent_level = strlen( $matches[1] );
 869
 870                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 871                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 872
 873                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 874                                 array_push( $td_history , false );
 875                                 array_push( $last_tag_history , '' );
 876                                 array_push( $tr_history , false );
 877                                 array_push( $tr_attributes , '' );
 878                                 array_push( $has_opened_tr , false );
 879                         } elseif ( count( $td_history ) == 0 ) {
 880                                 # Don't do any of the following
 881                                 $out .= $outLine."\n";
 882                                 continue;
 883                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 884                                 # We are ending a table
 885                                 $line = '</table>' . substr( $line , 2 );
 886                                 $last_tag = array_pop( $last_tag_history );
 887
 888                                 if ( !array_pop( $has_opened_tr ) ) {
 889                                         $line = "<tr><td></td></tr>{$line}";
 890                                 }
 891
 892                                 if ( array_pop( $tr_history ) ) {
 893                                         $line = "</tr>{$line}";
 894                                 }
 895
 896                                 if ( array_pop( $td_history ) ) {
 897                                         $line = "</{$last_tag}>{$line}";
 898                                 }
 899                                 array_pop( $tr_attributes );
 900                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 901                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 902                                 # Now we have a table row
 903                                 $line = preg_replace( '#^\|-+#', '', $line );
 904
 905                                 # Whats after the tag is now only attributes
 906                                 $attributes = $this->mStripState->unstripBoth( $line );
 907                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 908                                 array_pop( $tr_attributes );
 909                                 array_push( $tr_attributes, $attributes );
 910
 911                                 $line = '';
 912                                 $last_tag = array_pop( $last_tag_history );
 913                                 array_pop( $has_opened_tr );
 914                                 array_push( $has_opened_tr , true );
 915
 916                                 if ( array_pop( $tr_history ) ) {
 917                                         $line = '</tr>';
 918                                 }
 919
 920                                 if ( array_pop( $td_history ) ) {
 921                                         $line = "</{$last_tag}>{$line}";
 922                                 }
 923
 924                                 $outLine = $line;
 925                                 array_push( $tr_history , false );
 926                                 array_push( $td_history , false );
 927                                 array_push( $last_tag_history , '' );
 928                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 929                                 # This might be cell elements, td, th or captions
 930                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 931                                         $first_character = '+';
 932                                         $line = substr( $line , 1 );
 933                                 }
 934
 935                                 $line = substr( $line , 1 );
 936
 937                                 if ( $first_character === '!' ) {
 938                                         $line = str_replace( '!!' , '||' , $line );
 939                                 }
 940
 941                                 # Split up multiple cells on the same line.
 942                                 # FIXME : This can result in improper nesting of tags processed
 943                                 # by earlier parser steps, but should avoid splitting up eg
 944                                 # attribute values containing literal "||".
 945                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 946
 947                                 $outLine = '';
 948
 949                                 # Loop through each table cell
 950                                 foreach ( $cells as $cell ) {
 951                                         $previous = '';
 952                                         if ( $first_character !== '+' ) {
 953                                                 $tr_after = array_pop( $tr_attributes );
 954                                                 if ( !array_pop( $tr_history ) ) {
 955                                                         $previous = "<tr{$tr_after}>\n";
 956                                                 }
 957                                                 array_push( $tr_history , true );
 958                                                 array_push( $tr_attributes , '' );
 959                                                 array_pop( $has_opened_tr );
 960                                                 array_push( $has_opened_tr , true );
 961                                         }
 962
 963                                         $last_tag = array_pop( $last_tag_history );
 964
 965                                         if ( array_pop( $td_history ) ) {
 966                                                 $previous = "</{$last_tag}>\n{$previous}";
 967                                         }
 968
 969                                         if ( $first_character === '|' ) {
 970                                                 $last_tag = 'td';
 971                                         } elseif ( $first_character === '!' ) {
 972                                                 $last_tag = 'th';
 973                                         } elseif ( $first_character === '+' ) {
 974                                                 $last_tag = 'caption';
 975                                         } else {
 976                                                 $last_tag = '';
 977                                         }
 978
 979                                         array_push( $last_tag_history , $last_tag );
 980
 981                                         # A cell could contain both parameters and data
 982                                         $cell_data = explode( '|' , $cell , 2 );
 983
 984                                         # Bug 553: Note that a '|' inside an invalid link should not
 985                                         # be mistaken as delimiting cell parameters
 986                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 987                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 988                                         } elseif ( count( $cell_data ) == 1 ) {
 989                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 990                                         } else {
 991                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 992                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 993                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 994                                         }
 995
 996                                         $outLine .= $cell;
 997                                         array_push( $td_history , true );
 998                                 }
 999                         }
1000                         $out .= $outLine . "\n";
1001                 }
1002
1003                 # Closing open td, tr && table
1004                 while ( count( $td_history ) > 0 ) {
1005                         if ( array_pop( $td_history ) ) {
1006                                 $out .= "</td>\n";
1007                         }
1008                         if ( array_pop( $tr_history ) ) {
1009                                 $out .= "</tr>\n";
1010                         }
1011                         if ( !array_pop( $has_opened_tr ) ) {
1012                                 $out .= "<tr><td></td></tr>\n" ;
1013                         }
1014
1015                         $out .= "</table>\n";
1016                 }
1017
1018                 # Remove trailing line-ending (b/c)
1019                 if ( substr( $out, -1 ) === "\n" ) {
1020                         $out = substr( $out, 0, -1 );
1021                 }
1022
1023                 # special case: don't return empty table
1024                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1025                         $out = '';
1026                 }
1027
1028                 wfProfileOut( __METHOD__ );
1029
1030                 return $out;
1031         }
1032
1033         /**
1034          * Helper function for parse() that transforms wiki markup into
1035          * HTML. Only called for $mOutputType == self::OT_HTML.
1036          *
1037          * @private
1038          *
1039          * @param $text string
1040          * @param $isMain bool
1041          * @param $frame bool
1042          *
1043          * @return string
1044          */
1045         function internalParse( $text, $isMain = true, $frame = false ) {
1046                 wfProfileIn( __METHOD__ );
1047
1048                 $origText = $text;
1049
1050                 # Hook to suspend the parser in this state
1051                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1052                         wfProfileOut( __METHOD__ );
1053                         return $text ;
1054                 }
1055
1056                 # if $frame is provided, then use $frame for replacing any variables
1057                 if ( $frame ) {
1058                         # use frame depth to infer how include/noinclude tags should be handled
1059                         # depth=0 means this is the top-level document; otherwise it's an included document
1060                         if ( !$frame->depth ) {
1061                                 $flag = 0;
1062                         } else {
1063                                 $flag = Parser::PTD_FOR_INCLUSION;
1064                         }
1065                         $dom = $this->preprocessToDom( $text, $flag );
1066                         $text = $frame->expand( $dom );
1067                 } else {
1068                         # if $frame is not provided, then use old-style replaceVariables
1069                         $text = $this->replaceVariables( $text );
1070                 }
1071
1072                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1073                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1074
1075                 # Tables need to come after variable replacement for things to work
1076                 # properly; putting them before other transformations should keep
1077                 # exciting things like link expansions from showing up in surprising
1078                 # places.
1079                 $text = $this->doTableStuff( $text );
1080
1081                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1082
1083                 $text = $this->doDoubleUnderscore( $text );
1084
1085                 $text = $this->doHeadings( $text );
1086                 if ( $this->mOptions->getUseDynamicDates() ) {
1087                         $df = DateFormatter::getInstance();
1088                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1089                 }
1090                 $text = $this->replaceInternalLinks( $text );
1091                 $text = $this->doAllQuotes( $text );
1092                 $text = $this->replaceExternalLinks( $text );
1093
1094                 # replaceInternalLinks may sometimes leave behind
1095                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1096                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1097
1098                 $text = $this->doMagicLinks( $text );
1099                 $text = $this->formatHeadings( $text, $origText, $isMain );
1100
1101                 wfProfileOut( __METHOD__ );
1102                 return $text;
1103         }
1104
1105         /**
1106          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1107          * magic external links.
1108          *
1109          * DML
1110          * @private
1111          *
1112          * @param $text string
1113          *
1114          * @return string
1115          */
1116         function doMagicLinks( $text ) {
1117                 wfProfileIn( __METHOD__ );
1118                 $prots = wfUrlProtocolsWithoutProtRel();
1119                 $urlChar = self::EXT_LINK_URL_CLASS;
1120                 $text = preg_replace_callback(
1121                         '!(?:                           # Start cases
1122                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1123                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1124                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1125                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1126                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1127                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1128                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1129                                         [0-9Xx]                 # check digit
1130                                         \b)
1131                         )!xu', array( &$this, 'magicLinkCallback' ), $text );
1132                 wfProfileOut( __METHOD__ );
1133                 return $text;
1134         }
1135
1136         /**
1137          * @throws MWException
1138          * @param $m array
1139          * @return HTML|string
1140          */
1141         function magicLinkCallback( $m ) {
1142                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1143                         # Skip anchor
1144                         return $m[0];
1145                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1146                         # Skip HTML element
1147                         return $m[0];
1148                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1149                         # Free external link
1150                         return $this->makeFreeExternalLink( $m[0] );
1151                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1152                         # RFC or PMID
1153                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1154                                 $keyword = 'RFC';
1155                                 $urlmsg = 'rfcurl';
1156                                 $CssClass = 'mw-magiclink-rfc';
1157                                 $id = $m[4];
1158                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1159                                 $keyword = 'PMID';
1160                                 $urlmsg = 'pubmedurl';
1161                                 $CssClass = 'mw-magiclink-pmid';
1162                                 $id = $m[4];
1163                         } else {
1164                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1165                                         substr( $m[0], 0, 20 ) . '"' );
1166                         }
1167                         $url = wfMsgForContent( $urlmsg, $id );
1168                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1169                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1170                         # ISBN
1171                         $isbn = $m[5];
1172                         $num = strtr( $isbn, array(
1173                                 '-' => '',
1174                                 ' ' => '',
1175                                 'x' => 'X',
1176                         ));
1177                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1178                         return'<a href="' .
1179                                 $titleObj->escapeLocalUrl() .
1180                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1181                 } else {
1182                         return $m[0];
1183                 }
1184         }
1185
1186         /**
1187          * Make a free external link, given a user-supplied URL
1188          *
1189          * @param $url string
1190          *
1191          * @return string HTML
1192          * @private
1193          */
1194         function makeFreeExternalLink( $url ) {
1195                 wfProfileIn( __METHOD__ );
1196
1197                 $trail = '';
1198
1199                 # The characters '<' and '>' (which were escaped by
1200                 # removeHTMLtags()) should not be included in
1201                 # URLs, per RFC 2396.
1202                 $m2 = array();
1203                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1204                         $trail = substr( $url, $m2[0][1] ) . $trail;
1205                         $url = substr( $url, 0, $m2[0][1] );
1206                 }
1207
1208                 # Move trailing punctuation to $trail
1209                 $sep = ',;\.:!?';
1210                 # If there is no left bracket, then consider right brackets fair game too
1211                 if ( strpos( $url, '(' ) === false ) {
1212                         $sep .= ')';
1213                 }
1214
1215                 $numSepChars = strspn( strrev( $url ), $sep );
1216                 if ( $numSepChars ) {
1217                         $trail = substr( $url, -$numSepChars ) . $trail;
1218                         $url = substr( $url, 0, -$numSepChars );
1219                 }
1220
1221                 $url = Sanitizer::cleanUrl( $url );
1222
1223                 # Is this an external image?
1224                 $text = $this->maybeMakeExternalImage( $url );
1225                 if ( $text === false ) {
1226                         # Not an image, make a link
1227                         $text = Linker::makeExternalLink( $url, $this->getFunctionLang()->markNoConversion($url), true, 'free',
1228                                 $this->getExternalLinkAttribs( $url ) );
1229                         # Register it in the output object...
1230                         # Replace unnecessary URL escape codes with their equivalent characters
1231                         $pasteurized = self::replaceUnusualEscapes( $url );
1232                         $this->mOutput->addExternalLink( $pasteurized );
1233                 }
1234                 wfProfileOut( __METHOD__ );
1235                 return $text . $trail;
1236         }
1237
1238
1239         /**
1240          * Parse headers and return html
1241          *
1242          * @private
1243          *
1244          * @param $text string
1245          *
1246          * @return string
1247          */
1248         function doHeadings( $text ) {
1249                 wfProfileIn( __METHOD__ );
1250                 for ( $i = 6; $i >= 1; --$i ) {
1251                         $h = str_repeat( '=', $i );
1252                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1253                           "<h$i>\\1</h$i>", $text );
1254                 }
1255                 wfProfileOut( __METHOD__ );
1256                 return $text;
1257         }
1258
1259         /**
1260          * Replace single quotes with HTML markup
1261          * @private
1262          *
1263          * @param $text string
1264          *
1265          * @return string the altered text
1266          */
1267         function doAllQuotes( $text ) {
1268                 wfProfileIn( __METHOD__ );
1269                 $outtext = '';
1270                 $lines = StringUtils::explode( "\n", $text );
1271                 foreach ( $lines as $line ) {
1272                         $outtext .= $this->doQuotes( $line ) . "\n";
1273                 }
1274                 $outtext = substr( $outtext, 0,-1 );
1275                 wfProfileOut( __METHOD__ );
1276                 return $outtext;
1277         }
1278
1279         /**
1280          * Helper function for doAllQuotes()
1281          *
1282          * @param $text string
1283          *
1284          * @return string
1285          */
1286         public function doQuotes( $text ) {
1287                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1288                 if ( count( $arr ) == 1 ) {
1289                         return $text;
1290                 } else {
1291                         # First, do some preliminary work. This may shift some apostrophes from
1292                         # being mark-up to being text. It also counts the number of occurrences
1293                         # of bold and italics mark-ups.
1294                         $numbold = 0;
1295                         $numitalics = 0;
1296                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1297                                 if ( ( $i % 2 ) == 1 ) {
1298                                         # If there are ever four apostrophes, assume the first is supposed to
1299                                         # be text, and the remaining three constitute mark-up for bold text.
1300                                         if ( strlen( $arr[$i] ) == 4 ) {
1301                                                 $arr[$i-1] .= "'";
1302                                                 $arr[$i] = "'''";
1303                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1304                                                 # If there are more than 5 apostrophes in a row, assume they're all
1305                                                 # text except for the last 5.
1306                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1307                                                 $arr[$i] = "'''''";
1308                                         }
1309                                         # Count the number of occurrences of bold and italics mark-ups.
1310                                         # We are not counting sequences of five apostrophes.
1311                                         if ( strlen( $arr[$i] ) == 2 ) {
1312                                                 $numitalics++;
1313                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1314                                                 $numbold++;
1315                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1316                                                 $numitalics++;
1317                                                 $numbold++;
1318                                         }
1319                                 }
1320                         }
1321
1322                         # If there is an odd number of both bold and italics, it is likely
1323                         # that one of the bold ones was meant to be an apostrophe followed
1324                         # by italics. Which one we cannot know for certain, but it is more
1325                         # likely to be one that has a single-letter word before it.
1326                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1327                                 $i = 0;
1328                                 $firstsingleletterword = -1;
1329                                 $firstmultiletterword = -1;
1330                                 $firstspace = -1;
1331                                 foreach ( $arr as $r ) {
1332                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1333                                                 $x1 = substr( $arr[$i-1], -1 );
1334                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1335                                                 if ( $x1 === ' ' ) {
1336                                                         if ( $firstspace == -1 ) {
1337                                                                 $firstspace = $i;
1338                                                         }
1339                                                 } elseif ( $x2 === ' ') {
1340                                                         if ( $firstsingleletterword == -1 ) {
1341                                                                 $firstsingleletterword = $i;
1342                                                         }
1343                                                 } else {
1344                                                         if ( $firstmultiletterword == -1 ) {
1345                                                                 $firstmultiletterword = $i;
1346                                                         }
1347                                                 }
1348                                         }
1349                                         $i++;
1350                                 }
1351
1352                                 # If there is a single-letter word, use it!
1353                                 if ( $firstsingleletterword > -1 ) {
1354                                         $arr[$firstsingleletterword] = "''";
1355                                         $arr[$firstsingleletterword-1] .= "'";
1356                                 } elseif ( $firstmultiletterword > -1 ) {
1357                                         # If not, but there's a multi-letter word, use that one.
1358                                         $arr[$firstmultiletterword] = "''";
1359                                         $arr[$firstmultiletterword-1] .= "'";
1360                                 } elseif ( $firstspace > -1 ) {
1361                                         # ... otherwise use the first one that has neither.
1362                                         # (notice that it is possible for all three to be -1 if, for example,
1363                                         # there is only one pentuple-apostrophe in the line)
1364                                         $arr[$firstspace] = "''";
1365                                         $arr[$firstspace-1] .= "'";
1366                                 }
1367                         }
1368
1369                         # Now let's actually convert our apostrophic mush to HTML!
1370                         $output = '';
1371                         $buffer = '';
1372                         $state = '';
1373                         $i = 0;
1374                         foreach ( $arr as $r ) {
1375                                 if ( ( $i % 2 ) == 0 ) {
1376                                         if ( $state === 'both' ) {
1377                                                 $buffer .= $r;
1378                                         } else {
1379                                                 $output .= $r;
1380                                         }
1381                                 } else {
1382                                         if ( strlen( $r ) == 2 ) {
1383                                                 if ( $state === 'i' ) {
1384                                                         $output .= '</i>'; $state = '';
1385                                                 } elseif ( $state === 'bi' ) {
1386                                                         $output .= '</i>'; $state = 'b';
1387                                                 } elseif ( $state === 'ib' ) {
1388                                                         $output .= '</b></i><b>'; $state = 'b';
1389                                                 } elseif ( $state === 'both' ) {
1390                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1391                                                 } else { # $state can be 'b' or ''
1392                                                         $output .= '<i>'; $state .= 'i';
1393                                                 }
1394                                         } elseif ( strlen( $r ) == 3 ) {
1395                                                 if ( $state === 'b' ) {
1396                                                         $output .= '</b>'; $state = '';
1397                                                 } elseif ( $state === 'bi' ) {
1398                                                         $output .= '</i></b><i>'; $state = 'i';
1399                                                 } elseif ( $state === 'ib' ) {
1400                                                         $output .= '</b>'; $state = 'i';
1401                                                 } elseif ( $state === 'both' ) {
1402                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1403                                                 } else { # $state can be 'i' or ''
1404                                                         $output .= '<b>'; $state .= 'b';
1405                                                 }
1406                                         } elseif ( strlen( $r ) == 5 ) {
1407                                                 if ( $state === 'b' ) {
1408                                                         $output .= '</b><i>'; $state = 'i';
1409                                                 } elseif ( $state === 'i' ) {
1410                                                         $output .= '</i><b>'; $state = 'b';
1411                                                 } elseif ( $state === 'bi' ) {
1412                                                         $output .= '</i></b>'; $state = '';
1413                                                 } elseif ( $state === 'ib' ) {
1414                                                         $output .= '</b></i>'; $state = '';
1415                                                 } elseif ( $state === 'both' ) {
1416                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1417                                                 } else { # ($state == '')
1418                                                         $buffer = ''; $state = 'both';
1419                                                 }
1420                                         }
1421                                 }
1422                                 $i++;
1423                         }
1424                         # Now close all remaining tags.  Notice that the order is important.
1425                         if ( $state === 'b' || $state === 'ib' ) {
1426                                 $output .= '</b>';
1427                         }
1428                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1429                                 $output .= '</i>';
1430                         }
1431                         if ( $state === 'bi' ) {
1432                                 $output .= '</b>';
1433                         }
1434                         # There might be lonely ''''', so make sure we have a buffer
1435                         if ( $state === 'both' && $buffer ) {
1436                                 $output .= '<b><i>'.$buffer.'</i></b>';
1437                         }
1438                         return $output;
1439                 }
1440         }
1441
1442         /**
1443          * Replace external links (REL)
1444          *
1445          * Note: this is all very hackish and the order of execution matters a lot.
