includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *     produces HTML output
  21  * preSaveTransform().
  22  *     produces altered wiki markup.
  23  * preprocess()
  24  *     removes HTML comments and expands templates
  25  * cleanSig()
  26  *     Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *     Extracts sections from an article for section editing
  29  * getPreloadText()
  30  *     Removes <noinclude> sections, and <includeonly> tags.
  31  *
  32  * Globals used:
  33  *    objects:   $wgLang, $wgContLang
  34  *
  35  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  36  *
  37  * settings:
  38  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  39  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  40  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  41  *  $wgMaxArticleSize*
  42  *
  43  *  * only within ParserOptions
  44  * </pre>
  45  *
  46  * @ingroup Parser
  47  */
  48 class Parser {
  49         /**
  50          * Update this version number when the ParserOutput format
  51          * changes in an incompatible way, so the parser cache
  52          * can automatically discard old data.
  53          */
  54         const VERSION = '1.6.4';
  55
  56         # Flags for Parser::setFunctionHook
  57         # Also available as global constants from Defines.php
  58         const SFH_NO_HASH = 1;
  59         const SFH_OBJECT_ARGS = 2;
  60
  61         # Constants needed for external link processing
  62         # Everything except bracket, space, or control characters
  63         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  64         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  65                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  66
  67         # State constants for the definition list colon extraction
  68         const COLON_STATE_TEXT = 0;
  69         const COLON_STATE_TAG = 1;
  70         const COLON_STATE_TAGSTART = 2;
  71         const COLON_STATE_CLOSETAG = 3;
  72         const COLON_STATE_TAGSLASH = 4;
  73         const COLON_STATE_COMMENT = 5;
  74         const COLON_STATE_COMMENTDASH = 6;
  75         const COLON_STATE_COMMENTDASHDASH = 7;
  76
  77         # Flags for preprocessToDom
  78         const PTD_FOR_INCLUSION = 1;
  79
  80         # Allowed values for $this->mOutputType
  81         # Parameter to startExternalParse().
  82         const OT_HTML = 1; # like parse()
  83         const OT_WIKI = 2; # like preSaveTransform()
  84         const OT_PREPROCESS = 3; # like preprocess()
  85         const OT_MSG = 3;
  86         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  87
  88         # Marker Suffix needs to be accessible staticly.
  89         const MARKER_SUFFIX = "-QINU\x7f";
  90
  91         # Persistent:
  92         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  93         var $mSubstWords, $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex;
  94         var $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList;
  95         var $mVarCache, $mConf, $mFunctionTagHooks;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mLinkHolders, $mLinkID;
 102         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 103         var $mTplExpandCache; # empty-frame expansion cache
 104         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 105         var $mExpensiveFunctionCount; # number of expensive parser function calls
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions;      # ParserOptions object
 110         var $mTitle;        # Title context, used for self-link rendering and similar things
 111         var $mOutputType;   # Output type, one of the OT_xxx constants
 112         var $ot;            # Shortcut alias, see setOutputType()
 113         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 114         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 115         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 116
 117         /**
 118          * Constructor
 119          *
 120          * @public
 121          */
 122         function __construct( $conf = array() ) {
 123                 $this->mConf = $conf;
 124                 $this->mTagHooks = array();
 125                 $this->mTransparentTagHooks = array();
 126                 $this->mFunctionHooks = array();
 127                 $this->mFunctionTagHooks = array();
 128                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 129                 $this->mDefaultStripList = $this->mStripList = array();
 130                 $this->mUrlProtocols = wfUrlProtocols();
 131                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 132                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 133                 $this->mVarCache = array();
 134                 if ( isset( $conf['preprocessorClass'] ) ) {
 135                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 136                 } elseif ( extension_loaded( 'domxml' ) ) {
 137                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 138                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 139                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 140                 } elseif ( extension_loaded( 'dom' ) ) {
 141                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 142                 } else {
 143                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 144                 }
 145                 $this->mMarkerIndex = 0;
 146                 $this->mFirstCall = true;
 147         }
 148
 149         /**
 150          * Reduce memory usage to reduce the impact of circular references
 151          */
 152         function __destruct() {
 153                 if ( isset( $this->mLinkHolders ) ) {
 154                         $this->mLinkHolders->__destruct();
 155                 }
 156                 foreach ( $this as $name => $value ) {
 157                         unset( $this->$name );
 158                 }
 159         }
 160
 161         /**
 162          * Do various kinds of initialisation on the first call of the parser
 163          */
 164         function firstCallInit() {
 165                 if ( !$this->mFirstCall ) {
 166                         return;
 167                 }
 168                 $this->mFirstCall = false;
 169
 170                 wfProfileIn( __METHOD__ );
 171
 172                 CoreParserFunctions::register( $this );
 173                 CoreTagHooks::register( $this );
 174                 $this->initialiseVariables();
 175
 176                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 177                 wfProfileOut( __METHOD__ );
 178         }
 179
 180         /**
 181          * Clear Parser state
 182          *
 183          * @private
 184          */
 185         function clearState() {
 186                 wfProfileIn( __METHOD__ );
 187                 if ( $this->mFirstCall ) {
 188                         $this->firstCallInit();
 189                 }
 190                 $this->mOutput = new ParserOutput;
 191                 $this->mAutonumber = 0;
 192                 $this->mLastSection = '';
 193                 $this->mDTopen = false;
 194                 $this->mIncludeCount = array();
 195                 $this->mStripState = new StripState;
 196                 $this->mArgStack = false;
 197                 $this->mInPre = false;
 198                 $this->mLinkHolders = new LinkHolderArray( $this );
 199                 $this->mLinkID = 0;
 200                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 201                 $this->mVarCache = array();
 202
 203                 /**
 204                  * Prefix for temporary replacement strings for the multipass parser.
 205                  * \x07 should never appear in input as it's disallowed in XML.
 206                  * Using it at the front also gives us a little extra robustness
 207                  * since it shouldn't match when butted up against identifier-like
 208                  * string constructs.
 209                  *
 210                  * Must not consist of all title characters, or else it will change
 211                  * the behaviour of <nowiki> in a link.
 212                  */
 213                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 214                 # Changed to \x7f to allow XML double-parsing -- TS
 215                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 216
 217
 218                 # Clear these on every parse, bug 4549
 219                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 220
 221                 $this->mShowToc = true;
 222                 $this->mForceTocPosition = false;
 223                 $this->mIncludeSizes = array(
 224                         'post-expand' => 0,
 225                         'arg' => 0,
 226                 );
 227                 $this->mPPNodeCount = 0;
 228                 $this->mDefaultSort = false;
 229                 $this->mHeadings = array();
 230                 $this->mDoubleUnderscores = array();
 231                 $this->mExpensiveFunctionCount = 0;
 232
 233                 # Fix cloning
 234                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 235                         $this->mPreprocessor = null;
 236                 }
 237
 238                 wfRunHooks( 'ParserClearState', array( &$this ) );
 239                 wfProfileOut( __METHOD__ );
 240         }
 241
 242         /**
 243          * Convert wikitext to HTML
 244          * Do not call this function recursively.
 245          *
 246          * @param $text String: text we want to parse
 247          * @param $title A title object
 248          * @param $options ParserOptions
 249          * @param $linestart boolean
 250          * @param $clearState boolean
 251          * @param $revid Int: number to pass in {{REVISIONID}}
 252          * @return ParserOutput a ParserOutput
 253          */
 254         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 255                 /**
 256                  * First pass--just handle <nowiki> sections, pass the rest off
 257                  * to internalParse() which does all the real work.
 258                  */
 259
 260                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 261                 $fname = __METHOD__.'-' . wfGetCaller();
 262                 wfProfileIn( __METHOD__ );
 263                 wfProfileIn( $fname );
 264
 265                 if ( $clearState ) {
 266                         $this->clearState();
 267                 }
 268
 269                 $options->resetUsage();
 270                 $this->mOptions = $options;
 271                 $this->setTitle( $title ); # Page title has to be set for the pre-processor
 272
 273                 $oldRevisionId = $this->mRevisionId;
 274                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 275                 if ( $revid !== null ) {
 276                         $this->mRevisionId = $revid;
 277                         $this->mRevisionTimestamp = null;
 278                 }
 279                 $this->setOutputType( self::OT_HTML );
 280                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 281                 # No more strip!
 282                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 283                 $text = $this->internalParse( $text );
 284
 285                 $text = $this->mStripState->unstripGeneral( $text );
 286
 287                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 288                 $fixtags = array(
 289                         # french spaces, last one Guillemet-left
 290                         # only if there is something before the space
 291                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 292                         # french spaces, Guillemet-right
 293                         '/(\\302\\253) /' => '\\1&#160;',
 294                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 295                 );
 296                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 297
 298                 $text = $this->doBlockLevels( $text, $linestart );
 299
 300                 $this->replaceLinkHolders( $text );
 301
 302                 /**
 303                  * The page doesn't get language converted if
 304                  * a) It's disabled
 305                  * b) Content isn't converted
 306                  * c) It's a conversion table
 307                  */
 308                 if ( !( $wgDisableLangConversion
 309                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 310                                 || $this->mTitle->isConversionTable() ) ) {
 311
 312                         # The position of the convert() call should not be changed. it
 313                         # assumes that the links are all replaced and the only thing left
 314                         # is the <nowiki> mark.
 315
 316                         $text = $wgContLang->convert( $text );
 317                 }
 318
 319                 /**
 320                  * A page get its title converted except:
 321                  * a) Language conversion is globally disabled
 322                  * b) Title convert is globally disabled
 323                  * c) The page is a redirect page
 324                  * d) User request with a "linkconvert" set to "no"
 325                  * e) A "nocontentconvert" magic word has been set
 326                  * f) A "notitleconvert" magic word has been set
 327                  * g) User sets "noconvertlink" in his/her preference
 328                  *
 329                  * Note that if a user tries to set a title in a conversion
 330                  * rule but content conversion was not done, then the parser
 331                  * won't pick it up.  This is probably expected behavior.
 332                  */
 333                 if ( !( $wgDisableLangConversion
 334                                 || $wgDisableTitleConversion
 335                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 336                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 337                                 || $this->mOutput->getDisplayTitle() !== false ) )
 338                 {
 339                         $convruletitle = $wgContLang->getConvRuleTitle();
 340                         if ( $convruletitle ) {
 341                                 $this->mOutput->setTitleText( $convruletitle );
 342                         } else {
 343                                 $titleText = $wgContLang->convertTitle( $title );
 344                                 $this->mOutput->setTitleText( $titleText );
 345                         }
 346                 }
 347
 348                 $text = $this->mStripState->unstripNoWiki( $text );
 349
 350                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 351
 352 //!JF Move to its own function
 353
 354                 $uniq_prefix = $this->mUniqPrefix;
 355                 $matches = array();
 356                 $elements = array_keys( $this->mTransparentTagHooks );
 357                 $text = $this->extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 358
 359                 foreach ( $matches as $marker => $data ) {
 360                         list( $element, $content, $params, $tag ) = $data;
 361                         $tagName = strtolower( $element );
 362                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 363                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 364                         } else {
 365                                 $output = $tag;
 366                         }
 367                         $this->mStripState->general->setPair( $marker, $output );
 368                 }
 369                 $text = $this->mStripState->unstripGeneral( $text );
 370
 371                 $text = Sanitizer::normalizeCharReferences( $text );
 372
 373                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 374                         $text = MWTidy::tidy( $text );
 375                 } else {
 376                         # attempt to sanitize at least some nesting problems
 377                         # (bug #2702 and quite a few others)
 378                         $tidyregs = array(
 379                                 # ''Something [http://www.cool.com cool''] -->
 380                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 381                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 382                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 383                                 # fix up an anchor inside another anchor, only
 384                                 # at least for a single single nested link (bug 3695)
 385                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 386                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 387                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 388                                 # contains a div, so fix it up here; replace
 389                                 # div with escaped text
 390                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 391                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 392                                 # remove empty italic or bold tag pairs, some
 393                                 # introduced by rules above
 394                                 '/<([bi])><\/\\1>/' => '',
 395                         );
 396
 397                         $text = preg_replace(
 398                                 array_keys( $tidyregs ),
 399                                 array_values( $tidyregs ),
 400                                 $text );
 401                 }
 402                 global $wgExpensiveParserFunctionLimit;
 403                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 404                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 405                 }
 406
 407                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 408
 409                 # Information on include size limits, for the benefit of users who try to skirt them
 410                 if ( $this->mOptions->getEnableLimitReport() ) {
 411                         $max = $this->mOptions->getMaxIncludeSize();
 412                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 413                         $limitReport =
 414                                 "NewPP limit report\n" .
 415                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 416                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 417                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 418                                 $PFreport;
 419                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 420                         $text .= "\n<!-- \n$limitReport-->\n";
 421                 }
 422                 $this->mOutput->setText( $text );
 423
 424                 $this->mRevisionId = $oldRevisionId;
 425                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 426                 wfProfileOut( $fname );
 427                 wfProfileOut( __METHOD__ );
 428
 429                 return $this->mOutput;
 430         }
 431
 432         /**
 433          * Recursive parser entry point that can be called from an extension tag
 434          * hook.
 435          *
 436          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 437          *
 438          * @param $text String: text extension wants to have parsed
 439          * @param $frame PPFrame: The frame to use for expanding any template variables
 440          */
 441         function recursiveTagParse( $text, $frame=false ) {
 442                 wfProfileIn( __METHOD__ );
 443                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 444                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 445                 $text = $this->internalParse( $text, false, $frame );
 446                 wfProfileOut( __METHOD__ );
 447                 return $text;
 448         }
 449
 450         /**
 451          * Expand templates and variables in the text, producing valid, static wikitext.
 452          * Also removes comments.
 453          */
 454         function preprocess( $text, $title, $options, $revid = null ) {
 455                 wfProfileIn( __METHOD__ );
 456                 $this->clearState();
 457                 $this->setOutputType( self::OT_PREPROCESS );
 458                 $options->resetUsage();
 459                 $this->mOptions = $options;
 460                 $this->setTitle( $title );
 461                 if ( $revid !== null ) {
 462                         $this->mRevisionId = $revid;
 463                 }
 464                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 465                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 466                 $text = $this->replaceVariables( $text );
 467                 $text = $this->mStripState->unstripBoth( $text );
 468                 wfProfileOut( __METHOD__ );
 469                 return $text;
 470         }
 471
 472         /**
 473          * Process the wikitext for the ?preload= feature. (bug 5210)
 474          *
 475          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 476          * comments, templates, arguments, tags hooks and parser functions are untouched.
 477          */
 478         public function getPreloadText( $text, $title, $options ) {
 479                 # Parser (re)initialisation
 480                 $this->clearState();
 481                 $this->setOutputType( self::OT_PLAIN );
 482                 $options->resetUsage();
 483                 $this->mOptions = $options;
 484                 $this->setTitle( $title );
 485
 486                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 487                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 488                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 489         }
 490
 491         /**
 492          * Get a random string
 493          *
 494          * @private
 495          * @static
 496          */
 497         static private function getRandomString() {
 498                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 499         }
 500
 501         /**
 502          * Accessor for mUniqPrefix.
 503          *
 504          * @return String
 505          */
 506         public function uniqPrefix() {
 507                 if ( !isset( $this->mUniqPrefix ) ) {
 508                         # @todo Fixme: this is probably *horribly wrong*
 509                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 510                         # if this is called for a parser cache hit, the parser may not
 511                         # have ever been initialized in the first place.
 512                         # Not really sure what the heck is supposed to be going on here.
 513                         return '';
 514                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 515                 }
 516                 return $this->mUniqPrefix;
 517         }
 518
 519         /**
 520          * Set the context title
 521          */
 522         function setTitle( $t ) {
 523                 if ( !$t || $t instanceof FakeTitle ) {
 524                         $t = Title::newFromText( 'NO TITLE' );
 525                 }
 526
 527                 if ( strval( $t->getFragment() ) !== '' ) {
 528                         # Strip the fragment to avoid various odd effects
 529                         $this->mTitle = clone $t;
 530                         $this->mTitle->setFragment( '' );
 531                 } else {
 532                         $this->mTitle = $t;
 533                 }
 534         }
 535
 536         /**
 537          * Accessor for the Title object
 538          *
 539          * @return Title object
 540          */
 541         function getTitle() {
 542                 return $this->mTitle;
 543         }
 544
 545         /**
 546          * Accessor/mutator for the Title object
 547          *
 548          * @param $x New Title object or null to just get the current one
 549          * @return Title object
 550          */
 551         function Title( $x = null ) {
 552                 return wfSetVar( $this->mTitle, $x );
 553         }
 554
 555         /**
 556          * Set the output type
 557          *
 558          * @param $ot Integer: new value
 559          */
 560         function setOutputType( $ot ) {
 561                 $this->mOutputType = $ot;
 562                 # Shortcut alias
 563                 $this->ot = array(
 564                         'html' => $ot == self::OT_HTML,
 565                         'wiki' => $ot == self::OT_WIKI,
 566                         'pre' => $ot == self::OT_PREPROCESS,
 567                         'plain' => $ot == self::OT_PLAIN,
 568                 );
 569         }
 570
 571         /**
 572          * Accessor/mutator for the output type
 573          *
 574          * @param $x New value or null to just get the current one
 575          * @return Integer
 576          */
 577         function OutputType( $x = null ) {
 578                 return wfSetVar( $this->mOutputType, $x );
 579         }
 580
 581         /**
 582          * Get the ParserOutput object
 583          *
 584          * @return ParserOutput object
 585          */
 586         function getOutput() {
 587                 return $this->mOutput;
 588         }
 589
 590         /**
 591          * Get the ParserOptions object
 592          *
 593          * @return ParserOptions object
 594          */
 595         function getOptions() {
 596                 return $this->mOptions;
 597         }
 598
 599         /**
 600          * Accessor/mutator for the ParserOptions object
 601          *
 602          * @param $x New value or null to just get the current one
 603          * @return Current ParserOptions object
 604          */
 605         function Options( $x = null ) {
 606                 return wfSetVar( $this->mOptions, $x );
 607         }
 608
 609         function nextLinkID() {
 610                 return $this->mLinkID++;
 611         }
 612
 613         function getFunctionLang() {
 614                 global $wgLang, $wgContLang;
 615
 616                 $target = $this->mOptions->getTargetLanguage();
 617                 if ( $target !== null ) {
 618                         return $target;
 619                 } else {
 620                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 621                 }
 622         }
 623
 624         /**
 625          * Get a preprocessor object
 626          *
 627          * @return Preprocessor instance
 628          */
 629         function getPreprocessor() {
 630                 if ( !isset( $this->mPreprocessor ) ) {
 631                         $class = $this->mPreprocessorClass;
 632                         $this->mPreprocessor = new $class( $this );
 633                 }
 634                 return $this->mPreprocessor;
 635         }
 636
 637         /**
 638          * Replaces all occurrences of HTML-style comments and the given tags
 639          * in the text with a random marker and returns the next text. The output
 640          * parameter $matches will be an associative array filled with data in
 641          * the form:
 642          *   'UNIQ-xxxxx' => array(
 643          *     'element',
 644          *     'tag content',
 645          *     array( 'param' => 'x' ),
 646          *     '<element param="x">tag content</element>' ) )
 647          *
 648          * @param $elements list of element names. Comments are always extracted.
 649          * @param $text Source text string.
 650          * @param $matches Out parameter, Array: extracted tags
 651          * @param $uniq_prefix
 652          * @return String: stripped text
 653          *
 654          * @static
 655          */
 656         public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 657                 static $n = 1;
 658                 $stripped = '';
 659                 $matches = array();
 660
 661                 $taglist = implode( '|', $elements );
 662                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 663
 664                 while ( $text != '' ) {
 665                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 666                         $stripped .= $p[0];
 667                         if ( count( $p ) < 5 ) {
 668                                 break;
 669                         }
 670                         if ( count( $p ) > 5 ) {
 671                                 # comment
 672                                 $element    = $p[4];
 673                                 $attributes = '';
 674                                 $close      = '';
 675                                 $inside     = $p[5];
 676                         } else {
 677                                 # tag
 678                                 $element    = $p[1];
 679                                 $attributes = $p[2];
 680                                 $close      = $p[3];
 681                                 $inside     = $p[4];
 682                         }
 683
 684                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 685                         $stripped .= $marker;
 686
 687                         if ( $close === '/>' ) {
 688                                 # Empty element tag, <tag />
 689                                 $content = null;
 690                                 $text = $inside;
 691                                 $tail = null;
 692                         } else {
 693                                 if ( $element === '!--' ) {
 694                                         $end = '/(-->)/';
 695                                 } else {
 696                                         $end = "/(<\\/$element\\s*>)/i";
 697                                 }
 698                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 699                                 $content = $q[0];
 700                                 if ( count( $q ) < 3 ) {
 701                                         # No end tag -- let it run out to the end of the text.
