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         function setOutputType( $ot ) {
 243                 $this->mOutputType = $ot;
 244                 # Shortcut alias
 245                 $this->ot = array(
 246                         'html' => $ot == self::OT_HTML,
 247                         'wiki' => $ot == self::OT_WIKI,
 248                         'pre' => $ot == self::OT_PREPROCESS,
 249                         'plain' => $ot == self::OT_PLAIN,
 250                 );
 251         }
 252
 253         /**
 254          * Set the context title
 255          */
 256         function setTitle( $t ) {
 257                 if ( !$t || $t instanceof FakeTitle ) {
 258                         $t = Title::newFromText( 'NO TITLE' );
 259                 }
 260
 261                 if ( strval( $t->getFragment() ) !== '' ) {
 262                         # Strip the fragment to avoid various odd effects
 263                         $this->mTitle = clone $t;
 264                         $this->mTitle->setFragment( '' );
 265                 } else {
 266                         $this->mTitle = $t;
 267                 }
 268         }
 269
 270         /**
 271          * Accessor for mUniqPrefix.
 272          *
 273          * @public
 274          */
 275         function uniqPrefix() {
 276                 if ( !isset( $this->mUniqPrefix ) ) {
 277                         # @todo Fixme: this is probably *horribly wrong*
 278                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 279                         # if this is called for a parser cache hit, the parser may not
 280                         # have ever been initialized in the first place.
 281                         # Not really sure what the heck is supposed to be going on here.
 282                         return '';
 283                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 284                 }
 285                 return $this->mUniqPrefix;
 286         }
 287
 288         /**
 289          * Convert wikitext to HTML
 290          * Do not call this function recursively.
 291          *
 292          * @param $text String: text we want to parse
 293          * @param $title A title object
 294          * @param $options ParserOptions
 295          * @param $linestart boolean
 296          * @param $clearState boolean
 297          * @param $revid Int: number to pass in {{REVISIONID}}
 298          * @return ParserOutput a ParserOutput
 299          */
 300         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 301                 /**
 302                  * First pass--just handle <nowiki> sections, pass the rest off
 303                  * to internalParse() which does all the real work.
 304                  */
 305
 306                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 307                 $fname = __METHOD__.'-' . wfGetCaller();
 308                 wfProfileIn( __METHOD__ );
 309                 wfProfileIn( $fname );
 310
 311                 if ( $clearState ) {
 312                         $this->clearState();
 313                 }
 314
 315                 $this->mOptions = $options;
 316                 $this->setTitle( $title ); # Page title has to be set for the pre-processor
 317
 318                 $oldRevisionId = $this->mRevisionId;
 319                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 320                 if ( $revid !== null ) {
 321                         $this->mRevisionId = $revid;
 322                         $this->mRevisionTimestamp = null;
 323                 }
 324                 $this->setOutputType( self::OT_HTML );
 325                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 326                 # No more strip!
 327                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 328                 $text = $this->internalParse( $text );
 329
 330                 $text = $this->mStripState->unstripGeneral( $text );
 331
 332                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 333                 $fixtags = array(
 334                         # french spaces, last one Guillemet-left
 335                         # only if there is something before the space
 336                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 337                         # french spaces, Guillemet-right
 338                         '/(\\302\\253) /' => '\\1&nbsp;',
 339                         '/&nbsp;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 340                 );
 341                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 342
 343                 $text = $this->doBlockLevels( $text, $linestart );
 344
 345                 $this->replaceLinkHolders( $text );
 346
 347                 /**
 348                  * The page doesn't get language converted if
 349                  * a) It's disabled
 350                  * b) Content isn't converted
 351                  * c) It's a conversion table
 352                  */
 353                 if ( !( $wgDisableLangConversion
 354                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 355                                 || $this->mTitle->isConversionTable() ) ) {
 356
 357                         # The position of the convert() call should not be changed. it
 358                         # assumes that the links are all replaced and the only thing left
 359                         # is the <nowiki> mark.
 360
 361                         $text = $wgContLang->convert( $text );
 362                 }
 363
 364                 /**
 365                  * A page get its title converted except:
 366                  * a) Language conversion is globally disabled
 367                  * b) Title convert is globally disabled
 368                  * c) The page is a redirect page
 369                  * d) User request with a "linkconvert" set to "no"
 370                  * e) A "nocontentconvert" magic word has been set
 371                  * f) A "notitleconvert" magic word has been set
 372                  * g) User sets "noconvertlink" in his/her preference
 373                  *
 374                  * Note that if a user tries to set a title in a conversion
 375                  * rule but content conversion was not done, then the parser
 376                  * won't pick it up.  This is probably expected behavior.