1446          * Make sure to run maintenance/parserTests.php if you change this code.
1447          *
1448          * @private
1449          *
1450          * @param $text string
1451          *
1452          * @return string
1453          */
1454         function replaceExternalLinks( $text ) {
1455                 wfProfileIn( __METHOD__ );
1456
1457                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1458                 $s = array_shift( $bits );
1459
1460                 $i = 0;
1461                 while ( $i<count( $bits ) ) {
1462                         $url = $bits[$i++];
1463                         $protocol = $bits[$i++];
1464                         $text = $bits[$i++];
1465                         $trail = $bits[$i++];
1466
1467                         # The characters '<' and '>' (which were escaped by
1468                         # removeHTMLtags()) should not be included in
1469                         # URLs, per RFC 2396.
1470                         $m2 = array();
1471                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1472                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1473                                 $url = substr( $url, 0, $m2[0][1] );
1474                         }
1475
1476                         # If the link text is an image URL, replace it with an <img> tag
1477                         # This happened by accident in the original parser, but some people used it extensively
1478                         $img = $this->maybeMakeExternalImage( $text );
1479                         if ( $img !== false ) {
1480                                 $text = $img;
1481                         }
1482
1483                         $dtrail = '';
1484
1485                         # Set linktype for CSS - if URL==text, link is essentially free
1486                         $linktype = ( $text === $url ) ? 'free' : 'text';
1487
1488                         # No link text, e.g. [http://domain.tld/some.link]
1489                         if ( $text == '' ) {
1490                                 # Autonumber
1491                                 $langObj = $this->getFunctionLang();
1492                                 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1493                                 $linktype = 'autonumber';
1494                         } else {
1495                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1496                                 # Check for trail
1497                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1498                         }
1499
1500                         $text = $this->getFunctionLang()->markNoConversion( $text );
1501
1502                         $url = Sanitizer::cleanUrl( $url );
1503
1504                         # Use the encoded URL
1505                         # This means that users can paste URLs directly into the text
1506                         # Funny characters like ö aren't valid in URLs anyway
1507                         # This was changed in August 2004
1508                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1509                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1510
1511                         # Register link in the output object.
1512                         # Replace unnecessary URL escape codes with the referenced character
1513                         # This prevents spammers from hiding links from the filters
1514                         $pasteurized = self::replaceUnusualEscapes( $url );
1515                         $this->mOutput->addExternalLink( $pasteurized );
1516                 }
1517
1518                 wfProfileOut( __METHOD__ );
1519                 return $s;
1520         }
1521
1522         /**
1523          * Get an associative array of additional HTML attributes appropriate for a
1524          * particular external link.  This currently may include rel => nofollow
1525          * (depending on configuration, namespace, and the URL's domain) and/or a
1526          * target attribute (depending on configuration).
1527          *
1528          * @param $url String|bool optional URL, to extract the domain from for rel =>
1529          *   nofollow if appropriate
1530          * @return Array associative array of HTML attributes
1531          */
1532         function getExternalLinkAttribs( $url = false ) {
1533                 $attribs = array();
1534                 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1535                 $ns = $this->mTitle->getNamespace();
1536                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) &&
1537                                 !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions ) )
1538                 {
1539                         $attribs['rel'] = 'nofollow';
1540                 }
1541                 if ( $this->mOptions->getExternalLinkTarget() ) {
1542                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1543                 }
1544                 return $attribs;
1545         }
1546
1547         /**
1548          * Replace unusual URL escape codes with their equivalent characters
1549          *
1550          * @param $url String
1551          * @return String
1552          *
1553          * @todo  This can merge genuinely required bits in the path or query string,
1554          *        breaking legit URLs. A proper fix would treat the various parts of
1555          *        the URL differently; as a workaround, just use the output for
1556          *        statistical records, not for actual linking/output.
1557          */
1558         static function replaceUnusualEscapes( $url ) {
1559                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1560                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1561         }
1562
1563         /**
1564          * Callback function used in replaceUnusualEscapes().
1565          * Replaces unusual URL escape codes with their equivalent character
1566          *
1567          * @param $matches array
1568          *
1569          * @return string
1570          */
1571         private static function replaceUnusualEscapesCallback( $matches ) {
1572                 $char = urldecode( $matches[0] );
1573                 $ord = ord( $char );
1574                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1575                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1576                         # No, shouldn't be escaped
1577                         return $char;
1578                 } else {
1579                         # Yes, leave it escaped
1580                         return $matches[0];
1581                 }
1582         }
1583
1584         /**
1585          * make an image if it's allowed, either through the global
1586          * option, through the exception, or through the on-wiki whitelist
1587          * @private
1588          *
1589          * $param $url string
1590          *
1591          * @return string
1592          */
1593         function maybeMakeExternalImage( $url ) {
1594                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1595                 $imagesexception = !empty( $imagesfrom );
1596                 $text = false;
1597                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1598                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1599                         $imagematch = false;
1600                         foreach ( $imagesfrom as $match ) {
1601                                 if ( strpos( $url, $match ) === 0 ) {
1602                                         $imagematch = true;
1603                                         break;
1604                                 }
1605                         }
1606                 } elseif ( $imagesexception ) {
1607                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1608                 } else {
1609                         $imagematch = false;
1610                 }
1611                 if ( $this->mOptions->getAllowExternalImages()
1612                          || ( $imagesexception && $imagematch ) ) {
1613                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1614                                 # Image found
1615                                 $text = Linker::makeExternalImage( $url );
1616                         }
1617                 }
1618                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1619                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1620                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1621                         foreach ( $whitelist as $entry ) {
1622                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1623                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1624                                         continue;
1625                                 }
1626                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1627                                         # Image matches a whitelist entry
1628                                         $text = Linker::makeExternalImage( $url );
1629                                         break;
1630                                 }
1631                         }
1632                 }
1633                 return $text;
1634         }
1635
1636         /**
1637          * Process [[ ]] wikilinks
1638          *
1639          * @param $s string
1640          *
1641          * @return String: processed text
1642          *
1643          * @private
1644          */
1645         function replaceInternalLinks( $s ) {
1646                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1647                 return $s;
1648         }
1649
1650         /**
1651          * Process [[ ]] wikilinks (RIL)
1652          * @return LinkHolderArray
1653          *
1654          * @private
1655          */
1656         function replaceInternalLinks2( &$s ) {
1657                 wfProfileIn( __METHOD__ );
1658
1659                 wfProfileIn( __METHOD__.'-setup' );
1660                 static $tc = FALSE, $e1, $e1_img;
1661                 # the % is needed to support urlencoded titles as well
1662                 if ( !$tc ) {
1663                         $tc = Title::legalChars() . '#%';
1664                         # Match a link having the form [[namespace:link|alternate]]trail
1665                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1666                         # Match cases where there is no "]]", which might still be images
1667                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1668                 }
1669
1670                 $holders = new LinkHolderArray( $this );
1671
1672                 # split the entire text string on occurences of [[
1673                 $a = StringUtils::explode( '[[', ' ' . $s );
1674                 # get the first element (all text up to first [[), and remove the space we added
1675                 $s = $a->current();
1676                 $a->next();
1677                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1678                 $s = substr( $s, 1 );
1679
1680                 $useLinkPrefixExtension = $this->getFunctionLang()->linkPrefixExtension();
1681                 $e2 = null;
1682                 if ( $useLinkPrefixExtension ) {
1683                         # Match the end of a line for a word that's not followed by whitespace,
1684                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1685                         $e2 = wfMsgForContent( 'linkprefix' );
1686                 }
1687
1688                 if ( is_null( $this->mTitle ) ) {
1689                         wfProfileOut( __METHOD__.'-setup' );
1690                         wfProfileOut( __METHOD__ );
1691                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1692                 }
1693                 $nottalk = !$this->mTitle->isTalkPage();
1694
1695                 if ( $useLinkPrefixExtension ) {
1696                         $m = array();
1697                         if ( preg_match( $e2, $s, $m ) ) {
1698                                 $first_prefix = $m[2];
1699                         } else {
1700                                 $first_prefix = false;
1701                         }
1702                 } else {
1703                         $prefix = '';
1704                 }
1705
1706                 if ( $this->getFunctionLang()->hasVariants() ) {
1707                         $selflink = $this->getFunctionLang()->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1708                 } else {
1709                         $selflink = array( $this->mTitle->getPrefixedText() );
1710                 }
1711                 $useSubpages = $this->areSubpagesAllowed();
1712                 wfProfileOut( __METHOD__.'-setup' );
1713
1714                 # Loop for each link
1715                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1716                         # Check for excessive memory usage
1717                         if ( $holders->isBig() ) {
1718                                 # Too big
1719                                 # Do the existence check, replace the link holders and clear the array
1720                                 $holders->replace( $s );
1721                                 $holders->clear();
1722                         }
1723
1724                         if ( $useLinkPrefixExtension ) {
1725                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1726                                 if ( preg_match( $e2, $s, $m ) ) {
1727                                         $prefix = $m[2];
1728                                         $s = $m[1];
1729                                 } else {
1730                                         $prefix='';
1731                                 }
1732                                 # first link
1733                                 if ( $first_prefix ) {
1734                                         $prefix = $first_prefix;
1735                                         $first_prefix = false;
1736                                 }
1737                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1738                         }
1739
1740                         $might_be_img = false;
1741
1742                         wfProfileIn( __METHOD__."-e1" );
1743                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1744                                 $text = $m[2];
1745                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1746                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1747                                 # the real problem is with the $e1 regex
1748                                 # See bug 1300.
1749                                 #
1750                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1751                                 # and no image is in sight. See bug 2095.
1752                                 #
1753                                 if ( $text !== '' &&
1754                                         substr( $m[3], 0, 1 ) === ']' &&
1755                                         strpos( $text, '[' ) !== false
1756                                 )
1757                                 {
1758                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1759                                         $m[3] = substr( $m[3], 1 );
1760                                 }
1761                                 # fix up urlencoded title texts
1762                                 if ( strpos( $m[1], '%' ) !== false ) {
1763                                         # Should anchors '#' also be rejected?
1764                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1765                                 }
1766                                 $trail = $m[3];
1767                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1768                                 $might_be_img = true;
1769                                 $text = $m[2];
1770                                 if ( strpos( $m[1], '%' ) !== false ) {
1771                                         $m[1] = rawurldecode( $m[1] );
1772                                 }
1773                                 $trail = "";
1774                         } else { # Invalid form; output directly
1775                                 $s .= $prefix . '[[' . $line ;
1776                                 wfProfileOut( __METHOD__."-e1" );
1777                                 continue;
1778                         }
1779                         wfProfileOut( __METHOD__."-e1" );
1780                         wfProfileIn( __METHOD__."-misc" );
1781
1782                         # Don't allow internal links to pages containing
1783                         # PROTO: where PROTO is a valid URL protocol; these
1784                         # should be external links.
1785                         if ( preg_match( '/^(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1786                                 $s .= $prefix . '[[' . $line ;
1787                                 wfProfileOut( __METHOD__."-misc" );
1788                                 continue;
1789                         }
1790
1791                         # Make subpage if necessary
1792                         if ( $useSubpages ) {
1793                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1794                         } else {
1795                                 $link = $m[1];
1796                         }
1797
1798                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1799                         if ( !$noforce ) {
1800                                 # Strip off leading ':'
1801                                 $link = substr( $link, 1 );
1802                         }
1803
1804                         wfProfileOut( __METHOD__."-misc" );
1805                         wfProfileIn( __METHOD__."-title" );
1806                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1807                         if ( $nt === null ) {
1808                                 $s .= $prefix . '[[' . $line;
1809                                 wfProfileOut( __METHOD__."-title" );
1810                                 continue;
1811                         }
1812
1813                         $ns = $nt->getNamespace();
1814                         $iw = $nt->getInterWiki();
1815                         wfProfileOut( __METHOD__."-title" );
1816
1817                         if ( $might_be_img ) { # if this is actually an invalid link
1818                                 wfProfileIn( __METHOD__."-might_be_img" );
1819                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1820                                         $found = false;
1821                                         while ( true ) {
1822                                                 # look at the next 'line' to see if we can close it there
1823                                                 $a->next();
1824                                                 $next_line = $a->current();
1825                                                 if ( $next_line === false || $next_line === null ) {
1826                                                         break;
1827                                                 }
1828                                                 $m = explode( ']]', $next_line, 3 );
1829                                                 if ( count( $m ) == 3 ) {
1830                                                         # the first ]] closes the inner link, the second the image
1831                                                         $found = true;
1832                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1833                                                         $trail = $m[2];
1834                                                         break;
1835                                                 } elseif ( count( $m ) == 2 ) {
1836                                                         # if there's exactly one ]] that's fine, we'll keep looking
1837                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1838                                                 } else {
1839                                                         # if $next_line is invalid too, we need look no further
1840                                                         $text .= '[[' . $next_line;
1841                                                         break;
1842                                                 }
1843                                         }
1844                                         if ( !$found ) {
1845                                                 # we couldn't find the end of this imageLink, so output it raw
1846                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1847                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1848                                                 $s .= "{$prefix}[[$link|$text";
1849                                                 # note: no $trail, because without an end, there *is* no trail
1850                                                 wfProfileOut( __METHOD__."-might_be_img" );
1851                                                 continue;
1852                                         }
1853                                 } else { # it's not an image, so output it raw
1854                                         $s .= "{$prefix}[[$link|$text";
1855                                         # note: no $trail, because without an end, there *is* no trail
1856                                         wfProfileOut( __METHOD__."-might_be_img" );
1857                                         continue;
1858                                 }
1859                                 wfProfileOut( __METHOD__."-might_be_img" );
1860                         }
1861
1862                         $wasblank = ( $text  == '' );
1863                         if ( $wasblank ) {
1864                                 $text = $link;
1865                         } else {
1866                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1867                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1868                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1869                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1870                                 $text = $this->doQuotes( $text );
1871                         }
1872
1873                         # Link not escaped by : , create the various objects
1874                         if ( $noforce ) {
1875                                 global $wgContLang;
1876
1877                                 # Interwikis
1878                                 wfProfileIn( __METHOD__."-interwiki" );
1879                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1880                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1881                                         $s = rtrim( $s . $prefix );
1882                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1883                                         wfProfileOut( __METHOD__."-interwiki" );
1884                                         continue;
1885                                 }
1886                                 wfProfileOut( __METHOD__."-interwiki" );
1887
1888                                 if ( $ns == NS_FILE ) {
1889                                         wfProfileIn( __METHOD__."-image" );
1890                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1891                                                 if ( $wasblank ) {
1892                                                         # if no parameters were passed, $text
1893                                                         # becomes something like "File:Foo.png",
1894                                                         # which we don't want to pass on to the
1895                                                         # image generator
1896                                                         $text = '';
1897                                                 } else {
1898                                                         # recursively parse links inside the image caption
1899                                                         # actually, this will parse them in any other parameters, too,
1900                                                         # but it might be hard to fix that, and it doesn't matter ATM
1901                                                         $text = $this->replaceExternalLinks( $text );
1902                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1903                                                 }
1904                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1905                                                 $s .= $prefix . $this->armorLinks(
1906                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
1907                                         } else {
1908                                                 $s .= $prefix . $trail;
1909                                         }
1910                                         wfProfileOut( __METHOD__."-image" );
1911                                         continue;
1912                                 }
1913
1914                                 if ( $ns == NS_CATEGORY ) {
1915                                         wfProfileIn( __METHOD__."-category" );
1916                                         if( $this->getTitle()->isCssOrJsPage() ) {
1917                                                 # bug 32450 : js and script pages in MediaWiki: namespace do not want
1918                                                 # to get their code or comments altered. Think about js string:
1919                                                 # var foobar = "[[Category:" + $catname + "]];
1920                                                 $s .= "[[$text]]$trail";
1921                                                 wfProfileOut( __METHOD__."-category" );
1922                                                 continue;
1923                                         }
1924                                         $s = rtrim( $s . "\n" ); # bug 87
1925
1926                                         if ( $wasblank ) {
1927                                                 $sortkey = $this->getDefaultSort();
1928                                         } else {
1929                                                 $sortkey = $text;
1930                                         }
1931                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1932                                         $sortkey = str_replace( "\n", '', $sortkey );
1933                                         $sortkey = $this->getFunctionLang()->convertCategoryKey( $sortkey );
1934                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1935
1936                                         /**
1937                                          * Strip the whitespace Category links produce, see bug 87
1938                                          * @todo We might want to use trim($tmp, "\n") here.
1939                                          */
1940                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1941
1942                                         wfProfileOut( __METHOD__."-category" );
1943                                         continue;
1944                                 }
1945                         }
1946
1947                         # Self-link checking
1948                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1949                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1950                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
1951                                         continue;
1952                                 }
1953                         }
1954
1955                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1956                         # @todo FIXME: Should do batch file existence checks, see comment below
1957                         if ( $ns == NS_MEDIA ) {
1958                                 wfProfileIn( __METHOD__."-media" );
1959                                 # Give extensions a chance to select the file revision for us
1960                                 $options = array();
1961                                 $descQuery = false;
1962                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
1963                                         array( $this, $nt, &$options, &$descQuery ) );
1964                                 # Fetch and register the file (file title may be different via hooks)
1965                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
1966                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1967                                 $s .= $prefix . $this->armorLinks(
1968                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
1969                                 wfProfileOut( __METHOD__."-media" );
1970                                 continue;
1971                         }
1972
1973                         wfProfileIn( __METHOD__."-always_known" );
1974                         # Some titles, such as valid special pages or files in foreign repos, should
1975                         # be shown as bluelinks even though they're not included in the page table
1976                         #
1977                         # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1978                         # batch file existence checks for NS_FILE and NS_MEDIA
1979                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1980                                 $this->mOutput->addLink( $nt );
1981                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
1982                         } else {
1983                                 # Links will be added to the output link list after checking
1984                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
1985                         }
1986                         wfProfileOut( __METHOD__."-always_known" );
1987                 }
1988                 wfProfileOut( __METHOD__ );
1989                 return $holders;
1990         }
1991
1992         /**
1993          * Render a forced-blue link inline; protect against double expansion of
1994          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1995          * Since this little disaster has to split off the trail text to avoid
1996          * breaking URLs in the following text without breaking trails on the
1997          * wiki links, it's been made into a horrible function.
1998          *
1999          * @param $nt Title
2000          * @param $text String
2001          * @param $query Array or String
2002          * @param $trail String
2003          * @param $prefix String
2004          * @return String: HTML-wikitext mix oh yuck
2005          */
2006         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2007                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2008
2009                 if ( is_string( $query ) ) {
2010                         $query = wfCgiToArray( $query );
2011                 }
2012                 if ( $text == '' ) {
2013                         $text = htmlspecialchars( $nt->getPrefixedText() );
2014                 }
2015
2016                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2017
2018                 return $this->armorLinks( $link ) . $trail;
2019         }
2020
2021         /**
2022          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2023          * going to go through further parsing steps before inline URL expansion.
2024          *
2025          * Not needed quite as much as it used to be since free links are a bit
2026          * more sensible these days. But bracketed links are still an issue.
2027          *
2028          * @param $text String: more-or-less HTML
2029          * @return String: less-or-more HTML with NOPARSE bits
2030          */
2031         function armorLinks( $text ) {
2032                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2033                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2034         }
2035
2036         /**
2037          * Return true if subpage links should be expanded on this page.