 702                                         $tail = '';
 703                                         $text = '';
 704                                 } else {
 705                                         $tail = $q[1];
 706                                         $text = $q[2];
 707                                 }
 708                         }
 709
 710                         $matches[$marker] = array( $element,
 711                                 $content,
 712                                 Sanitizer::decodeTagAttributes( $attributes ),
 713                                 "<$element$attributes$close$content$tail" );
 714                 }
 715                 return $stripped;
 716         }
 717
 718         /**
 719          * Get a list of strippable XML-like elements
 720          */
 721         function getStripList() {
 722                 return $this->mStripList;
 723         }
 724
 725         /**
 726          * @deprecated use replaceVariables
 727          */
 728         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 729                 return $text;
 730         }
 731
 732         /**
 733          * Restores pre, math, and other extensions removed by strip()
 734          *
 735          * always call unstripNoWiki() after this one
 736          * @private
 737          * @deprecated use $this->mStripState->unstrip()
 738          */
 739         function unstrip( $text, $state ) {
 740                 return $state->unstripGeneral( $text );
 741         }
 742
 743         /**
 744          * Always call this after unstrip() to preserve the order
 745          *
 746          * @private
 747          * @deprecated use $this->mStripState->unstrip()
 748          */
 749         function unstripNoWiki( $text, $state ) {
 750                 return $state->unstripNoWiki( $text );
 751         }
 752
 753         /**
 754          * @deprecated use $this->mStripState->unstripBoth()
 755          */
 756         function unstripForHTML( $text ) {
 757                 return $this->mStripState->unstripBoth( $text );
 758         }
 759
 760         /**
 761          * Add an item to the strip state
 762          * Returns the unique tag which must be inserted into the stripped text
 763          * The tag will be replaced with the original text in unstrip()
 764          *
 765          * @private
 766          */
 767         function insertStripItem( $text ) {
 768                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 769                 $this->mMarkerIndex++;
 770                 $this->mStripState->general->setPair( $rnd, $text );
 771                 return $rnd;
 772         }
 773
 774         /**
 775          * Interface with html tidy
 776          * @deprecated Use MWTidy::tidy()
 777          */
 778         public static function tidy( $text ) {
 779                 wfDeprecated( __METHOD__ );
 780                 return MWTidy::tidy( $text );
 781         }
 782
 783         /**
 784          * parse the wiki syntax used to render tables
 785          *
 786          * @private
 787          */
 788         function doTableStuff( $text ) {
 789                 wfProfileIn( __METHOD__ );
 790
 791                 $lines = StringUtils::explode( "\n", $text );
 792                 $out = '';
 793                 $td_history = array(); # Is currently a td tag open?
 794                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 795                 $tr_history = array(); # Is currently a tr tag open?
 796                 $tr_attributes = array(); # history of tr attributes
 797                 $has_opened_tr = array(); # Did this table open a <tr> element?
 798                 $indent_level = 0; # indent level of the table
 799
 800                 foreach ( $lines as $outLine ) {
 801                         $line = trim( $outLine );
 802
 803                         if ( $line === '' ) { # empty line, go to next line
 804                                 $out .= $outLine."\n";
 805                                 continue;
 806                         }
 807
 808                         $first_character = $line[0];
 809                         $matches = array();
 810
 811                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 812                                 # First check if we are starting a new table
 813                                 $indent_level = strlen( $matches[1] );
 814
 815                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 816                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 817
 818                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 819                                 array_push( $td_history , false );
 820                                 array_push( $last_tag_history , '' );
 821                                 array_push( $tr_history , false );
 822                                 array_push( $tr_attributes , '' );
 823                                 array_push( $has_opened_tr , false );
 824                         } elseif ( count( $td_history ) == 0 ) {
 825                                 # Don't do any of the following
 826                                 $out .= $outLine."\n";
 827                                 continue;
 828                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 829                                 # We are ending a table
 830                                 $line = '</table>' . substr( $line , 2 );
 831                                 $last_tag = array_pop( $last_tag_history );
 832
 833                                 if ( !array_pop( $has_opened_tr ) ) {
 834                                         $line = "<tr><td></td></tr>{$line}";
 835                                 }
 836
 837                                 if ( array_pop( $tr_history ) ) {
 838                                         $line = "</tr>{$line}";
 839                                 }
 840
 841                                 if ( array_pop( $td_history ) ) {
 842                                         $line = "</{$last_tag}>{$line}";
 843                                 }
 844                                 array_pop( $tr_attributes );
 845                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 846                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 847                                 # Now we have a table row
 848                                 $line = preg_replace( '#^\|-+#', '', $line );
 849
 850                                 # Whats after the tag is now only attributes
 851                                 $attributes = $this->mStripState->unstripBoth( $line );
 852                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 853                                 array_pop( $tr_attributes );
 854                                 array_push( $tr_attributes, $attributes );
 855
 856                                 $line = '';
 857                                 $last_tag = array_pop( $last_tag_history );
 858                                 array_pop( $has_opened_tr );
 859                                 array_push( $has_opened_tr , true );
 860
 861                                 if ( array_pop( $tr_history ) ) {
 862                                         $line = '</tr>';
 863                                 }
 864
 865                                 if ( array_pop( $td_history ) ) {
 866                                         $line = "</{$last_tag}>{$line}";
 867                                 }
 868
 869                                 $outLine = $line;
 870                                 array_push( $tr_history , false );
 871                                 array_push( $td_history , false );
 872                                 array_push( $last_tag_history , '' );
 873                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 874                                 # This might be cell elements, td, th or captions
 875                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 876                                         $first_character = '+';
 877                                         $line = substr( $line , 1 );
 878                                 }
 879
 880                                 $line = substr( $line , 1 );
 881
 882                                 if ( $first_character === '!' ) {
 883                                         $line = str_replace( '!!' , '||' , $line );
 884                                 }
 885
 886                                 # Split up multiple cells on the same line.
 887                                 # FIXME : This can result in improper nesting of tags processed
 888                                 # by earlier parser steps, but should avoid splitting up eg
 889                                 # attribute values containing literal "||".
 890                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 891
 892                                 $outLine = '';
 893
 894                                 # Loop through each table cell
 895                                 foreach ( $cells as $cell ) {
 896                                         $previous = '';
 897                                         if ( $first_character !== '+' ) {
 898                                                 $tr_after = array_pop( $tr_attributes );
 899                                                 if ( !array_pop( $tr_history ) ) {
 900                                                         $previous = "<tr{$tr_after}>\n";
 901                                                 }
 902                                                 array_push( $tr_history , true );
 903                                                 array_push( $tr_attributes , '' );
 904                                                 array_pop( $has_opened_tr );
 905                                                 array_push( $has_opened_tr , true );
 906                                         }
 907
 908                                         $last_tag = array_pop( $last_tag_history );
 909
 910                                         if ( array_pop( $td_history ) ) {
 911                                                 $previous = "</{$last_tag}>\n{$previous}";
 912                                         }
 913
 914                                         if ( $first_character === '|' ) {
 915                                                 $last_tag = 'td';
 916                                         } elseif ( $first_character === '!' ) {
 917                                                 $last_tag = 'th';
 918                                         } elseif ( $first_character === '+' ) {
 919                                                 $last_tag = 'caption';
 920                                         } else {
 921                                                 $last_tag = '';
 922                                         }
 923
 924                                         array_push( $last_tag_history , $last_tag );
 925
 926                                         # A cell could contain both parameters and data
 927                                         $cell_data = explode( '|' , $cell , 2 );
 928
 929                                         # Bug 553: Note that a '|' inside an invalid link should not
 930                                         # be mistaken as delimiting cell parameters
 931                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 932                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 933                                         } elseif ( count( $cell_data ) == 1 ) {
 934                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 935                                         } else {
 936                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 937                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 938                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 939                                         }
 940
 941                                         $outLine .= $cell;
 942                                         array_push( $td_history , true );
 943                                 }
 944                         }
 945                         $out .= $outLine . "\n";
 946                 }
 947
 948                 # Closing open td, tr && table
 949                 while ( count( $td_history ) > 0 ) {
 950                         if ( array_pop( $td_history ) ) {
 951                                 $out .= "</td>\n";
 952                         }
 953                         if ( array_pop( $tr_history ) ) {
 954                                 $out .= "</tr>\n";
 955                         }
 956                         if ( !array_pop( $has_opened_tr ) ) {
 957                                 $out .= "<tr><td></td></tr>\n" ;
 958                         }
 959
 960                         $out .= "</table>\n";
 961                 }
 962
 963                 # Remove trailing line-ending (b/c)
 964                 if ( substr( $out, -1 ) === "\n" ) {
 965                         $out = substr( $out, 0, -1 );
 966                 }
 967
 968                 # special case: don't return empty table
 969                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 970                         $out = '';
 971                 }
 972
 973                 wfProfileOut( __METHOD__ );
 974
 975                 return $out;
 976         }
 977
 978         /**
 979          * Helper function for parse() that transforms wiki markup into
 980          * HTML. Only called for $mOutputType == self::OT_HTML.
 981          *
 982          * @private
 983          */
 984         function internalParse( $text, $isMain = true, $frame=false ) {
 985                 wfProfileIn( __METHOD__ );
 986
 987                 $origText = $text;
 988
 989                 # Hook to suspend the parser in this state
 990                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 991                         wfProfileOut( __METHOD__ );
 992                         return $text ;
 993                 }
 994
 995                 # if $frame is provided, then use $frame for replacing any variables
 996                 if ( $frame ) {
 997                         # use frame depth to infer how include/noinclude tags should be handled
 998                         # depth=0 means this is the top-level document; otherwise it's an included document
 999                         if ( !$frame->depth ) {
1000                                 $flag = 0;
1001                         } else {
1002                                 $flag = Parser::PTD_FOR_INCLUSION;
1003                         }
1004                         $dom = $this->preprocessToDom( $text, $flag );
1005                         $text = $frame->expand( $dom );
1006                 } else {
1007                         # if $frame is not provided, then use old-style replaceVariables
1008                         $text = $this->replaceVariables( $text );
1009                 }
1010
1011                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1012                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1013
1014                 # Tables need to come after variable replacement for things to work
1015                 # properly; putting them before other transformations should keep
1016                 # exciting things like link expansions from showing up in surprising
1017                 # places.
1018                 $text = $this->doTableStuff( $text );
1019
1020                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1021
1022                 $text = $this->doDoubleUnderscore( $text );
1023
1024                 $text = $this->doHeadings( $text );
1025                 if ( $this->mOptions->getUseDynamicDates() ) {
1026                         $df = DateFormatter::getInstance();
1027                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1028                 }
1029                 $text = $this->replaceInternalLinks( $text );
1030                 $text = $this->doAllQuotes( $text );
1031                 $text = $this->replaceExternalLinks( $text );
1032
1033                 # replaceInternalLinks may sometimes leave behind
1034                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1035                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1036
1037                 $text = $this->doMagicLinks( $text );
1038                 $text = $this->formatHeadings( $text, $origText, $isMain );
1039
1040                 wfProfileOut( __METHOD__ );
1041                 return $text;
1042         }
1043
1044         /**
1045          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1046          * magic external links.
1047          *
1048          * DML
1049          * @private
1050          */
1051         function doMagicLinks( $text ) {
1052                 wfProfileIn( __METHOD__ );
1053                 $prots = $this->mUrlProtocols;
1054                 $urlChar = self::EXT_LINK_URL_CLASS;
1055                 $text = preg_replace_callback(
1056                         '!(?:                           # Start cases
1057                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1058                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1059                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1060                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1061                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1062                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1063                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1064                                     [0-9Xx]                 # check digit
1065                                     \b)
1066                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1067                 wfProfileOut( __METHOD__ );
1068                 return $text;
1069         }
1070
1071         function magicLinkCallback( $m ) {
1072                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1073                         # Skip anchor
1074                         return $m[0];
1075                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1076                         # Skip HTML element
1077                         return $m[0];
1078                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1079                         # Free external link
1080                         return $this->makeFreeExternalLink( $m[0] );
1081                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1082                         # RFC or PMID
1083                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1084                                 $keyword = 'RFC';
1085                                 $urlmsg = 'rfcurl';
1086                                 $CssClass = 'mw-magiclink-rfc';
1087                                 $id = $m[4];
1088                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1089                                 $keyword = 'PMID';
1090                                 $urlmsg = 'pubmedurl';
1091                                 $CssClass = 'mw-magiclink-pmid';
1092                                 $id = $m[4];
1093                         } else {
1094                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1095                                         substr( $m[0], 0, 20 ) . '"' );
1096                         }
1097                         $url = wfMsgForContent( $urlmsg, $id);
1098                         $sk = $this->mOptions->getSkin( $this->mTitle );
1099                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1100                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1101                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1102                         # ISBN
1103                         $isbn = $m[5];
1104                         $num = strtr( $isbn, array(
1105                                 '-' => '',
1106                                 ' ' => '',
1107                                 'x' => 'X',
1108                         ));
1109                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1110                         return'<a href="' .
1111                                 $titleObj->escapeLocalUrl() .
1112                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1113                 } else {
1114                         return $m[0];
1115                 }
1116         }
1117
1118         /**
1119          * Make a free external link, given a user-supplied URL
1120          * @return HTML
1121          * @private
1122          */
1123         function makeFreeExternalLink( $url ) {
1124                 global $wgContLang;
1125                 wfProfileIn( __METHOD__ );
1126
1127                 $sk = $this->mOptions->getSkin( $this->mTitle );
1128                 $trail = '';
1129
1130                 # The characters '<' and '>' (which were escaped by
1131                 # removeHTMLtags()) should not be included in
1132                 # URLs, per RFC 2396.
1133                 $m2 = array();
1134                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1135                         $trail = substr( $url, $m2[0][1] ) . $trail;
1136                         $url = substr( $url, 0, $m2[0][1] );
1137                 }
1138
1139                 # Move trailing punctuation to $trail
1140                 $sep = ',;\.:!?';
1141                 # If there is no left bracket, then consider right brackets fair game too
1142                 if ( strpos( $url, '(' ) === false ) {
1143                         $sep .= ')';
1144                 }
1145
1146                 $numSepChars = strspn( strrev( $url ), $sep );
1147                 if ( $numSepChars ) {
1148                         $trail = substr( $url, -$numSepChars ) . $trail;
1149                         $url = substr( $url, 0, -$numSepChars );
1150                 }
1151
1152                 $url = Sanitizer::cleanUrl( $url );
1153
1154                 # Is this an external image?
1155                 $text = $this->maybeMakeExternalImage( $url );
1156                 if ( $text === false ) {
1157                         # Not an image, make a link
1158                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1159                                 $this->getExternalLinkAttribs( $url ) );
1160                         # Register it in the output object...
1161                         # Replace unnecessary URL escape codes with their equivalent characters
1162                         $pasteurized = self::replaceUnusualEscapes( $url );
1163                         $this->mOutput->addExternalLink( $pasteurized );
1164                 }
1165                 wfProfileOut( __METHOD__ );
1166                 return $text . $trail;
1167         }
1168
1169
1170         /**
1171          * Parse headers and return html
1172          *
1173          * @private
1174          */
1175         function doHeadings( $text ) {
1176                 wfProfileIn( __METHOD__ );
1177                 for ( $i = 6; $i >= 1; --$i ) {
1178                         $h = str_repeat( '=', $i );
1179                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1180                           "<h$i>\\1</h$i>", $text );
1181                 }
1182                 wfProfileOut( __METHOD__ );
1183                 return $text;
1184         }
1185
1186         /**
1187          * Replace single quotes with HTML markup
1188          * @private
1189          * @return string the altered text
1190          */
1191         function doAllQuotes( $text ) {
1192                 wfProfileIn( __METHOD__ );
1193                 $outtext = '';
1194                 $lines = StringUtils::explode( "\n", $text );
1195                 foreach ( $lines as $line ) {
1196                         $outtext .= $this->doQuotes( $line ) . "\n";
1197                 }
1198                 $outtext = substr( $outtext, 0,-1 );
1199                 wfProfileOut( __METHOD__ );
1200                 return $outtext;
1201         }
1202
1203         /**
1204          * Helper function for doAllQuotes()
1205          */
1206         public function doQuotes( $text ) {
1207                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1208                 if ( count( $arr ) == 1 ) {
1209                         return $text;
1210                 } else {
1211                         # First, do some preliminary work. This may shift some apostrophes from
1212                         # being mark-up to being text. It also counts the number of occurrences
1213                         # of bold and italics mark-ups.
1214                         $numbold = 0;
1215                         $numitalics = 0;
1216                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1217                                 if ( ( $i % 2 ) == 1 ) {
1218                                         # If there are ever four apostrophes, assume the first is supposed to
1219                                         # be text, and the remaining three constitute mark-up for bold text.
1220                                         if ( strlen( $arr[$i] ) == 4 ) {
1221                                                 $arr[$i-1] .= "'";
1222                                                 $arr[$i] = "'''";
1223                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1224                                                 # If there are more than 5 apostrophes in a row, assume they're all
1225                                                 # text except for the last 5.
1226                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1227                                                 $arr[$i] = "'''''";
1228                                         }
1229                                         # Count the number of occurrences of bold and italics mark-ups.
1230                                         # We are not counting sequences of five apostrophes.
1231                                         if ( strlen( $arr[$i] ) == 2 ) {
1232                                                 $numitalics++;
1233                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1234                                                 $numbold++;
1235                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1236                                                 $numitalics++;
1237                                                 $numbold++;
1238                                         }
1239                                 }
1240                         }
1241
1242                         # If there is an odd number of both bold and italics, it is likely
1243                         # that one of the bold ones was meant to be an apostrophe followed
1244                         # by italics. Which one we cannot know for certain, but it is more
1245                         # likely to be one that has a single-letter word before it.
1246                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1247                                 $i = 0;
1248                                 $firstsingleletterword = -1;
1249                                 $firstmultiletterword = -1;
1250                                 $firstspace = -1;
1251                                 foreach ( $arr as $r ) {
1252                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1253                                                 $x1 = substr( $arr[$i-1], -1 );
1254                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1255                                                 if ( $x1 === ' ' ) {
1256                                                         if ( $firstspace == -1 ) {
1257                                                                 $firstspace = $i;
1258                                                         }
1259                                                 } elseif ( $x2 === ' ') {
1260                                                         if ( $firstsingleletterword == -1 ) {
1261                                                                 $firstsingleletterword = $i;
1262                                                         }
1263                                                 } else {
1264                                                         if ( $firstmultiletterword == -1 ) {
1265                                                                 $firstmultiletterword = $i;
1266                                                         }
1267                                                 }
1268                                         }
1269                                         $i++;
1270                                 }
1271
1272                                 # If there is a single-letter word, use it!
1273                                 if ( $firstsingleletterword > -1 ) {
1274                                         $arr[$firstsingleletterword] = "''";
1275                                         $arr[$firstsingleletterword-1] .= "'";
1276                                 } elseif ( $firstmultiletterword > -1 ) {
1277                                         # If not, but there's a multi-letter word, use that one.
1278                                         $arr[$firstmultiletterword] = "''";
1279                                         $arr[$firstmultiletterword-1] .= "'";
1280                                 } elseif ( $firstspace > -1 ) {
1281                                         # ... otherwise use the first one that has neither.
1282                                         # (notice that it is possible for all three to be -1 if, for example,
1283                                         # there is only one pentuple-apostrophe in the line)
1284                                         $arr[$firstspace] = "''";
1285                                         $arr[$firstspace-1] .= "'";
1286                                 }
1287                         }
1288
1289                         # Now let's actually convert our apostrophic mush to HTML!
1290                         $output = '';
1291                         $buffer = '';
1292                         $state = '';
1293                         $i = 0;
1294                         foreach ( $arr as $r ) {
1295                                 if ( ( $i % 2 ) == 0 ) {
1296                                         if ( $state === 'both' ) {
1297                                                 $buffer .= $r;
1298                                         } else {
1299                                                 $output .= $r;
1300                                         }
1301                                 } else {
1302                                         if ( strlen( $r ) == 2 ) {
1303                                                 if ( $state === 'i' ) {
1304                                                         $output .= '</i>'; $state = '';
1305                                                 } elseif ( $state === 'bi' ) {
1306                                                         $output .= '</i>'; $state = 'b';
1307                                                 } elseif ( $state === 'ib' ) {
1308                                                         $output .= '</b></i><b>'; $state = 'b';
1309                                                 } elseif ( $state === 'both' ) {
1310                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1311                                                 } else { # $state can be 'b' or ''
1312                                                         $output .= '<i>'; $state .= 'i';
1313                                                 }
1314                                         } elseif ( strlen( $r ) == 3 ) {
1315                                                 if ( $state === 'b' ) {
1316                                                         $output .= '</b>'; $state = '';
1317                                                 } elseif ( $state === 'bi' ) {
1318                                                         $output .= '</i></b><i>'; $state = 'i';
1319                                                 } elseif ( $state === 'ib' ) {
1320                                                         $output .= '</b>'; $state = 'i';
1321                                                 } elseif ( $state === 'both' ) {
1322                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1323                                                 } else { # $state can be 'i' or ''
1324                                                         $output .= '<b>'; $state .= 'b';
1325                                                 }
1326                                         } elseif ( strlen( $r ) == 5 ) {
1327                                                 if ( $state === 'b' ) {
1328                                                         $output .= '</b><i>'; $state = 'i';
1329                                                 } elseif ( $state === 'i' ) {
1330                                                         $output .= '</i><b>'; $state = 'b';
1331                                                 } elseif ( $state === 'bi' ) {
1332                                                         $output .= '</i></b>'; $state = '';
1333                                                 } elseif ( $state === 'ib' ) {
1334                                                         $output .= '</b></i>'; $state = '';
1335                                                 } elseif ( $state === 'both' ) {
1336                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1337                                                 } else { # ($state == '')
1338                                                         $buffer = ''; $state = 'both';
1339                                                 }
1340                                         }
1341                                 }
1342                                 $i++;
1343                         }
1344                         # Now close all remaining tags.  Notice that the order is important.