 377                  */
 378                 if ( !( $wgDisableLangConversion
 379                                 || $wgDisableTitleConversion
 380                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 381                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 382                                 || $this->mOutput->getDisplayTitle() !== false ) )
 383                 {
 384                         $convruletitle = $wgContLang->getConvRuleTitle();
 385                         if ( $convruletitle ) {
 386                                 $this->mOutput->setTitleText( $convruletitle );
 387                         } else {
 388                                 $titleText = $wgContLang->convertTitle( $title );
 389                                 $this->mOutput->setTitleText( $titleText );
 390                         }
 391                 }
 392
 393                 $text = $this->mStripState->unstripNoWiki( $text );
 394
 395                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 396
 397 //!JF Move to its own function
 398
 399                 $uniq_prefix = $this->mUniqPrefix;
 400                 $matches = array();
 401                 $elements = array_keys( $this->mTransparentTagHooks );
 402                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 403
 404                 foreach ( $matches as $marker => $data ) {
 405                         list( $element, $content, $params, $tag ) = $data;
 406                         $tagName = strtolower( $element );
 407                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 408                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 409                         } else {
 410                                 $output = $tag;
 411                         }
 412                         $this->mStripState->general->setPair( $marker, $output );
 413                 }
 414                 $text = $this->mStripState->unstripGeneral( $text );
 415
 416                 $text = Sanitizer::normalizeCharReferences( $text );
 417
 418                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 419                         $text = MWTidy::tidy( $text );
 420                 } else {
 421                         # attempt to sanitize at least some nesting problems
 422                         # (bug #2702 and quite a few others)
 423                         $tidyregs = array(
 424                                 # ''Something [http://www.cool.com cool''] -->
 425                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 426                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 427                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 428                                 # fix up an anchor inside another anchor, only
 429                                 # at least for a single single nested link (bug 3695)
 430                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 431                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 432                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 433                                 # contains a div, so fix it up here; replace
 434                                 # div with escaped text
 435                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 436                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 437                                 # remove empty italic or bold tag pairs, some
 438                                 # introduced by rules above
 439                                 '/<([bi])><\/\\1>/' => '',
 440                         );
 441
 442                         $text = preg_replace(
 443                                 array_keys( $tidyregs ),
 444                                 array_values( $tidyregs ),
 445                                 $text );
 446                 }
 447                 global $wgExpensiveParserFunctionLimit;
 448                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 449                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 450                 }
 451
 452                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 453
 454                 # Information on include size limits, for the benefit of users who try to skirt them
 455                 if ( $this->mOptions->getEnableLimitReport() ) {
 456                         $max = $this->mOptions->getMaxIncludeSize();
 457                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 458                         $limitReport =
 459                                 "NewPP limit report\n" .
 460                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 461                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 462                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 463                                 $PFreport;
 464                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 465                         $text .= "\n<!-- \n$limitReport-->\n";
 466                 }
 467                 $this->mOutput->setText( $text );
 468
 469                 $this->mRevisionId = $oldRevisionId;
 470                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 471                 wfProfileOut( $fname );
 472                 wfProfileOut( __METHOD__ );
 473
 474                 return $this->mOutput;
 475         }
 476
 477         /**
 478          * Recursive parser entry point that can be called from an extension tag
 479          * hook.
 480          *
 481          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 482          *
 483          * @param $text String: text extension wants to have parsed
 484          * @param PPFrame $frame: The frame to use for expanding any template variables
 485          */
 486         function recursiveTagParse( $text, $frame=false ) {
 487                 wfProfileIn( __METHOD__ );
 488                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 489                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 490                 $text = $this->internalParse( $text, false, $frame );
 491                 wfProfileOut( __METHOD__ );
 492                 return $text;
 493         }
 494
 495         /**
 496          * Expand templates and variables in the text, producing valid, static wikitext.
 497          * Also removes comments.
 498          */
 499         function preprocess( $text, $title, $options, $revid = null ) {
 500                 wfProfileIn( __METHOD__ );
 501                 $this->clearState();
 502                 $this->setOutputType( self::OT_PREPROCESS );
 503                 $this->mOptions = $options;
 504                 $this->setTitle( $title );
 505                 if ( $revid !== null ) {
 506                         $this->mRevisionId = $revid;
 507                 }
 508                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 509                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 510                 $text = $this->replaceVariables( $text );
 511                 $text = $this->mStripState->unstripBoth( $text );
 512                 wfProfileOut( __METHOD__ );
 513                 return $text;
 514         }
 515
 516         /**
 517          * Process the wikitext for the ?preload= feature. (bug 5210)
 518          *
 519          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 520          * comments, templates, arguments, tags hooks and parser functions are untouched.