2038          * @return Boolean
2039          */
2040         function areSubpagesAllowed() {
2041                 # Some namespaces don't allow subpages
2042                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2043         }
2044
2045         /**
2046          * Handle link to subpage if necessary
2047          *
2048          * @param $target String: the source of the link
2049          * @param &$text String: the link text, modified as necessary
2050          * @return string the full name of the link
2051          * @private
2052          */
2053         function maybeDoSubpageLink( $target, &$text ) {
2054                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2055         }
2056
2057         /**#@+
2058          * Used by doBlockLevels()
2059          * @private
2060          *
2061          * @return string
2062          */
2063         function closeParagraph() {
2064                 $result = '';
2065                 if ( $this->mLastSection != '' ) {
2066                         $result = '</' . $this->mLastSection  . ">\n";
2067                 }
2068                 $this->mInPre = false;
2069                 $this->mLastSection = '';
2070                 return $result;
2071         }
2072
2073         /**
2074          * getCommon() returns the length of the longest common substring
2075          * of both arguments, starting at the beginning of both.
2076          * @private
2077          *
2078          * @param $st1 string
2079          * @param $st2 string
2080          *
2081          * @return int
2082          */
2083         function getCommon( $st1, $st2 ) {
2084                 $fl = strlen( $st1 );
2085                 $shorter = strlen( $st2 );
2086                 if ( $fl < $shorter ) {
2087                         $shorter = $fl;
2088                 }
2089
2090                 for ( $i = 0; $i < $shorter; ++$i ) {
2091                         if ( $st1[$i] != $st2[$i] ) {
2092                                 break;
2093                         }
2094                 }
2095                 return $i;
2096         }
2097
2098         /**
2099          * These next three functions open, continue, and close the list
2100          * element appropriate to the prefix character passed into them.
2101          * @private
2102          *
2103          * @param $char char
2104          *
2105          * @return string
2106          */
2107         function openList( $char ) {
2108                 $result = $this->closeParagraph();
2109
2110                 if ( '*' === $char ) {
2111                         $result .= '<ul><li>';
2112                 } elseif ( '#' === $char ) {
2113                         $result .= '<ol><li>';
2114                 } elseif ( ':' === $char ) {
2115                         $result .= '<dl><dd>';
2116                 } elseif ( ';' === $char ) {
2117                         $result .= '<dl><dt>';
2118                         $this->mDTopen = true;
2119                 } else {
2120                         $result = '<!-- ERR 1 -->';
2121                 }
2122
2123                 return $result;
2124         }
2125
2126         /**
2127          * TODO: document
2128          * @param $char String
2129          * @private
2130          *
2131          * @return string
2132          */
2133         function nextItem( $char ) {
2134                 if ( '*' === $char || '#' === $char ) {
2135                         return '</li><li>';
2136                 } elseif ( ':' === $char || ';' === $char ) {
2137                         $close = '</dd>';
2138                         if ( $this->mDTopen ) {
2139                                 $close = '</dt>';
2140                         }
2141                         if ( ';' === $char ) {
2142                                 $this->mDTopen = true;
2143                                 return $close . '<dt>';
2144                         } else {
2145                                 $this->mDTopen = false;
2146                                 return $close . '<dd>';
2147                         }
2148                 }
2149                 return '<!-- ERR 2 -->';
2150         }
2151
2152         /**
2153          * TODO: document
2154          * @param $char String
2155          * @private
2156          *
2157          * @return string
2158          */
2159         function closeList( $char ) {
2160                 if ( '*' === $char ) {
2161                         $text = '</li></ul>';
2162                 } elseif ( '#' === $char ) {
2163                         $text = '</li></ol>';
2164                 } elseif ( ':' === $char ) {
2165                         if ( $this->mDTopen ) {
2166                                 $this->mDTopen = false;
2167                                 $text = '</dt></dl>';
2168                         } else {
2169                                 $text = '</dd></dl>';
2170                         }
2171                 } else {
2172                         return '<!-- ERR 3 -->';
2173                 }
2174                 return $text."\n";
2175         }
2176         /**#@-*/
2177
2178         /**
2179          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2180          *
2181          * @param $text String
2182          * @param $linestart Boolean: whether or not this is at the start of a line.
2183          * @private
2184          * @return string the lists rendered as HTML
2185          */
2186         function doBlockLevels( $text, $linestart ) {
2187                 wfProfileIn( __METHOD__ );
2188
2189                 # Parsing through the text line by line.  The main thing
2190                 # happening here is handling of block-level elements p, pre,
2191                 # and making lists from lines starting with * # : etc.
2192                 #
2193                 $textLines = StringUtils::explode( "\n", $text );
2194
2195                 $lastPrefix = $output = '';
2196                 $this->mDTopen = $inBlockElem = false;
2197                 $prefixLength = 0;
2198                 $paragraphStack = false;
2199
2200                 foreach ( $textLines as $oLine ) {
2201                         # Fix up $linestart
2202                         if ( !$linestart ) {
2203                                 $output .= $oLine;
2204                                 $linestart = true;
2205                                 continue;
2206                         }
2207                         # * = ul
2208                         # # = ol
2209                         # ; = dt
2210                         # : = dd
2211
2212                         $lastPrefixLength = strlen( $lastPrefix );
2213                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2214                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2215                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2216                         if ( !$this->mInPre ) {
2217                                 # Multiple prefixes may abut each other for nested lists.
2218                                 $prefixLength = strspn( $oLine, '*#:;' );
2219                                 $prefix = substr( $oLine, 0, $prefixLength );
2220
2221                                 # eh?
2222                                 # ; and : are both from definition-lists, so they're equivalent
2223                                 #  for the purposes of determining whether or not we need to open/close
2224                                 #  elements.
2225                                 $prefix2 = str_replace( ';', ':', $prefix );
2226                                 $t = substr( $oLine, $prefixLength );
2227                                 $this->mInPre = (bool)$preOpenMatch;
2228                         } else {
2229                                 # Don't interpret any other prefixes in preformatted text
2230                                 $prefixLength = 0;
2231                                 $prefix = $prefix2 = '';
2232                                 $t = $oLine;
2233                         }
2234
2235                         # List generation
2236                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2237                                 # Same as the last item, so no need to deal with nesting or opening stuff
2238                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2239                                 $paragraphStack = false;
2240
2241                                 if ( substr( $prefix, -1 ) === ';') {
2242                                         # The one nasty exception: definition lists work like this:
2243                                         # ; title : definition text
2244                                         # So we check for : in the remainder text to split up the
2245                                         # title and definition, without b0rking links.
2246                                         $term = $t2 = '';
2247                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2248                                                 $t = $t2;
2249                                                 $output .= $term . $this->nextItem( ':' );
2250                                         }
2251                                 }
2252                         } elseif ( $prefixLength || $lastPrefixLength ) {
2253                                 # We need to open or close prefixes, or both.
2254
2255                                 # Either open or close a level...
2256                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2257                                 $paragraphStack = false;
2258
2259                                 # Close all the prefixes which aren't shared.
2260                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2261                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2262                                         --$lastPrefixLength;
2263                                 }
2264
2265                                 # Continue the current prefix if appropriate.
2266                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2267                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2268                                 }
2269
2270                                 # Open prefixes where appropriate.
2271                                 while ( $prefixLength > $commonPrefixLength ) {
2272                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2273                                         $output .= $this->openList( $char );
2274
2275                                         if ( ';' === $char ) {
2276                                                 # @todo FIXME: This is dupe of code above
2277                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2278                                                         $t = $t2;
2279                                                         $output .= $term . $this->nextItem( ':' );
2280                                                 }
2281                                         }
2282                                         ++$commonPrefixLength;
2283                                 }
2284                                 $lastPrefix = $prefix2;
2285                         }
2286
2287                         # If we have no prefixes, go to paragraph mode.
2288                         if ( 0 == $prefixLength ) {
2289                                 wfProfileIn( __METHOD__."-paragraph" );
2290                                 # No prefix (not in list)--go to paragraph mode
2291                                 # XXX: use a stack for nestable elements like span, table and div
2292                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2293                                 $closematch = preg_match(
2294                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2295                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2296                                 if ( $openmatch or $closematch ) {
2297                                         $paragraphStack = false;
2298                                         # TODO bug 5718: paragraph closed
2299                                         $output .= $this->closeParagraph();
2300                                         if ( $preOpenMatch and !$preCloseMatch ) {
2301                                                 $this->mInPre = true;
2302                                         }
2303                                         $inBlockElem = !$closematch;
2304                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2305                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2306                                                 # pre
2307                                                 if ( $this->mLastSection !== 'pre' ) {
2308                                                         $paragraphStack = false;
2309                                                         $output .= $this->closeParagraph().'<pre>';
2310                                                         $this->mLastSection = 'pre';
2311                                                 }
2312                                                 $t = substr( $t, 1 );
2313                                         } else {
2314                                                 # paragraph
2315                                                 if ( trim( $t ) === '' ) {
2316                                                         if ( $paragraphStack ) {
2317                                                                 $output .= $paragraphStack.'<br />';
2318                                                                 $paragraphStack = false;
2319                                                                 $this->mLastSection = 'p';
2320                                                         } else {
2321                                                                 if ( $this->mLastSection !== 'p' ) {
2322                                                                         $output .= $this->closeParagraph();
2323                                                                         $this->mLastSection = '';
2324                                                                         $paragraphStack = '<p>';
2325                                                                 } else {
2326                                                                         $paragraphStack = '</p><p>';
2327                                                                 }
2328                                                         }
2329                                                 } else {
2330                                                         if ( $paragraphStack ) {
2331                                                                 $output .= $paragraphStack;
2332                                                                 $paragraphStack = false;
2333                                                                 $this->mLastSection = 'p';
2334                                                         } elseif ( $this->mLastSection !== 'p' ) {
2335                                                                 $output .= $this->closeParagraph().'<p>';
2336                                                                 $this->mLastSection = 'p';
2337                                                         }
2338                                                 }
2339                                         }
2340                                 }
2341                                 wfProfileOut( __METHOD__."-paragraph" );
2342                         }
2343                         # somewhere above we forget to get out of pre block (bug 785)
2344                         if ( $preCloseMatch && $this->mInPre ) {
2345                                 $this->mInPre = false;
2346                         }
2347                         if ( $paragraphStack === false ) {
2348                                 $output .= $t."\n";
2349                         }
2350                 }
2351                 while ( $prefixLength ) {
2352                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2353                         --$prefixLength;
2354                 }
2355                 if ( $this->mLastSection != '' ) {
2356                         $output .= '</' . $this->mLastSection . '>';
2357                         $this->mLastSection = '';
2358                 }
2359
2360                 wfProfileOut( __METHOD__ );
2361                 return $output;
2362         }
2363
2364         /**
2365          * Split up a string on ':', ignoring any occurences inside tags
2366          * to prevent illegal overlapping.
2367          *
2368          * @param $str String the string to split
2369          * @param &$before String set to everything before the ':'
2370          * @param &$after String set to everything after the ':'
2371          * @return String the position of the ':', or false if none found
2372          */
2373         function findColonNoLinks( $str, &$before, &$after ) {
2374                 wfProfileIn( __METHOD__ );
2375
2376                 $pos = strpos( $str, ':' );
2377                 if ( $pos === false ) {
2378                         # Nothing to find!
2379                         wfProfileOut( __METHOD__ );
2380                         return false;
2381                 }
2382
2383                 $lt = strpos( $str, '<' );
2384                 if ( $lt === false || $lt > $pos ) {
2385                         # Easy; no tag nesting to worry about
2386                         $before = substr( $str, 0, $pos );
2387                         $after = substr( $str, $pos+1 );
2388                         wfProfileOut( __METHOD__ );
2389                         return $pos;
2390                 }
2391
2392                 # Ugly state machine to walk through avoiding tags.
2393                 $state = self::COLON_STATE_TEXT;
2394                 $stack = 0;
2395                 $len = strlen( $str );
2396                 for( $i = 0; $i < $len; $i++ ) {
2397                         $c = $str[$i];
2398
2399                         switch( $state ) {
2400                         # (Using the number is a performance hack for common cases)
2401                         case 0: # self::COLON_STATE_TEXT:
2402                                 switch( $c ) {
2403                                 case "<":
2404                                         # Could be either a <start> tag or an </end> tag
2405                                         $state = self::COLON_STATE_TAGSTART;
2406                                         break;
2407                                 case ":":
2408                                         if ( $stack == 0 ) {
2409                                                 # We found it!
2410                                                 $before = substr( $str, 0, $i );
2411                                                 $after = substr( $str, $i + 1 );
2412                                                 wfProfileOut( __METHOD__ );
2413                                                 return $i;
2414                                         }
2415                                         # Embedded in a tag; don't break it.
2416                                         break;
2417                                 default:
2418                                         # Skip ahead looking for something interesting
2419                                         $colon = strpos( $str, ':', $i );
2420                                         if ( $colon === false ) {
2421                                                 # Nothing else interesting
2422                                                 wfProfileOut( __METHOD__ );
2423                                                 return false;
2424                                         }
2425                                         $lt = strpos( $str, '<', $i );
2426                                         if ( $stack === 0 ) {
2427                                                 if ( $lt === false || $colon < $lt ) {
2428                                                         # We found it!
2429                                                         $before = substr( $str, 0, $colon );
2430                                                         $after = substr( $str, $colon + 1 );
2431                                                         wfProfileOut( __METHOD__ );
2432                                                         return $i;
2433                                                 }
2434                                         }
2435                                         if ( $lt === false ) {
2436                                                 # Nothing else interesting to find; abort!
2437                                                 # We're nested, but there's no close tags left. Abort!
2438                                                 break 2;
2439                                         }
2440                                         # Skip ahead to next tag start
2441                                         $i = $lt;
2442                                         $state = self::COLON_STATE_TAGSTART;
2443                                 }
2444                                 break;
2445                         case 1: # self::COLON_STATE_TAG:
2446                                 # In a <tag>
2447                                 switch( $c ) {
2448                                 case ">":
2449                                         $stack++;
2450                                         $state = self::COLON_STATE_TEXT;
2451                                         break;
2452                                 case "/":
2453                                         # Slash may be followed by >?
2454                                         $state = self::COLON_STATE_TAGSLASH;
2455                                         break;
2456                                 default:
2457                                         # ignore
2458                                 }
2459                                 break;
2460                         case 2: # self::COLON_STATE_TAGSTART:
2461                                 switch( $c ) {
2462                                 case "/":
2463                                         $state = self::COLON_STATE_CLOSETAG;
2464                                         break;
2465                                 case "!":
2466                                         $state = self::COLON_STATE_COMMENT;
2467                                         break;
2468                                 case ">":
2469                                         # Illegal early close? This shouldn't happen D:
2470                                         $state = self::COLON_STATE_TEXT;
2471                                         break;
2472                                 default:
2473                                         $state = self::COLON_STATE_TAG;
2474                                 }
2475                                 break;
2476                         case 3: # self::COLON_STATE_CLOSETAG:
2477                                 # In a </tag>
2478                                 if ( $c === ">" ) {
2479                                         $stack--;
2480                                         if ( $stack < 0 ) {
2481                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2482                                                 wfProfileOut( __METHOD__ );
2483                                                 return false;
2484                                         }
2485                                         $state = self::COLON_STATE_TEXT;
2486                                 }
2487                                 break;
2488                         case self::COLON_STATE_TAGSLASH:
2489                                 if ( $c === ">" ) {
2490                                         # Yes, a self-closed tag <blah/>
2491                                         $state = self::COLON_STATE_TEXT;
2492                                 } else {
2493                                         # Probably we're jumping the gun, and this is an attribute
2494                                         $state = self::COLON_STATE_TAG;
2495                                 }
2496                                 break;
2497                         case 5: # self::COLON_STATE_COMMENT:
2498                                 if ( $c === "-" ) {
2499                                         $state = self::COLON_STATE_COMMENTDASH;
2500                                 }
2501                                 break;
2502                         case self::COLON_STATE_COMMENTDASH:
2503                                 if ( $c === "-" ) {
2504                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2505                                 } else {
2506                                         $state = self::COLON_STATE_COMMENT;
2507                                 }
2508                                 break;
2509                         case self::COLON_STATE_COMMENTDASHDASH:
2510                                 if ( $c === ">" ) {
2511                                         $state = self::COLON_STATE_TEXT;
2512                                 } else {
2513                                         $state = self::COLON_STATE_COMMENT;
2514                                 }
2515                                 break;
2516                         default:
2517                                 throw new MWException( "State machine error in " . __METHOD__ );
2518                         }
2519                 }
2520                 if ( $stack > 0 ) {
2521                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2522                         wfProfileOut( __METHOD__ );
2523                         return false;
2524                 }
2525                 wfProfileOut( __METHOD__ );
2526                 return false;
2527         }
2528
2529         /**
2530          * Return value of a magic variable (like PAGENAME)
2531          *
2532          * @private
2533          *
2534          * @param $index integer
2535          * @param $frame PPFrame
2536          *
2537          * @return string
2538          */
2539         function getVariableValue( $index, $frame = false ) {
2540                 global $wgContLang, $wgSitename, $wgServer;
2541                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2542
2543                 if ( is_null( $this->mTitle ) ) {
2544                         // If no title set, bad things are going to happen
2545                         // later. Title should always be set since this
2546                         // should only be called in the middle of a parse
2547                         // operation (but the unit-tests do funky stuff)
2548                         throw new MWException( __METHOD__ . ' Should only be '
2549                                 . ' called while parsing (no title set)' );
2550                 }
2551
2552                 /**
2553                  * Some of these require message or data lookups and can be
2554                  * expensive to check many times.
2555                  */
2556                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2557                         if ( isset( $this->mVarCache[$index] ) ) {
2558                                 return $this->mVarCache[$index];
2559                         }
2560                 }
2561
2562                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2563                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2564
2565                 # Use the time zone
2566                 global $wgLocaltimezone;
2567                 if ( isset( $wgLocaltimezone ) ) {
2568                         $oldtz = date_default_timezone_get();
2569                         date_default_timezone_set( $wgLocaltimezone );
2570                 }
2571
2572                 $localTimestamp = date( 'YmdHis', $ts );
2573                 $localMonth = date( 'm', $ts );
2574                 $localMonth1 = date( 'n', $ts );
2575                 $localMonthName = date( 'n', $ts );
2576                 $localDay = date( 'j', $ts );
2577                 $localDay2 = date( 'd', $ts );
2578                 $localDayOfWeek = date( 'w', $ts );
2579                 $localWeek = date( 'W', $ts );
2580                 $localYear = date( 'Y', $ts );
2581                 $localHour = date( 'H', $ts );
2582                 if ( isset( $wgLocaltimezone ) ) {
2583                         date_default_timezone_set( $oldtz );
2584                 }
2585
2586                 $pageLang = $this->getFunctionLang();
2587
2588                 switch ( $index ) {
2589                         case 'currentmonth':
2590                                 $value = $pageLang->formatNum( gmdate( 'm', $ts ) );
2591                                 break;
2592                         case 'currentmonth1':
2593                                 $value = $pageLang->formatNum( gmdate( 'n', $ts ) );
2594                                 break;
2595                         case 'currentmonthname':
2596                                 $value = $pageLang->getMonthName( gmdate( 'n', $ts ) );
2597                                 break;
2598                         case 'currentmonthnamegen':
2599                                 $value = $pageLang->getMonthNameGen( gmdate( 'n', $ts ) );
2600                                 break;
2601                         case 'currentmonthabbrev':
2602                                 $value = $pageLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2603                                 break;
2604                         case 'currentday':
2605                                 $value = $pageLang->formatNum( gmdate( 'j', $ts ) );
2606                                 break;
2607                         case 'currentday2':
2608                                 $value = $pageLang->formatNum( gmdate( 'd', $ts ) );
2609                                 break;
2610                         case 'localmonth':
2611                                 $value = $pageLang->formatNum( $localMonth );
2612                                 break;
2613                         case 'localmonth1':
2614                                 $value = $pageLang->formatNum( $localMonth1 );
2615                                 break;
2616                         case 'localmonthname':
2617                                 $value = $pageLang->getMonthName( $localMonthName );
2618                                 break;
2619                         case 'localmonthnamegen':
2620                                 $value = $pageLang->getMonthNameGen( $localMonthName );
2621                                 break;
2622                         case 'localmonthabbrev':
2623                                 $value = $pageLang->getMonthAbbreviation( $localMonthName );
2624                                 break;
2625                         case 'localday':
2626                                 $value = $pageLang->formatNum( $localDay );
2627                                 break;
2628                         case 'localday2':
2629                                 $value = $pageLang->formatNum( $localDay2 );
2630                                 break;
2631                         case 'pagename':
2632                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2633                                 break;
2634                         case 'pagenamee':
2635                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2636                                 break;
2637                         case 'fullpagename':
2638                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2639                                 break;
2640                         case 'fullpagenamee':
2641                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2642                                 break;
2643                         case 'subpagename':
2644                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2645                                 break;
2646                         case 'subpagenamee':
2647                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2648                                 break;
2649                         case 'basepagename':
2650                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2651                                 break;
2652                         case 'basepagenamee':
2653                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2654                                 break;
2655                         case 'talkpagename':
2656                                 if ( $this->mTitle->canTalk() ) {
2657                                         $talkPage = $this->mTitle->getTalkPage();
2658                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2659                                 } else {
2660                                         $value = '';
2661                                 }
2662                                 break;
2663                         case 'talkpagenamee':
2664                                 if ( $this->mTitle->canTalk() ) {
2665                                         $talkPage = $this->mTitle->getTalkPage();
2666                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2667                                 } else {
2668                                         $value = '';
2669                                 }
2670                                 break;
2671                         case 'subjectpagename':
2672                                 $subjPage = $this->mTitle->getSubjectPage();
2673                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2674                                 break;
2675                         case 'subjectpagenamee':
2676                                 $subjPage = $this->mTitle->getSubjectPage();
2677                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2678                                 break;
2679                         case 'revisionid':
2680                                 # Let the edit saving system know we should parse the page
2681                                 # *after* a revision ID has been assigned.