1345                         if ( $state === 'b' || $state === 'ib' ) {
1346                                 $output .= '</b>';
1347                         }
1348                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1349                                 $output .= '</i>';
1350                         }
1351                         if ( $state === 'bi' ) {
1352                                 $output .= '</b>';
1353                         }
1354                         # There might be lonely ''''', so make sure we have a buffer
1355                         if ( $state === 'both' && $buffer ) {
1356                                 $output .= '<b><i>'.$buffer.'</i></b>';
1357                         }
1358                         return $output;
1359                 }
1360         }
1361
1362         /**
1363          * Replace external links (REL)
1364          *
1365          * Note: this is all very hackish and the order of execution matters a lot.
1366          * Make sure to run maintenance/parserTests.php if you change this code.
1367          *
1368          * @private
1369          */
1370         function replaceExternalLinks( $text ) {
1371                 global $wgContLang;
1372                 wfProfileIn( __METHOD__ );
1373
1374                 $sk = $this->mOptions->getSkin( $this->mTitle );
1375
1376                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1377                 $s = array_shift( $bits );
1378
1379                 $i = 0;
1380                 while ( $i<count( $bits ) ) {
1381                         $url = $bits[$i++];
1382                         $protocol = $bits[$i++];
1383                         $text = $bits[$i++];
1384                         $trail = $bits[$i++];
1385
1386                         # The characters '<' and '>' (which were escaped by
1387                         # removeHTMLtags()) should not be included in
1388                         # URLs, per RFC 2396.
1389                         $m2 = array();
1390                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1391                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1392                                 $url = substr( $url, 0, $m2[0][1] );
1393                         }
1394
1395                         # If the link text is an image URL, replace it with an <img> tag
1396                         # This happened by accident in the original parser, but some people used it extensively
1397                         $img = $this->maybeMakeExternalImage( $text );
1398                         if ( $img !== false ) {
1399                                 $text = $img;
1400                         }
1401
1402                         $dtrail = '';
1403
1404                         # Set linktype for CSS - if URL==text, link is essentially free
1405                         $linktype = ( $text === $url ) ? 'free' : 'text';
1406
1407                         # No link text, e.g. [http://domain.tld/some.link]
1408                         if ( $text == '' ) {
1409                                 # Autonumber if allowed. See bug #5918
1410                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1411                                         $langObj = $this->getFunctionLang();
1412                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1413                                         $linktype = 'autonumber';
1414                                 } else {
1415                                         # Otherwise just use the URL
1416                                         $text = htmlspecialchars( $url );
1417                                         $linktype = 'free';
1418                                 }
1419                         } else {
1420                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1421                                 # Check for trail
1422                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1423                         }
1424
1425                         $text = $wgContLang->markNoConversion( $text );
1426
1427                         $url = Sanitizer::cleanUrl( $url );
1428
1429                         # Use the encoded URL
1430                         # This means that users can paste URLs directly into the text
1431                         # Funny characters like ö aren't valid in URLs anyway
1432                         # This was changed in August 2004
1433                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1434                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1435
1436                         # Register link in the output object.
1437                         # Replace unnecessary URL escape codes with the referenced character
1438                         # This prevents spammers from hiding links from the filters
1439                         $pasteurized = self::replaceUnusualEscapes( $url );
1440                         $this->mOutput->addExternalLink( $pasteurized );
1441                 }
1442
1443                 wfProfileOut( __METHOD__ );
1444                 return $s;
1445         }
1446
1447         /**
1448          * Get an associative array of additional HTML attributes appropriate for a
1449          * particular external link.  This currently may include rel => nofollow
1450          * (depending on configuration, namespace, and the URL's domain) and/or a
1451          * target attribute (depending on configuration).
1452          *
1453          * @param $url String: optional URL, to extract the domain from for rel =>
1454          *   nofollow if appropriate
1455          * @return Array: associative array of HTML attributes
1456          */
1457         function getExternalLinkAttribs( $url = false ) {
1458                 $attribs = array();
1459                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1460                 $ns = $this->mTitle->getNamespace();
1461                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1462                         $attribs['rel'] = 'nofollow';
1463
1464                         global $wgNoFollowDomainExceptions;
1465                         if ( $wgNoFollowDomainExceptions ) {
1466                                 $bits = wfParseUrl( $url );
1467                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1468                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1469                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1470                                                         unset( $attribs['rel'] );
1471                                                         break;
1472                                                 }
1473                                         }
1474                                 }
1475                         }
1476                 }
1477                 if ( $this->mOptions->getExternalLinkTarget() ) {
1478                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1479                 }
1480                 return $attribs;
1481         }
1482
1483
1484         /**
1485          * Replace unusual URL escape codes with their equivalent characters
1486          *
1487          * @param $url String
1488          * @return String
1489          *
1490          * @todo  This can merge genuinely required bits in the path or query string,
1491          *        breaking legit URLs. A proper fix would treat the various parts of
1492          *        the URL differently; as a workaround, just use the output for
1493          *        statistical records, not for actual linking/output.
1494          */
1495         static function replaceUnusualEscapes( $url ) {
1496                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1497                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1498         }
1499
1500         /**
1501          * Callback function used in replaceUnusualEscapes().
1502          * Replaces unusual URL escape codes with their equivalent character
1503          */
1504         private static function replaceUnusualEscapesCallback( $matches ) {
1505                 $char = urldecode( $matches[0] );
1506                 $ord = ord( $char );
1507                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1508                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1509                         # No, shouldn't be escaped
1510                         return $char;
1511                 } else {
1512                         # Yes, leave it escaped
1513                         return $matches[0];
1514                 }
1515         }
1516
1517         /**
1518          * make an image if it's allowed, either through the global
1519          * option, through the exception, or through the on-wiki whitelist
1520          * @private
1521          */
1522         function maybeMakeExternalImage( $url ) {
1523                 $sk = $this->mOptions->getSkin( $this->mTitle );
1524                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1525                 $imagesexception = !empty( $imagesfrom );
1526                 $text = false;
1527                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1528                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1529                         $imagematch = false;
1530                         foreach ( $imagesfrom as $match ) {
1531                                 if ( strpos( $url, $match ) === 0 ) {
1532                                         $imagematch = true;
1533                                         break;
1534                                 }
1535                         }
1536                 } elseif ( $imagesexception ) {
1537                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1538                 } else {
1539                         $imagematch = false;
1540                 }
1541                 if ( $this->mOptions->getAllowExternalImages()
1542                      || ( $imagesexception && $imagematch ) ) {
1543                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1544                                 # Image found
1545                                 $text = $sk->makeExternalImage( $url );
1546                         }
1547                 }
1548                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1549                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1550                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1551                         foreach ( $whitelist as $entry ) {
1552                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1553                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1554                                         continue;
1555                                 }
1556                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1557                                         # Image matches a whitelist entry
1558                                         $text = $sk->makeExternalImage( $url );
1559                                         break;
1560                                 }
1561                         }
1562                 }
1563                 return $text;
1564         }
1565
1566         /**
1567          * Process [[ ]] wikilinks
1568          * @return String: processed text
1569          *
1570          * @private
1571          */
1572         function replaceInternalLinks( $s ) {
1573                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1574                 return $s;
1575         }
1576
1577         /**
1578          * Process [[ ]] wikilinks (RIL)
1579          * @return LinkHolderArray
1580          *
1581          * @private
1582          */
1583         function replaceInternalLinks2( &$s ) {
1584                 global $wgContLang;
1585
1586                 wfProfileIn( __METHOD__ );
1587
1588                 wfProfileIn( __METHOD__.'-setup' );
1589                 static $tc = FALSE, $e1, $e1_img;
1590                 # the % is needed to support urlencoded titles as well
1591                 if ( !$tc ) {
1592                         $tc = Title::legalChars() . '#%';
1593                         # Match a link having the form [[namespace:link|alternate]]trail
1594                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1595                         # Match cases where there is no "]]", which might still be images
1596                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1597                 }
1598
1599                 $sk = $this->mOptions->getSkin( $this->mTitle );
1600                 $holders = new LinkHolderArray( $this );
1601
1602                 # split the entire text string on occurences of [[
1603                 $a = StringUtils::explode( '[[', ' ' . $s );
1604                 # get the first element (all text up to first [[), and remove the space we added
1605                 $s = $a->current();
1606                 $a->next();
1607                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1608                 $s = substr( $s, 1 );
1609
1610                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1611                 $e2 = null;
1612                 if ( $useLinkPrefixExtension ) {
1613                         # Match the end of a line for a word that's not followed by whitespace,
1614                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1615                         $e2 = wfMsgForContent( 'linkprefix' );
1616                 }
1617
1618                 if ( is_null( $this->mTitle ) ) {
1619                         wfProfileOut( __METHOD__.'-setup' );
1620                         wfProfileOut( __METHOD__ );
1621                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1622                 }
1623                 $nottalk = !$this->mTitle->isTalkPage();
1624
1625                 if ( $useLinkPrefixExtension ) {
1626                         $m = array();
1627                         if ( preg_match( $e2, $s, $m ) ) {
1628                                 $first_prefix = $m[2];
1629                         } else {
1630                                 $first_prefix = false;
1631                         }
1632                 } else {
1633                         $prefix = '';
1634                 }
1635
1636                 if ( $wgContLang->hasVariants() ) {
1637                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1638                 } else {
1639                         $selflink = array( $this->mTitle->getPrefixedText() );
1640                 }
1641                 $useSubpages = $this->areSubpagesAllowed();
1642                 wfProfileOut( __METHOD__.'-setup' );
1643
1644                 # Loop for each link
1645                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1646                         # Check for excessive memory usage
1647                         if ( $holders->isBig() ) {
1648                                 # Too big
1649                                 # Do the existence check, replace the link holders and clear the array
1650                                 $holders->replace( $s );
1651                                 $holders->clear();
1652                         }
1653
1654                         if ( $useLinkPrefixExtension ) {
1655                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1656                                 if ( preg_match( $e2, $s, $m ) ) {
1657                                         $prefix = $m[2];
1658                                         $s = $m[1];
1659                                 } else {
1660                                         $prefix='';
1661                                 }
1662                                 # first link
1663                                 if ( $first_prefix ) {
1664                                         $prefix = $first_prefix;
1665                                         $first_prefix = false;
1666                                 }
1667                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1668                         }
1669
1670                         $might_be_img = false;
1671
1672                         wfProfileIn( __METHOD__."-e1" );
1673                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1674                                 $text = $m[2];
1675                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1676                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1677                                 # the real problem is with the $e1 regex
1678                                 # See bug 1300.
1679                                 #
1680                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1681                                 # and no image is in sight. See bug 2095.
1682                                 #
1683                                 if ( $text !== '' &&
1684                                         substr( $m[3], 0, 1 ) === ']' &&
1685                                         strpos( $text, '[' ) !== false
1686                                 )
1687                                 {
1688                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1689                                         $m[3] = substr( $m[3], 1 );
1690                                 }
1691                                 # fix up urlencoded title texts
1692                                 if ( strpos( $m[1], '%' ) !== false ) {
1693                                         # Should anchors '#' also be rejected?
1694                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode( $m[1] ) );
1695                                 }
1696                                 $trail = $m[3];
1697                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1698                                 $might_be_img = true;
1699                                 $text = $m[2];
1700                                 if ( strpos( $m[1], '%' ) !== false ) {
1701                                         $m[1] = urldecode( $m[1] );
1702                                 }
1703                                 $trail = "";
1704                         } else { # Invalid form; output directly
1705                                 $s .= $prefix . '[[' . $line ;
1706                                 wfProfileOut( __METHOD__."-e1" );
1707                                 continue;
1708                         }
1709                         wfProfileOut( __METHOD__."-e1" );
1710                         wfProfileIn( __METHOD__."-misc" );
1711
1712                         # Don't allow internal links to pages containing
1713                         # PROTO: where PROTO is a valid URL protocol; these
1714                         # should be external links.
1715                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1716                                 $s .= $prefix . '[[' . $line ;
1717                                 wfProfileOut( __METHOD__."-misc" );
1718                                 continue;
1719                         }
1720
1721                         # Make subpage if necessary
1722                         if ( $useSubpages ) {
1723                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1724                         } else {
1725                                 $link = $m[1];
1726                         }
1727
1728                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1729                         if ( !$noforce ) {
1730                                 # Strip off leading ':'
1731                                 $link = substr( $link, 1 );
1732                         }
1733
1734                         wfProfileOut( __METHOD__."-misc" );
1735                         wfProfileIn( __METHOD__."-title" );
1736                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1737                         if ( $nt === null ) {
1738                                 $s .= $prefix . '[[' . $line;
1739                                 wfProfileOut( __METHOD__."-title" );
1740                                 continue;
1741                         }
1742
1743                         $ns = $nt->getNamespace();
1744                         $iw = $nt->getInterWiki();
1745                         wfProfileOut( __METHOD__."-title" );
1746
1747                         if ( $might_be_img ) { # if this is actually an invalid link
1748                                 wfProfileIn( __METHOD__."-might_be_img" );
1749                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1750                                         $found = false;
1751                                         while ( true ) {
1752                                                 # look at the next 'line' to see if we can close it there
1753                                                 $a->next();
1754                                                 $next_line = $a->current();
1755                                                 if ( $next_line === false || $next_line === null ) {
1756                                                         break;
1757                                                 }
1758                                                 $m = explode( ']]', $next_line, 3 );
1759                                                 if ( count( $m ) == 3 ) {
1760                                                         # the first ]] closes the inner link, the second the image
1761                                                         $found = true;
1762                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1763                                                         $trail = $m[2];
1764                                                         break;
1765                                                 } elseif ( count( $m ) == 2 ) {
1766                                                         # if there's exactly one ]] that's fine, we'll keep looking
1767                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1768                                                 } else {
1769                                                         # if $next_line is invalid too, we need look no further
1770                                                         $text .= '[[' . $next_line;
1771                                                         break;
1772                                                 }
1773                                         }
1774                                         if ( !$found ) {
1775                                                 # we couldn't find the end of this imageLink, so output it raw
1776                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1777                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1778                                                 $s .= "{$prefix}[[$link|$text";
1779                                                 # note: no $trail, because without an end, there *is* no trail
1780                                                 wfProfileOut( __METHOD__."-might_be_img" );
1781                                                 continue;
1782                                         }
1783                                 } else { # it's not an image, so output it raw
1784                                         $s .= "{$prefix}[[$link|$text";
1785                                         # note: no $trail, because without an end, there *is* no trail
1786                                         wfProfileOut( __METHOD__."-might_be_img" );
1787                                         continue;
1788                                 }
1789                                 wfProfileOut( __METHOD__."-might_be_img" );
1790                         }
1791
1792                         $wasblank = ( $text  == '' );
1793                         if ( $wasblank ) {
1794                                 $text = $link;
1795                         } else {
1796                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1797                                 # [[Lista d''e paise d''o munno]] -> <a href="">Lista d''e paise d''o munno</a>
1798                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']] -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1799                                 $text = $this->doQuotes($text);
1800                         }
1801
1802                         # Link not escaped by : , create the various objects
1803                         if ( $noforce ) {
1804
1805                                 # Interwikis
1806                                 wfProfileIn( __METHOD__."-interwiki" );
1807                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1808                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1809                                         $s = rtrim( $s . $prefix );
1810                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1811                                         wfProfileOut( __METHOD__."-interwiki" );
1812                                         continue;
1813                                 }
1814                                 wfProfileOut( __METHOD__."-interwiki" );
1815
1816                                 if ( $ns == NS_FILE ) {
1817                                         wfProfileIn( __METHOD__."-image" );
1818                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1819                                                 if ( $wasblank ) {
1820                                                         # if no parameters were passed, $text
1821                                                         # becomes something like "File:Foo.png",
1822                                                         # which we don't want to pass on to the
1823                                                         # image generator
1824                                                         $text = '';
1825                                                 } else {
1826                                                         # recursively parse links inside the image caption
1827                                                         # actually, this will parse them in any other parameters, too,
1828                                                         # but it might be hard to fix that, and it doesn't matter ATM
1829                                                         $text = $this->replaceExternalLinks( $text );
1830                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1831                                                 }
1832                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1833                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1834                                         } else {
1835                                                 $s .= $prefix . $trail;
1836                                         }
1837                                         $this->mOutput->addImage( $nt->getDBkey() );
1838                                         wfProfileOut( __METHOD__."-image" );
1839                                         continue;
1840
1841                                 }
1842
1843                                 if ( $ns == NS_CATEGORY ) {
1844                                         wfProfileIn( __METHOD__."-category" );
1845                                         $s = rtrim( $s . "\n" ); # bug 87
1846
1847                                         if ( $wasblank ) {
1848                                                 $sortkey = $this->getDefaultSort();
1849                                         } else {
1850                                                 $sortkey = $text;
1851                                         }
1852                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1853                                         $sortkey = str_replace( "\n", '', $sortkey );
1854                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1855                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1856
1857                                         /**
1858                                          * Strip the whitespace Category links produce, see bug 87
1859                                          * @todo We might want to use trim($tmp, "\n") here.
1860                                          */
1861                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '': $prefix . $trail;
1862
1863                                         wfProfileOut( __METHOD__."-category" );
1864                                         continue;
1865                                 }
1866                         }
1867
1868                         # Self-link checking
1869                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1870                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1871                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1872                                         continue;
1873                                 }
1874                         }
1875
1876                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1877                         # FIXME: Should do batch file existence checks, see comment below
1878                         if ( $ns == NS_MEDIA ) {
1879                                 wfProfileIn( __METHOD__."-media" );
1880                                 # Give extensions a chance to select the file revision for us
1881                                 $skip = $time = false;
1882                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1883                                 if ( $skip ) {
1884                                         $link = $sk->link( $nt );
1885                                 } else {
1886                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1887                                 }
1888                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1889                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1890                                 $this->mOutput->addImage( $nt->getDBkey() );
1891                                 wfProfileOut( __METHOD__."-media" );
1892                                 continue;
1893                         }
1894
1895                         wfProfileIn( __METHOD__."-always_known" );
1896                         # Some titles, such as valid special pages or files in foreign repos, should
1897                         # be shown as bluelinks even though they're not included in the page table
1898                         #
1899                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1900                         # batch file existence checks for NS_FILE and NS_MEDIA
1901                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1902                                 $this->mOutput->addLink( $nt );
1903                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1904                         } else {
1905                                 # Links will be added to the output link list after checking
1906                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1907                         }
1908                         wfProfileOut( __METHOD__."-always_known" );
1909                 }
1910                 wfProfileOut( __METHOD__ );
1911                 return $holders;
1912         }
1913
1914         /**
1915          * Make a link placeholder. The text returned can be later resolved to a real link with
1916          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1917          * parsing of interwiki links, and secondly to allow all existence checks and
1918          * article length checks (for stub links) to be bundled into a single query.
1919          *
1920          * @deprecated
1921          */
1922         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1923                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1924         }
1925
1926         /**
1927          * Render a forced-blue link inline; protect against double expansion of
1928          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1929          * Since this little disaster has to split off the trail text to avoid
1930          * breaking URLs in the following text without breaking trails on the
1931          * wiki links, it's been made into a horrible function.
1932          *
1933          * @param $nt Title
1934          * @param $text String
1935          * @param $query String
1936          * @param $trail String
1937          * @param $prefix String
1938          * @return String: HTML-wikitext mix oh yuck
1939          */
1940         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1941                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1942                 $sk = $this->mOptions->getSkin( $this->mTitle );
1943                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1944                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1945                 return $this->armorLinks( $link ) . $trail;
1946         }
1947
1948         /**
1949          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1950          * going to go through further parsing steps before inline URL expansion.
1951          *
1952          * Not needed quite as much as it used to be since free links are a bit
1953          * more sensible these days. But bracketed links are still an issue.
1954          *
1955          * @param $text String: more-or-less HTML
1956          * @return String: less-or-more HTML with NOPARSE bits
1957          */
1958         function armorLinks( $text ) {
1959                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1960                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1961         }
1962
1963         /**
1964          * Return true if subpage links should be expanded on this page.
1965          * @return Boolean
1966          */
1967         function areSubpagesAllowed() {
1968                 # Some namespaces don't allow subpages
1969                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1970         }
1971
1972         /**
1973          * Handle link to subpage if necessary
1974          *
1975          * @param $target String: the source of the link
1976          * @param &$text String: the link text, modified as necessary
1977          * @return string the full name of the link
1978          * @private
1979          */
1980         function maybeDoSubpageLink( $target, &$text ) {
1981                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1982         }
1983
1984         /**#@+
1985          * Used by doBlockLevels()
1986          * @private
1987          */
1988         function closeParagraph() {
1989                 $result = '';
1990                 if ( $this->mLastSection != '' ) {
1991                         $result = '</' . $this->mLastSection  . ">\n";
1992                 }
1993                 $this->mInPre = false;
1994                 $this->mLastSection = '';
1995                 return $result;
1996         }
1997
1998         /**
1999          * getCommon() returns the length of the longest common substring
2000          * of both arguments, starting at the beginning of both.
2001          * @private
2002          */
2003         function getCommon( $st1, $st2 ) {
2004                 $fl = strlen( $st1 );
2005                 $shorter = strlen( $st2 );
2006                 if ( $fl < $shorter ) {
2007                         $shorter = $fl;
2008                 }
2009
2010                 for ( $i = 0; $i < $shorter; ++$i ) {
2011                         if ( $st1{$i} != $st2{$i} ) {
2012                                 break;
2013                         }
2014                 }
2015                 return $i;
2016         }
2017
2018         /**
2019          * These next three functions open, continue, and close the list
2020          * element appropriate to the prefix character passed into them.