 521          */
 522         public function getPreloadText( $text, $title, $options ) {
 523                 # Parser (re)initialisation
 524                 $this->clearState();
 525                 $this->setOutputType( self::OT_PLAIN );
 526                 $this->mOptions = $options;
 527                 $this->setTitle( $title );
 528
 529                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 530                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 531                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 532         }
 533
 534         /**
 535          * Get a random string
 536          *
 537          * @private
 538          * @static
 539          */
 540         function getRandomString() {
 541                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 542         }
 543
 544         function &getTitle() { return $this->mTitle; }
 545         function getOptions() { return $this->mOptions; }
 546         function getRevisionId() { return $this->mRevisionId; }
 547         function getOutput() { return $this->mOutput; }
 548         function nextLinkID() { return $this->mLinkID++; }
 549
 550         function getFunctionLang() {
 551                 global $wgLang, $wgContLang;
 552
 553                 $target = $this->mOptions->getTargetLanguage();
 554                 if ( $target !== null ) {
 555                         return $target;
 556                 } else {
 557                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 558                 }
 559         }
 560
 561         /**
 562          * Get a preprocessor object
 563          */
 564         function getPreprocessor() {
 565                 if ( !isset( $this->mPreprocessor ) ) {
 566                         $class = $this->mPreprocessorClass;
 567                         $this->mPreprocessor = new $class( $this );
 568                 }
 569                 return $this->mPreprocessor;
 570         }
 571
 572         /**
 573          * Replaces all occurrences of HTML-style comments and the given tags
 574          * in the text with a random marker and returns the next text. The output
 575          * parameter $matches will be an associative array filled with data in
 576          * the form:
 577          *   'UNIQ-xxxxx' => array(
 578          *     'element',
 579          *     'tag content',
 580          *     array( 'param' => 'x' ),
 581          *     '<element param="x">tag content</element>' ) )
 582          *
 583          * @param $elements list of element names. Comments are always extracted.
 584          * @param $text Source text string.
 585          * @param $uniq_prefix
 586          *
 587          * @public
 588          * @static
 589          */
 590         function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 591                 static $n = 1;
 592                 $stripped = '';
 593                 $matches = array();
 594
 595                 $taglist = implode( '|', $elements );
 596                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 597
 598                 while ( $text != '' ) {
 599                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 600                         $stripped .= $p[0];
 601                         if ( count( $p ) < 5 ) {
 602                                 break;
 603                         }
 604                         if ( count( $p ) > 5 ) {
 605                                 # comment
 606                                 $element    = $p[4];
 607                                 $attributes = '';
 608                                 $close      = '';
 609                                 $inside     = $p[5];
 610                         } else {
 611                                 # tag
 612                                 $element    = $p[1];
 613                                 $attributes = $p[2];
 614                                 $close      = $p[3];
 615                                 $inside     = $p[4];
 616                         }
 617
 618                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 619                         $stripped .= $marker;
 620
 621                         if ( $close === '/>' ) {
 622                                 # Empty element tag, <tag />
 623                                 $content = null;
 624                                 $text = $inside;
 625                                 $tail = null;
 626                         } else {
 627                                 if ( $element === '!--' ) {
 628                                         $end = '/(-->)/';
 629                                 } else {
 630                                         $end = "/(<\\/$element\\s*>)/i";
 631                                 }
 632                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 633                                 $content = $q[0];
 634                                 if ( count( $q ) < 3 ) {
 635                                         # No end tag -- let it run out to the end of the text.