2682                                 $this->mOutput->setFlag( 'vary-revision' );
2683                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2684                                 $value = $this->mRevisionId;
2685                                 break;
2686                         case 'revisionday':
2687                                 # Let the edit saving system know we should parse the page
2688                                 # *after* a revision ID has been assigned. This is for null edits.
2689                                 $this->mOutput->setFlag( 'vary-revision' );
2690                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2691                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2692                                 break;
2693                         case 'revisionday2':
2694                                 # Let the edit saving system know we should parse the page
2695                                 # *after* a revision ID has been assigned. This is for null edits.
2696                                 $this->mOutput->setFlag( 'vary-revision' );
2697                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2698                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2699                                 break;
2700                         case 'revisionmonth':
2701                                 # Let the edit saving system know we should parse the page
2702                                 # *after* a revision ID has been assigned. This is for null edits.
2703                                 $this->mOutput->setFlag( 'vary-revision' );
2704                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2705                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2706                                 break;
2707                         case 'revisionmonth1':
2708                                 # Let the edit saving system know we should parse the page
2709                                 # *after* a revision ID has been assigned. This is for null edits.
2710                                 $this->mOutput->setFlag( 'vary-revision' );
2711                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2712                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2713                                 break;
2714                         case 'revisionyear':
2715                                 # Let the edit saving system know we should parse the page
2716                                 # *after* a revision ID has been assigned. This is for null edits.
2717                                 $this->mOutput->setFlag( 'vary-revision' );
2718                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2719                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2720                                 break;
2721                         case 'revisiontimestamp':
2722                                 # Let the edit saving system know we should parse the page
2723                                 # *after* a revision ID has been assigned. This is for null edits.
2724                                 $this->mOutput->setFlag( 'vary-revision' );
2725                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2726                                 $value = $this->getRevisionTimestamp();
2727                                 break;
2728                         case 'revisionuser':
2729                                 # Let the edit saving system know we should parse the page
2730                                 # *after* a revision ID has been assigned. This is for null edits.
2731                                 $this->mOutput->setFlag( 'vary-revision' );
2732                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2733                                 $value = $this->getRevisionUser();
2734                                 break;
2735                         case 'namespace':
2736                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2737                                 break;
2738                         case 'namespacee':
2739                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2740                                 break;
2741                         case 'talkspace':
2742                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2743                                 break;
2744                         case 'talkspacee':
2745                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2746                                 break;
2747                         case 'subjectspace':
2748                                 $value = $this->mTitle->getSubjectNsText();
2749                                 break;
2750                         case 'subjectspacee':
2751                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2752                                 break;
2753                         case 'currentdayname':
2754                                 $value = $pageLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2755                                 break;
2756                         case 'currentyear':
2757                                 $value = $pageLang->formatNum( gmdate( 'Y', $ts ), true );
2758                                 break;
2759                         case 'currenttime':
2760                                 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2761                                 break;
2762                         case 'currenthour':
2763                                 $value = $pageLang->formatNum( gmdate( 'H', $ts ), true );
2764                                 break;
2765                         case 'currentweek':
2766                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2767                                 # int to remove the padding
2768                                 $value = $pageLang->formatNum( (int)gmdate( 'W', $ts ) );
2769                                 break;
2770                         case 'currentdow':
2771                                 $value = $pageLang->formatNum( gmdate( 'w', $ts ) );
2772                                 break;
2773                         case 'localdayname':
2774                                 $value = $pageLang->getWeekdayName( $localDayOfWeek + 1 );
2775                                 break;
2776                         case 'localyear':
2777                                 $value = $pageLang->formatNum( $localYear, true );
2778                                 break;
2779                         case 'localtime':
2780                                 $value = $pageLang->time( $localTimestamp, false, false );
2781                                 break;
2782                         case 'localhour':
2783                                 $value = $pageLang->formatNum( $localHour, true );
2784                                 break;
2785                         case 'localweek':
2786                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2787                                 # int to remove the padding
2788                                 $value = $pageLang->formatNum( (int)$localWeek );
2789                                 break;
2790                         case 'localdow':
2791                                 $value = $pageLang->formatNum( $localDayOfWeek );
2792                                 break;
2793                         case 'numberofarticles':
2794                                 $value = $pageLang->formatNum( SiteStats::articles() );
2795                                 break;
2796                         case 'numberoffiles':
2797                                 $value = $pageLang->formatNum( SiteStats::images() );
2798                                 break;
2799                         case 'numberofusers':
2800                                 $value = $pageLang->formatNum( SiteStats::users() );
2801                                 break;
2802                         case 'numberofactiveusers':
2803                                 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2804                                 break;
2805                         case 'numberofpages':
2806                                 $value = $pageLang->formatNum( SiteStats::pages() );
2807                                 break;
2808                         case 'numberofadmins':
2809                                 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2810                                 break;
2811                         case 'numberofedits':
2812                                 $value = $pageLang->formatNum( SiteStats::edits() );
2813                                 break;
2814                         case 'numberofviews':
2815                                 $value = $pageLang->formatNum( SiteStats::views() );
2816                                 break;
2817                         case 'currenttimestamp':
2818                                 $value = wfTimestamp( TS_MW, $ts );
2819                                 break;
2820                         case 'localtimestamp':
2821                                 $value = $localTimestamp;
2822                                 break;
2823                         case 'currentversion':
2824                                 $value = SpecialVersion::getVersion();
2825                                 break;
2826                         case 'articlepath':
2827                                 return $wgArticlePath;
2828                         case 'sitename':
2829                                 return $wgSitename;
2830                         case 'server':
2831                                 return $wgServer;
2832                         case 'servername':
2833                                 $serverParts = wfParseUrl( $wgServer );
2834                                 return $serverParts && isset( $serverParts['host'] ) ? $serverParts['host'] : $wgServer;
2835                         case 'scriptpath':
2836                                 return $wgScriptPath;
2837                         case 'stylepath':
2838                                 return $wgStylePath;
2839                         case 'directionmark':
2840                                 return $pageLang->getDirMark();
2841                         case 'contentlanguage':
2842                                 global $wgLanguageCode;
2843                                 return $wgLanguageCode;
2844                         default:
2845                                 $ret = null;
2846                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2847                                         return $ret;
2848                                 } else {
2849                                         return null;
2850                                 }
2851                 }
2852
2853                 if ( $index ) {
2854                         $this->mVarCache[$index] = $value;
2855                 }
2856
2857                 return $value;
2858         }
2859
2860         /**
2861          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2862          *
2863          * @private
2864          */
2865         function initialiseVariables() {
2866                 wfProfileIn( __METHOD__ );
2867                 $variableIDs = MagicWord::getVariableIDs();
2868                 $substIDs = MagicWord::getSubstIDs();
2869
2870                 $this->mVariables = new MagicWordArray( $variableIDs );
2871                 $this->mSubstWords = new MagicWordArray( $substIDs );
2872                 wfProfileOut( __METHOD__ );
2873         }
2874
2875         /**
2876          * Preprocess some wikitext and return the document tree.
2877          * This is the ghost of replace_variables().
2878          *
2879          * @param $text String: The text to parse
2880          * @param $flags Integer: bitwise combination of:
2881          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2882          *                                     included. Default is to assume a direct page view.
2883          *
2884          * The generated DOM tree must depend only on the input text and the flags.
2885          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2886          *
2887          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2888          * change in the DOM tree for a given text, must be passed through the section identifier
2889          * in the section edit link and thus back to extractSections().
2890          *
2891          * The output of this function is currently only cached in process memory, but a persistent
2892          * cache may be implemented at a later date which takes further advantage of these strict
2893          * dependency requirements.
2894          *
2895          * @private
2896          *
2897          * @return PPNode
2898          */
2899         function preprocessToDom( $text, $flags = 0 ) {
2900                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2901                 return $dom;
2902         }
2903
2904         /**
2905          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2906          *
2907          * @param $s string
2908          *
2909          * @return array
2910          */
2911         public static function splitWhitespace( $s ) {
2912                 $ltrimmed = ltrim( $s );
2913                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2914                 $trimmed = rtrim( $ltrimmed );
2915                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2916                 if ( $diff > 0 ) {
2917                         $w2 = substr( $ltrimmed, -$diff );
2918                 } else {
2919                         $w2 = '';
2920                 }
2921                 return array( $w1, $trimmed, $w2 );
2922         }
2923
2924         /**
2925          * Replace magic variables, templates, and template arguments
2926          * with the appropriate text. Templates are substituted recursively,
2927          * taking care to avoid infinite loops.
2928          *
2929          * Note that the substitution depends on value of $mOutputType:
2930          *  self::OT_WIKI: only {{subst:}} templates
2931          *  self::OT_PREPROCESS: templates but not extension tags
2932          *  self::OT_HTML: all templates and extension tags
2933          *
2934          * @param $text String the text to transform
2935          * @param $frame PPFrame Object describing the arguments passed to the template.
2936          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2937          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2938          * @param $argsOnly Boolean only do argument (triple-brace) expansion, not double-brace expansion
2939          * @private
2940          *
2941          * @return string
2942          */
2943         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2944                 # Is there any text? Also, Prevent too big inclusions!
2945                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2946                         return $text;
2947                 }
2948                 wfProfileIn( __METHOD__ );
2949
2950                 if ( $frame === false ) {
2951                         $frame = $this->getPreprocessor()->newFrame();
2952                 } elseif ( !( $frame instanceof PPFrame ) ) {
2953                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2954                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2955                 }
2956
2957                 $dom = $this->preprocessToDom( $text );
2958                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2959                 $text = $frame->expand( $dom, $flags );
2960
2961                 wfProfileOut( __METHOD__ );
2962                 return $text;
2963         }
2964
2965         /**
2966          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2967          *
2968          * @param $args array
2969          *
2970          * @return array
2971          */
2972         static function createAssocArgs( $args ) {
2973                 $assocArgs = array();
2974                 $index = 1;
2975                 foreach ( $args as $arg ) {
2976                         $eqpos = strpos( $arg, '=' );
2977                         if ( $eqpos === false ) {
2978                                 $assocArgs[$index++] = $arg;
2979                         } else {
2980                                 $name = trim( substr( $arg, 0, $eqpos ) );
2981                                 $value = trim( substr( $arg, $eqpos+1 ) );
2982                                 if ( $value === false ) {
2983                                         $value = '';
2984                                 }
2985                                 if ( $name !== false ) {
2986                                         $assocArgs[$name] = $value;
2987                                 }
2988                         }
2989                 }
2990
2991                 return $assocArgs;
2992         }
2993
2994         /**
2995          * Warn the user when a parser limitation is reached
2996          * Will warn at most once the user per limitation type
2997          *
2998          * @param $limitationType String: should be one of:
2999          *   'expensive-parserfunction' (corresponding messages:
3000          *       'expensive-parserfunction-warning',
3001          *       'expensive-parserfunction-category')
3002          *   'post-expand-template-argument' (corresponding messages:
3003          *       'post-expand-template-argument-warning',
3004          *       'post-expand-template-argument-category')
3005          *   'post-expand-template-inclusion' (corresponding messages:
3006          *       'post-expand-template-inclusion-warning',
3007          *       'post-expand-template-inclusion-category')
3008          * @param $current Current value
3009          * @param $max Maximum allowed, when an explicit limit has been
3010          *       exceeded, provide the values (optional)
3011          */
3012         function limitationWarn( $limitationType, $current=null, $max=null) {
3013                 # does no harm if $current and $max are present but are unnecessary for the message
3014                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3015                 $this->mOutput->addWarning( $warning );
3016                 $this->addTrackingCategory( "$limitationType-category" );
3017         }
3018
3019         /**
3020          * Return the text of a template, after recursively
3021          * replacing any variables or templates within the template.
3022          *
3023          * @param $piece Array: the parts of the template
3024          *  $piece['title']: the title, i.e. the part before the |
3025          *  $piece['parts']: the parameter array
3026          *  $piece['lineStart']: whether the brace was at the start of a line
3027          * @param $frame PPFrame The current frame, contains template arguments
3028          * @return String: the text of the template
3029          * @private
3030          */
3031         function braceSubstitution( $piece, $frame ) {
3032                 global $wgNonincludableNamespaces;
3033                 wfProfileIn( __METHOD__ );
3034                 wfProfileIn( __METHOD__.'-setup' );
3035
3036                 # Flags
3037                 $found = false;             # $text has been filled
3038                 $nowiki = false;            # wiki markup in $text should be escaped
3039                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3040                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3041                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3042                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3043
3044                 # Title object, where $text came from
3045                 $title = false;
3046
3047                 # $part1 is the bit before the first |, and must contain only title characters.
3048                 # Various prefixes will be stripped from it later.
3049                 $titleWithSpaces = $frame->expand( $piece['title'] );
3050                 $part1 = trim( $titleWithSpaces );
3051                 $titleText = false;
3052
3053                 # Original title text preserved for various purposes
3054                 $originalTitle = $part1;
3055
3056                 # $args is a list of argument nodes, starting from index 0, not including $part1
3057                 # @todo FIXME: If piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3058                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3059                 wfProfileOut( __METHOD__.'-setup' );
3060
3061                 $titleProfileIn = null; // profile templates
3062
3063                 # SUBST
3064                 wfProfileIn( __METHOD__.'-modifiers' );
3065                 if ( !$found ) {
3066
3067                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3068
3069                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3070                         # Decide whether to expand template or keep wikitext as-is.
3071                         if ( $this->ot['wiki'] ) {
3072                                 if ( $substMatch === false ) {
3073                                         $literal = true;  # literal when in PST with no prefix
3074                                 } else {
3075                                         $literal = false; # expand when in PST with subst: or safesubst:
3076                                 }
3077                         } else {
3078                                 if ( $substMatch == 'subst' ) {
3079                                         $literal = true;  # literal when not in PST with plain subst:
3080                                 } else {
3081                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3082                                 }
3083                         }
3084                         if ( $literal ) {
3085                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3086                                 $isLocalObj = true;
3087                                 $found = true;
3088                         }
3089                 }
3090
3091                 # Variables
3092                 if ( !$found && $args->getLength() == 0 ) {
3093                         $id = $this->mVariables->matchStartToEnd( $part1 );
3094                         if ( $id !== false ) {
3095                                 $text = $this->getVariableValue( $id, $frame );
3096                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3097                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3098                                 }
3099                                 $found = true;
3100                         }
3101                 }
3102
3103                 # MSG, MSGNW and RAW
3104                 if ( !$found ) {
3105                         # Check for MSGNW:
3106                         $mwMsgnw = MagicWord::get( 'msgnw' );
3107                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3108                                 $nowiki = true;
3109                         } else {
3110                                 # Remove obsolete MSG:
3111                                 $mwMsg = MagicWord::get( 'msg' );
3112                                 $mwMsg->matchStartAndRemove( $part1 );
3113                         }
3114
3115                         # Check for RAW:
3116                         $mwRaw = MagicWord::get( 'raw' );
3117                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3118                                 $forceRawInterwiki = true;
3119                         }
3120                 }
3121                 wfProfileOut( __METHOD__.'-modifiers' );
3122
3123                 # Parser functions
3124                 if ( !$found ) {
3125                         wfProfileIn( __METHOD__ . '-pfunc' );
3126
3127                         $colonPos = strpos( $part1, ':' );
3128                         if ( $colonPos !== false ) {
3129                                 # Case sensitive functions
3130                                 $function = substr( $part1, 0, $colonPos );
3131                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3132                                         $function = $this->mFunctionSynonyms[1][$function];
3133                                 } else {
3134                                         # Case insensitive functions
3135                                         $function = $this->getFunctionLang()->lc( $function );
3136                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3137                                                 $function = $this->mFunctionSynonyms[0][$function];
3138                                         } else {
3139                                                 $function = false;
3140                                         }
3141                                 }
3142                                 if ( $function ) {
3143                                         wfProfileIn( __METHOD__ . '-pfunc-' . $function );
3144                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3145                                         $initialArgs = array( &$this );
3146                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3147                                         if ( $flags & SFH_OBJECT_ARGS ) {
3148                                                 # Add a frame parameter, and pass the arguments as an array
3149                                                 $allArgs = $initialArgs;
3150                                                 $allArgs[] = $frame;
3151                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3152                                                         $funcArgs[] = $args->item( $i );
3153                                                 }
3154                                                 $allArgs[] = $funcArgs;
3155                                         } else {
3156                                                 # Convert arguments to plain text
3157                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3158                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3159                                                 }
3160                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3161                                         }
3162
3163                                         # Workaround for PHP bug 35229 and similar
3164                                         if ( !is_callable( $callback ) ) {
3165                                                 wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3166                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3167                                                 wfProfileOut( __METHOD__ );
3168                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3169                                         }
3170                                         $result = call_user_func_array( $callback, $allArgs );
3171                                         $found = true;
3172                                         $noparse = true;
3173                                         $preprocessFlags = 0;
3174
3175                                         if ( is_array( $result ) ) {
3176                                                 if ( isset( $result[0] ) ) {
3177                                                         $text = $result[0];
3178                                                         unset( $result[0] );
3179                                                 }
3180
3181                                                 # Extract flags into the local scope
3182                                                 # This allows callers to set flags such as nowiki, found, etc.
3183                                                 extract( $result );
3184                                         } else {
3185                                                 $text = $result;
3186                                         }
3187                                         if ( !$noparse ) {
3188                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3189                                                 $isChildObj = true;
3190                                         }
3191                                         wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3192                                 }
3193                         }
3194                         wfProfileOut( __METHOD__ . '-pfunc' );
3195                 }
3196
3197                 # Finish mangling title and then check for loops.