2021          * @private
2022          */
2023         function openList( $char ) {
2024                 $result = $this->closeParagraph();
2025
2026                 if ( '*' === $char ) {
2027                         $result .= '<ul><li>';
2028                 } elseif ( '#' === $char ) {
2029                         $result .= '<ol><li>';
2030                 } elseif ( ':' === $char ) {
2031                         $result .= '<dl><dd>';
2032                 } elseif ( ';' === $char ) {
2033                         $result .= '<dl><dt>';
2034                         $this->mDTopen = true;
2035                 } else {
2036                         $result = '<!-- ERR 1 -->';
2037                 }
2038
2039                 return $result;
2040         }
2041
2042         /**
2043          * TODO: document
2044          * @param $char String
2045          * @private
2046          */
2047         function nextItem( $char ) {
2048                 if ( '*' === $char || '#' === $char ) {
2049                         return '</li><li>';
2050                 } elseif ( ':' === $char || ';' === $char ) {
2051                         $close = '</dd>';
2052                         if ( $this->mDTopen ) {
2053                                 $close = '</dt>';
2054                         }
2055                         if ( ';' === $char ) {
2056                                 $this->mDTopen = true;
2057                                 return $close . '<dt>';
2058                         } else {
2059                                 $this->mDTopen = false;
2060                                 return $close . '<dd>';
2061                         }
2062                 }
2063                 return '<!-- ERR 2 -->';
2064         }
2065
2066         /**
2067          * TODO: document
2068          * @param $char String
2069          * @private
2070          */
2071         function closeList( $char ) {
2072                 if ( '*' === $char ) {
2073                         $text = '</li></ul>';
2074                 } elseif ( '#' === $char ) {
2075                         $text = '</li></ol>';
2076                 } elseif ( ':' === $char ) {
2077                         if ( $this->mDTopen ) {
2078                                 $this->mDTopen = false;
2079                                 $text = '</dt></dl>';
2080                         } else {
2081                                 $text = '</dd></dl>';
2082                         }
2083                 } else {
2084                         return '<!-- ERR 3 -->';
2085                 }
2086                 return $text."\n";
2087         }
2088         /**#@-*/
2089
2090         /**
2091          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2092          *
2093          * @param $text String
2094          * @param $linestart Boolean: whether or not this is at the start of a line.
2095          * @private
2096          * @return string the lists rendered as HTML
2097          */
2098         function doBlockLevels( $text, $linestart ) {
2099                 wfProfileIn( __METHOD__ );
2100
2101                 # Parsing through the text line by line.  The main thing
2102                 # happening here is handling of block-level elements p, pre,
2103                 # and making lists from lines starting with * # : etc.
2104                 #
2105                 $textLines = StringUtils::explode( "\n", $text );
2106
2107                 $lastPrefix = $output = '';
2108                 $this->mDTopen = $inBlockElem = false;
2109                 $prefixLength = 0;
2110                 $paragraphStack = false;
2111
2112                 foreach ( $textLines as $oLine ) {
2113                         # Fix up $linestart
2114                         if ( !$linestart ) {
2115                                 $output .= $oLine;
2116                                 $linestart = true;
2117                                 continue;
2118                         }
2119                         # * = ul
2120                         # # = ol
2121                         # ; = dt
2122                         # : = dd
2123
2124                         $lastPrefixLength = strlen( $lastPrefix );
2125                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2126                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2127                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2128                         if ( !$this->mInPre ) {
2129                                 # Multiple prefixes may abut each other for nested lists.
2130                                 $prefixLength = strspn( $oLine, '*#:;' );
2131                                 $prefix = substr( $oLine, 0, $prefixLength );
2132
2133                                 # eh?
2134                                 # ; and : are both from definition-lists, so they're equivalent
2135                                 #  for the purposes of determining whether or not we need to open/close
2136                                 #  elements.
2137                                 $prefix2 = str_replace( ';', ':', $prefix );
2138                                 $t = substr( $oLine, $prefixLength );
2139                                 $this->mInPre = (bool)$preOpenMatch;
2140                         } else {
2141                                 # Don't interpret any other prefixes in preformatted text
2142                                 $prefixLength = 0;
2143                                 $prefix = $prefix2 = '';
2144                                 $t = $oLine;
2145                         }
2146
2147                         # List generation
2148                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2149                                 # Same as the last item, so no need to deal with nesting or opening stuff
2150                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2151                                 $paragraphStack = false;
2152
2153                                 if ( substr( $prefix, -1 ) === ';') {
2154                                         # The one nasty exception: definition lists work like this:
2155                                         # ; title : definition text
2156                                         # So we check for : in the remainder text to split up the
2157                                         # title and definition, without b0rking links.
2158                                         $term = $t2 = '';
2159                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2160                                                 $t = $t2;
2161                                                 $output .= $term . $this->nextItem( ':' );
2162                                         }
2163                                 }
2164                         } elseif ( $prefixLength || $lastPrefixLength ) {
2165                                 # We need to open or close prefixes, or both.
2166
2167                                 # Either open or close a level...
2168                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2169                                 $paragraphStack = false;
2170
2171                                 # Close all the prefixes which aren't shared.
2172                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2173                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2174                                         --$lastPrefixLength;
2175                                 }
2176
2177                                 # Continue the current prefix if appropriate.
2178                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2179                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2180                                 }
2181
2182                                 # Open prefixes where appropriate.
2183                                 while ( $prefixLength > $commonPrefixLength ) {
2184                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2185                                         $output .= $this->openList( $char );
2186
2187                                         if ( ';' === $char ) {
2188                                                 # FIXME: This is dupe of code above
2189                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2190                                                         $t = $t2;
2191                                                         $output .= $term . $this->nextItem( ':' );
2192                                                 }
2193                                         }
2194                                         ++$commonPrefixLength;
2195                                 }
2196                                 $lastPrefix = $prefix2;
2197                         }
2198
2199                         # If we have no prefixes, go to paragraph mode.
2200                         if ( 0 == $prefixLength ) {
2201                                 wfProfileIn( __METHOD__."-paragraph" );
2202                                 # No prefix (not in list)--go to paragraph mode
2203                                 # XXX: use a stack for nestable elements like span, table and div
2204                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2205                                 $closematch = preg_match(
2206                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2207                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2208                                 if ( $openmatch or $closematch ) {
2209                                         $paragraphStack = false;
2210                                         # TODO bug 5718: paragraph closed
2211                                         $output .= $this->closeParagraph();
2212                                         if ( $preOpenMatch and !$preCloseMatch ) {
2213                                                 $this->mInPre = true;
2214                                         }
2215                                         $inBlockElem = !$closematch;
2216                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2217                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2218                                                 # pre
2219                                                 if ( $this->mLastSection !== 'pre' ) {
2220                                                         $paragraphStack = false;
2221                                                         $output .= $this->closeParagraph().'<pre>';
2222                                                         $this->mLastSection = 'pre';
2223                                                 }
2224                                                 $t = substr( $t, 1 );
2225                                         } else {
2226                                                 # paragraph
2227                                                 if ( trim( $t ) === '' ) {
2228                                                         if ( $paragraphStack ) {
2229                                                                 $output .= $paragraphStack.'<br />';
2230                                                                 $paragraphStack = false;
2231                                                                 $this->mLastSection = 'p';
2232                                                         } else {
2233                                                                 if ( $this->mLastSection !== 'p' ) {
2234                                                                         $output .= $this->closeParagraph();
2235                                                                         $this->mLastSection = '';
2236                                                                         $paragraphStack = '<p>';
2237                                                                 } else {
2238                                                                         $paragraphStack = '</p><p>';
2239                                                                 }
2240                                                         }
2241                                                 } else {
2242                                                         if ( $paragraphStack ) {
2243                                                                 $output .= $paragraphStack;
2244                                                                 $paragraphStack = false;
2245                                                                 $this->mLastSection = 'p';
2246                                                         } elseif ( $this->mLastSection !== 'p' ) {
2247                                                                 $output .= $this->closeParagraph().'<p>';
2248                                                                 $this->mLastSection = 'p';
2249                                                         }
2250                                                 }
2251                                         }
2252                                 }
2253                                 wfProfileOut( __METHOD__."-paragraph" );
2254                         }
2255                         # somewhere above we forget to get out of pre block (bug 785)
2256                         if ( $preCloseMatch && $this->mInPre ) {
2257                                 $this->mInPre = false;
2258                         }
2259                         if ( $paragraphStack === false ) {
2260                                 $output .= $t."\n";
2261                         }
2262                 }
2263                 while ( $prefixLength ) {
2264                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2265                         --$prefixLength;
2266                 }
2267                 if ( $this->mLastSection != '' ) {
2268                         $output .= '</' . $this->mLastSection . '>';
2269                         $this->mLastSection = '';
2270                 }
2271
2272                 wfProfileOut( __METHOD__ );
2273                 return $output;
2274         }
2275
2276         /**
2277          * Split up a string on ':', ignoring any occurences inside tags
2278          * to prevent illegal overlapping.
2279          *
2280          * @param $str String: the string to split
2281          * @param &$before String: set to everything before the ':'
2282          * @param &$after String: set to everything after the ':'
2283          * return String: the position of the ':', or false if none found
2284          */
2285         function findColonNoLinks( $str, &$before, &$after ) {
2286                 wfProfileIn( __METHOD__ );
2287
2288                 $pos = strpos( $str, ':' );
2289                 if ( $pos === false ) {
2290                         # Nothing to find!
2291                         wfProfileOut( __METHOD__ );
2292                         return false;
2293                 }
2294
2295                 $lt = strpos( $str, '<' );
2296                 if ( $lt === false || $lt > $pos ) {
2297                         # Easy; no tag nesting to worry about
2298                         $before = substr( $str, 0, $pos );
2299                         $after = substr( $str, $pos+1 );
2300                         wfProfileOut( __METHOD__ );
2301                         return $pos;
2302                 }
2303
2304                 # Ugly state machine to walk through avoiding tags.
2305                 $state = self::COLON_STATE_TEXT;
2306                 $stack = 0;
2307                 $len = strlen( $str );
2308                 for( $i = 0; $i < $len; $i++ ) {
2309                         $c = $str{$i};
2310
2311                         switch( $state ) {
2312                         # (Using the number is a performance hack for common cases)
2313                         case 0: # self::COLON_STATE_TEXT:
2314                                 switch( $c ) {
2315                                 case "<":
2316                                         # Could be either a <start> tag or an </end> tag
2317                                         $state = self::COLON_STATE_TAGSTART;
2318                                         break;
2319                                 case ":":
2320                                         if ( $stack == 0 ) {
2321                                                 # We found it!
2322                                                 $before = substr( $str, 0, $i );
2323                                                 $after = substr( $str, $i + 1 );
2324                                                 wfProfileOut( __METHOD__ );
2325                                                 return $i;
2326                                         }
2327                                         # Embedded in a tag; don't break it.
2328                                         break;
2329                                 default:
2330                                         # Skip ahead looking for something interesting
2331                                         $colon = strpos( $str, ':', $i );
2332                                         if ( $colon === false ) {
2333                                                 # Nothing else interesting
2334                                                 wfProfileOut( __METHOD__ );
2335                                                 return false;
2336                                         }
2337                                         $lt = strpos( $str, '<', $i );
2338                                         if ( $stack === 0 ) {
2339                                                 if ( $lt === false || $colon < $lt ) {
2340                                                         # We found it!
2341                                                         $before = substr( $str, 0, $colon );
2342                                                         $after = substr( $str, $colon + 1 );
2343                                                         wfProfileOut( __METHOD__ );
2344                                                         return $i;
2345                                                 }
2346                                         }
2347                                         if ( $lt === false ) {
2348                                                 # Nothing else interesting to find; abort!
2349                                                 # We're nested, but there's no close tags left. Abort!
2350                                                 break 2;
2351                                         }
2352                                         # Skip ahead to next tag start
2353                                         $i = $lt;
2354                                         $state = self::COLON_STATE_TAGSTART;
2355                                 }
2356                                 break;
2357                         case 1: # self::COLON_STATE_TAG:
2358                                 # In a <tag>
2359                                 switch( $c ) {
2360                                 case ">":
2361                                         $stack++;
2362                                         $state = self::COLON_STATE_TEXT;
2363                                         break;
2364                                 case "/":
2365                                         # Slash may be followed by >?
2366                                         $state = self::COLON_STATE_TAGSLASH;
2367                                         break;
2368                                 default:
2369                                         # ignore
2370                                 }
2371                                 break;
2372                         case 2: # self::COLON_STATE_TAGSTART:
2373                                 switch( $c ) {
2374                                 case "/":
2375                                         $state = self::COLON_STATE_CLOSETAG;
2376                                         break;
2377                                 case "!":
2378                                         $state = self::COLON_STATE_COMMENT;
2379                                         break;
2380                                 case ">":
2381                                         # Illegal early close? This shouldn't happen D:
2382                                         $state = self::COLON_STATE_TEXT;
2383                                         break;
2384                                 default:
2385                                         $state = self::COLON_STATE_TAG;
2386                                 }
2387                                 break;
2388                         case 3: # self::COLON_STATE_CLOSETAG:
2389                                 # In a </tag>
2390                                 if ( $c === ">" ) {
2391                                         $stack--;
2392                                         if ( $stack < 0 ) {
2393                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2394                                                 wfProfileOut( __METHOD__ );
2395                                                 return false;
2396                                         }
2397                                         $state = self::COLON_STATE_TEXT;
2398                                 }
2399                                 break;
2400                         case self::COLON_STATE_TAGSLASH:
2401                                 if ( $c === ">" ) {
2402                                         # Yes, a self-closed tag <blah/>
2403                                         $state = self::COLON_STATE_TEXT;
2404                                 } else {
2405                                         # Probably we're jumping the gun, and this is an attribute
2406                                         $state = self::COLON_STATE_TAG;
2407                                 }
2408                                 break;
2409                         case 5: # self::COLON_STATE_COMMENT:
2410                                 if ( $c === "-" ) {
2411                                         $state = self::COLON_STATE_COMMENTDASH;
2412                                 }
2413                                 break;
2414                         case self::COLON_STATE_COMMENTDASH:
2415                                 if ( $c === "-" ) {
2416                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2417                                 } else {
2418                                         $state = self::COLON_STATE_COMMENT;
2419                                 }
2420                                 break;
2421                         case self::COLON_STATE_COMMENTDASHDASH:
2422                                 if ( $c === ">" ) {
2423                                         $state = self::COLON_STATE_TEXT;
2424                                 } else {
2425                                         $state = self::COLON_STATE_COMMENT;
2426                                 }
2427                                 break;
2428                         default:
2429                                 throw new MWException( "State machine error in " . __METHOD__ );
2430                         }
2431                 }
2432                 if ( $stack > 0 ) {
2433                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2434                         return false;
2435                 }
2436                 wfProfileOut( __METHOD__ );
2437                 return false;
2438         }
2439
2440         /**
2441          * Return value of a magic variable (like PAGENAME)
2442          *
2443          * @private
2444          */
2445         function getVariableValue( $index, $frame=false ) {
2446                 global $wgContLang, $wgSitename, $wgServer;
2447                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2448
2449                 /**
2450                  * Some of these require message or data lookups and can be
2451                  * expensive to check many times.
2452                  */
2453                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2454                         if ( isset( $this->mVarCache[$index] ) ) {
2455                                 return $this->mVarCache[$index];
2456                         }
2457                 }
2458
2459                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2460                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2461
2462                 # Use the time zone
2463                 global $wgLocaltimezone;
2464                 if ( isset( $wgLocaltimezone ) ) {
2465                         $oldtz = date_default_timezone_get();
2466                         date_default_timezone_set( $wgLocaltimezone );
2467                 }
2468
2469                 $localTimestamp = date( 'YmdHis', $ts );
2470                 $localMonth = date( 'm', $ts );
2471                 $localMonth1 = date( 'n', $ts );
2472                 $localMonthName = date( 'n', $ts );
2473                 $localDay = date( 'j', $ts );
2474                 $localDay2 = date( 'd', $ts );
2475                 $localDayOfWeek = date( 'w', $ts );
2476                 $localWeek = date( 'W', $ts );
2477                 $localYear = date( 'Y', $ts );
2478                 $localHour = date( 'H', $ts );
2479                 if ( isset( $wgLocaltimezone ) ) {
2480                         date_default_timezone_set( $oldtz );
2481                 }
2482
2483                 switch ( $index ) {
2484                         case 'currentmonth':
2485                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2486                                 break;
2487                         case 'currentmonth1':
2488                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2489                                 break;
2490                         case 'currentmonthname':
2491                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2492                                 break;
2493                         case 'currentmonthnamegen':
2494                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2495                                 break;
2496                         case 'currentmonthabbrev':
2497                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2498                                 break;
2499                         case 'currentday':
2500                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2501                                 break;
2502                         case 'currentday2':
2503                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2504                                 break;
2505                         case 'localmonth':
2506                                 $value = $wgContLang->formatNum( $localMonth );
2507                                 break;
2508                         case 'localmonth1':
2509                                 $value = $wgContLang->formatNum( $localMonth1 );
2510                                 break;
2511                         case 'localmonthname':
2512                                 $value = $wgContLang->getMonthName( $localMonthName );
2513                                 break;
2514                         case 'localmonthnamegen':
2515                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2516                                 break;
2517                         case 'localmonthabbrev':
2518                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2519                                 break;
2520                         case 'localday':
2521                                 $value = $wgContLang->formatNum( $localDay );
2522                                 break;
2523                         case 'localday2':
2524                                 $value = $wgContLang->formatNum( $localDay2 );
2525                                 break;
2526                         case 'pagename':
2527                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2528                                 break;
2529                         case 'pagenamee':
2530                                 $value = $this->mTitle->getPartialURL();
2531                                 break;
2532                         case 'fullpagename':
2533                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2534                                 break;
2535                         case 'fullpagenamee':
2536                                 $value = $this->mTitle->getPrefixedURL();
2537                                 break;
2538                         case 'subpagename':
2539                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2540                                 break;
2541                         case 'subpagenamee':
2542                                 $value = $this->mTitle->getSubpageUrlForm();
2543                                 break;
2544                         case 'basepagename':
2545                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2546                                 break;
2547                         case 'basepagenamee':
2548                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2549                                 break;
2550                         case 'talkpagename':
2551                                 if ( $this->mTitle->canTalk() ) {
2552                                         $talkPage = $this->mTitle->getTalkPage();
2553                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2554                                 } else {
2555                                         $value = '';
2556                                 }
2557                                 break;
2558                         case 'talkpagenamee':
2559                                 if ( $this->mTitle->canTalk() ) {
2560                                         $talkPage = $this->mTitle->getTalkPage();
2561                                         $value = $talkPage->getPrefixedUrl();
2562                                 } else {
2563                                         $value = '';
2564                                 }
2565                                 break;
2566                         case 'subjectpagename':
2567                                 $subjPage = $this->mTitle->getSubjectPage();
2568                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2569                                 break;
2570                         case 'subjectpagenamee':
2571                                 $subjPage = $this->mTitle->getSubjectPage();
2572                                 $value = $subjPage->getPrefixedUrl();
2573                                 break;
2574                         case 'revisionid':
2575                                 # Let the edit saving system know we should parse the page
2576                                 # *after* a revision ID has been assigned.
2577                                 $this->mOutput->setFlag( 'vary-revision' );
2578                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2579                                 $value = $this->mRevisionId;
2580                                 break;
2581                         case 'revisionday':
2582                                 # Let the edit saving system know we should parse the page
2583                                 # *after* a revision ID has been assigned. This is for null edits.
2584                                 $this->mOutput->setFlag( 'vary-revision' );
2585                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2586                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2587                                 break;
2588                         case 'revisionday2':
2589                                 # Let the edit saving system know we should parse the page
2590                                 # *after* a revision ID has been assigned. This is for null edits.
2591                                 $this->mOutput->setFlag( 'vary-revision' );
2592                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2593                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2594                                 break;
2595                         case 'revisionmonth':
2596                                 # Let the edit saving system know we should parse the page
2597                                 # *after* a revision ID has been assigned. This is for null edits.
2598                                 $this->mOutput->setFlag( 'vary-revision' );
2599                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2600                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2601                                 break;
2602                         case 'revisionmonth1':
2603                                 # Let the edit saving system know we should parse the page
2604                                 # *after* a revision ID has been assigned. This is for null edits.
2605                                 $this->mOutput->setFlag( 'vary-revision' );
2606                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2607                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2608                                 break;
2609                         case 'revisionyear':
2610                                 # Let the edit saving system know we should parse the page
2611                                 # *after* a revision ID has been assigned. This is for null edits.
2612                                 $this->mOutput->setFlag( 'vary-revision' );
2613                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2614                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2615                                 break;
2616                         case 'revisiontimestamp':
2617                                 # Let the edit saving system know we should parse the page
2618                                 # *after* a revision ID has been assigned. This is for null edits.
2619                                 $this->mOutput->setFlag( 'vary-revision' );
2620                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2621                                 $value = $this->getRevisionTimestamp();
2622                                 break;
2623                         case 'revisionuser':
2624                                 # Let the edit saving system know we should parse the page
2625                                 # *after* a revision ID has been assigned. This is for null edits.