 636                                         $tail = '';
 637                                         $text = '';
 638                                 } else {
 639                                         $tail = $q[1];
 640                                         $text = $q[2];
 641                                 }
 642                         }
 643
 644                         $matches[$marker] = array( $element,
 645                                 $content,
 646                                 Sanitizer::decodeTagAttributes( $attributes ),
 647                                 "<$element$attributes$close$content$tail" );
 648                 }
 649                 return $stripped;
 650         }
 651
 652         /**
 653          * Get a list of strippable XML-like elements
 654          */
 655         function getStripList() {
 656                 return $this->mStripList;
 657         }
 658
 659         /**
 660          * @deprecated use replaceVariables
 661          */
 662         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 663                 return $text;
 664         }
 665
 666         /**
 667          * Restores pre, math, and other extensions removed by strip()
 668          *
 669          * always call unstripNoWiki() after this one
 670          * @private
 671          * @deprecated use $this->mStripState->unstrip()
 672          */
 673         function unstrip( $text, $state ) {
 674                 return $state->unstripGeneral( $text );
 675         }
 676
 677         /**
 678          * Always call this after unstrip() to preserve the order
 679          *
 680          * @private
 681          * @deprecated use $this->mStripState->unstrip()
 682          */
 683         function unstripNoWiki( $text, $state ) {
 684                 return $state->unstripNoWiki( $text );
 685         }
 686
 687         /**
 688          * @deprecated use $this->mStripState->unstripBoth()
 689          */
 690         function unstripForHTML( $text ) {
 691                 return $this->mStripState->unstripBoth( $text );
 692         }
 693
 694         /**
 695          * Add an item to the strip state
 696          * Returns the unique tag which must be inserted into the stripped text
 697          * The tag will be replaced with the original text in unstrip()
 698          *
 699          * @private
 700          */
 701         function insertStripItem( $text ) {
 702                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 703                 $this->mMarkerIndex++;
 704                 $this->mStripState->general->setPair( $rnd, $text );
 705                 return $rnd;
 706         }
 707
 708         /**
 709          * Interface with html tidy
 710          * @deprecated Use MWTidy::tidy()
 711          */
 712         public static function tidy( $text ) {
 713                 wfDeprecated( __METHOD__ );
 714                 return MWTidy::tidy( $text );
 715         }
 716
 717         /**
 718          * parse the wiki syntax used to render tables
 719          *
 720          * @private
 721          */
 722         function doTableStuff( $text ) {
 723                 wfProfileIn( __METHOD__ );
 724
 725                 $lines = StringUtils::explode( "\n", $text );
 726                 $out = '';
 727                 $td_history = array(); # Is currently a td tag open?
 728                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 729                 $tr_history = array(); # Is currently a tr tag open?
 730                 $tr_attributes = array(); # history of tr attributes
 731                 $has_opened_tr = array(); # Did this table open a <tr> element?
 732                 $indent_level = 0; # indent level of the table
 733
 734                 foreach ( $lines as $outLine ) {
 735                         $line = trim( $outLine );
 736
 737                         if ( $line == '' ) { # empty line, go to next line
 738                                 $out .= $outLine."\n";
 739                                 continue;
 740                         }
 741                         $first_character = $line[0];
 742                         $matches = array();
 743
 744                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 745                                 # First check if we are starting a new table
 746                                 $indent_level = strlen( $matches[1] );
 747
 748                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 749                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 750
 751                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 752                                 array_push( $td_history , false );
 753                                 array_push( $last_tag_history , '' );
 754                                 array_push( $tr_history , false );
 755                                 array_push( $tr_attributes , '' );
 756                                 array_push( $has_opened_tr , false );
 757                         } elseif ( count( $td_history ) == 0 ) {
 758                                 # Don't do any of the following
 759                                 $out .= $outLine."\n";
 760                                 continue;
 761                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 762                                 # We are ending a table
 763                                 $line = '</table>' . substr( $line , 2 );
 764                                 $last_tag = array_pop( $last_tag_history );
 765
 766                                 if ( !array_pop( $has_opened_tr ) ) {
 767                                         $line = "<tr><td></td></tr>{$line}";
 768                                 }
 769
 770                                 if ( array_pop( $tr_history ) ) {
 771                                         $line = "</tr>{$line}";
 772                                 }
 773
 774                                 if ( array_pop( $td_history ) ) {
 775                                         $line = "</{$last_tag}>{$line}";
 776                                 }
 777                                 array_pop( $tr_attributes );
 778                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 779                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 780                                 # Now we have a table row
 781                                 $line = preg_replace( '#^\|-+#', '', $line );
 782
 783                                 # Whats after the tag is now only attributes
 784                                 $attributes = $this->mStripState->unstripBoth( $line );
 785                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 786                                 array_pop( $tr_attributes );
 787                                 array_push( $tr_attributes, $attributes );
 788
 789                                 $line = '';
 790                                 $last_tag = array_pop( $last_tag_history );
 791                                 array_pop( $has_opened_tr );
 792                                 array_push( $has_opened_tr , true );
 793
 794                                 if ( array_pop( $tr_history ) ) {
 795                                         $line = '</tr>';
 796                                 }
 797
 798                                 if ( array_pop( $td_history ) ) {
 799                                         $line = "</{$last_tag}>{$line}";
 800                                 }
 801
 802                                 $outLine = $line;
 803                                 array_push( $tr_history , false );
 804                                 array_push( $td_history , false );
 805                                 array_push( $last_tag_history , '' );
 806                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 807                                 # This might be cell elements, td, th or captions
 808                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 809                                         $first_character = '+';
 810                                         $line = substr( $line , 1 );
 811                                 }
 812
 813                                 $line = substr( $line , 1 );
 814
 815                                 if ( $first_character === '!' ) {
 816                                         $line = str_replace( '!!' , '||' , $line );
 817                                 }
 818
 819                                 # Split up multiple cells on the same line.