3198                 # Set $title to a Title object and $titleText to the PDBK
3199                 if ( !$found ) {
3200                         $ns = NS_TEMPLATE;
3201                         # Split the title into page and subpage
3202                         $subpage = '';
3203                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3204                         if ( $subpage !== '' ) {
3205                                 $ns = $this->mTitle->getNamespace();
3206                         }
3207                         $title = Title::newFromText( $part1, $ns );
3208                         if ( $title ) {
3209                                 $titleText = $title->getPrefixedText();
3210                                 # Check for language variants if the template is not found
3211                                 if ( $this->getFunctionLang()->hasVariants() && $title->getArticleID() == 0 ) {
3212                                         $this->getFunctionLang()->findVariantLink( $part1, $title, true );
3213                                 }
3214                                 # Do recursion depth check
3215                                 $limit = $this->mOptions->getMaxTemplateDepth();
3216                                 if ( $frame->depth >= $limit ) {
3217                                         $found = true;
3218                                         $text = '<span class="error">'
3219                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3220                                                 . '</span>';
3221                                 }
3222                         }
3223                 }
3224
3225                 # Load from database
3226                 if ( !$found && $title ) {
3227                         $titleProfileIn = __METHOD__ . "-title-" . $title->getDBKey();
3228                         wfProfileIn( $titleProfileIn ); // template in
3229                         wfProfileIn( __METHOD__ . '-loadtpl' );
3230                         if ( !$title->isExternal() ) {
3231                                 if ( $title->isSpecialPage()
3232                                         && $this->mOptions->getAllowSpecialInclusion()
3233                                         && $this->ot['html'] )
3234                                 {
3235                                         // Pass the template arguments as URL parameters.
3236                                         // "uselang" will have no effect since the Language object
3237                                         // is forced to the one defined in ParserOptions.
3238                                         $pageArgs = array();
3239                                         for ( $i = 0; $i < $args->getLength(); $i++ ) {
3240                                                 $bits = $args->item( $i )->splitArg();
3241                                                 if ( strval( $bits['index'] ) === '' ) {
3242                                                         $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3243                                                         $value = trim( $frame->expand( $bits['value'] ) );
3244                                                         $pageArgs[$name] = $value;
3245                                                 }
3246                                         }
3247
3248                                         // Create a new context to execute the special page
3249                                         $context = new RequestContext;
3250                                         $context->setTitle( $title );
3251                                         $context->setRequest( new FauxRequest( $pageArgs ) );
3252                                         $context->setUser( $this->getUser() );
3253                                         $context->setLang( $this->mOptions->getUserLangObj() );
3254                                         $ret = SpecialPageFactory::capturePath( $title, $context );
3255                                         if ( $ret ) {
3256                                                 $text = $context->getOutput()->getHTML();
3257                                                 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3258                                                 $found = true;
3259                                                 $isHTML = true;
3260                                                 $this->disableCache();
3261                                         }
3262                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3263                                         $found = false; # access denied
3264                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3265                                 } else {
3266                                         list( $text, $title ) = $this->getTemplateDom( $title );
3267                                         if ( $text !== false ) {
3268                                                 $found = true;
3269                                                 $isChildObj = true;
3270                                         }
3271                                 }
3272
3273                                 # If the title is valid but undisplayable, make a link to it
3274                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3275                                         $text = "[[:$titleText]]";
3276                                         $found = true;
3277                                 }
3278                         } elseif ( $title->isTrans() ) {
3279                                 # Interwiki transclusion
3280                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3281                                         $text = $this->interwikiTransclude( $title, 'render' );
3282                                         $isHTML = true;
3283                                 } else {
3284                                         $text = $this->interwikiTransclude( $title, 'raw' );
3285                                         # Preprocess it like a template
3286                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3287                                         $isChildObj = true;
3288                                 }
3289                                 $found = true;
3290                         }
3291
3292                         # Do infinite loop check
3293                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3294                         if ( !$frame->loopCheck( $title ) ) {
3295                                 $found = true;
3296                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3297                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3298                         }
3299                         wfProfileOut( __METHOD__ . '-loadtpl' );
3300                 }
3301
3302                 # If we haven't found text to substitute by now, we're done
3303                 # Recover the source wikitext and return it
3304                 if ( !$found ) {
3305                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3306                         if ( $titleProfileIn ) {
3307                                 wfProfileOut( $titleProfileIn ); // template out
3308                         }
3309                         wfProfileOut( __METHOD__ );
3310                         return array( 'object' => $text );
3311                 }
3312
3313                 # Expand DOM-style return values in a child frame
3314                 if ( $isChildObj ) {
3315                         # Clean up argument array
3316                         $newFrame = $frame->newChild( $args, $title );
3317
3318                         if ( $nowiki ) {
3319                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3320                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3321                                 # Expansion is eligible for the empty-frame cache
3322                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3323                                         $text = $this->mTplExpandCache[$titleText];
3324                                 } else {
3325                                         $text = $newFrame->expand( $text );
3326                                         $this->mTplExpandCache[$titleText] = $text;
3327                                 }
3328                         } else {
3329                                 # Uncached expansion
3330                                 $text = $newFrame->expand( $text );
3331                         }
3332                 }
3333                 if ( $isLocalObj && $nowiki ) {
3334                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3335                         $isLocalObj = false;
3336                 }
3337
3338                 if ( $titleProfileIn ) {
3339                         wfProfileOut( $titleProfileIn ); // template out
3340                 }
3341
3342                 # Replace raw HTML by a placeholder
3343                 # Add a blank line preceding, to prevent it from mucking up
3344                 # immediately preceding headings
3345                 if ( $isHTML ) {
3346                         $text = "\n\n" . $this->insertStripItem( $text );
3347                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3348                         # Escape nowiki-style return values
3349                         $text = wfEscapeWikiText( $text );
3350                 } elseif ( is_string( $text )
3351                         && !$piece['lineStart']
3352                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3353                 {
3354                         # Bug 529: if the template begins with a table or block-level
3355                         # element, it should be treated as beginning a new line.
3356                         # This behaviour is somewhat controversial.
3357                         $text = "\n" . $text;
3358                 }
3359
3360                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3361                         # Error, oversize inclusion
3362                         if ( $titleText !== false ) {
3363                                 # Make a working, properly escaped link if possible (bug 23588)
3364                                 $text = "[[:$titleText]]";
3365                         } else {
3366                                 # This will probably not be a working link, but at least it may
3367                                 # provide some hint of where the problem is
3368                                 preg_replace( '/^:/', '', $originalTitle );
3369                                 $text = "[[:$originalTitle]]";
3370                         }
3371                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3372                         $this->limitationWarn( 'post-expand-template-inclusion' );
3373                 }
3374
3375                 if ( $isLocalObj ) {
3376                         $ret = array( 'object' => $text );
3377                 } else {
3378                         $ret = array( 'text' => $text );
3379                 }
3380
3381                 wfProfileOut( __METHOD__ );
3382                 return $ret;
3383         }
3384
3385         /**
3386          * Get the semi-parsed DOM representation of a template with a given title,
3387          * and its redirect destination title. Cached.
3388          *
3389          * @param $title Title
3390          *
3391          * @return array
3392          */
3393         function getTemplateDom( $title ) {
3394                 $cacheTitle = $title;
3395                 $titleText = $title->getPrefixedDBkey();
3396
3397                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3398                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3399                         $title = Title::makeTitle( $ns, $dbk );
3400                         $titleText = $title->getPrefixedDBkey();
3401                 }
3402                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3403                         return array( $this->mTplDomCache[$titleText], $title );
3404                 }
3405
3406                 # Cache miss, go to the database
3407                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3408
3409                 if ( $text === false ) {
3410                         $this->mTplDomCache[$titleText] = false;
3411                         return array( false, $title );
3412                 }
3413
3414                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3415                 $this->mTplDomCache[ $titleText ] = $dom;
3416
3417                 if ( !$title->equals( $cacheTitle ) ) {
3418                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3419                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3420                 }
3421
3422                 return array( $dom, $title );
3423         }
3424
3425         /**
3426          * Fetch the unparsed text of a template and register a reference to it.
3427          * @param Title $title
3428          * @return Array ( string or false, Title )
3429          */
3430         function fetchTemplateAndTitle( $title ) {
3431                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3432                 $stuff = call_user_func( $templateCb, $title, $this );
3433                 $text = $stuff['text'];
3434                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3435                 if ( isset( $stuff['deps'] ) ) {
3436                         foreach ( $stuff['deps'] as $dep ) {
3437                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3438                         }
3439                 }
3440                 return array( $text, $finalTitle );
3441         }
3442
3443         /**
3444          * Fetch the unparsed text of a template and register a reference to it.
3445          * @param Title $title
3446          * @return mixed string or false
3447          */
3448         function fetchTemplate( $title ) {
3449                 $rv = $this->fetchTemplateAndTitle( $title );
3450                 return $rv[0];
3451         }
3452
3453         /**
3454          * Static function to get a template
3455          * Can be overridden via ParserOptions::setTemplateCallback().
3456          *
3457          * @parma $title Title
3458          * @param $parser Parser
3459          *
3460          * @return array
3461          */
3462         static function statelessFetchTemplate( $title, $parser = false ) {
3463                 $text = $skip = false;
3464                 $finalTitle = $title;
3465                 $deps = array();
3466
3467                 # Loop to fetch the article, with up to 1 redirect
3468                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3469                         # Give extensions a chance to select the revision instead
3470                         $id = false; # Assume current
3471                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3472                                 array( $parser, $title, &$skip, &$id ) );
3473
3474                         if ( $skip ) {
3475                                 $text = false;
3476                                 $deps[] = array(
3477                                         'title'         => $title,
3478                                         'page_id'       => $title->getArticleID(),
3479                                         'rev_id'        => null
3480                                 );
3481                                 break;
3482                         }
3483                         # Get the revision
3484                         $rev = $id
3485                                 ? Revision::newFromId( $id )
3486                                 : Revision::newFromTitle( $title );
3487                         $rev_id = $rev ? $rev->getId() : 0;
3488                         # If there is no current revision, there is no page
3489                         if ( $id === false && !$rev ) {
3490                                 $linkCache = LinkCache::singleton();
3491                                 $linkCache->addBadLinkObj( $title );
3492                         }
3493
3494                         $deps[] = array(
3495                                 'title'         => $title,
3496                                 'page_id'       => $title->getArticleID(),
3497                                 'rev_id'        => $rev_id );
3498                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3499                                 # We fetched a rev from a different title; register it too...
3500                                 $deps[] = array(
3501                                         'title'         => $rev->getTitle(),
3502                                         'page_id'       => $rev->getPage(),
3503                                         'rev_id'        => $rev_id );
3504                         }
3505
3506                         if ( $rev ) {
3507                                 $text = $rev->getText();
3508                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3509                                 global $wgContLang;
3510                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3511                                 if ( !$message->exists() ) {
3512                                         $text = false;
3513                                         break;
3514                                 }
3515                                 $text = $message->plain();
3516                         } else {
3517                                 break;
3518                         }
3519                         if ( $text === false ) {
3520                                 break;
3521                         }
3522                         # Redirect?
3523                         $finalTitle = $title;
3524                         $title = Title::newFromRedirect( $text );
3525                 }
3526                 return array(
3527                         'text' => $text,
3528                         'finalTitle' => $finalTitle,
3529                         'deps' => $deps );
3530         }
3531
3532         /**
3533          * Fetch a file and its title and register a reference to it.
3534          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3535          * @param Title $title
3536          * @param Array $options Array of options to RepoGroup::findFile
3537          * @return File|false
3538          */
3539         function fetchFile( $title, $options = array() ) {
3540                 $res = $this->fetchFileAndTitle( $title, $options );
3541                 return $res[0];
3542         }
3543
3544         /**
3545          * Fetch a file and its title and register a reference to it.
3546          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3547          * @param Title $title
3548          * @param Array $options Array of options to RepoGroup::findFile
3549          * @return Array ( File or false, Title of file )
3550          */
3551         function fetchFileAndTitle( $title, $options = array() ) {
3552                 if ( isset( $options['broken'] ) ) {
3553                         $file = false; // broken thumbnail forced by hook
3554                 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3555                         $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3556                 } else { // get by (name,timestamp)
3557                         $file = wfFindFile( $title, $options );
3558                 }
3559                 $time = $file ? $file->getTimestamp() : false;
3560                 $sha1 = $file ? $file->getSha1() : false;
3561                 # Register the file as a dependency...
3562                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3563                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3564                         # Update fetched file title
3565                         $title = $file->getTitle();
3566                         if ( is_null( $file->getRedirectedTitle() ) ) {
3567                                 # This file was not a redirect, but the title does not match.
3568                                 # Register under the new name because otherwise the link will
3569                                 # get lost.
3570                                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3571                         }
3572                 }
3573                 return array( $file, $title );
3574         }
3575
3576         /**
3577          * Transclude an interwiki link.
3578          *
3579          * @param $title Title
3580          * @param $action
3581          *
3582          * @return string
3583          */
3584         function interwikiTransclude( $title, $action ) {
3585                 global $wgEnableScaryTranscluding;
3586
3587                 if ( !$wgEnableScaryTranscluding ) {
3588                         return wfMsgForContent('scarytranscludedisabled');
3589                 }
3590
3591                 $url = $title->getFullUrl( "action=$action" );
3592
3593                 if ( strlen( $url ) > 255 ) {
3594                         return wfMsgForContent( 'scarytranscludetoolong' );
3595                 }
3596                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3597         }
3598
3599         /**
3600          * @param $url string
3601          * @return Mixed|String
3602          */
3603         function fetchScaryTemplateMaybeFromCache( $url ) {
3604                 global $wgTranscludeCacheExpiry;
3605                 $dbr = wfGetDB( DB_SLAVE );
3606                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3607                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3608                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3609                 if ( $obj ) {
3610                         return $obj->tc_contents;
3611                 }
3612
3613                 $text = Http::get( $url );
3614                 if ( !$text ) {
3615                         return wfMsgForContent( 'scarytranscludefailed', $url );
3616                 }
3617
3618                 $dbw = wfGetDB( DB_MASTER );
3619                 $dbw->replace( 'transcache', array('tc_url'), array(
3620                         'tc_url' => $url,
3621                         'tc_time' => $dbw->timestamp( time() ),
3622                         'tc_contents' => $text)
3623                 );
3624                 return $text;
3625         }
3626
3627         /**
3628          * Triple brace replacement -- used for template arguments
3629          * @private
3630          *
3631          * @param $peice array
3632          * @param $frame PPFrame
3633          *
3634          * @return array
3635          */
3636         function argSubstitution( $piece, $frame ) {
3637                 wfProfileIn( __METHOD__ );
3638
3639                 $error = false;
3640                 $parts = $piece['parts'];
3641                 $nameWithSpaces = $frame->expand( $piece['title'] );
3642                 $argName = trim( $nameWithSpaces );
3643                 $object = false;
3644                 $text = $frame->getArgument( $argName );
3645                 if (  $text === false && $parts->getLength() > 0
3646                   && (
3647                         $this->ot['html']
3648                         || $this->ot['pre']
3649                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3650                   )
3651                 ) {
3652                         # No match in frame, use the supplied default
3653                         $object = $parts->item( 0 )->getChildren();
3654                 }
3655                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3656                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3657                         $this->limitationWarn( 'post-expand-template-argument' );
3658                 }
3659
3660                 if ( $text === false && $object === false ) {
3661                         # No match anywhere
3662                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3663                 }
3664                 if ( $error !== false ) {
3665                         $text .= $error;
3666                 }
3667                 if ( $object !== false ) {
3668                         $ret = array( 'object' => $object );
3669                 } else {
3670                         $ret = array( 'text' => $text );
3671                 }
3672
3673                 wfProfileOut( __METHOD__ );
3674                 return $ret;
3675         }
3676
3677         /**
3678          * Return the text to be used for a given extension tag.
3679          * This is the ghost of strip().
3680          *
3681          * @param $params Associative array of parameters:
3682          *     name       PPNode for the tag name
3683          *     attr       PPNode for unparsed text where tag attributes are thought to be
3684          *     attributes Optional associative array of parsed attributes
3685          *     inner      Contents of extension element
3686          *     noClose    Original text did not have a close tag
3687          * @param $frame PPFrame
3688          *
3689          * @return string
3690          */
3691         function extensionSubstitution( $params, $frame ) {
3692                 $name = $frame->expand( $params['name'] );
3693                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3694                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3695                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3696
3697                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3698                         ( $this->ot['html'] || $this->ot['pre'] );
3699                 if ( $isFunctionTag ) {
3700                         $markerType = 'none';
3701                 } else {
3702                         $markerType = 'general';
3703                 }
3704                 if ( $this->ot['html'] || $isFunctionTag ) {
3705                         $name = strtolower( $name );
3706                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3707                         if ( isset( $params['attributes'] ) ) {
3708                                 $attributes = $attributes + $params['attributes'];
3709                         }
3710
3711                         if ( isset( $this->mTagHooks[$name] ) ) {
3712                                 # Workaround for PHP bug 35229 and similar
3713                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3714                                         throw new MWException( "Tag hook for $name is not callable\n" );
3715                                 }
3716                                 $output = call_user_func_array( $this->mTagHooks[$name],
3717                                         array( $content, $attributes, $this, $frame ) );
3718                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3719                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3720                                 if ( !is_callable( $callback ) ) {
3721                                         throw new MWException( "Tag hook for $name is not callable\n" );
3722                                 }
3723
3724                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3725                         } else {
3726                                 $output = '<span class="error">Invalid tag extension name: ' .
3727                                         htmlspecialchars( $name ) . '</span>';
3728                         }
3729
3730                         if ( is_array( $output ) ) {
3731                                 # Extract flags to local scope (to override $markerType)
3732                                 $flags = $output;
3733                                 $output = $flags[0];
3734                                 unset( $flags[0] );
3735                                 extract( $flags );
3736                         }
3737                 } else {
3738                         if ( is_null( $attrText ) ) {
3739                                 $attrText = '';
3740                         }
3741                         if ( isset( $params['attributes'] ) ) {
3742                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3743                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3744                                                 htmlspecialchars( $attrValue ) . '"';
3745                                 }
3746                         }
3747                         if ( $content === null ) {
3748                                 $output = "<$name$attrText/>";
3749                         } else {
3750                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3751                                 $output = "<$name$attrText>$content$close";
3752                         }
3753                 }
3754
3755                 if ( $markerType === 'none' ) {
3756                         return $output;
3757                 } elseif ( $markerType === 'nowiki' ) {
3758                         $this->mStripState->addNoWiki( $marker, $output );
3759                 } elseif ( $markerType === 'general' ) {
3760                         $this->mStripState->addGeneral( $marker, $output );
3761                 } else {
3762                         throw new MWException( __METHOD__.': invalid marker type' );
3763                 }
3764                 return $marker;
3765         }
3766
3767         /**
3768          * Increment an include size counter
3769          *
3770          * @param $type String: the type of expansion
3771          * @param $size Integer: the size of the text
3772          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3773          */
3774         function incrementIncludeSize( $type, $size ) {
3775                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3776                         return false;
3777                 } else {
3778                         $this->mIncludeSizes[$type] += $size;
3779                         return true;
3780                 }
3781         }
3782
3783         /**
3784          * Increment the expensive function count
3785          *
3786          * @return Boolean: false if the limit has been exceeded
3787          */
3788         function incrementExpensiveFunctionCount() {
3789                 global $wgExpensiveParserFunctionLimit;
3790                 $this->mExpensiveFunctionCount++;
3791                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3792                         return true;
3793                 }
3794                 return false;
3795         }
3796
3797         /**
3798          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3799          * Fills $this->mDoubleUnderscores, returns the modified text
3800          *
3801          * @param $text string
3802          *
3803          * @return string
3804          */
3805         function doDoubleUnderscore( $text ) {
3806                 wfProfileIn( __METHOD__ );
3807
3808                 # The position of __TOC__ needs to be recorded
3809                 $mw = MagicWord::get( 'toc' );
3810                 if ( $mw->match( $text ) ) {
3811                         $this->mShowToc = true;
3812                         $this->mForceTocPosition = true;
3813
3814                         # Set a placeholder. At the end we'll fill it in with the TOC.
3815                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3816
3817                         # Only keep the first one.