2626                                 $this->mOutput->setFlag( 'vary-revision' );
2627                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2628                                 $value = $this->getRevisionUser();
2629                                 break;
2630                         case 'namespace':
2631                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2632                                 break;
2633                         case 'namespacee':
2634                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2635                                 break;
2636                         case 'talkspace':
2637                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2638                                 break;
2639                         case 'talkspacee':
2640                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2641                                 break;
2642                         case 'subjectspace':
2643                                 $value = $this->mTitle->getSubjectNsText();
2644                                 break;
2645                         case 'subjectspacee':
2646                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2647                                 break;
2648                         case 'currentdayname':
2649                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2650                                 break;
2651                         case 'currentyear':
2652                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2653                                 break;
2654                         case 'currenttime':
2655                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2656                                 break;
2657                         case 'currenthour':
2658                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2659                                 break;
2660                         case 'currentweek':
2661                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2662                                 # int to remove the padding
2663                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2664                                 break;
2665                         case 'currentdow':
2666                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2667                                 break;
2668                         case 'localdayname':
2669                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2670                                 break;
2671                         case 'localyear':
2672                                 $value = $wgContLang->formatNum( $localYear, true );
2673                                 break;
2674                         case 'localtime':
2675                                 $value = $wgContLang->time( $localTimestamp, false, false );
2676                                 break;
2677                         case 'localhour':
2678                                 $value = $wgContLang->formatNum( $localHour, true );
2679                                 break;
2680                         case 'localweek':
2681                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2682                                 # int to remove the padding
2683                                 $value = $wgContLang->formatNum( (int)$localWeek );
2684                                 break;
2685                         case 'localdow':
2686                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2687                                 break;
2688                         case 'numberofarticles':
2689                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2690                                 break;
2691                         case 'numberoffiles':
2692                                 $value = $wgContLang->formatNum( SiteStats::images() );
2693                                 break;
2694                         case 'numberofusers':
2695                                 $value = $wgContLang->formatNum( SiteStats::users() );
2696                                 break;
2697                         case 'numberofactiveusers':
2698                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2699                                 break;
2700                         case 'numberofpages':
2701                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2702                                 break;
2703                         case 'numberofadmins':
2704                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2705                                 break;
2706                         case 'numberofedits':
2707                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2708                                 break;
2709                         case 'numberofviews':
2710                                 $value = $wgContLang->formatNum( SiteStats::views() );
2711                                 break;
2712                         case 'currenttimestamp':
2713                                 $value = wfTimestamp( TS_MW, $ts );
2714                                 break;
2715                         case 'localtimestamp':
2716                                 $value = $localTimestamp;
2717                                 break;
2718                         case 'currentversion':
2719                                 $value = SpecialVersion::getVersion();
2720                                 break;
2721                         case 'articlepath':
2722                                 return $wgArticlePath;
2723                         case 'sitename':
2724                                 return $wgSitename;
2725                         case 'server':
2726                                 return $wgServer;
2727                         case 'servername':
2728                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2729                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2730                                 wfRestoreWarnings();
2731                                 return $serverName ? $serverName : $wgServer;
2732                         case 'scriptpath':
2733                                 return $wgScriptPath;
2734                         case 'stylepath':
2735                                 return $wgStylePath;
2736                         case 'directionmark':
2737                                 return $wgContLang->getDirMark();
2738                         case 'contentlanguage':
2739                                 global $wgLanguageCode;
2740                                 return $wgLanguageCode;
2741                         default:
2742                                 $ret = null;
2743                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2744                                         return $ret;
2745                                 } else {
2746                                         return null;
2747                                 }
2748                 }
2749
2750                 if ( $index )
2751                         $this->mVarCache[$index] = $value;
2752
2753                 return $value;
2754         }
2755
2756         /**
2757          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2758          *
2759          * @private
2760          */
2761         function initialiseVariables() {
2762                 wfProfileIn( __METHOD__ );
2763                 $variableIDs = MagicWord::getVariableIDs();
2764                 $substIDs = MagicWord::getSubstIDs();
2765
2766                 $this->mVariables = new MagicWordArray( $variableIDs );
2767                 $this->mSubstWords = new MagicWordArray( $substIDs );
2768                 wfProfileOut( __METHOD__ );
2769         }
2770
2771         /**
2772          * Preprocess some wikitext and return the document tree.
2773          * This is the ghost of replace_variables().
2774          *
2775          * @param $text String: The text to parse
2776          * @param $flags Integer: bitwise combination of:
2777          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2778          *                                     included. Default is to assume a direct page view.
2779          *
2780          * The generated DOM tree must depend only on the input text and the flags.
2781          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2782          *
2783          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2784          * change in the DOM tree for a given text, must be passed through the section identifier
2785          * in the section edit link and thus back to extractSections().
2786          *
2787          * The output of this function is currently only cached in process memory, but a persistent
2788          * cache may be implemented at a later date which takes further advantage of these strict
2789          * dependency requirements.
2790          *
2791          * @private
2792          */
2793         function preprocessToDom( $text, $flags = 0 ) {
2794                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2795                 return $dom;
2796         }
2797
2798         /**
2799          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2800          */
2801         public static function splitWhitespace( $s ) {
2802                 $ltrimmed = ltrim( $s );
2803                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2804                 $trimmed = rtrim( $ltrimmed );
2805                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2806                 if ( $diff > 0 ) {
2807                         $w2 = substr( $ltrimmed, -$diff );
2808                 } else {
2809                         $w2 = '';
2810                 }
2811                 return array( $w1, $trimmed, $w2 );
2812         }
2813
2814         /**
2815          * Replace magic variables, templates, and template arguments
2816          * with the appropriate text. Templates are substituted recursively,
2817          * taking care to avoid infinite loops.
2818          *
2819          * Note that the substitution depends on value of $mOutputType:
2820          *  self::OT_WIKI: only {{subst:}} templates
2821          *  self::OT_PREPROCESS: templates but not extension tags
2822          *  self::OT_HTML: all templates and extension tags
2823          *
2824          * @param $text String: the text to transform
2825          * @param $frame PPFrame Object describing the arguments passed to the template.
2826          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2827          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2828          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2829          * @private
2830          */
2831         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2832                 # Is there any text? Also, Prevent too big inclusions!
2833                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2834                         return $text;
2835                 }
2836                 wfProfileIn( __METHOD__ );
2837
2838                 if ( $frame === false ) {
2839                         $frame = $this->getPreprocessor()->newFrame();
2840                 } elseif ( !( $frame instanceof PPFrame ) ) {
2841                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2842                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2843                 }
2844
2845                 $dom = $this->preprocessToDom( $text );
2846                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2847                 $text = $frame->expand( $dom, $flags );
2848
2849                 wfProfileOut( __METHOD__ );
2850                 return $text;
2851         }
2852
2853         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2854         static function createAssocArgs( $args ) {
2855                 $assocArgs = array();
2856                 $index = 1;
2857                 foreach ( $args as $arg ) {
2858                         $eqpos = strpos( $arg, '=' );
2859                         if ( $eqpos === false ) {
2860                                 $assocArgs[$index++] = $arg;
2861                         } else {
2862                                 $name = trim( substr( $arg, 0, $eqpos ) );
2863                                 $value = trim( substr( $arg, $eqpos+1 ) );
2864                                 if ( $value === false ) {
2865                                         $value = '';
2866                                 }
2867                                 if ( $name !== false ) {
2868                                         $assocArgs[$name] = $value;
2869                                 }
2870                         }
2871                 }
2872
2873                 return $assocArgs;
2874         }
2875
2876         /**
2877          * Warn the user when a parser limitation is reached
2878          * Will warn at most once the user per limitation type
2879          *
2880          * @param $limitationType String: should be one of:
2881          *   'expensive-parserfunction' (corresponding messages:
2882          *       'expensive-parserfunction-warning',
2883          *       'expensive-parserfunction-category')
2884          *   'post-expand-template-argument' (corresponding messages:
2885          *       'post-expand-template-argument-warning',
2886          *       'post-expand-template-argument-category')
2887          *   'post-expand-template-inclusion' (corresponding messages:
2888          *       'post-expand-template-inclusion-warning',
2889          *       'post-expand-template-inclusion-category')
2890          * @param $current Current value
2891          * @param $max Maximum allowed, when an explicit limit has been
2892          *       exceeded, provide the values (optional)
2893          */
2894         function limitationWarn( $limitationType, $current=null, $max=null) {
2895                 # does no harm if $current and $max are present but are unnecessary for the message
2896                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2897                 $this->mOutput->addWarning( $warning );
2898                 $this->addTrackingCategory( "$limitationType-category" );
2899         }
2900
2901         /**
2902          * Return the text of a template, after recursively
2903          * replacing any variables or templates within the template.
2904          *
2905          * @param $piece Array: the parts of the template
2906          *  $piece['title']: the title, i.e. the part before the |
2907          *  $piece['parts']: the parameter array
2908          *  $piece['lineStart']: whether the brace was at the start of a line
2909          * @param $frame PPFrame The current frame, contains template arguments
2910          * @return String: the text of the template
2911          * @private
2912          */
2913         function braceSubstitution( $piece, $frame ) {
2914                 global $wgContLang, $wgNonincludableNamespaces;
2915                 wfProfileIn( __METHOD__ );
2916                 wfProfileIn( __METHOD__.'-setup' );
2917
2918                 # Flags
2919                 $found = false;             # $text has been filled
2920                 $nowiki = false;            # wiki markup in $text should be escaped
2921                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2922                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2923                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2924                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2925
2926                 # Title object, where $text came from
2927                 $title = null;
2928
2929                 # $part1 is the bit before the first |, and must contain only title characters.
2930                 # Various prefixes will be stripped from it later.
2931                 $titleWithSpaces = $frame->expand( $piece['title'] );
2932                 $part1 = trim( $titleWithSpaces );
2933                 $titleText = false;
2934
2935                 # Original title text preserved for various purposes
2936                 $originalTitle = $part1;
2937
2938                 # $args is a list of argument nodes, starting from index 0, not including $part1
2939                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2940                 wfProfileOut( __METHOD__.'-setup' );
2941
2942                 # SUBST
2943                 wfProfileIn( __METHOD__.'-modifiers' );
2944                 if ( !$found ) {
2945
2946                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2947
2948                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2949                         # Decide whether to expand template or keep wikitext as-is.
2950                         if ( $this->ot['wiki'] ) {
2951                                 if ( $substMatch === false ) {
2952                                         $literal = true;  # literal when in PST with no prefix
2953                                 } else {
2954                                         $literal = false; # expand when in PST with subst: or safesubst:
2955                                 }
2956                         } else {
2957                                 if ( $substMatch == 'subst' ) {
2958                                         $literal = true;  # literal when not in PST with plain subst:
2959                                 } else {
2960                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2961                                 }
2962                         }
2963                         if ( $literal ) {
2964                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2965                                 $isLocalObj = true;
2966                                 $found = true;
2967                         }
2968                 }
2969
2970                 # Variables
2971                 if ( !$found && $args->getLength() == 0 ) {
2972                         $id = $this->mVariables->matchStartToEnd( $part1 );
2973                         if ( $id !== false ) {
2974                                 $text = $this->getVariableValue( $id, $frame );
2975                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
2976                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
2977                                 }
2978                                 $found = true;
2979                         }
2980                 }
2981
2982                 # MSG, MSGNW and RAW
2983                 if ( !$found ) {
2984                         # Check for MSGNW:
2985                         $mwMsgnw = MagicWord::get( 'msgnw' );
2986                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2987                                 $nowiki = true;
2988                         } else {
2989                                 # Remove obsolete MSG:
2990                                 $mwMsg = MagicWord::get( 'msg' );
2991                                 $mwMsg->matchStartAndRemove( $part1 );
2992                         }
2993
2994                         # Check for RAW:
2995                         $mwRaw = MagicWord::get( 'raw' );
2996                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2997                                 $forceRawInterwiki = true;
2998                         }
2999                 }
3000                 wfProfileOut( __METHOD__.'-modifiers' );
3001
3002                 # Parser functions
3003                 if ( !$found ) {
3004                         wfProfileIn( __METHOD__ . '-pfunc' );
3005
3006                         $colonPos = strpos( $part1, ':' );
3007                         if ( $colonPos !== false ) {
3008                                 # Case sensitive functions
3009                                 $function = substr( $part1, 0, $colonPos );
3010                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3011                                         $function = $this->mFunctionSynonyms[1][$function];
3012                                 } else {
3013                                         # Case insensitive functions
3014                                         $function = $wgContLang->lc( $function );
3015                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3016                                                 $function = $this->mFunctionSynonyms[0][$function];
3017                                         } else {
3018                                                 $function = false;
3019                                         }
3020                                 }
3021                                 if ( $function ) {
3022                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3023                                         $initialArgs = array( &$this );
3024                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3025                                         if ( $flags & SFH_OBJECT_ARGS ) {
3026                                                 # Add a frame parameter, and pass the arguments as an array
3027                                                 $allArgs = $initialArgs;
3028                                                 $allArgs[] = $frame;
3029                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3030                                                         $funcArgs[] = $args->item( $i );
3031                                                 }
3032                                                 $allArgs[] = $funcArgs;
3033                                         } else {
3034                                                 # Convert arguments to plain text
3035                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3036                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3037                                                 }
3038                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3039                                         }
3040
3041                                         # Workaround for PHP bug 35229 and similar
3042                                         if ( !is_callable( $callback ) ) {
3043                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3044                                                 wfProfileOut( __METHOD__ );
3045                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3046                                         }
3047                                         $result = call_user_func_array( $callback, $allArgs );
3048                                         $found = true;
3049                                         $noparse = true;
3050                                         $preprocessFlags = 0;
3051
3052                                         if ( is_array( $result ) ) {
3053                                                 if ( isset( $result[0] ) ) {
3054                                                         $text = $result[0];
3055                                                         unset( $result[0] );
3056                                                 }
3057
3058                                                 # Extract flags into the local scope
3059                                                 # This allows callers to set flags such as nowiki, found, etc.
3060                                                 extract( $result );
3061                                         } else {
3062                                                 $text = $result;
3063                                         }
3064                                         if ( !$noparse ) {
3065                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3066                                                 $isChildObj = true;
3067                                         }
3068                                 }
3069                         }
3070                         wfProfileOut( __METHOD__ . '-pfunc' );
3071                 }
3072
3073                 # Finish mangling title and then check for loops.
3074                 # Set $title to a Title object and $titleText to the PDBK
3075                 if ( !$found ) {
3076                         $ns = NS_TEMPLATE;
3077                         # Split the title into page and subpage
3078                         $subpage = '';
3079                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3080                         if ( $subpage !== '' ) {
3081                                 $ns = $this->mTitle->getNamespace();
3082                         }
3083                         $title = Title::newFromText( $part1, $ns );
3084                         if ( $title ) {
3085                                 $titleText = $title->getPrefixedText();
3086                                 # Check for language variants if the template is not found
3087                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3088                                         $wgContLang->findVariantLink( $part1, $title, true );
3089                                 }
3090                                 # Do recursion depth check
3091                                 $limit = $this->mOptions->getMaxTemplateDepth();
3092                                 if ( $frame->depth >= $limit ) {
3093                                         $found = true;
3094                                         $text = '<span class="error">'
3095                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3096                                                 . '</span>';
3097                                 }
3098                         }
3099                 }
3100
3101                 # Load from database
3102                 if ( !$found && $title ) {
3103                         wfProfileIn( __METHOD__ . '-loadtpl' );
3104                         if ( !$title->isExternal() ) {
3105                                 if ( $title->getNamespace() == NS_SPECIAL
3106                                         && $this->mOptions->getAllowSpecialInclusion()
3107                                         && $this->ot['html'] )
3108                                 {
3109                                         $text = SpecialPage::capturePath( $title );
3110                                         if ( is_string( $text ) ) {
3111                                                 $found = true;
3112                                                 $isHTML = true;
3113                                                 $this->disableCache();
3114                                         }
3115                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3116                                         $found = false; # access denied
3117                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3118                                 } else {
3119                                         list( $text, $title ) = $this->getTemplateDom( $title );
3120                                         if ( $text !== false ) {
3121                                                 $found = true;
3122                                                 $isChildObj = true;
3123                                         }
3124                                 }
3125
3126                                 # If the title is valid but undisplayable, make a link to it
3127                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3128                                         $text = "[[:$titleText]]";
3129                                         $found = true;
3130                                 }
3131                         } elseif ( $title->isTrans() ) {
3132                                 # Interwiki transclusion
3133                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3134                                         $text = $this->interwikiTransclude( $title, 'render' );
3135                                         $isHTML = true;
3136                                 } else {
3137                                         $text = $this->interwikiTransclude( $title, 'raw' );
3138                                         # Preprocess it like a template
3139                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3140                                         $isChildObj = true;
3141                                 }
3142                                 $found = true;
3143                         }
3144
3145                         # Do infinite loop check
3146                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3147                         if ( !$frame->loopCheck( $title ) ) {
3148                                 $found = true;
3149                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3150                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3151                         }
3152                         wfProfileOut( __METHOD__ . '-loadtpl' );
3153                 }
3154
3155                 # If we haven't found text to substitute by now, we're done
3156                 # Recover the source wikitext and return it
3157                 if ( !$found ) {
3158                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3159                         wfProfileOut( __METHOD__ );
3160                         return array( 'object' => $text );
3161                 }
3162
3163                 # Expand DOM-style return values in a child frame
3164                 if ( $isChildObj ) {
3165                         # Clean up argument array
3166                         $newFrame = $frame->newChild( $args, $title );
3167
3168                         if ( $nowiki ) {
3169                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3170                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3171                                 # Expansion is eligible for the empty-frame cache
3172                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3173                                         $text = $this->mTplExpandCache[$titleText];
3174                                 } else {
3175                                         $text = $newFrame->expand( $text );
3176                                         $this->mTplExpandCache[$titleText] = $text;
3177                                 }
3178                         } else {
3179                                 # Uncached expansion
3180                                 $text = $newFrame->expand( $text );
3181                         }
3182                 }
3183                 if ( $isLocalObj && $nowiki ) {
3184                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3185                         $isLocalObj = false;
3186                 }
3187
3188                 # Replace raw HTML by a placeholder
3189                 # Add a blank line preceding, to prevent it from mucking up
3190                 # immediately preceding headings
3191                 if ( $isHTML ) {
3192                         $text = "\n\n" . $this->insertStripItem( $text );
3193                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3194                         # Escape nowiki-style return values
3195                         $text = wfEscapeWikiText( $text );
3196                 } elseif ( is_string( $text )
3197                         && !$piece['lineStart']
3198                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3199                 {
3200                         # Bug 529: if the template begins with a table or block-level
3201                         # element, it should be treated as beginning a new line.
3202                         # This behaviour is somewhat controversial.
3203                         $text = "\n" . $text;
3204                 }
3205
3206                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3207                         # Error, oversize inclusion
3208                         if ( $titleText !== false ) {
3209                                 # Make a working, properly escaped link if possible (bug 23588)
3210                                 $text = "[[:$titleText]]";
3211                         } else {
3212                                 # This will probably not be a working link, but at least it may
3213                                 # provide some hint of where the problem is
3214                                 preg_replace( '/^:/', '', $originalTitle );
3215                                 $text = "[[:$originalTitle]]";
3216                         }
3217                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3218                         $this->limitationWarn( 'post-expand-template-inclusion' );
3219                 }
3220
3221                 if ( $isLocalObj ) {
3222                         $ret = array( 'object' => $text );
3223                 } else {
3224                         $ret = array( 'text' => $text );
3225                 }
3226
3227                 wfProfileOut( __METHOD__ );
3228                 return $ret;
3229         }
3230
3231         /**
3232          * Get the semi-parsed DOM representation of a template with a given title,
3233          * and its redirect destination title. Cached.
3234          */
3235         function getTemplateDom( $title ) {
3236                 $cacheTitle = $title;
3237                 $titleText = $title->getPrefixedDBkey();
3238
3239                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3240                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3241                         $title = Title::makeTitle( $ns, $dbk );
3242                         $titleText = $title->getPrefixedDBkey();
3243                 }
3244                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3245                         return array( $this->mTplDomCache[$titleText], $title );
3246                 }
3247
3248                 # Cache miss, go to the database
3249                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3250
3251                 if ( $text === false ) {
3252                         $this->mTplDomCache[$titleText] = false;
3253                         return array( false, $title );
3254                 }
3255
3256                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3257                 $this->mTplDomCache[ $titleText ] = $dom;
3258
3259                 if ( !$title->equals( $cacheTitle ) ) {
3260                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3261                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3262                 }
3263
3264                 return array( $dom, $title );
3265         }
3266
3267         /**
3268          * Fetch the unparsed text of a template and register a reference to it.
3269          */
3270         function fetchTemplateAndTitle( $title ) {
3271                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3272                 $stuff = call_user_func( $templateCb, $title, $this );
3273                 $text = $stuff['text'];
3274                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3275                 if ( isset( $stuff['deps'] ) ) {
3276                         foreach ( $stuff['deps'] as $dep ) {
3277                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3278                         }
3279                 }
3280                 return array( $text, $finalTitle );
3281         }
3282
3283         function fetchTemplate( $title ) {
3284                 $rv = $this->fetchTemplateAndTitle( $title );
3285                 return $rv[0];
3286         }
3287
3288         /**
3289          * Static function to get a template
3290          * Can be overridden via ParserOptions::setTemplateCallback().