3818                         $text = $mw->replace( '', $text );
3819                 }
3820
3821                 # Now match and remove the rest of them
3822                 $mwa = MagicWord::getDoubleUnderscoreArray();
3823                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3824
3825                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3826                         $this->mOutput->mNoGallery = true;
3827                 }
3828                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3829                         $this->mShowToc = false;
3830                 }
3831                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3832                         $this->addTrackingCategory( 'hidden-category-category' );
3833                 }
3834                 # (bug 8068) Allow control over whether robots index a page.
3835                 #
3836                 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here!  This
3837                 # is not desirable, the last one on the page should win.
3838                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3839                         $this->mOutput->setIndexPolicy( 'noindex' );
3840                         $this->addTrackingCategory( 'noindex-category' );
3841                 }
3842                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3843                         $this->mOutput->setIndexPolicy( 'index' );
3844                         $this->addTrackingCategory( 'index-category' );
3845                 }
3846
3847                 # Cache all double underscores in the database
3848                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3849                         $this->mOutput->setProperty( $key, '' );
3850                 }
3851
3852                 wfProfileOut( __METHOD__ );
3853                 return $text;
3854         }
3855
3856         /**
3857          * Add a tracking category, getting the title from a system message,
3858          * or print a debug message if the title is invalid.
3859          *
3860          * @param $msg String: message key
3861          * @return Boolean: whether the addition was successful
3862          */
3863         protected function addTrackingCategory( $msg ) {
3864                 if ( $this->mTitle->getNamespace() === NS_SPECIAL ) {
3865                         wfDebug( __METHOD__.": Not adding tracking category $msg to special page!\n" );
3866                         return false;
3867                 }
3868                 $cat = wfMsgForContent( $msg );
3869
3870                 # Allow tracking categories to be disabled by setting them to "-"
3871                 if ( $cat === '-' ) {
3872                         return false;
3873                 }
3874
3875                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3876                 if ( $containerCategory ) {
3877                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3878                         return true;
3879                 } else {
3880                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3881                         return false;
3882                 }
3883         }
3884
3885         /**
3886          * This function accomplishes several tasks:
3887          * 1) Auto-number headings if that option is enabled
3888          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3889          * 3) Add a Table of contents on the top for users who have enabled the option
3890          * 4) Auto-anchor headings
3891          *
3892          * It loops through all headlines, collects the necessary data, then splits up the
3893          * string and re-inserts the newly formatted headlines.
3894          *
3895          * @param $text String
3896          * @param $origText String: original, untouched wikitext
3897          * @param $isMain Boolean
3898          * @private
3899          */
3900         function formatHeadings( $text, $origText, $isMain=true ) {
3901                 global $wgMaxTocLevel, $wgHtml5, $wgExperimentalHtmlIds;
3902
3903                 # Inhibit editsection links if requested in the page
3904                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3905                         $maybeShowEditLink = $showEditLink = false;
3906                 } else {
3907                         $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
3908                         $showEditLink = $this->mOptions->getEditSection();
3909                 }
3910                 if ( $showEditLink ) {
3911                         $this->mOutput->setEditSectionTokens( true );
3912                 }
3913
3914                 # Get all headlines for numbering them and adding funky stuff like [edit]
3915                 # links - this is for later, but we need the number of headlines right now
3916                 $matches = array();
3917                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3918
3919                 # if there are fewer than 4 headlines in the article, do not show TOC
3920                 # unless it's been explicitly enabled.
3921                 $enoughToc = $this->mShowToc &&
3922                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3923
3924                 # Allow user to stipulate that a page should have a "new section"
3925                 # link added via __NEWSECTIONLINK__
3926                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3927                         $this->mOutput->setNewSection( true );
3928                 }
3929
3930                 # Allow user to remove the "new section"
3931                 # link via __NONEWSECTIONLINK__
3932                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3933                         $this->mOutput->hideNewSection( true );
3934                 }
3935
3936                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3937                 # override above conditions and always show TOC above first header
3938                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3939                         $this->mShowToc = true;
3940                         $enoughToc = true;
3941                 }
3942
3943                 # headline counter
3944                 $headlineCount = 0;
3945                 $numVisible = 0;
3946
3947                 # Ugh .. the TOC should have neat indentation levels which can be
3948                 # passed to the skin functions. These are determined here
3949                 $toc = '';
3950                 $full = '';
3951                 $head = array();
3952                 $sublevelCount = array();
3953                 $levelCount = array();
3954                 $level = 0;
3955                 $prevlevel = 0;
3956                 $toclevel = 0;
3957                 $prevtoclevel = 0;
3958                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3959                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3960                 $oldType = $this->mOutputType;
3961                 $this->setOutputType( self::OT_WIKI );
3962                 $frame = $this->getPreprocessor()->newFrame();
3963                 $root = $this->preprocessToDom( $origText );
3964                 $node = $root->getFirstChild();
3965                 $byteOffset = 0;
3966                 $tocraw = array();
3967                 $refers = array();
3968
3969                 foreach ( $matches[3] as $headline ) {
3970                         $isTemplate = false;
3971                         $titleText = false;
3972                         $sectionIndex = false;
3973                         $numbering = '';
3974                         $markerMatches = array();
3975                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3976                                 $serial = $markerMatches[1];
3977                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3978                                 $isTemplate = ( $titleText != $baseTitleText );
3979                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3980                         }
3981
3982                         if ( $toclevel ) {
3983                                 $prevlevel = $level;
3984                         }
3985                         $level = $matches[1][$headlineCount];
3986
3987                         if ( $level > $prevlevel ) {
3988                                 # Increase TOC level
3989                                 $toclevel++;
3990                                 $sublevelCount[$toclevel] = 0;
3991                                 if ( $toclevel<$wgMaxTocLevel ) {
3992                                         $prevtoclevel = $toclevel;
3993                                         $toc .= Linker::tocIndent();
3994                                         $numVisible++;
3995                                 }
3996                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3997                                 # Decrease TOC level, find level to jump to
3998
3999                                 for ( $i = $toclevel; $i > 0; $i-- ) {
4000                                         if ( $levelCount[$i] == $level ) {
4001                                                 # Found last matching level
4002                                                 $toclevel = $i;
4003                                                 break;
4004                                         } elseif ( $levelCount[$i] < $level ) {
4005                                                 # Found first matching level below current level
4006                                                 $toclevel = $i + 1;
4007                                                 break;
4008                                         }
4009                                 }
4010                                 if ( $i == 0 ) {
4011                                         $toclevel = 1;
4012                                 }
4013                                 if ( $toclevel<$wgMaxTocLevel ) {
4014                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
4015                                                 # Unindent only if the previous toc level was shown :p
4016                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4017                                                 $prevtoclevel = $toclevel;
4018                                         } else {
4019                                                 $toc .= Linker::tocLineEnd();
4020                                         }
4021                                 }
4022                         } else {
4023                                 # No change in level, end TOC line
4024                                 if ( $toclevel<$wgMaxTocLevel ) {
4025                                         $toc .= Linker::tocLineEnd();
4026                                 }
4027                         }
4028
4029                         $levelCount[$toclevel] = $level;
4030
4031                         # count number of headlines for each level
4032                         @$sublevelCount[$toclevel]++;
4033                         $dot = 0;
4034                         for( $i = 1; $i <= $toclevel; $i++ ) {
4035                                 if ( !empty( $sublevelCount[$i] ) ) {
4036                                         if ( $dot ) {
4037                                                 $numbering .= '.';
4038                                         }
4039                                         $numbering .= $this->getFunctionLang()->formatNum( $sublevelCount[$i] );
4040                                         $dot = 1;
4041                                 }
4042                         }
4043
4044                         # The safe header is a version of the header text safe to use for links
4045                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4046                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
4047
4048                         # Remove link placeholders by the link text.
4049                         #     <!--LINK number-->
4050                         # turns into
4051                         #     link text with suffix
4052                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
4053
4054                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393, or <i>: bug 26375)
4055                         $tocline = preg_replace(
4056                                 array( '#<(?!/?(sup|sub|i)).*?'.'>#', '#<(/?(sup|sub|i)).*?'.'>#' ),
4057                                 array( '',                          '<$1>' ),
4058                                 $safeHeadline
4059                         );
4060                         $tocline = trim( $tocline );
4061
4062                         # For the anchor, strip out HTML-y stuff period
4063                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4064                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4065
4066                         # Save headline for section edit hint before it's escaped
4067                         $headlineHint = $safeHeadline;
4068
4069                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4070                                 # For reverse compatibility, provide an id that's
4071                                 # HTML4-compatible, like we used to.
4072                                 #
4073                                 # It may be worth noting, academically, that it's possible for
4074                                 # the legacy anchor to conflict with a non-legacy headline
4075                                 # anchor on the page.  In this case likely the "correct" thing
4076                                 # would be to either drop the legacy anchors or make sure
4077                                 # they're numbered first.  However, this would require people
4078                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4079                                 # manually, so let's not bother worrying about it.
4080                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4081                                         array( 'noninitial', 'legacy' ) );
4082                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4083
4084                                 if ( $legacyHeadline == $safeHeadline ) {
4085                                         # No reason to have both (in fact, we can't)
4086                                         $legacyHeadline = false;
4087                                 }
4088                         } else {
4089                                 $legacyHeadline = false;
4090                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4091                                         'noninitial' );
4092                         }
4093
4094                         # HTML names must be case-insensitively unique (bug 10721).
4095                         # This does not apply to Unicode characters per
4096                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4097                         # @todo FIXME: We may be changing them depending on the current locale.
4098                         $arrayKey = strtolower( $safeHeadline );
4099                         if ( $legacyHeadline === false ) {
4100                                 $legacyArrayKey = false;
4101                         } else {
4102                                 $legacyArrayKey = strtolower( $legacyHeadline );
4103                         }
4104
4105                         # count how many in assoc. array so we can track dupes in anchors
4106                         if ( isset( $refers[$arrayKey] ) ) {
4107                                 $refers[$arrayKey]++;
4108                         } else {
4109                                 $refers[$arrayKey] = 1;
4110                         }
4111                         if ( isset( $refers[$legacyArrayKey] ) ) {
4112                                 $refers[$legacyArrayKey]++;
4113                         } else {
4114                                 $refers[$legacyArrayKey] = 1;
4115                         }
4116
4117                         # Don't number the heading if it is the only one (looks silly)
4118                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4119                                 # the two are different if the line contains a link
4120                                 $headline = $numbering . ' ' . $headline;
4121                         }
4122
4123                         # Create the anchor for linking from the TOC to the section
4124                         $anchor = $safeHeadline;
4125                         $legacyAnchor = $legacyHeadline;
4126                         if ( $refers[$arrayKey] > 1 ) {
4127                                 $anchor .= '_' . $refers[$arrayKey];
4128                         }
4129                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4130                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4131                         }
4132                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4133                                 $toc .= Linker::tocLine( $anchor, $tocline,
4134                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4135                         }
4136
4137                         # Add the section to the section tree
4138                         # Find the DOM node for this header
4139                         while ( $node && !$isTemplate ) {
4140                                 if ( $node->getName() === 'h' ) {
4141                                         $bits = $node->splitHeading();
4142                                         if ( $bits['i'] == $sectionIndex ) {
4143                                                 break;
4144                                         }
4145                                 }
4146                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4147                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4148                                 $node = $node->getNextSibling();
4149                         }
4150                         $tocraw[] = array(
4151                                 'toclevel' => $toclevel,
4152                                 'level' => $level,
4153                                 'line' => $tocline,
4154                                 'number' => $numbering,
4155                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4156                                 'fromtitle' => $titleText,
4157                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4158                                 'anchor' => $anchor,
4159                         );
4160
4161                         # give headline the correct <h#> tag
4162                         if ( $maybeShowEditLink && $sectionIndex !== false ) {
4163                                 // Output edit section links as markers with styles that can be customized by skins
4164                                 if ( $isTemplate ) {
4165                                         # Put a T flag in the section identifier, to indicate to extractSections()
4166                                         # that sections inside <includeonly> should be counted.
4167                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4168                                 } else {
4169                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4170                                 }
4171                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4172                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4173                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4174                                 // so we don't have to worry about a user trying to input one of these markers directly.
4175                                 // We use a page and section attribute to stop the language converter from converting these important bits
4176                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4177                                 // be able to convert that piece of data.
4178                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4179                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4180                                 if ( isset($editlinkArgs[2]) ) {
4181                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4182                                 } else {
4183                                         $editlink .= '/>';
4184                                 }
4185                         } else {
4186                                 $editlink = '';
4187                         }
4188                         $head[$headlineCount] = Linker::makeHeadline( $level,
4189                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4190                                 $editlink, $legacyAnchor );
4191
4192                         $headlineCount++;
4193                 }
4194
4195                 $this->setOutputType( $oldType );
4196
4197                 # Never ever show TOC if no headers
4198                 if ( $numVisible < 1 ) {
4199                         $enoughToc = false;
4200                 }
4201
4202                 if ( $enoughToc ) {
4203                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4204                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4205                         }
4206                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4207                         $this->mOutput->setTOCHTML( $toc );
4208                 }
4209
4210                 if ( $isMain ) {
4211                         $this->mOutput->setSections( $tocraw );
4212                 }
4213
4214                 # split up and insert constructed headlines
4215                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4216                 $i = 0;
4217
4218                 // build an array of document sections
4219                 $sections = array();
4220                 foreach ( $blocks as $block ) {
4221                         // $head is zero-based, sections aren't.
4222                         if ( empty( $head[$i - 1] ) ) {
4223                                 $sections[$i] = $block;
4224                         } else {
4225                                 $sections[$i] = $head[$i - 1] . $block;
4226                         }
4227
4228                         /**
4229                          * Send a hook, one per section.
4230                          * The idea here is to be able to make section-level DIVs, but to do so in a
4231                          * lower-impact, more correct way than r50769
4232                          *
4233                          * $this : caller
4234                          * $section : the section number
4235                          * &$sectionContent : ref to the content of the section
4236                          * $showEditLinks : boolean describing whether this section has an edit link
4237                          */
4238                         wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4239
4240                         $i++;
4241                 }
4242
4243                 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4244                         // append the TOC at the beginning
4245                         // Top anchor now in skin
4246                         $sections[0] = $sections[0] . $toc . "\n";
4247                 }
4248
4249                 $full .= join( '', $sections );
4250
4251                 if ( $this->mForceTocPosition ) {
4252                         return str_replace( '<!--MWTOC-->', $toc, $full );
4253                 } else {
4254                         return $full;
4255                 }
4256         }
4257
4258         /**
4259          * Transform wiki markup when saving a page by doing \r\n -> \n
4260          * conversion, substitting signatures, {{subst:}} templates, etc.
4261          *
4262          * @param $text String: the text to transform
4263          * @param $title Title: the Title object for the current article
4264          * @param $user User: the User object describing the current user
4265          * @param $options ParserOptions: parsing options
4266          * @param $clearState Boolean: whether to clear the parser state first
4267          * @return String: the altered wiki markup
4268          */
4269         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4270                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4271                 $this->setUser( $user );
4272
4273                 $pairs = array(
4274                         "\r\n" => "\n",
4275                 );
4276                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4277                 if( $options->getPreSaveTransform() ) {
4278                         $text = $this->pstPass2( $text, $user );
4279                 }
4280                 $text = $this->mStripState->unstripBoth( $text );
4281
4282                 $this->setUser( null ); #Reset
4283
4284                 return $text;
4285         }
4286
4287         /**
4288          * Pre-save transform helper function
4289          * @private
4290          *
4291          * @param $text string
4292          * @param $user User
4293          *
4294          * @return string
4295          */
4296         function pstPass2( $text, $user ) {
4297                 global $wgContLang, $wgLocaltimezone;
4298
4299                 # Note: This is the timestamp saved as hardcoded wikitext to
4300                 # the database, we use $wgContLang here in order to give
4301                 # everyone the same signature and use the default one rather
4302                 # than the one selected in each user's preferences.
4303                 # (see also bug 12815)
4304                 $ts = $this->mOptions->getTimestamp();
4305                 if ( isset( $wgLocaltimezone ) ) {
4306                         $tz = $wgLocaltimezone;
4307                 } else {
4308                         $tz = date_default_timezone_get();
4309                 }
4310
4311                 $unixts = wfTimestamp( TS_UNIX, $ts );
4312                 $oldtz = date_default_timezone_get();
4313                 date_default_timezone_set( $tz );
4314                 $ts = date( 'YmdHis', $unixts );
4315                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4316
4317                 # Allow translation of timezones through wiki. date() can return
4318                 # whatever crap the system uses, localised or not, so we cannot
4319                 # ship premade translations.
4320                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4321                 $msg = wfMessage( $key )->inContentLanguage();
4322                 if ( $msg->exists() ) {
4323                         $tzMsg = $msg->text();
4324                 }
4325
4326                 date_default_timezone_set( $oldtz );
4327
4328                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4329
4330                 # Variable replacement
4331                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4332                 $text = $this->replaceVariables( $text );
4333
4334                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4335                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4336
4337                 # Signatures
4338                 $sigText = $this->getUserSig( $user );
4339                 $text = strtr( $text, array(
4340                         '~~~~~' => $d,
4341                         '~~~~' => "$sigText $d",
4342                         '~~~' => $sigText
4343                 ) );
4344
4345                 # Context links: [[|name]] and [[name (context)|]]
4346                 global $wgLegalTitleChars;
4347                 $tc = "[$wgLegalTitleChars]";
4348                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4349
4350                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";           # [[ns:page (context)|]]
4351                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";           # [[ns:page（context）|]]
4352                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)(, $tc+|)\\|]]/"; # [[ns:page (context), context|]]
4353                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4354
4355                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4356                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4357                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4358                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4359
4360                 $t = $this->mTitle->getText();
4361                 $m = array();
4362                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4363                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4364                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4365                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4366                 } else {
4367                         # if there's no context, don't bother duplicating the title
4368                         $text = preg_replace( $p2, '[[\\1]]', $text );
4369                 }
4370
4371                 # Trim trailing whitespace
4372                 $text = rtrim( $text );
4373
4374                 return $text;
4375         }
4376
4377         /**
4378          * Fetch the user's signature text, if any, and normalize to
4379          * validated, ready-to-insert wikitext.
4380          * If you have pre-fetched the nickname or the fancySig option, you can
4381          * specify them here to save a database query.
4382          * Do not reuse this parser instance after calling getUserSig(),
4383          * as it may have changed if it's the $wgParser.
4384          *
4385          * @param $user User
4386          * @param $nickname String|bool nickname to use or false to use user's default nickname
4387          * @param $fancySig Boolean|null whether the nicknname is the complete signature
4388          *                  or null to use default value
4389          * @return string
4390          */
4391         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4392                 global $wgMaxSigChars;
4393
4394                 $username = $user->getName();
4395
4396                 # If not given, retrieve from the user object.
4397                 if ( $nickname === false )
4398                         $nickname = $user->getOption( 'nickname' );
4399
4400                 if ( is_null( $fancySig ) ) {
4401                         $fancySig = $user->getBoolOption( 'fancysig' );
4402                 }
4403
4404                 $nickname = $nickname == null ? $username : $nickname;
4405
4406                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4407                         $nickname = $username;
4408                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4409                 } elseif ( $fancySig !== false ) {
4410                         # Sig. might contain markup; validate this
4411                         if ( $this->validateSig( $nickname ) !== false ) {
4412                                 # Validated; clean up (if needed) and return it
4413                                 return $this->cleanSig( $nickname, true );
4414                         } else {
4415                                 # Failed to validate; fall back to the default
4416                                 $nickname = $username;
4417                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4418                         }
4419                 }
4420
4421                 # Make sure nickname doesnt get a sig in a sig
4422                 $nickname = $this->cleanSigInSig( $nickname );
4423
4424                 # If we're still here, make it a link to the user page
4425                 $userText = wfEscapeWikiText( $username );
4426                 $nickText = wfEscapeWikiText( $nickname );
4427                 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4428
4429                 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()->title( $this->getTitle() )->text();
4430         }
4431
4432         /**
4433          * Check that the user's signature contains no bad XML
4434          *
4435          * @param $text String
4436          * @return mixed An expanded string, or false if invalid.
4437          */
4438         function validateSig( $text ) {
4439                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4440         }
4441
4442         /**
4443          * Clean up signature text
4444          *
4445          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4446          * 2) Substitute all transclusions
4447          *
4448          * @param $text String
4449          * @param $parsing bool Whether we're cleaning (preferences save) or parsing
4450          * @return String: signature text
4451          */
4452         function cleanSig( $text, $parsing = false ) {
4453                 if ( !$parsing ) {
4454                         global $wgTitle;
4455                         $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4456                 }
4457
4458                 # Option to disable this feature
4459                 if ( !$this->mOptions->getCleanSignatures() ) {
4460                         return $text;
4461                 }
4462
4463                 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4464                 #  => Move this logic to braceSubstitution()
4465                 $substWord = MagicWord::get( 'subst' );
4466                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4467                 $substText = '{{' . $substWord->getSynonym( 0 );
4468
4469                 $text = preg_replace( $substRegex, $substText, $text );
4470                 $text = $this->cleanSigInSig( $text );
4471                 $dom = $this->preprocessToDom( $text );
4472                 $frame = $this->getPreprocessor()->newFrame();
4473                 $text = $frame->expand( $dom );
4474
4475                 if ( !$parsing ) {
4476                         $text = $this->mStripState->unstripBoth( $text );
4477                 }
4478
4479                 return $text;
4480         }
4481
4482         /**
4483          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4484          *
4485          * @param $text String
4486          * @return String: signature text with /~{3,5}/ removed
4487          */
4488         function cleanSigInSig( $text ) {
4489                 $text = preg_replace( '/~{3,5}/', '', $text );
4490                 return $text;
4491         }
4492
4493         /**
4494          * Set up some variables which are usually set up in parse()
4495          * so that an external function can call some class members with confidence
4496          *
4497          * @param $title Title|null
4498          * @param $options ParserOptions
4499          * @param $outputType
4500          * @param $clearState bool
4501          */
4502         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4503                 $this->startParse( $title, $options, $outputType, $clearState );
4504         }
4505
4506         /**
4507          * @param $title Title|null
4508          * @param $options ParserOptions
4509          * @param $outputType
4510          * @param $clearState bool
4511          */
4512         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4513                 $this->setTitle( $title );
4514                 $this->mOptions = $options;
4515                 $this->setOutputType( $outputType );
4516                 if ( $clearState ) {
4517                         $this->clearState();
4518                 }
4519         }
4520
4521         /**
4522          * Wrapper for preprocess()
4523          *
4524          * @param $text String: the text to preprocess
4525          * @param $options ParserOptions: options
4526          * @param $title Title object or null to use $wgTitle
4527          * @return String
4528          */
4529         public function transformMsg( $text, $options, $title = null ) {
4530                 static $executing = false;
4531
4532                 # Guard against infinite recursion
4533                 if ( $executing ) {
4534                         return $text;
4535                 }
4536                 $executing = true;
4537
4538                 wfProfileIn( __METHOD__ );
4539                 if ( !$title ) {
4540                         global $wgTitle;
4541                         $title = $wgTitle;
4542                 }
4543                 if ( !$title ) {
4544                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4545                         # Create a ghost title in such case
4546                         $title = Title::newFromText( 'Dwimmerlaik' );
4547                 }
4548                 $text = $this->preprocess( $text, $title, $options );
4549
4550                 $executing = false;
4551                 wfProfileOut( __METHOD__ );
4552                 return $text;
4553         }
4554
4555         /**
4556          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4557          * The callback should have the following form:
4558          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4559          *
4560          * Transform and return $text. Use $parser for any required context, e.g. use
4561          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4562          *
4563          * Hooks may return extended information by returning an array, of which the
4564          * first numbered element (index 0) must be the return string, and all other
4565          * entries are extracted into local variables within an internal function
4566          * in the Parser class.
4567          *
4568          * This interface (introduced r61913) appears to be undocumented, but
4569          * 'markerName' is used by some core tag hooks to override which strip
4570          * array their results are placed in. **Use great caution if attempting
4571          * this interface, as it is not documented and injudicious use could smash
4572          * private variables.**
4573          *
4574          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4575          * @param $callback Mixed: the callback function (and object) to use for the tag
4576          * @return The old value of the mTagHooks array associated with the hook
4577          */
4578         public function setHook( $tag, $callback ) {
4579                 $tag = strtolower( $tag );
4580                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4581                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4582                 $this->mTagHooks[$tag] = $callback;
4583                 if ( !in_array( $tag, $this->mStripList ) ) {
4584                         $this->mStripList[] = $tag;
4585                 }
4586
4587                 return $oldVal;
4588         }
4589
4590         /**
4591          * As setHook(), but letting the contents be parsed.
4592          *
4593          * Transparent tag hooks are like regular XML-style tag hooks, except they
4594          * operate late in the transformation sequence, on HTML instead of wikitext.
4595          *
4596          * This is probably obsoleted by things dealing with parser frames?
4597          * The only extension currently using it is geoserver.
4598          *
4599          * @since 1.10
4600          * @todo better document or deprecate this
4601          *
4602          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4603          * @param $callback Mixed: the callback function (and object) to use for the tag
4604          * @return The old value of the mTagHooks array associated with the hook
4605          */
4606         function setTransparentTagHook( $tag, $callback ) {
4607                 $tag = strtolower( $tag );
4608                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4609                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4610                 $this->mTransparentTagHooks[$tag] = $callback;
4611
4612                 return $oldVal;
4613         }
4614
4615         /**
4616          * Remove all tag hooks
4617          */
4618         function clearTagHooks() {
4619                 $this->mTagHooks = array();
4620                 $this->mStripList = $this->mDefaultStripList;
4621         }
4622
4623         /**
4624          * Create a function, e.g. {{sum:1|2|3}}
4625          * The callback function should have the form:
4626          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4627          *
4628          * Or with SFH_OBJECT_ARGS:
4629          *    function myParserFunction( $parser, $frame, $args ) { ... }
4630          *
4631          * The callback may either return the text result of the function, or an array with the text
4632          * in element 0, and a number of flags in the other elements. The names of the flags are
4633          * specified in the keys. Valid flags are:
4634          *   found                     The text returned is valid, stop processing the template. This
4635          *                             is on by default.
4636          *   nowiki                    Wiki markup in the return value should be escaped
4637          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4638          *
4639          * @param $id String: The magic word ID
4640          * @param $callback Mixed: the callback function (and object) to use
4641          * @param $flags Integer: a combination of the following flags:
4642          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4643          *
4644          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4645          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4646          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4647          *     the arguments, and to control the way they are expanded.
4648          *
4649          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4650          *     arguments, for instance:
4651          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4652          *
4653          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4654          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4655          *     working if/when this is changed.
4656          *
4657          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4658          *     expansion.
4659          *
4660          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4661          *     about the methods available in PPFrame and PPNode.
4662          *
4663          * @return The old callback function for this name, if any
4664          */
4665         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4666                 global $wgContLang;
4667
4668                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4669                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4670
4671                 # Add to function cache
4672                 $mw = MagicWord::get( $id );
4673                 if ( !$mw )
4674                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4675
4676                 $synonyms = $mw->getSynonyms();
4677                 $sensitive = intval( $mw->isCaseSensitive() );
4678
4679                 foreach ( $synonyms as $syn ) {
4680                         # Case
4681                         if ( !$sensitive ) {
4682                                 $syn = $wgContLang->lc( $syn );
4683                         }
4684                         # Add leading hash
4685                         if ( !( $flags & SFH_NO_HASH ) ) {
4686                                 $syn = '#' . $syn;
4687                         }
4688                         # Remove trailing colon
4689                         if ( substr( $syn, -1, 1 ) === ':' ) {
4690                                 $syn = substr( $syn, 0, -1 );
4691                         }
4692                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4693                 }
4694                 return $oldVal;
4695         }
4696
4697         /**
4698          * Get all registered function hook identifiers
4699          *
4700          * @return Array
4701          */
4702         function getFunctionHooks() {
4703                 return array_keys( $this->mFunctionHooks );
4704         }
4705
4706         /**
4707          * Create a tag function, e.g. <test>some stuff</test>.
4708          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4709          * Unlike parser functions, their content is not preprocessed.
4710          */
4711         function setFunctionTagHook( $tag, $callback, $flags ) {
4712                 $tag = strtolower( $tag );
4713                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4714                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4715                         $this->mFunctionTagHooks[$tag] : null;
4716                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4717
4718                 if ( !in_array( $tag, $this->mStripList ) ) {
4719                         $this->mStripList[] = $tag;
4720                 }
4721
4722                 return $old;
4723         }
4724
4725         /**
4726          * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
4727          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4728          * Placeholders created in Skin::makeLinkObj()
4729          *
4730          * @param $text string
4731          * @param $options int
4732          *
4733          * @return array of link CSS classes, indexed by PDBK.
4734          */
4735         function replaceLinkHolders( &$text, $options = 0 ) {
4736                 return $this->mLinkHolders->replace( $text );
4737         }
4738
4739         /**
4740          * Replace <!--LINK--> link placeholders with plain text of links
4741          * (not HTML-formatted).
4742          *
4743          * @param $text String
4744          * @return String
4745          */
4746         function replaceLinkHoldersText( $text ) {
4747                 return $this->mLinkHolders->replaceText( $text );
4748         }
4749
4750         /**
4751          * Renders an image gallery from a text with one line per image.
4752          * text labels may be given by using |-style alternative text. E.g.
4753          *   Image:one.jpg|The number "1"
4754          *   Image:tree.jpg|A tree
4755          * given as text will return the HTML of a gallery with two images,
4756          * labeled 'The number "1"' and
4757          * 'A tree'.
4758          *
4759          * @param string $text
4760          * @param array $params
4761          * @return string HTML
4762          */
4763         function renderImageGallery( $text, $params ) {
4764                 $ig = new ImageGallery();
4765                 $ig->setContextTitle( $this->mTitle );
4766                 $ig->setShowBytes( false );
4767                 $ig->setShowFilename( false );
4768                 $ig->setParser( $this );
4769                 $ig->setHideBadImages();
4770                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4771
4772                 if ( isset( $params['showfilename'] ) ) {
4773                         $ig->setShowFilename( true );
4774                 } else {
4775                         $ig->setShowFilename( false );
4776                 }
4777                 if ( isset( $params['caption'] ) ) {
4778                         $caption = $params['caption'];
4779                         $caption = htmlspecialchars( $caption );
4780                         $caption = $this->replaceInternalLinks( $caption );
4781                         $ig->setCaptionHtml( $caption );
4782                 }
4783                 if ( isset( $params['perrow'] ) ) {
4784                         $ig->setPerRow( $params['perrow'] );
4785                 }
4786                 if ( isset( $params['widths'] ) ) {
4787                         $ig->setWidths( $params['widths'] );
4788                 }
4789                 if ( isset( $params['heights'] ) ) {
4790                         $ig->setHeights( $params['heights'] );
4791                 }
4792
4793                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4794
4795                 $lines = StringUtils::explode( "\n", $text );
4796                 foreach ( $lines as $line ) {
4797                         # match lines like these:
4798                         # Image:someimage.jpg|This is some image
4799                         $matches = array();
4800                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4801                         # Skip empty lines
4802                         if ( count( $matches ) == 0 ) {
4803                                 continue;
4804                         }
4805
4806                         if ( strpos( $matches[0], '%' ) !== false ) {
4807                                 $matches[1] = rawurldecode( $matches[1] );
4808                         }
4809                         $title = Title::newFromText( $matches[1], NS_FILE );
4810                         if ( is_null( $title ) ) {
4811                                 # Bogus title. Ignore these so we don't bomb out later.
4812                                 continue;
4813                         }
4814
4815                         $label = '';
4816                         $alt = '';
4817                         if ( isset( $matches[3] ) ) {
4818                                 // look for an |alt= definition while trying not to break existing
4819                                 // captions with multiple pipes (|) in it, until a more sensible grammar
4820                                 // is defined for images in galleries
4821
4822                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4823                                 $altmatches = StringUtils::explode('|', $matches[3]);
4824                                 $magicWordAlt = MagicWord::get( 'img_alt' );
4825
4826                                 foreach ( $altmatches as $altmatch ) {
4827                                         $match = $magicWordAlt->matchVariableStartToEnd( $altmatch );
4828                                         if ( $match ) {
4829                                                 $alt = $this->stripAltText( $match, false );
4830                                         }
4831                                         else {
4832                                                 // concatenate all other pipes
4833                                                 $label .= '|' . $altmatch;
4834                                         }
4835                                 }
4836                                 // remove the first pipe
4837                                 $label = substr( $label, 1 );
4838                         }
4839
4840                         $ig->add( $title, $label, $alt );
4841                 }
4842                 return $ig->toHTML();
4843         }
4844
4845         /**
4846          * @param $handler
4847          * @return array
4848          */
4849         function getImageParams( $handler ) {
4850                 if ( $handler ) {
4851                         $handlerClass = get_class( $handler );
4852                 } else {
4853                         $handlerClass = '';
4854                 }
4855                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4856                         # Initialise static lists
4857                         static $internalParamNames = array(
4858                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4859                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4860                                         'bottom', 'text-bottom' ),
4861                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4862                                         'upright', 'border', 'link', 'alt' ),
4863                         );
4864                         static $internalParamMap;
4865                         if ( !$internalParamMap ) {
4866                                 $internalParamMap = array();
4867                                 foreach ( $internalParamNames as $type => $names ) {
4868                                         foreach ( $names as $name ) {
4869                                                 $magicName = str_replace( '-', '_', "img_$name" );
4870                                                 $internalParamMap[$magicName] = array( $type, $name );
4871                                         }
4872                                 }
4873                         }
4874
4875                         # Add handler params
4876                         $paramMap = $internalParamMap;
4877                         if ( $handler ) {
4878                                 $handlerParamMap = $handler->getParamMap();
4879                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4880                                         $paramMap[$magic] = array( 'handler', $paramName );
4881                                 }
4882                         }
4883                         $this->mImageParams[$handlerClass] = $paramMap;
4884                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4885                 }
4886                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4887         }
4888
4889         /**
4890          * Parse image options text and use it to make an image
4891          *
4892          * @param $title Title
4893          * @param $options String
4894          * @param $holders LinkHolderArray|false
4895          * @return string HTML
4896          */
4897         function makeImage( $title, $options, $holders = false ) {
4898                 # Check if the options text is of the form "options|alt text"
4899                 # Options are:
4900                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4901                 #  * left       no resizing, just left align. label is used for alt= only
4902                 #  * right      same, but right aligned
4903                 #  * none       same, but not aligned
4904                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4905                 #  * center     center the image
4906                 #  * frame      Keep original image size, no magnify-button.
4907                 #  * framed     Same as "frame"
4908                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4909                 #  * upright    reduce width for upright images, rounded to full __0 px
4910                 #  * border     draw a 1px border around the image
4911                 #  * alt        Text for HTML alt attribute (defaults to empty)
4912                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4913                 # vertical-align values (no % or length right now):
4914                 #  * baseline
4915                 #  * sub
4916                 #  * super
4917                 #  * top
4918                 #  * text-top
4919                 #  * middle
4920                 #  * bottom
4921                 #  * text-bottom
4922
4923                 $parts = StringUtils::explode( "|", $options );
4924
4925                 # Give extensions a chance to select the file revision for us
4926                 $options = array();
4927                 $descQuery = false;
4928                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
4929                         array( $this, $title, &$options, &$descQuery ) );
4930                 # Fetch and register the file (file title may be different via hooks)
4931                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
4932
4933                 # Get parameter map
4934                 $handler = $file ? $file->getHandler() : false;
4935
4936                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4937
4938                 if ( !$file ) {
4939                         $this->addTrackingCategory( 'broken-file-category' );
4940                 }
4941
4942                 # Process the input parameters
4943                 $caption = '';
4944                 $params = array( 'frame' => array(), 'handler' => array(),
4945                         'horizAlign' => array(), 'vertAlign' => array() );
4946                 foreach ( $parts as $part ) {
4947                         $part = trim( $part );
4948                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4949                         $validated = false;
4950                         if ( isset( $paramMap[$magicName] ) ) {
4951                                 list( $type, $paramName ) = $paramMap[$magicName];
4952
4953                                 # Special case; width and height come in one variable together
4954                                 if ( $type === 'handler' && $paramName === 'width' ) {
4955                                         $m = array();
4956                                         # (bug 13500) In both cases (width/height and width only),
4957                                         # permit trailing "px" for backward compatibility.
4958                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4959                                                 $width = intval( $m[1] );
4960                                                 $height = intval( $m[2] );
4961                                                 if ( $handler->validateParam( 'width', $width ) ) {
4962                                                         $params[$type]['width'] = $width;
4963                                                         $validated = true;
4964                                                 }
4965                                                 if ( $handler->validateParam( 'height', $height ) ) {
4966                                                         $params[$type]['height'] = $height;
4967                                                         $validated = true;
4968                                                 }
4969                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4970                                                 $width = intval( $value );
4971                                                 if ( $handler->validateParam( 'width', $width ) ) {
4972                                                         $params[$type]['width'] = $width;
4973                                                         $validated = true;
4974                                                 }
4975                                         } # else no validation -- bug 13436
4976                                 } else {
4977                                         if ( $type === 'handler' ) {
4978                                                 # Validate handler parameter
4979                                                 $validated = $handler->validateParam( $paramName, $value );
4980                                         } else {
4981                                                 # Validate internal parameters
4982                                                 switch( $paramName ) {
4983                                                 case 'manualthumb':
4984                                                 case 'alt':
4985                                                         # @todo FIXME: Possibly check validity here for
4986                                                         # manualthumb? downstream behavior seems odd with
4987                                                         # missing manual thumbs.
4988                                                         $validated = true;
4989                                                         $value = $this->stripAltText( $value, $holders );
4990                                                         break;
4991                                                 case 'link':
4992                                                         $chars = self::EXT_LINK_URL_CLASS;
4993                                                         $prots = $this->mUrlProtocols;
4994                                                         if ( $value === '' ) {
4995                                                                 $paramName = 'no-link';
4996                                                                 $value = true;
4997                                                                 $validated = true;
4998                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4999                                                                 if ( preg_match( "/^($prots)$chars+$/u", $value, $m ) ) {
5000                                                                         $paramName = 'link-url';
5001                                                                         $this->mOutput->addExternalLink( $value );
5002                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
5003                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5004                                                                         }
5005                                                                         $validated = true;
5006                                                                 }
5007                                                         } else {
5008                                                                 $linkTitle = Title::newFromText( $value );
5009                                                                 if ( $linkTitle ) {
5010                                                                         $paramName = 'link-title';
5011                                                                         $value = $linkTitle;
5012                                                                         $this->mOutput->addLink( $linkTitle );
5013                                                                         $validated = true;
5014                                                                 }
5015                                                         }
5016                                                         break;
5017                                                 default:
5018                                                         # Most other things appear to be empty or numeric...
5019                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
5020                                                 }
5021                                         }
5022
5023                                         if ( $validated ) {
5024                                                 $params[$type][$paramName] = $value;
5025                                         }
5026                                 }
5027                         }
5028                         if ( !$validated ) {
5029                                 $caption = $part;
5030                         }
5031                 }
5032
5033                 # Process alignment parameters
5034                 if ( $params['horizAlign'] ) {
5035                         $params['frame']['align'] = key( $params['horizAlign'] );
5036                 }
5037                 if ( $params['vertAlign'] ) {
5038                         $params['frame']['valign'] = key( $params['vertAlign'] );
5039                 }
5040
5041                 $params['frame']['caption'] = $caption;
5042
5043                 # Will the image be presented in a frame, with the caption below?