3291          */
3292         static function statelessFetchTemplate( $title, $parser=false ) {
3293                 $text = $skip = false;
3294                 $finalTitle = $title;
3295                 $deps = array();
3296
3297                 # Loop to fetch the article, with up to 1 redirect
3298                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3299                         # Give extensions a chance to select the revision instead
3300                         $id = false; # Assume current
3301                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3302
3303                         if ( $skip ) {
3304                                 $text = false;
3305                                 $deps[] = array(
3306                                         'title' => $title,
3307                                         'page_id' => $title->getArticleID(),
3308                                         'rev_id' => null );
3309                                 break;
3310                         }
3311                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3312                         $rev_id = $rev ? $rev->getId() : 0;
3313                         # If there is no current revision, there is no page
3314                         if ( $id === false && !$rev ) {
3315                                 $linkCache = LinkCache::singleton();
3316                                 $linkCache->addBadLinkObj( $title );
3317                         }
3318
3319                         $deps[] = array(
3320                                 'title' => $title,
3321                                 'page_id' => $title->getArticleID(),
3322                                 'rev_id' => $rev_id );
3323
3324                         if ( $rev ) {
3325                                 $text = $rev->getText();
3326                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3327                                 global $wgContLang;
3328                                 $message = $wgContLang->lcfirst( $title->getText() );
3329                                 $text = wfMsgForContentNoTrans( $message );
3330                                 if ( wfEmptyMsg( $message, $text ) ) {
3331                                         $text = false;
3332                                         break;
3333                                 }
3334                         } else {
3335                                 break;
3336                         }
3337                         if ( $text === false ) {
3338                                 break;
3339                         }
3340                         # Redirect?
3341                         $finalTitle = $title;
3342                         $title = Title::newFromRedirect( $text );
3343                 }
3344                 return array(
3345                         'text' => $text,
3346                         'finalTitle' => $finalTitle,
3347                         'deps' => $deps );
3348         }
3349
3350         /**
3351          * Transclude an interwiki link.
3352          */
3353         function interwikiTransclude( $title, $action ) {
3354                 global $wgEnableScaryTranscluding;
3355
3356                 if ( !$wgEnableScaryTranscluding ) {
3357                         return wfMsgForContent('scarytranscludedisabled');
3358                 }
3359
3360                 $url = $title->getFullUrl( "action=$action" );
3361
3362                 if ( strlen( $url ) > 255 ) {
3363                         return wfMsgForContent( 'scarytranscludetoolong' );
3364                 }
3365                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3366         }
3367
3368         function fetchScaryTemplateMaybeFromCache( $url ) {
3369                 global $wgTranscludeCacheExpiry;
3370                 $dbr = wfGetDB( DB_SLAVE );
3371                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3372                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3373                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3374                 if ( $obj ) {
3375                         return $obj->tc_contents;
3376                 }
3377
3378                 $text = Http::get( $url );
3379                 if ( !$text ) {
3380                         return wfMsgForContent( 'scarytranscludefailed', $url );
3381                 }
3382
3383                 $dbw = wfGetDB( DB_MASTER );
3384                 $dbw->replace( 'transcache', array('tc_url'), array(
3385                         'tc_url' => $url,
3386                         'tc_time' => $dbw->timestamp( time() ),
3387                         'tc_contents' => $text)
3388                 );
3389                 return $text;
3390         }
3391
3392
3393         /**
3394          * Triple brace replacement -- used for template arguments
3395          * @private
3396          */
3397         function argSubstitution( $piece, $frame ) {
3398                 wfProfileIn( __METHOD__ );
3399
3400                 $error = false;
3401                 $parts = $piece['parts'];
3402                 $nameWithSpaces = $frame->expand( $piece['title'] );
3403                 $argName = trim( $nameWithSpaces );
3404                 $object = false;
3405                 $text = $frame->getArgument( $argName );
3406                 if (  $text === false && $parts->getLength() > 0
3407                   && (
3408                     $this->ot['html']
3409                     || $this->ot['pre']
3410                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3411                   )
3412                 ) {
3413                         # No match in frame, use the supplied default
3414                         $object = $parts->item( 0 )->getChildren();
3415                 }
3416                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3417                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3418                         $this->limitationWarn( 'post-expand-template-argument' );
3419                 }
3420
3421                 if ( $text === false && $object === false ) {
3422                         # No match anywhere
3423                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3424                 }
3425                 if ( $error !== false ) {
3426                         $text .= $error;
3427                 }
3428                 if ( $object !== false ) {
3429                         $ret = array( 'object' => $object );
3430                 } else {
3431                         $ret = array( 'text' => $text );
3432                 }
3433
3434                 wfProfileOut( __METHOD__ );
3435                 return $ret;
3436         }
3437
3438         /**
3439          * Return the text to be used for a given extension tag.
3440          * This is the ghost of strip().
3441          *
3442          * @param $params Associative array of parameters:
3443          *     name       PPNode for the tag name
3444          *     attr       PPNode for unparsed text where tag attributes are thought to be
3445          *     attributes Optional associative array of parsed attributes
3446          *     inner      Contents of extension element
3447          *     noClose    Original text did not have a close tag
3448          * @param $frame PPFrame
3449          */
3450         function extensionSubstitution( $params, $frame ) {
3451                 $name = $frame->expand( $params['name'] );
3452                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3453                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3454                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3455
3456                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3457                         ( $this->ot['html'] || $this->ot['pre'] );
3458                 if ( $isFunctionTag ) {
3459                         $markerType = 'none';
3460                 } else {
3461                         $markerType = 'general';
3462                 }
3463                 if ( $this->ot['html'] || $isFunctionTag ) {
3464                         $name = strtolower( $name );
3465                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3466                         if ( isset( $params['attributes'] ) ) {
3467                                 $attributes = $attributes + $params['attributes'];
3468                         }
3469
3470                         if ( isset( $this->mTagHooks[$name] ) ) {
3471                                 # Workaround for PHP bug 35229 and similar
3472                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3473                                         throw new MWException( "Tag hook for $name is not callable\n" );
3474                                 }
3475                                 $output = call_user_func_array( $this->mTagHooks[$name],
3476                                         array( $content, $attributes, $this, $frame ) );
3477                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3478                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3479                                 if ( !is_callable( $callback ) ) {
3480                                         throw new MWException( "Tag hook for $name is not callable\n" );
3481                                 }
3482
3483                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3484                         } else {
3485                                 $output = '<span class="error">Invalid tag extension name: ' .
3486                                         htmlspecialchars( $name ) . '</span>';
3487                         }
3488
3489                         if ( is_array( $output ) ) {
3490                                 # Extract flags to local scope (to override $markerType)
3491                                 $flags = $output;
3492                                 $output = $flags[0];
3493                                 unset( $flags[0] );
3494                                 extract( $flags );
3495                         }
3496                 } else {
3497                         if ( is_null( $attrText ) ) {
3498                                 $attrText = '';
3499                         }
3500                         if ( isset( $params['attributes'] ) ) {
3501                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3502                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3503                                                 htmlspecialchars( $attrValue ) . '"';
3504                                 }
3505                         }
3506                         if ( $content === null ) {
3507                                 $output = "<$name$attrText/>";
3508                         } else {
3509                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3510                                 $output = "<$name$attrText>$content$close";
3511                         }
3512                 }
3513
3514                 if ( $markerType === 'none' ) {
3515                         return $output;
3516                 } elseif ( $markerType === 'nowiki' ) {
3517                         $this->mStripState->nowiki->setPair( $marker, $output );
3518                 } elseif ( $markerType === 'general' ) {
3519                         $this->mStripState->general->setPair( $marker, $output );
3520                 } else {
3521                         throw new MWException( __METHOD__.': invalid marker type' );
3522                 }
3523                 return $marker;
3524         }
3525
3526         /**
3527          * Increment an include size counter
3528          *
3529          * @param $type String: the type of expansion
3530          * @param $size Integer: the size of the text
3531          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3532          */
3533         function incrementIncludeSize( $type, $size ) {
3534                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3535                         return false;
3536                 } else {
3537                         $this->mIncludeSizes[$type] += $size;
3538                         return true;
3539                 }
3540         }
3541
3542         /**
3543          * Increment the expensive function count
3544          *
3545          * @return Boolean: false if the limit has been exceeded
3546          */
3547         function incrementExpensiveFunctionCount() {
3548                 global $wgExpensiveParserFunctionLimit;
3549                 $this->mExpensiveFunctionCount++;
3550                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3551                         return true;
3552                 }
3553                 return false;
3554         }
3555
3556         /**
3557          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3558          * Fills $this->mDoubleUnderscores, returns the modified text
3559          */
3560         function doDoubleUnderscore( $text ) {
3561                 wfProfileIn( __METHOD__ );
3562
3563                 # The position of __TOC__ needs to be recorded
3564                 $mw = MagicWord::get( 'toc' );
3565                 if ( $mw->match( $text ) ) {
3566                         $this->mShowToc = true;
3567                         $this->mForceTocPosition = true;
3568
3569                         # Set a placeholder. At the end we'll fill it in with the TOC.
3570                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3571
3572                         # Only keep the first one.
3573                         $text = $mw->replace( '', $text );
3574                 }
3575
3576                 # Now match and remove the rest of them
3577                 $mwa = MagicWord::getDoubleUnderscoreArray();
3578                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3579
3580                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3581                         $this->mOutput->mNoGallery = true;
3582                 }
3583                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3584                         $this->mShowToc = false;
3585                 }
3586                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3587                         $this->addTrackingCategory( 'hidden-category-category' );
3588                 }
3589                 # (bug 8068) Allow control over whether robots index a page.
3590                 #
3591                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3592                 # is not desirable, the last one on the page should win.
3593                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3594                         $this->mOutput->setIndexPolicy( 'noindex' );
3595                         $this->addTrackingCategory( 'noindex-category' );
3596                 }
3597                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3598                         $this->mOutput->setIndexPolicy( 'index' );
3599                         $this->addTrackingCategory( 'index-category' );
3600                 }
3601
3602                 # Cache all double underscores in the database
3603                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3604                         $this->mOutput->setProperty( $key, '' );
3605                 }
3606
3607                 wfProfileOut( __METHOD__ );
3608                 return $text;
3609         }
3610
3611         /**
3612          * Add a tracking category, getting the title from a system message,
3613          * or print a debug message if the title is invalid.
3614          *
3615          * @param $msg String: message key
3616          * @return Boolean: whether the addition was successful
3617          */
3618         protected function addTrackingCategory( $msg ) {
3619                 $cat = wfMsgForContent( $msg );
3620
3621                 # Allow tracking categories to be disabled by setting them to "-"
3622                 if ( $cat === '-' ) {
3623                         return false;
3624                 }
3625
3626                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3627                 if ( $containerCategory ) {
3628                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3629                         return true;
3630                 } else {
3631                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3632                         return false;
3633                 }
3634         }
3635
3636         /**
3637          * This function accomplishes several tasks:
3638          * 1) Auto-number headings if that option is enabled
3639          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3640          * 3) Add a Table of contents on the top for users who have enabled the option
3641          * 4) Auto-anchor headings
3642          *
3643          * It loops through all headlines, collects the necessary data, then splits up the
3644          * string and re-inserts the newly formatted headlines.
3645          *
3646          * @param $text String
3647          * @param $origText String: original, untouched wikitext
3648          * @param $isMain Boolean
3649          * @private
3650          */
3651         function formatHeadings( $text, $origText, $isMain=true ) {
3652                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3653
3654                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3655
3656                 # Inhibit editsection links if requested in the page
3657                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3658                         $showEditLink = 0;
3659                 } else {
3660                         $showEditLink = $this->mOptions->getEditSection();
3661                 }
3662
3663                 # Get all headlines for numbering them and adding funky stuff like [edit]
3664                 # links - this is for later, but we need the number of headlines right now
3665                 $matches = array();
3666                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3667
3668                 # if there are fewer than 4 headlines in the article, do not show TOC
3669                 # unless it's been explicitly enabled.
3670                 $enoughToc = $this->mShowToc &&
3671                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3672
3673                 # Allow user to stipulate that a page should have a "new section"
3674                 # link added via __NEWSECTIONLINK__
3675                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3676                         $this->mOutput->setNewSection( true );
3677                 }
3678
3679                 # Allow user to remove the "new section"
3680                 # link via __NONEWSECTIONLINK__
3681                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3682                         $this->mOutput->hideNewSection( true );
3683                 }
3684
3685                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3686                 # override above conditions and always show TOC above first header
3687                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3688                         $this->mShowToc = true;
3689                         $enoughToc = true;
3690                 }
3691
3692                 # We need this to perform operations on the HTML
3693                 $sk = $this->mOptions->getSkin( $this->mTitle );
3694
3695                 # headline counter
3696                 $headlineCount = 0;
3697                 $numVisible = 0;
3698
3699                 # Ugh .. the TOC should have neat indentation levels which can be
3700                 # passed to the skin functions. These are determined here
3701                 $toc = '';
3702                 $full = '';
3703                 $head = array();
3704                 $sublevelCount = array();
3705                 $levelCount = array();
3706                 $level = 0;
3707                 $prevlevel = 0;
3708                 $toclevel = 0;
3709                 $prevtoclevel = 0;
3710                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3711                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3712                 $oldType = $this->mOutputType;
3713                 $this->setOutputType( self::OT_WIKI );
3714                 $frame = $this->getPreprocessor()->newFrame();
3715                 $root = $this->preprocessToDom( $origText );
3716                 $node = $root->getFirstChild();
3717                 $byteOffset = 0;
3718                 $tocraw = array();
3719                 $refers = array();
3720
3721                 foreach ( $matches[3] as $headline ) {
3722                         $isTemplate = false;
3723                         $titleText = false;
3724                         $sectionIndex = false;
3725                         $numbering = '';
3726                         $markerMatches = array();
3727                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3728                                 $serial = $markerMatches[1];
3729                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3730                                 $isTemplate = ( $titleText != $baseTitleText );
3731                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3732                         }
3733
3734                         if ( $toclevel ) {
3735                                 $prevlevel = $level;
3736                         }
3737                         $level = $matches[1][$headlineCount];
3738
3739                         if ( $level > $prevlevel ) {
3740                                 # Increase TOC level
3741                                 $toclevel++;
3742                                 $sublevelCount[$toclevel] = 0;
3743                                 if ( $toclevel<$wgMaxTocLevel ) {
3744                                         $prevtoclevel = $toclevel;
3745                                         $toc .= $sk->tocIndent();
3746                                         $numVisible++;
3747                                 }
3748                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3749                                 # Decrease TOC level, find level to jump to
3750
3751                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3752                                         if ( $levelCount[$i] == $level ) {
3753                                                 # Found last matching level
3754                                                 $toclevel = $i;
3755                                                 break;
3756                                         } elseif ( $levelCount[$i] < $level ) {
3757                                                 # Found first matching level below current level
3758                                                 $toclevel = $i + 1;
3759                                                 break;
3760                                         }
3761                                 }
3762                                 if ( $i == 0 ) {
3763                                         $toclevel = 1;
3764                                 }
3765                                 if ( $toclevel<$wgMaxTocLevel ) {
3766                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3767                                                 # Unindent only if the previous toc level was shown :p
3768                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3769                                                 $prevtoclevel = $toclevel;
3770                                         } else {
3771                                                 $toc .= $sk->tocLineEnd();
3772                                         }
3773                                 }
3774                         } else {
3775                                 # No change in level, end TOC line
3776                                 if ( $toclevel<$wgMaxTocLevel ) {
3777                                         $toc .= $sk->tocLineEnd();
3778                                 }
3779                         }
3780
3781                         $levelCount[$toclevel] = $level;
3782
3783                         # count number of headlines for each level
3784                         @$sublevelCount[$toclevel]++;
3785                         $dot = 0;
3786                         for( $i = 1; $i <= $toclevel; $i++ ) {
3787                                 if ( !empty( $sublevelCount[$i] ) ) {
3788                                         if ( $dot ) {
3789                                                 $numbering .= '.';
3790                                         }
3791                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3792                                         $dot = 1;
3793                                 }
3794                         }
3795
3796                         # The safe header is a version of the header text safe to use for links
3797                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3798                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3799
3800                         # Remove link placeholders by the link text.
3801                         #     <!--LINK number-->
3802                         # turns into
3803                         #     link text with suffix
3804                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3805
3806                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3807                         $tocline = preg_replace(
3808                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3809                                 array( '',                          '<$1>' ),
3810                                 $safeHeadline
3811                         );
3812                         $tocline = trim( $tocline );
3813
3814                         # For the anchor, strip out HTML-y stuff period
3815                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3816                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
3817
3818                         # Save headline for section edit hint before it's escaped
3819                         $headlineHint = $safeHeadline;
3820
3821                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3822                                 # For reverse compatibility, provide an id that's
3823                                 # HTML4-compatible, like we used to.
3824                                 #
3825                                 # It may be worth noting, academically, that it's possible for
3826                                 # the legacy anchor to conflict with a non-legacy headline
3827                                 # anchor on the page.  In this case likely the "correct" thing
3828                                 # would be to either drop the legacy anchors or make sure
3829                                 # they're numbered first.  However, this would require people
3830                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3831                                 # manually, so let's not bother worrying about it.
3832                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3833                                         array( 'noninitial', 'legacy' ) );
3834                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3835
3836                                 if ( $legacyHeadline == $safeHeadline ) {
3837                                         # No reason to have both (in fact, we can't)
3838                                         $legacyHeadline = false;
3839                                 }
3840                         } else {
3841                                 $legacyHeadline = false;
3842                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3843                                         'noninitial' );
3844                         }
3845
3846                         # HTML names must be case-insensitively unique (bug 10721).
3847                         # This does not apply to Unicode characters per
3848                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
3849                         # FIXME: We may be changing them depending on the current locale.
3850                         $arrayKey = strtolower( $safeHeadline );
3851                         if ( $legacyHeadline === false ) {
3852                                 $legacyArrayKey = false;
3853                         } else {
3854                                 $legacyArrayKey = strtolower( $legacyHeadline );
3855                         }
3856
3857                         # count how many in assoc. array so we can track dupes in anchors
3858                         if ( isset( $refers[$arrayKey] ) ) {
3859                                 $refers[$arrayKey]++;
3860                         } else {
3861                                 $refers[$arrayKey] = 1;
3862                         }
3863                         if ( isset( $refers[$legacyArrayKey] ) ) {
3864                                 $refers[$legacyArrayKey]++;
3865                         } else {
3866                                 $refers[$legacyArrayKey] = 1;
3867                         }
3868
3869                         # Don't number the heading if it is the only one (looks silly)
3870                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3871                                 # the two are different if the line contains a link
3872                                 $headline = $numbering . ' ' . $headline;
3873                         }
3874
3875                         # Create the anchor for linking from the TOC to the section
3876                         $anchor = $safeHeadline;
3877                         $legacyAnchor = $legacyHeadline;
3878                         if ( $refers[$arrayKey] > 1 ) {
3879                                 $anchor .= '_' . $refers[$arrayKey];
3880                         }
3881                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3882                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3883                         }
3884                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3885                                 $toc .= $sk->tocLine( $anchor, $tocline,
3886                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3887                         }
3888
3889                         # Add the section to the section tree
3890                         # Find the DOM node for this header
3891                         while ( $node && !$isTemplate ) {
3892                                 if ( $node->getName() === 'h' ) {
3893                                         $bits = $node->splitHeading();
3894                                         if ( $bits['i'] == $sectionIndex ) {
3895                                                 break;
3896                                         }
3897                                 }
3898                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3899                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3900                                 $node = $node->getNextSibling();
3901                         }
3902                         $tocraw[] = array(
3903                                 'toclevel' => $toclevel,
3904                                 'level' => $level,
3905                                 'line' => $tocline,
3906                                 'number' => $numbering,
3907                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3908                                 'fromtitle' => $titleText,
3909                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3910                                 'anchor' => $anchor,
3911                         );
3912
3913                         # give headline the correct <h#> tag
3914                         if ( $showEditLink && $sectionIndex !== false ) {
3915                                 if ( $isTemplate ) {
3916                                         # Put a T flag in the section identifier, to indicate to extractSections()
3917                                         # that sections inside <includeonly> should be counted.
3918                                         $editlink = $sk->doEditSectionLink( Title::newFromText( $titleText ), "T-$sectionIndex", null, $this->mOptions->getUserLang() );
3919                                 } else {
3920                                         $editlink = $sk->doEditSectionLink( $this->mTitle, $sectionIndex, $headlineHint, $this->mOptions->getUserLang() );
3921                                 }
3922                         } else {
3923                                 $editlink = '';
3924                         }
3925                         $head[$headlineCount] = $sk->makeHeadline( $level,
3926                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3927                                 $editlink, $legacyAnchor );
3928
3929                         $headlineCount++;
3930                 }
3931
3932                 $this->setOutputType( $oldType );
3933
3934                 # Never ever show TOC if no headers
3935                 if ( $numVisible < 1 ) {
3936                         $enoughToc = false;
3937                 }
3938
3939                 if ( $enoughToc ) {
3940                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3941                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3942                         }
3943                         $toc = $sk->tocList( $toc );
3944                         $this->mOutput->setTOCHTML( $toc );
3945                 }
3946
3947                 if ( $isMain ) {
3948                         $this->mOutput->setSections( $tocraw );
3949                 }
3950
3951                 # split up and insert constructed headlines
3952
3953                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3954                 $i = 0;
3955
3956                 foreach ( $blocks as $block ) {
3957                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3958                                 # This is the [edit] link that appears for the top block of text when
3959                                 # section editing is enabled
3960
3961                                 # Disabled because it broke block formatting
3962                                 # For example, a bullet point in the top line
3963                                 # $full .= $sk->editSectionLink(0);
3964                         }
3965                         $full .= $block;
3966                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3967                                 # Top anchor now in skin
3968                                 $full = $full.$toc;
3969                         }
3970
3971                         if ( !empty( $head[$i] ) ) {
3972                                 $full .= $head[$i];
3973                         }
3974                         $i++;
3975                 }
3976                 if ( $this->mForceTocPosition ) {
3977                         return str_replace( '<!--MWTOC-->', $toc, $full );
3978                 } else {
3979                         return $full;
3980                 }
3981         }
3982
3983         /**
3984          * Transform wiki markup when saving a page by doing \r\n -> \n
3985          * conversion, substitting signatures, {{subst:}} templates, etc.
3986          *
3987          * @param $text String: the text to transform
3988          * @param $title Title: the Title object for the current article
3989          * @param $user User: the User object describing the current user
3990          * @param $options ParserOptions: parsing options
3991          * @param $clearState Boolean: whether to clear the parser state first
3992          * @return String: the altered wiki markup
3993          */
3994         public function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3995                 $options->resetUsage();
3996                 $this->mOptions = $options;
3997                 $this->setTitle( $title );
3998                 $this->setOutputType( self::OT_WIKI );
3999
4000                 if ( $clearState ) {
4001                         $this->clearState();
4002                 }
4003
4004                 $pairs = array(
4005                         "\r\n" => "\n",
4006                 );
4007                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4008                 $text = $this->pstPass2( $text, $user );
4009                 $text = $this->mStripState->unstripBoth( $text );
4010                 return $text;
4011         }
4012
4013         /**
4014          * Pre-save transform helper function
4015          * @private
4016          */
4017         function pstPass2( $text, $user ) {
4018                 global $wgContLang, $wgLocaltimezone;
4019
4020                 # Note: This is the timestamp saved as hardcoded wikitext to
4021                 # the database, we use $wgContLang here in order to give
4022                 # everyone the same signature and use the default one rather
4023                 # than the one selected in each user's preferences.
4024                 # (see also bug 12815)
4025                 $ts = $this->mOptions->getTimestamp();
4026                 if ( isset( $wgLocaltimezone ) ) {
4027                         $tz = $wgLocaltimezone;
4028                 } else {
4029                         $tz = date_default_timezone_get();
4030                 }
4031
4032                 $unixts = wfTimestamp( TS_UNIX, $ts );
4033                 $oldtz = date_default_timezone_get();
4034                 date_default_timezone_set( $tz );
4035                 $ts = date( 'YmdHis', $unixts );
4036                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4037
4038                 # Allow translation of timezones through wiki. date() can return
4039                 # whatever crap the system uses, localised or not, so we cannot
4040                 # ship premade translations.
4041                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4042                 $value = wfMsgForContent( $key );
4043                 if ( !wfEmptyMsg( $key, $value ) ) {
4044                         $tzMsg = $value;
4045                 }
4046
4047                 date_default_timezone_set( $oldtz );
4048
4049                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4050
4051                 # Variable replacement
4052                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4053                 $text = $this->replaceVariables( $text );
4054
4055                 # Signatures
4056                 $sigText = $this->getUserSig( $user );
4057                 $text = strtr( $text, array(
4058                         '~~~~~' => $d,
4059                         '~~~~' => "$sigText $d",
4060                         '~~~' => $sigText
4061                 ) );
4062
4063                 # Context links: [[|name]] and [[name (context)|]]
4064                 global $wgLegalTitleChars;
4065                 $tc = "[$wgLegalTitleChars]";
4066                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4067
4068                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4069                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4070                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4071                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4072
4073                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4074                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4075                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4076                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4077
4078                 $t = $this->mTitle->getText();
4079                 $m = array();
4080                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4081                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4082                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4083                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4084                 } else {
4085                         # if there's no context, don't bother duplicating the title
4086                         $text = preg_replace( $p2, '[[\\1]]', $text );
4087                 }
4088
4089                 # Trim trailing whitespace
4090                 $text = rtrim( $text );
4091
4092                 return $text;
4093         }
4094
4095         /**
4096          * Fetch the user's signature text, if any, and normalize to
4097          * validated, ready-to-insert wikitext.
4098          * If you have pre-fetched the nickname or the fancySig option, you can
4099          * specify them here to save a database query.
4100          *
4101          * @param $user User
4102          * @param $nickname String: nickname to use or false to use user's default nickname
4103          * @param $fancySig Boolean: whether the nicknname is the complete signature
4104          *                  or null to use default value
4105          * @return string
4106          */
4107         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4108                 global $wgMaxSigChars;
4109
4110                 $username = $user->getName();
4111
4112                 # If not given, retrieve from the user object.
4113                 if ( $nickname === false )
4114                         $nickname = $user->getOption( 'nickname' );
4115
4116                 if ( is_null( $fancySig ) ) {
4117                         $fancySig = $user->getBoolOption( 'fancysig' );
4118                 }
4119
4120                 $nickname = $nickname == null ? $username : $nickname;
4121
4122                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4123                         $nickname = $username;
4124                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4125                 } elseif ( $fancySig !== false ) {
4126                         # Sig. might contain markup; validate this
4127                         if ( $this->validateSig( $nickname ) !== false ) {
4128                                 # Validated; clean up (if needed) and return it
4129                                 return $this->cleanSig( $nickname, true );
4130                         } else {
4131                                 # Failed to validate; fall back to the default
4132                                 $nickname = $username;
4133                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4134                         }
4135                 }
4136
4137                 # Make sure nickname doesnt get a sig in a sig
4138                 $nickname = $this->cleanSigInSig( $nickname );
4139
4140                 # If we're still here, make it a link to the user page
4141                 $userText = wfEscapeWikiText( $username );
4142                 $nickText = wfEscapeWikiText( $nickname );
4143                 if ( $user->isAnon() )  {
4144                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4145                 } else {
4146                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4147                 }
4148         }
4149
4150         /**
4151          * Check that the user's signature contains no bad XML
4152          *
4153          * @param $text String
4154          * @return mixed An expanded string, or false if invalid.
4155          */
4156         function validateSig( $text ) {
4157                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4158         }
4159
4160         /**
4161          * Clean up signature text
4162          *
4163          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4164          * 2) Substitute all transclusions
4165          *
4166          * @param $text String
4167          * @param $parsing Whether we're cleaning (preferences save) or parsing
4168          * @return String: signature text
4169          */
4170         function cleanSig( $text, $parsing = false ) {
4171                 if ( !$parsing ) {
4172                         global $wgTitle;
4173                         $this->clearState();
4174                         $this->setTitle( $wgTitle );
4175                         $this->mOptions = new ParserOptions;
4176                         $this->setOutputType = self::OT_PREPROCESS;
4177                 }
4178
4179                 # Option to disable this feature
4180                 if ( !$this->mOptions->getCleanSignatures() ) {
4181                         return $text;
4182                 }
4183
4184                 # FIXME: regex doesn't respect extension tags or nowiki
4185                 #  => Move this logic to braceSubstitution()
4186                 $substWord = MagicWord::get( 'subst' );
4187                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4188                 $substText = '{{' . $substWord->getSynonym( 0 );
4189
4190                 $text = preg_replace( $substRegex, $substText, $text );
4191                 $text = $this->cleanSigInSig( $text );
4192                 $dom = $this->preprocessToDom( $text );
4193                 $frame = $this->getPreprocessor()->newFrame();
4194                 $text = $frame->expand( $dom );
4195
4196                 if ( !$parsing ) {
4197                         $text = $this->mStripState->unstripBoth( $text );
4198                 }
4199
4200                 return $text;
4201         }
4202
4203         /**
4204          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4205          *
4206          * @param $text String
4207          * @return String: signature text with /~{3,5}/ removed
4208          */
4209         function cleanSigInSig( $text ) {
4210                 $text = preg_replace( '/~{3,5}/', '', $text );
4211                 return $text;
4212         }
4213
4214         /**
4215          * Set up some variables which are usually set up in parse()
4216          * so that an external function can call some class members with confidence
4217          */
4218         public function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4219                 $this->setTitle( $title );
4220                 $options->resetUsage();
4221                 $this->mOptions = $options;
4222                 $this->setOutputType( $outputType );
4223                 if ( $clearState ) {
4224                         $this->clearState();
4225                 }
4226         }
4227
4228         /**
4229          * Wrapper for preprocess()
4230          *
4231          * @param $text String: the text to preprocess
4232          * @param $options ParserOptions: options
4233          * @return String
4234          */
4235         public function transformMsg( $text, $options ) {
4236                 global $wgTitle;
4237                 static $executing = false;
4238
4239                 # Guard against infinite recursion
4240                 if ( $executing ) {
4241                         return $text;
4242                 }
4243                 $executing = true;
4244
4245                 wfProfileIn( __METHOD__ );
4246                 $text = $this->preprocess( $text, $wgTitle, $options );
4247
4248                 $executing = false;
4249                 wfProfileOut( __METHOD__ );
4250                 return $text;
4251         }
4252
4253         /**
4254          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4255          * The callback should have the following form:
4256          *    function myParserHook( $text, $params, $parser ) { ... }
4257          *
4258          * Transform and return $text. Use $parser for any required context, e.g. use
4259          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4260          *
4261          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4262          * @param $callback Mixed: the callback function (and object) to use for the tag
4263          * @return The old value of the mTagHooks array associated with the hook
4264          */
4265         public function setHook( $tag, $callback ) {
4266                 $tag = strtolower( $tag );
4267                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4268                 $this->mTagHooks[$tag] = $callback;
4269                 if ( !in_array( $tag, $this->mStripList ) ) {
4270                         $this->mStripList[] = $tag;
4271                 }
4272
4273                 return $oldVal;
4274         }
4275
4276         function setTransparentTagHook( $tag, $callback ) {
4277                 $tag = strtolower( $tag );
4278                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4279                 $this->mTransparentTagHooks[$tag] = $callback;
4280
4281                 return $oldVal;
4282         }
4283
4284         /**
4285          * Remove all tag hooks
4286          */
4287         function clearTagHooks() {
4288                 $this->mTagHooks = array();
4289                 $this->mStripList = $this->mDefaultStripList;
4290         }
4291
4292         /**
4293          * Create a function, e.g. {{sum:1|2|3}}
4294          * The callback function should have the form:
4295          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4296          *
4297          * Or with SFH_OBJECT_ARGS:
4298          *    function myParserFunction( $parser, $frame, $args ) { ... }
4299          *
4300          * The callback may either return the text result of the function, or an array with the text
4301          * in element 0, and a number of flags in the other elements. The names of the flags are
4302          * specified in the keys. Valid flags are:
4303          *   found                     The text returned is valid, stop processing the template. This
4304          *                             is on by default.
4305          *   nowiki                    Wiki markup in the return value should be escaped
4306          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4307          *
4308          * @param $id String: The magic word ID
4309          * @param $callback Mixed: the callback function (and object) to use
4310          * @param $flags Integer: a combination of the following flags:
4311          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4312          *
4313          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4314          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4315          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4316          *     the arguments, and to control the way they are expanded.
4317          *
4318          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4319          *     arguments, for instance:
4320          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4321          *
4322          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4323          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4324          *     working if/when this is changed.
4325          *
4326          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4327          *     expansion.
4328          *
4329          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4330          *     about the methods available in PPFrame and PPNode.
4331          *
4332          * @return The old callback function for this name, if any
4333          */
4334         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4335                 global $wgContLang;
4336
4337                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4338                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4339
4340                 # Add to function cache
4341                 $mw = MagicWord::get( $id );
4342                 if ( !$mw )
4343                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4344
4345                 $synonyms = $mw->getSynonyms();
4346                 $sensitive = intval( $mw->isCaseSensitive() );
4347
4348                 foreach ( $synonyms as $syn ) {
4349                         # Case
4350                         if ( !$sensitive ) {
4351                                 $syn = $wgContLang->lc( $syn );
4352                         }
4353                         # Add leading hash
4354                         if ( !( $flags & SFH_NO_HASH ) ) {
4355                                 $syn = '#' . $syn;
4356                         }
4357                         # Remove trailing colon
4358                         if ( substr( $syn, -1, 1 ) === ':' ) {
4359                                 $syn = substr( $syn, 0, -1 );
4360                         }
4361                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4362                 }
4363                 return $oldVal;
4364         }
4365
4366         /**
4367          * Get all registered function hook identifiers
4368          *
4369          * @return Array
4370          */
4371         function getFunctionHooks() {
4372                 return array_keys( $this->mFunctionHooks );
4373         }
4374
4375         /**
4376          * Create a tag function, e.g. <test>some stuff</test>.
4377          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4378          * Unlike parser functions, their content is not preprocessed.
4379          */
4380         function setFunctionTagHook( $tag, $callback, $flags ) {
4381                 $tag = strtolower( $tag );
4382                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4383                         $this->mFunctionTagHooks[$tag] : null;
4384                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4385
4386                 if ( !in_array( $tag, $this->mStripList ) ) {
4387                         $this->mStripList[] = $tag;
4388                 }
4389
4390                 return $old;
4391         }
4392
4393         /**
4394          * FIXME: update documentation. makeLinkObj() is deprecated.
4395          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4396          * Placeholders created in Skin::makeLinkObj()
4397          * Returns an array of link CSS classes, indexed by PDBK.
4398          */
4399         function replaceLinkHolders( &$text, $options = 0 ) {
4400                 return $this->mLinkHolders->replace( $text );
4401         }
4402
4403         /**
4404          * Replace <!--LINK--> link placeholders with plain text of links
4405          * (not HTML-formatted).
4406          *
4407          * @param $text String
4408          * @return String
4409          */
4410         function replaceLinkHoldersText( $text ) {
4411                 return $this->mLinkHolders->replaceText( $text );
4412         }
4413
4414         /**
4415          * Renders an image gallery from a text with one line per image.
4416          * text labels may be given by using |-style alternative text. E.g.
4417          *   Image:one.jpg|The number "1"
4418          *   Image:tree.jpg|A tree
4419          * given as text will return the HTML of a gallery with two images,
4420          * labeled 'The number "1"' and
4421          * 'A tree'.
4422          */
4423         function renderImageGallery( $text, $params ) {
4424                 $ig = new ImageGallery();
4425                 $ig->setContextTitle( $this->mTitle );
4426                 $ig->setShowBytes( false );
4427                 $ig->setShowFilename( false );
4428                 $ig->setParser( $this );
4429                 $ig->setHideBadImages();
4430                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4431                 $ig->useSkin( $this->mOptions->getSkin( $this->mTitle ) );
4432                 $ig->mRevisionId = $this->mRevisionId;
4433
4434                 if ( isset( $params['showfilename'] ) ) {
4435                         $ig->setShowFilename( true );
4436                 } else {
4437                         $ig->setShowFilename( false );
4438                 }
4439                 if ( isset( $params['caption'] ) ) {
4440                         $caption = $params['caption'];
4441                         $caption = htmlspecialchars( $caption );
4442                         $caption = $this->replaceInternalLinks( $caption );
4443                         $ig->setCaptionHtml( $caption );
4444                 }
4445                 if ( isset( $params['perrow'] ) ) {
4446                         $ig->setPerRow( $params['perrow'] );
4447                 }
4448                 if ( isset( $params['widths'] ) ) {
4449                         $ig->setWidths( $params['widths'] );
4450                 }
4451                 if ( isset( $params['heights'] ) ) {
4452                         $ig->setHeights( $params['heights'] );
4453                 }
4454
4455                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4456
4457                 $lines = StringUtils::explode( "\n", $text );
4458                 foreach ( $lines as $line ) {
4459                         # match lines like these:
4460                         # Image:someimage.jpg|This is some image
4461                         $matches = array();
4462                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4463                         # Skip empty lines
4464                         if ( count( $matches ) == 0 ) {
4465                                 continue;
4466                         }
4467
4468                         if ( strpos( $matches[0], '%' ) !== false ) {
4469                                 $matches[1] = urldecode( $matches[1] );
4470                         }
4471                         $tp = Title::newFromText( $matches[1] );
4472                         $nt =& $tp;
4473                         if ( is_null( $nt ) ) {
4474                                 # Bogus title. Ignore these so we don't bomb out later.
4475                                 continue;
4476                         }
4477                         if ( isset( $matches[3] ) ) {
4478                                 $label = $matches[3];
4479                         } else {
4480                                 $label = '';
4481                         }
4482
4483                         $html = $this->recursiveTagParse( trim( $label ) );
4484
4485                         $ig->add( $nt, $html );
4486
4487                         # Only add real images (bug #5586)
4488                         if ( $nt->getNamespace() == NS_FILE ) {
4489                                 $this->mOutput->addImage( $nt->getDBkey() );
4490                         }
4491                 }
4492                 return $ig->toHTML();
4493         }
4494
4495         function getImageParams( $handler ) {
4496                 if ( $handler ) {
4497                         $handlerClass = get_class( $handler );
4498                 } else {
4499                         $handlerClass = '';
4500                 }
4501                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4502                         # Initialise static lists
4503                         static $internalParamNames = array(
4504                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4505                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4506                                         'bottom', 'text-bottom' ),
4507                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4508                                         'upright', 'border', 'link', 'alt' ),
4509                         );
4510                         static $internalParamMap;
4511                         if ( !$internalParamMap ) {
4512                                 $internalParamMap = array();
4513                                 foreach ( $internalParamNames as $type => $names ) {
4514                                         foreach ( $names as $name ) {
4515                                                 $magicName = str_replace( '-', '_', "img_$name" );
4516                                                 $internalParamMap[$magicName] = array( $type, $name );
4517                                         }
4518                                 }
4519                         }
4520
4521                         # Add handler params
4522                         $paramMap = $internalParamMap;
4523                         if ( $handler ) {
4524                                 $handlerParamMap = $handler->getParamMap();
4525                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4526                                         $paramMap[$magic] = array( 'handler', $paramName );
4527                                 }
4528                         }
4529                         $this->mImageParams[$handlerClass] = $paramMap;
4530                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4531                 }
4532                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4533         }
4534
4535         /**
4536          * Parse image options text and use it to make an image
4537          *
4538          * @param $title Title
4539          * @param $options String
4540          * @param $holders LinkHolderArray
4541          */
4542         function makeImage( $title, $options, $holders = false ) {
4543                 # Check if the options text is of the form "options|alt text"
4544                 # Options are:
4545                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4546                 #  * left       no resizing, just left align. label is used for alt= only
4547                 #  * right      same, but right aligned
4548                 #  * none       same, but not aligned
4549                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4550                 #  * center     center the image
4551                 #  * frame      Keep original image size, no magnify-button.
4552                 #  * framed     Same as "frame"
4553                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4554                 #  * upright    reduce width for upright images, rounded to full __0 px
4555                 #  * border     draw a 1px border around the image
4556                 #  * alt        Text for HTML alt attribute (defaults to empty)
4557                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4558                 # vertical-align values (no % or length right now):
4559                 #  * baseline
4560                 #  * sub
4561                 #  * super
4562                 #  * top
4563                 #  * text-top
4564                 #  * middle
4565                 #  * bottom
4566                 #  * text-bottom
4567
4568                 $parts = StringUtils::explode( "|", $options );
4569                 $sk = $this->mOptions->getSkin( $this->mTitle );
4570
4571                 # Give extensions a chance to select the file revision for us
4572                 $skip = $time = $descQuery = false;
4573                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4574
4575                 if ( $skip ) {
4576                         return $sk->link( $title );
4577                 }
4578
4579                 # Get the file
4580                 $file = wfFindFile( $title, array( 'time' => $time ) );
4581                 # Get parameter map
4582                 $handler = $file ? $file->getHandler() : false;
4583
4584                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4585
4586                 # Process the input parameters
4587                 $caption = '';
4588                 $params = array( 'frame' => array(), 'handler' => array(),
4589                         'horizAlign' => array(), 'vertAlign' => array() );
4590                 foreach ( $parts as $part ) {
4591                         $part = trim( $part );
4592                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4593                         $validated = false;
4594                         if ( isset( $paramMap[$magicName] ) ) {
4595                                 list( $type, $paramName ) = $paramMap[$magicName];
4596
4597                                 # Special case; width and height come in one variable together
4598                                 if ( $type === 'handler' && $paramName === 'width' ) {
4599                                         $m = array();
4600                                         # (bug 13500) In both cases (width/height and width only),
4601                                         # permit trailing "px" for backward compatibility.