5044                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5045                                                  isset( $params['frame']['framed'] ) ||
5046                                                  isset( $params['frame']['thumbnail'] ) ||
5047                                                  isset( $params['frame']['manualthumb'] );
5048
5049                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
5050                 # came to also set the caption, ordinary text after the image -- which
5051                 # makes no sense, because that just repeats the text multiple times in
5052                 # screen readers.  It *also* came to set the title attribute.
5053                 #
5054                 # Now that we have an alt attribute, we should not set the alt text to
5055                 # equal the caption: that's worse than useless, it just repeats the
5056                 # text.  This is the framed/thumbnail case.  If there's no caption, we
5057                 # use the unnamed parameter for alt text as well, just for the time be-
5058                 # ing, if the unnamed param is set and the alt param is not.
5059                 #
5060                 # For the future, we need to figure out if we want to tweak this more,
5061                 # e.g., introducing a title= parameter for the title; ignoring the un-
5062                 # named parameter entirely for images without a caption; adding an ex-
5063                 # plicit caption= parameter and preserving the old magic unnamed para-
5064                 # meter for BC; ...
5065                 if ( $imageIsFramed ) { # Framed image
5066                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5067                                 # No caption or alt text, add the filename as the alt text so
5068                                 # that screen readers at least get some description of the image
5069                                 $params['frame']['alt'] = $title->getText();
5070                         }
5071                         # Do not set $params['frame']['title'] because tooltips don't make sense
5072                         # for framed images
5073                 } else { # Inline image
5074                         if ( !isset( $params['frame']['alt'] ) ) {
5075                                 # No alt text, use the "caption" for the alt text
5076                                 if ( $caption !== '') {
5077                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5078                                 } else {
5079                                         # No caption, fall back to using the filename for the
5080                                         # alt text
5081                                         $params['frame']['alt'] = $title->getText();
5082                                 }
5083                         }
5084                         # Use the "caption" for the tooltip text
5085                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5086                 }
5087
5088                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
5089
5090                 # Linker does the rest
5091                 $time = isset( $options['time'] ) ? $options['time'] : false;
5092                 $ret = Linker::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
5093                         $time, $descQuery, $this->mOptions->getThumbSize() );
5094
5095                 # Give the handler a chance to modify the parser object
5096                 if ( $handler ) {
5097                         $handler->parserTransformHook( $this, $file );
5098                 }
5099
5100                 return $ret;
5101         }
5102
5103         /**
5104          * @param $caption
5105          * @param $holders LinkHolderArray
5106          * @return mixed|String
5107          */
5108         protected function stripAltText( $caption, $holders ) {
5109                 # Strip bad stuff out of the title (tooltip).  We can't just use
5110                 # replaceLinkHoldersText() here, because if this function is called
5111                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5112                 if ( $holders ) {
5113                         $tooltip = $holders->replaceText( $caption );
5114                 } else {
5115                         $tooltip = $this->replaceLinkHoldersText( $caption );
5116                 }
5117
5118                 # make sure there are no placeholders in thumbnail attributes
5119                 # that are later expanded to html- so expand them now and
5120                 # remove the tags
5121                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5122                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5123
5124                 return $tooltip;
5125         }
5126
5127         /**
5128          * Set a flag in the output object indicating that the content is dynamic and
5129          * shouldn't be cached.
5130          */
5131         function disableCache() {
5132                 wfDebug( "Parser output marked as uncacheable.\n" );
5133                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5134                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5135         }
5136
5137         /**
5138          * Callback from the Sanitizer for expanding items found in HTML attribute
5139          * values, so they can be safely tested and escaped.
5140          *
5141          * @param $text String
5142          * @param $frame PPFrame
5143          * @return String
5144          */
5145         function attributeStripCallback( &$text, $frame = false ) {
5146                 $text = $this->replaceVariables( $text, $frame );
5147                 $text = $this->mStripState->unstripBoth( $text );
5148                 return $text;
5149         }
5150
5151         /**
5152          * Accessor
5153          *
5154          * @return array
5155          */
5156         function getTags() {
5157                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
5158         }
5159
5160         /**
5161          * Replace transparent tags in $text with the values given by the callbacks.
5162          *
5163          * Transparent tag hooks are like regular XML-style tag hooks, except they
5164          * operate late in the transformation sequence, on HTML instead of wikitext.
5165          *
5166          * @param $text string
5167          *
5168          * @return string
5169          */
5170         function replaceTransparentTags( $text ) {
5171                 $matches = array();
5172                 $elements = array_keys( $this->mTransparentTagHooks );
5173                 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5174                 $replacements = array();
5175
5176                 foreach ( $matches as $marker => $data ) {
5177                         list( $element, $content, $params, $tag ) = $data;
5178                         $tagName = strtolower( $element );
5179                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5180                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5181                         } else {
5182                                 $output = $tag;
5183                         }
5184                         $replacements[$marker] = $output;
5185                 }
5186                 return strtr( $text, $replacements );
5187         }
5188
5189         /**
5190          * Break wikitext input into sections, and either pull or replace
5191          * some particular section's text.
5192          *
5193          * External callers should use the getSection and replaceSection methods.
5194          *
5195          * @param $text String: Page wikitext
5196          * @param $section String: a section identifier string of the form:
5197          *   <flag1> - <flag2> - ... - <section number>
5198          *
5199          * Currently the only recognised flag is "T", which means the target section number
5200          * was derived during a template inclusion parse, in other words this is a template
5201          * section edit link. If no flags are given, it was an ordinary section edit link.
5202          * This flag is required to avoid a section numbering mismatch when a section is
5203          * enclosed by <includeonly> (bug 6563).
5204          *
5205          * The section number 0 pulls the text before the first heading; other numbers will
5206          * pull the given section along with its lower-level subsections. If the section is
5207          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5208          *
5209          * Section 0 is always considered to exist, even if it only contains the empty
5210          * string. If $text is the empty string and section 0 is replaced, $newText is
5211          * returned.
5212          *
5213          * @param $mode String: one of "get" or "replace"
5214          * @param $newText String: replacement text for section data.
5215          * @return String: for "get", the extracted section text.
5216          *                 for "replace", the whole page with the section replaced.
5217          */
5218         private function extractSections( $text, $section, $mode, $newText='' ) {
5219                 global $wgTitle; # not generally used but removes an ugly failure mode
5220                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5221                 $outText = '';
5222                 $frame = $this->getPreprocessor()->newFrame();
5223
5224                 # Process section extraction flags
5225                 $flags = 0;
5226                 $sectionParts = explode( '-', $section );
5227                 $sectionIndex = array_pop( $sectionParts );
5228                 foreach ( $sectionParts as $part ) {
5229                         if ( $part === 'T' ) {
5230                                 $flags |= self::PTD_FOR_INCLUSION;
5231                         }
5232                 }
5233
5234                 # Check for empty input
5235                 if ( strval( $text ) === '' ) {
5236                         # Only sections 0 and T-0 exist in an empty document
5237                         if ( $sectionIndex == 0 ) {
5238                                 if ( $mode === 'get' ) {
5239                                         return '';
5240                                 } else {
5241                                         return $newText;
5242                                 }
5243                         } else {
5244                                 if ( $mode === 'get' ) {
5245                                         return $newText;
5246                                 } else {
5247                                         return $text;
5248                                 }
5249                         }
5250                 }
5251
5252                 # Preprocess the text
5253                 $root = $this->preprocessToDom( $text, $flags );
5254
5255                 # <h> nodes indicate section breaks
5256                 # They can only occur at the top level, so we can find them by iterating the root's children
5257                 $node = $root->getFirstChild();
5258
5259                 # Find the target section
5260                 if ( $sectionIndex == 0 ) {
5261                         # Section zero doesn't nest, level=big
5262                         $targetLevel = 1000;
5263                 } else {
5264                         while ( $node ) {
5265                                 if ( $node->getName() === 'h' ) {
5266                                         $bits = $node->splitHeading();
5267                                         if ( $bits['i'] == $sectionIndex ) {
5268                                                 $targetLevel = $bits['level'];
5269                                                 break;
5270                                         }
5271                                 }
5272                                 if ( $mode === 'replace' ) {
5273                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5274                                 }
5275                                 $node = $node->getNextSibling();
5276                         }
5277                 }
5278
5279                 if ( !$node ) {
5280                         # Not found
5281                         if ( $mode === 'get' ) {
5282                                 return $newText;
5283                         } else {
5284                                 return $text;
5285                         }
5286                 }
5287
5288                 # Find the end of the section, including nested sections
5289                 do {
5290                         if ( $node->getName() === 'h' ) {
5291                                 $bits = $node->splitHeading();
5292                                 $curLevel = $bits['level'];
5293                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5294                                         break;
5295                                 }
5296                         }
5297                         if ( $mode === 'get' ) {
5298                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5299                         }
5300                         $node = $node->getNextSibling();
5301                 } while ( $node );
5302
5303                 # Write out the remainder (in replace mode only)
5304                 if ( $mode === 'replace' ) {
5305                         # Output the replacement text
5306                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5307                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5308                         # Only add trailing whitespace if there is newText
5309                         if ( $newText != "" ) {
5310                                 $outText .= $newText . "\n\n";
5311                         }
5312
5313                         while ( $node ) {
5314                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5315                                 $node = $node->getNextSibling();
5316                         }
5317                 }
5318
5319                 if ( is_string( $outText ) ) {
5320                         # Re-insert stripped tags
5321                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5322                 }
5323
5324                 return $outText;
5325         }
5326
5327         /**
5328          * This function returns the text of a section, specified by a number ($section).
5329          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5330          * the first section before any such heading (section 0).
5331          *
5332          * If a section contains subsections, these are also returned.
5333          *
5334          * @param $text String: text to look in
5335          * @param $section String: section identifier
5336          * @param $deftext String: default to return if section is not found
5337          * @return string text of the requested section
5338          */
5339         public function getSection( $text, $section, $deftext='' ) {
5340                 return $this->extractSections( $text, $section, "get", $deftext );
5341         }
5342
5343         /**
5344          * This function returns $oldtext after the content of the section
5345          * specified by $section has been replaced with $text. If the target
5346          * section does not exist, $oldtext is returned unchanged.
5347          *
5348          * @param $oldtext String: former text of the article
5349          * @param $section Numeric: section identifier
5350          * @param $text String: replacing text
5351          * @return String: modified text
5352          */
5353         public function replaceSection( $oldtext, $section, $text ) {
5354                 return $this->extractSections( $oldtext, $section, "replace", $text );
5355         }
5356
5357         /**
5358          * Get the ID of the revision we are parsing
5359          *
5360          * @return Mixed: integer or null
5361          */
5362         function getRevisionId() {
5363                 return $this->mRevisionId;
5364         }
5365
5366         /**
5367          * Get the revision object for $this->mRevisionId
5368          *
5369          * @return Revision|null either a Revision object or null
5370          */
5371         protected function getRevisionObject() {
5372                 if ( !is_null( $this->mRevisionObject ) ) {
5373                         return $this->mRevisionObject;
5374                 }
5375                 if ( is_null( $this->mRevisionId ) ) {
5376                         return null;
5377                 }
5378
5379                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5380                 return $this->mRevisionObject;
5381         }
5382
5383         /**
5384          * Get the timestamp associated with the current revision, adjusted for
5385          * the default server-local timestamp
5386          */
5387         function getRevisionTimestamp() {
5388                 if ( is_null( $this->mRevisionTimestamp ) ) {
5389                         wfProfileIn( __METHOD__ );
5390
5391                         global $wgContLang;
5392
5393                         $revObject = $this->getRevisionObject();
5394                         $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5395
5396                         # The cryptic '' timezone parameter tells to use the site-default
5397                         # timezone offset instead of the user settings.
5398                         #
5399                         # Since this value will be saved into the parser cache, served
5400                         # to other users, and potentially even used inside links and such,
5401                         # it needs to be consistent for all visitors.
5402                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5403
5404                         wfProfileOut( __METHOD__ );
5405                 }
5406                 return $this->mRevisionTimestamp;
5407         }
5408
5409         /**
5410          * Get the name of the user that edited the last revision
5411          *
5412          * @return String: user name
5413          */
5414         function getRevisionUser() {
5415                 if( is_null( $this->mRevisionUser ) ) {
5416                         $revObject = $this->getRevisionObject();
5417
5418                         # if this template is subst: the revision id will be blank,
5419                         # so just use the current user's name
5420                         if( $revObject ) {
5421                                 $this->mRevisionUser = $revObject->getUserText();
5422                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5423                                 $this->mRevisionUser = $this->getUser()->getName();
5424                         }
5425                 }
5426                 return $this->mRevisionUser;
5427         }
5428
5429         /**
5430          * Mutator for $mDefaultSort
5431          *
5432          * @param $sort New value
5433          */
5434         public function setDefaultSort( $sort ) {
5435                 $this->mDefaultSort = $sort;
5436                 $this->mOutput->setProperty( 'defaultsort', $sort );
5437         }
5438
5439         /**
5440          * Accessor for $mDefaultSort
5441          * Will use the empty string if none is set.
5442          *
5443          * This value is treated as a prefix, so the
5444          * empty string is equivalent to sorting by
5445          * page name.
5446          *
5447          * @return string
5448          */
5449         public function getDefaultSort() {
5450                 if ( $this->mDefaultSort !== false ) {
5451                         return $this->mDefaultSort;
5452                 } else {
5453                         return '';
5454                 }
5455         }
5456
5457         /**
5458          * Accessor for $mDefaultSort
5459          * Unlike getDefaultSort(), will return false if none is set
5460          *
5461          * @return string or false
5462          */
5463         public function getCustomDefaultSort() {
5464                 return $this->mDefaultSort;
5465         }
5466
5467         /**
5468          * Try to guess the section anchor name based on a wikitext fragment
5469          * presumably extracted from a heading, for example "Header" from
5470          * "== Header ==".
5471          *
5472          * @param $text string
5473          *
5474          * @return string
5475          */
5476         public function guessSectionNameFromWikiText( $text ) {
5477                 # Strip out wikitext links(they break the anchor)
5478                 $text = $this->stripSectionName( $text );
5479                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5480                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5481         }
5482
5483         /**
5484          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5485          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5486          * as something other than UTF-8 (apparently?), resulting in breakage.
5487          *
5488          * @param $text String: The section name
5489          * @return string An anchor
5490          */
5491         public function guessLegacySectionNameFromWikiText( $text ) {
5492                 # Strip out wikitext links(they break the anchor)
5493                 $text = $this->stripSectionName( $text );
5494                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5495                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5496         }
5497
5498         /**
5499          * Strips a text string of wikitext for use in a section anchor
5500          *
5501          * Accepts a text string and then removes all wikitext from the
5502          * string and leaves only the resultant text (i.e. the result of
5503          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5504          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5505          * to create valid section anchors by mimicing the output of the
5506          * parser when headings are parsed.
5507          *
5508          * @param $text String: text string to be stripped of wikitext
5509          * for use in a Section anchor
5510          * @return Filtered text string
5511          */
5512         public function stripSectionName( $text ) {
5513                 # Strip internal link markup
5514                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5515                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5516
5517                 # Strip external link markup
5518                 # @todo FIXME: Not tolerant to blank link text
5519                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5520                 # on how many empty links there are on the page - need to figure that out.
5521                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5522
5523                 # Parse wikitext quotes (italics & bold)
5524                 $text = $this->doQuotes( $text );
5525
5526                 # Strip HTML tags
5527                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5528                 return $text;
5529         }
5530
5531         /**
5532          * strip/replaceVariables/unstrip for preprocessor regression testing
5533          *
5534          * @param $text string
5535          * @param $title Title
5536          * @param $options ParserOptions
5537          * @param $outputType int
5538          *
5539          * @return string
5540          */
5541         function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5542                 $this->startParse( $title, $options, $outputType, true );
5543
5544                 $text = $this->replaceVariables( $text );
5545                 $text = $this->mStripState->unstripBoth( $text );
5546                 $text = Sanitizer::removeHTMLtags( $text );
5547                 return $text;
5548         }
5549
5550         /**
5551          * @param $text string
5552          * @param $title Title
5553          * @param $options ParserOptions
5554          * @return string
5555          */
5556         function testPst( $text, Title $title, ParserOptions $options ) {
5557                 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5558         }
5559
5560         /**
5561          * @param $text
5562          * @param $title Title
5563          * @param $options ParserOptions
5564          * @return string
5565          */
5566         function testPreprocess( $text, Title $title, ParserOptions $options ) {
5567                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5568         }
5569
5570         /**
5571          * Call a callback function on all regions of the given text that are not
5572          * inside strip markers, and replace those regions with the return value
5573          * of the callback. For example, with input:
5574          *
5575          *  aaa<MARKER>bbb
5576          *
5577          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5578          * two strings will be replaced with the value returned by the callback in
5579          * each case.
5580          *
5581          * @param $s string
5582          * @param $callback
5583          *
5584          * @return string
5585          */
5586         function markerSkipCallback( $s, $callback ) {
5587                 $i = 0;
5588                 $out = '';
5589                 while ( $i < strlen( $s ) ) {
5590                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5591                         if ( $markerStart === false ) {
5592                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5593                                 break;
5594                         } else {
5595                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5596                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5597                                 if ( $markerEnd === false ) {
5598                                         $out .= substr( $s, $markerStart );
5599                                         break;
5600                                 } else {
5601                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5602                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5603                                         $i = $markerEnd;
5604                                 }
5605                         }
5606                 }
5607                 return $out;
5608         }
5609
5610         /**
5611          * Save the parser state required to convert the given half-parsed text to
5612          * HTML. "Half-parsed" in this context means the output of
5613          * recursiveTagParse() or internalParse(). This output has strip markers
5614          * from replaceVariables (extensionSubstitution() etc.), and link
5615          * placeholders from replaceLinkHolders().
5616          *
5617          * Returns an array which can be serialized and stored persistently. This
5618          * array can later be loaded into another parser instance with
5619          * unserializeHalfParsedText(). The text can then be safely incorporated into
5620          * the return value of a parser hook.
5621          *
5622          * @param $text string
5623          *
5624          * @return array
5625          */
5626         function serializeHalfParsedText( $text ) {
5627                 wfProfileIn( __METHOD__ );
5628                 $data = array(
5629                         'text' => $text,
5630                         'version' => self::HALF_PARSED_VERSION,
5631                         'stripState' => $this->mStripState->getSubState( $text ),
5632                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5633                 );
5634                 wfProfileOut( __METHOD__ );
5635                 return $data;
5636         }
5637
5638         /**
5639          * Load the parser state given in the $data array, which is assumed to
5640          * have been generated by serializeHalfParsedText(). The text contents is
5641          * extracted from the array, and its markers are transformed into markers
5642          * appropriate for the current Parser instance. This transformed text is
5643          * returned, and can be safely included in the return value of a parser
5644          * hook.
5645          *
5646          * If the $data array has been stored persistently, the caller should first
5647          * check whether it is still valid, by calling isValidHalfParsedText().
5648          *
5649          * @param $data Serialized data
5650          * @return String
5651          */
5652         function unserializeHalfParsedText( $data ) {
5653                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5654                         throw new MWException( __METHOD__.': invalid version' );
5655                 }
5656
5657                 # First, extract the strip state.
5658                 $texts = array( $data['text'] );
5659                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5660
5661                 # Now renumber links
5662                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5663
5664                 # Should be good to go.
5665                 return $texts[0];
5666         }
5667
5668         /**
5669          * Returns true if the given array, presumed to be generated by
5670          * serializeHalfParsedText(), is compatible with the current version of the
5671          * parser.
5672          *
5673          * @param $data Array
5674          *
5675          * @return bool
5676          */
5677         function isValidHalfParsedText( $data ) {
5678                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5679         }
5680 }