4602                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4603                                                 $width = intval( $m[1] );
4604                                                 $height = intval( $m[2] );
4605                                                 if ( $handler->validateParam( 'width', $width ) ) {
4606                                                         $params[$type]['width'] = $width;
4607                                                         $validated = true;
4608                                                 }
4609                                                 if ( $handler->validateParam( 'height', $height ) ) {
4610                                                         $params[$type]['height'] = $height;
4611                                                         $validated = true;
4612                                                 }
4613                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4614                                                 $width = intval( $value );
4615                                                 if ( $handler->validateParam( 'width', $width ) ) {
4616                                                         $params[$type]['width'] = $width;
4617                                                         $validated = true;
4618                                                 }
4619                                         } # else no validation -- bug 13436
4620                                 } else {
4621                                         if ( $type === 'handler' ) {
4622                                                 # Validate handler parameter
4623                                                 $validated = $handler->validateParam( $paramName, $value );
4624                                         } else {
4625                                                 # Validate internal parameters
4626                                                 switch( $paramName ) {
4627                                                 case 'manualthumb':
4628                                                 case 'alt':
4629                                                         # @todo Fixme: possibly check validity here for
4630                                                         # manualthumb? downstream behavior seems odd with
4631                                                         # missing manual thumbs.
4632                                                         $validated = true;
4633                                                         $value = $this->stripAltText( $value, $holders );
4634                                                         break;
4635                                                 case 'link':
4636                                                         $chars = self::EXT_LINK_URL_CLASS;
4637                                                         $prots = $this->mUrlProtocols;
4638                                                         if ( $value === '' ) {
4639                                                                 $paramName = 'no-link';
4640                                                                 $value = true;
4641                                                                 $validated = true;
4642                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4643                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4644                                                                         $paramName = 'link-url';
4645                                                                         $this->mOutput->addExternalLink( $value );
4646                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4647                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4648                                                                         }
4649                                                                         $validated = true;
4650                                                                 }
4651                                                         } else {
4652                                                                 $linkTitle = Title::newFromText( $value );
4653                                                                 if ( $linkTitle ) {
4654                                                                         $paramName = 'link-title';
4655                                                                         $value = $linkTitle;
4656                                                                         $this->mOutput->addLink( $linkTitle );
4657                                                                         $validated = true;
4658                                                                 }
4659                                                         }
4660                                                         break;
4661                                                 default:
4662                                                         # Most other things appear to be empty or numeric...
4663                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4664                                                 }
4665                                         }
4666
4667                                         if ( $validated ) {
4668                                                 $params[$type][$paramName] = $value;
4669                                         }
4670                                 }
4671                         }
4672                         if ( !$validated ) {
4673                                 $caption = $part;
4674                         }
4675                 }
4676
4677                 # Process alignment parameters
4678                 if ( $params['horizAlign'] ) {
4679                         $params['frame']['align'] = key( $params['horizAlign'] );
4680                 }
4681                 if ( $params['vertAlign'] ) {
4682                         $params['frame']['valign'] = key( $params['vertAlign'] );
4683                 }
4684
4685                 $params['frame']['caption'] = $caption;
4686
4687                 # Will the image be presented in a frame, with the caption below?
4688                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4689                                  isset( $params['frame']['framed'] ) ||
4690                                  isset( $params['frame']['thumbnail'] ) ||
4691                                  isset( $params['frame']['manualthumb'] );
4692
4693                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4694                 # came to also set the caption, ordinary text after the image -- which
4695                 # makes no sense, because that just repeats the text multiple times in
4696                 # screen readers.  It *also* came to set the title attribute.
4697                 #
4698                 # Now that we have an alt attribute, we should not set the alt text to
4699                 # equal the caption: that's worse than useless, it just repeats the
4700                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4701                 # use the unnamed parameter for alt text as well, just for the time be-
4702                 # ing, if the unnamed param is set and the alt param is not.
4703                 #
4704                 # For the future, we need to figure out if we want to tweak this more,
4705                 # e.g., introducing a title= parameter for the title; ignoring the un-
4706                 # named parameter entirely for images without a caption; adding an ex-
4707                 # plicit caption= parameter and preserving the old magic unnamed para-
4708                 # meter for BC; ...
4709                 if ( $imageIsFramed ) { # Framed image
4710                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4711                                 # No caption or alt text, add the filename as the alt text so
4712                                 # that screen readers at least get some description of the image
4713                                 $params['frame']['alt'] = $title->getText();
4714                         }
4715                         # Do not set $params['frame']['title'] because tooltips don't make sense
4716                         # for framed images
4717                 } else { # Inline image
4718                         if ( !isset( $params['frame']['alt'] ) ) {
4719                                 # No alt text, use the "caption" for the alt text
4720                                 if ( $caption !== '') {
4721                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4722                                 } else {
4723                                         # No caption, fall back to using the filename for the
4724                                         # alt text
4725                                         $params['frame']['alt'] = $title->getText();
4726                                 }
4727                         }
4728                         # Use the "caption" for the tooltip text
4729                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4730                 }
4731
4732                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4733
4734                 # Linker does the rest
4735                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery, $this->mOptions->getThumbSize() );
4736
4737                 # Give the handler a chance to modify the parser object
4738                 if ( $handler ) {
4739                         $handler->parserTransformHook( $this, $file );
4740                 }
4741
4742                 return $ret;
4743         }
4744
4745         protected function stripAltText( $caption, $holders ) {
4746                 # Strip bad stuff out of the title (tooltip).  We can't just use
4747                 # replaceLinkHoldersText() here, because if this function is called
4748                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4749                 if ( $holders ) {
4750                         $tooltip = $holders->replaceText( $caption );
4751                 } else {
4752                         $tooltip = $this->replaceLinkHoldersText( $caption );
4753                 }
4754
4755                 # make sure there are no placeholders in thumbnail attributes
4756                 # that are later expanded to html- so expand them now and
4757                 # remove the tags
4758                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4759                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4760
4761                 return $tooltip;
4762         }
4763
4764         /**
4765          * Set a flag in the output object indicating that the content is dynamic and
4766          * shouldn't be cached.
4767          */
4768         function disableCache() {
4769                 wfDebug( "Parser output marked as uncacheable.\n" );
4770                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
4771                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
4772         }
4773
4774         /**
4775          * Callback from the Sanitizer for expanding items found in HTML attribute
4776          * values, so they can be safely tested and escaped.
4777          *
4778          * @param $text String
4779          * @param $frame PPFrame
4780          * @return String
4781          * @private
4782          */
4783         function attributeStripCallback( &$text, $frame = false ) {
4784                 $text = $this->replaceVariables( $text, $frame );
4785                 $text = $this->mStripState->unstripBoth( $text );
4786                 return $text;
4787         }
4788
4789         /**
4790          * Accessor
4791          */
4792         function getTags() {
4793                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4794         }
4795
4796         /**
4797          * Break wikitext input into sections, and either pull or replace
4798          * some particular section's text.
4799          *
4800          * External callers should use the getSection and replaceSection methods.
4801          *
4802          * @param $text String: Page wikitext
4803          * @param $section String: a section identifier string of the form:
4804          *   <flag1> - <flag2> - ... - <section number>
4805          *
4806          * Currently the only recognised flag is "T", which means the target section number
4807          * was derived during a template inclusion parse, in other words this is a template
4808          * section edit link. If no flags are given, it was an ordinary section edit link.
4809          * This flag is required to avoid a section numbering mismatch when a section is
4810          * enclosed by <includeonly> (bug 6563).
4811          *
4812          * The section number 0 pulls the text before the first heading; other numbers will
4813          * pull the given section along with its lower-level subsections. If the section is
4814          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4815          *
4816          * @param $mode String: one of "get" or "replace"
4817          * @param $newText String: replacement text for section data.
4818          * @return String: for "get", the extracted section text.
4819          *                 for "replace", the whole page with the section replaced.
4820          */
4821         private function extractSections( $text, $section, $mode, $newText='' ) {
4822                 global $wgTitle;
4823                 $this->clearState();
4824                 $this->setTitle( $wgTitle ); # not generally used but removes an ugly failure mode
4825                 $this->mOptions = new ParserOptions;
4826                 $this->setOutputType( self::OT_PLAIN );
4827                 $outText = '';
4828                 $frame = $this->getPreprocessor()->newFrame();
4829
4830                 # Process section extraction flags
4831                 $flags = 0;
4832                 $sectionParts = explode( '-', $section );
4833                 $sectionIndex = array_pop( $sectionParts );
4834                 foreach ( $sectionParts as $part ) {
4835                         if ( $part === 'T' ) {
4836                                 $flags |= self::PTD_FOR_INCLUSION;
4837                         }
4838                 }
4839                 # Preprocess the text
4840                 $root = $this->preprocessToDom( $text, $flags );
4841
4842                 # <h> nodes indicate section breaks
4843                 # They can only occur at the top level, so we can find them by iterating the root's children
4844                 $node = $root->getFirstChild();
4845
4846                 # Find the target section
4847                 if ( $sectionIndex == 0 ) {
4848                         # Section zero doesn't nest, level=big
4849                         $targetLevel = 1000;
4850                 } else {
4851                         while ( $node ) {
4852                                 if ( $node->getName() === 'h' ) {
4853                                         $bits = $node->splitHeading();
4854                                         if ( $bits['i'] == $sectionIndex ) {
4855                                                 $targetLevel = $bits['level'];
4856                                                 break;
4857                                         }
4858                                 }
4859                                 if ( $mode === 'replace' ) {
4860                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4861                                 }
4862                                 $node = $node->getNextSibling();
4863                         }
4864                 }
4865
4866                 if ( !$node ) {
4867                         # Not found
4868                         if ( $mode === 'get' ) {
4869                                 return $newText;
4870                         } else {
4871                                 return $text;
4872                         }
4873                 }
4874
4875                 # Find the end of the section, including nested sections
4876                 do {
4877                         if ( $node->getName() === 'h' ) {
4878                                 $bits = $node->splitHeading();
4879                                 $curLevel = $bits['level'];
4880                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4881                                         break;
4882                                 }
4883                         }
4884                         if ( $mode === 'get' ) {
4885                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4886                         }
4887                         $node = $node->getNextSibling();
4888                 } while ( $node );
4889
4890                 # Write out the remainder (in replace mode only)
4891                 if ( $mode === 'replace' ) {
4892                         # Output the replacement text
4893                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4894                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4895                         # Only add trailing whitespace if there is newText
4896                         if ( $newText != "" ) {
4897                                 $outText .= $newText . "\n\n";
4898                         }
4899
4900                         while ( $node ) {
4901                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4902                                 $node = $node->getNextSibling();
4903                         }
4904                 }
4905
4906                 if ( is_string( $outText ) ) {
4907                         # Re-insert stripped tags
4908                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4909                 }
4910
4911                 return $outText;
4912         }
4913
4914         /**
4915          * This function returns the text of a section, specified by a number ($section).
4916          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4917          * the first section before any such heading (section 0).
4918          *
4919          * If a section contains subsections, these are also returned.
4920          *
4921          * @param $text String: text to look in
4922          * @param $section String: section identifier
4923          * @param $deftext String: default to return if section is not found
4924          * @return string text of the requested section
4925          */
4926         public function getSection( $text, $section, $deftext='' ) {
4927                 return $this->extractSections( $text, $section, "get", $deftext );
4928         }
4929
4930         public function replaceSection( $oldtext, $section, $text ) {
4931                 return $this->extractSections( $oldtext, $section, "replace", $text );
4932         }
4933
4934         /**
4935          * Get the ID of the revision we are parsing
4936          *
4937          * @return Mixed: integer or null
4938          */
4939         function getRevisionId() {
4940                 return $this->mRevisionId;
4941         }
4942
4943         /**
4944          * Get the timestamp associated with the current revision, adjusted for
4945          * the default server-local timestamp
4946          */
4947         function getRevisionTimestamp() {
4948                 if ( is_null( $this->mRevisionTimestamp ) ) {
4949                         wfProfileIn( __METHOD__ );
4950                         global $wgContLang;
4951                         $dbr = wfGetDB( DB_SLAVE );
4952                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4953                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4954
4955                         # Normalize timestamp to internal MW format for timezone processing.
4956                         # This has the added side-effect of replacing a null value with
4957                         # the current time, which gives us more sensible behavior for
4958                         # previews.
4959                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4960
4961                         # The cryptic '' timezone parameter tells to use the site-default
4962                         # timezone offset instead of the user settings.
4963                         #
4964                         # Since this value will be saved into the parser cache, served
4965                         # to other users, and potentially even used inside links and such,
4966                         # it needs to be consistent for all visitors.
4967                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4968
4969                         wfProfileOut( __METHOD__ );
4970                 }
4971                 return $this->mRevisionTimestamp;
4972         }
4973
4974         /**
4975          * Get the name of the user that edited the last revision
4976          *
4977          * @return String: user name
4978          */
4979         function getRevisionUser() {
4980                 # if this template is subst: the revision id will be blank,
4981                 # so just use the current user's name
4982                 if ( $this->mRevisionId ) {
4983                         $revision = Revision::newFromId( $this->mRevisionId );
4984                         $revuser = $revision->getUserText();
4985                 } else {
4986                         global $wgUser;
4987                         $revuser = $wgUser->getName();
4988                 }
4989                 return $revuser;
4990         }
4991
4992         /**
4993          * Mutator for $mDefaultSort
4994          *
4995          * @param $sort New value
4996          */
4997         public function setDefaultSort( $sort ) {
4998                 $this->mDefaultSort = $sort;
4999                 $this->mOutput->setProperty( 'defaultsort', $sort );
5000         }
5001
5002         /**
5003          * Accessor for $mDefaultSort
5004          * Will use the title/prefixed title if none is set
5005          *
5006          * @return string
5007          */
5008         public function getDefaultSort() {
5009                 if ( $this->mDefaultSort !== false ) {
5010                         return $this->mDefaultSort;
5011                 } else {
5012                         return $this->mTitle->getCategorySortkey();
5013                 }
5014         }
5015
5016         /**
5017          * Accessor for $mDefaultSort
5018          * Unlike getDefaultSort(), will return false if none is set
5019          *
5020          * @return string or false
5021          */
5022         public function getCustomDefaultSort() {
5023                 return $this->mDefaultSort;
5024         }
5025
5026         /**
5027          * Try to guess the section anchor name based on a wikitext fragment
5028          * presumably extracted from a heading, for example "Header" from
5029          * "== Header ==".
5030          */
5031         public function guessSectionNameFromWikiText( $text ) {
5032                 # Strip out wikitext links(they break the anchor)
5033                 $text = $this->stripSectionName( $text );
5034                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5035                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5036         }
5037
5038         /**
5039          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5040          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5041          * as something other than UTF-8 (apparently?), resulting in breakage.
5042          *
5043          * @param $text String: The section name
5044          * @return string An anchor
5045          */
5046         public function guessLegacySectionNameFromWikiText( $text ) {
5047                 # Strip out wikitext links(they break the anchor)
5048                 $text = $this->stripSectionName( $text );
5049                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5050                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5051         }
5052
5053         /**
5054          * Strips a text string of wikitext for use in a section anchor
5055          *
5056          * Accepts a text string and then removes all wikitext from the
5057          * string and leaves only the resultant text (i.e. the result of
5058          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5059          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5060          * to create valid section anchors by mimicing the output of the
5061          * parser when headings are parsed.
5062          *
5063          * @param $text String: text string to be stripped of wikitext
5064          * for use in a Section anchor
5065          * @return Filtered text string
5066          */
5067         public function stripSectionName( $text ) {
5068                 # Strip internal link markup
5069                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5070                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5071
5072                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5073                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5074                 # on how many empty links there are on the page - need to figure that out.
5075                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5076
5077                 # Parse wikitext quotes (italics & bold)
5078                 $text = $this->doQuotes( $text );
5079
5080                 # Strip HTML tags
5081                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5082                 return $text;
5083         }
5084
5085         /**
5086          * strip/replaceVariables/unstrip for preprocessor regression testing
5087          */
5088         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5089                 $this->clearState();
5090                 if ( !$title instanceof Title ) {
5091                         $title = Title::newFromText( $title );
5092                 }
5093                 $this->mTitle = $title;
5094                 $options->resetUsage();
5095                 $this->mOptions = $options;
5096                 $this->setOutputType( $outputType );
5097                 $text = $this->replaceVariables( $text );
5098                 $text = $this->mStripState->unstripBoth( $text );
5099                 $text = Sanitizer::removeHTMLtags( $text );
5100                 return $text;
5101         }
5102
5103         function testPst( $text, $title, $options ) {
5104                 global $wgUser;
5105                 if ( !$title instanceof Title ) {
5106                         $title = Title::newFromText( $title );
5107                 }
5108                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5109         }
5110
5111         function testPreprocess( $text, $title, $options ) {
5112                 if ( !$title instanceof Title ) {
5113                         $title = Title::newFromText( $title );
5114                 }
5115                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5116         }
5117
5118         function markerSkipCallback( $s, $callback ) {
5119                 $i = 0;
5120                 $out = '';
5121                 while ( $i < strlen( $s ) ) {
5122                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5123                         if ( $markerStart === false ) {
5124                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5125                                 break;
5126                         } else {
5127                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5128                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5129                                 if ( $markerEnd === false ) {
5130                                         $out .= substr( $s, $markerStart );
5131                                         break;
5132                                 } else {
5133                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5134                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5135                                         $i = $markerEnd;
5136                                 }
5137                         }
5138                 }
5139                 return $out;
5140         }
5141
5142         function serialiseHalfParsedText( $text ) {
5143                 $data = array();
5144                 $data['text'] = $text;
5145
5146                 # First, find all strip markers, and store their
5147                 #  data in an array.
5148                 $stripState = new StripState;
5149                 $pos = 0;
5150                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5151                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5152                 {
5153                         $end_pos += strlen( self::MARKER_SUFFIX );
5154                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5155
5156                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5157                                 $replaceArray = $stripState->general;
5158                                 $stripText = $this->mStripState->general->data[$marker];
5159                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5160                                 $replaceArray = $stripState->nowiki;
5161                                 $stripText = $this->mStripState->nowiki->data[$marker];
5162                         } else {
5163                                 throw new MWException( "Hanging strip marker: '$marker'." );
5164                         }
5165
5166                         $replaceArray->setPair( $marker, $stripText );
5167                         $pos = $end_pos;
5168                 }
5169                 $data['stripstate'] = $stripState;
5170
5171                 # Now, find all of our links, and store THEIR
5172                 #  data in an array! :)
5173                 $links = array( 'internal' => array(), 'interwiki' => array() );
5174                 $pos = 0;
5175
5176                 # Internal links
5177                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5178                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5179
5180                         $ns = trim( $ns );
5181                         if ( empty( $links['internal'][$ns] ) ) {
5182                                 $links['internal'][$ns] = array();
5183                         }
5184
5185                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5186                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5187                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5188                 }
5189
5190                 $pos = 0;
5191
5192                 # Interwiki links
5193                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5194                         $data = substr( $text, $start_pos );
5195                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5196                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5197                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5198                 }
5199
5200                 $data['linkholder'] = $links;
5201
5202                 return $data;
5203         }
5204
5205         /**
5206          * TODO: document
5207          * @param $data Array
5208          * @param $intPrefix String unique identifying prefix
5209          * @return String
5210          */
5211         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5212                 if ( !$intPrefix ) {
5213                         $intPrefix = self::getRandomString();
5214                 }
5215
5216                 # First, extract the strip state.
5217                 $stripState = $data['stripstate'];
5218                 $this->mStripState->general->merge( $stripState->general );
5219                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5220
5221                 # Now, extract the text, and renumber links
5222                 $text = $data['text'];
5223                 $links = $data['linkholder'];
5224
5225                 # Internal...
5226                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5227                         foreach ( $nsLinks as $key => $entry ) {
5228                                 $newKey = $intPrefix . '-' . $key;
5229                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5230
5231                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5232                         }
5233                 }
5234
5235                 # Interwiki...
5236                 foreach ( $links['interwiki'] as $key => $entry ) {
5237                         $newKey = "$intPrefix-$key";
5238                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5239
5240                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5241                 }
5242
5243                 # Should be good to go.
5244                 return $text;
5245         }
5246 }
5247
5248 /**
5249  * @todo document, briefly.
5250  * @ingroup Parser
5251  */
5252 class StripState {
5253         var $general, $nowiki;
5254
5255         function __construct() {
5256                 $this->general = new ReplacementArray;
5257                 $this->nowiki = new ReplacementArray;
5258         }
5259
5260         function unstripGeneral( $text ) {
5261                 wfProfileIn( __METHOD__ );
5262                 do {
5263                         $oldText = $text;
5264                         $text = $this->general->replace( $text );
5265                 } while ( $text !== $oldText );
5266                 wfProfileOut( __METHOD__ );
5267                 return $text;
5268         }
5269
5270         function unstripNoWiki( $text ) {
5271                 wfProfileIn( __METHOD__ );
5272                 do {
5273                         $oldText = $text;
5274                         $text = $this->nowiki->replace( $text );
5275                 } while ( $text !== $oldText );
5276                 wfProfileOut( __METHOD__ );
5277                 return $text;
5278         }
5279
5280         function unstripBoth( $text ) {
5281                 wfProfileIn( __METHOD__ );
5282                 do {
5283                         $oldText = $text;
5284                         $text = $this->general->replace( $text );
5285                         $text = $this->nowiki->replace( $text );
5286                 } while ( $text !== $oldText );
5287                 wfProfileOut( __METHOD__ );
5288                 return $text;
5289         }
5290 }
5291
5292 /**
5293  * @todo document, briefly.
5294  * @ingroup Parser
5295  */
5296 class OnlyIncludeReplacer {
5297         var $output = '';
5298
5299         function replace( $matches ) {
5300                 if ( substr( $matches[1], -1 ) === "\n" ) {
5301                         $this->output .= substr( $matches[1], 0, -1 );
5302                 } else {
5303                         $this->output .= $matches[1];
5304                 }
5305         }
5306 }