 820                                 # FIXME : This can result in improper nesting of tags processed
 821                                 # by earlier parser steps, but should avoid splitting up eg
 822                                 # attribute values containing literal "||".
 823                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 824
 825                                 $outLine = '';
 826
 827                                 # Loop through each table cell
 828                                 foreach ( $cells as $cell ) {
 829                                         $previous = '';
 830                                         if ( $first_character !== '+' ) {
 831                                                 $tr_after = array_pop( $tr_attributes );
 832                                                 if ( !array_pop( $tr_history ) ) {
 833                                                         $previous = "<tr{$tr_after}>\n";
 834                                                 }
 835                                                 array_push( $tr_history , true );
 836                                                 array_push( $tr_attributes , '' );
 837                                                 array_pop( $has_opened_tr );
 838                                                 array_push( $has_opened_tr , true );
 839                                         }
 840
 841                                         $last_tag = array_pop( $last_tag_history );
 842
 843                                         if ( array_pop( $td_history ) ) {
 844                                                 $previous = "</{$last_tag}>{$previous}";
 845                                         }
 846
 847                                         if ( $first_character === '|' ) {
 848                                                 $last_tag = 'td';
 849                                         } elseif ( $first_character === '!' ) {
 850                                                 $last_tag = 'th';
 851                                         } elseif ( $first_character === '+' ) {
 852                                                 $last_tag = 'caption';
 853                                         } else {
 854                                                 $last_tag = '';
 855                                         }
 856
 857                                         array_push( $last_tag_history , $last_tag );
 858
 859                                         # A cell could contain both parameters and data
 860                                         $cell_data = explode( '|' , $cell , 2 );
 861
 862                                         # Bug 553: Note that a '|' inside an invalid link should not
 863                                         # be mistaken as delimiting cell parameters
 864                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 865                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 866                                         } elseif ( count( $cell_data ) == 1 ) {
 867                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 868                                         } else {
 869                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 870                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 871                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 872                                         }
 873
 874                                         $outLine .= $cell;
 875                                         array_push( $td_history , true );
 876                                 }
 877                         }
 878                         $out .= $outLine . "\n";
 879                 }
 880
 881                 # Closing open td, tr && table
 882                 while ( count( $td_history ) > 0 ) {
 883                         if ( array_pop( $td_history ) ) {
 884                                 $out .= "</td>\n";
 885                         }
 886                         if ( array_pop( $tr_history ) ) {
 887                                 $out .= "</tr>\n";
 888                         }
 889                         if ( !array_pop( $has_opened_tr ) ) {
 890                                 $out .= "<tr><td></td></tr>\n" ;
 891                         }
 892
 893                         $out .= "</table>\n";
 894                 }
 895
 896                 # Remove trailing line-ending (b/c)
 897                 if ( substr( $out, -1 ) === "\n" ) {
 898                         $out = substr( $out, 0, -1 );
 899                 }
 900
 901                 # special case: don't return empty table
 902                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 903                         $out = '';
 904                 }
 905
 906                 wfProfileOut( __METHOD__ );
 907
 908                 return $out;
 909         }
 910
 911         /**
 912          * Helper function for parse() that transforms wiki markup into
 913          * HTML. Only called for $mOutputType == self::OT_HTML.
 914          *
 915          * @private
 916          */
 917         function internalParse( $text, $isMain = true, $frame=false ) {
 918                 wfProfileIn( __METHOD__ );
 919
 920                 $origText = $text;
 921
 922                 # Hook to suspend the parser in this state
 923                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 924                         wfProfileOut( __METHOD__ );
 925                         return $text ;
 926                 }
 927
 928                 # if $frame is provided, then use $frame for replacing any variables
 929                 if ( $frame ) {
 930                         # use frame depth to infer how include/noinclude tags should be handled
 931                         # depth=0 means this is the top-level document; otherwise it's an included document
 932                         if ( !$frame->depth ) {
 933                                 $flag = 0;
 934                         } else {
 935                                 $flag = Parser::PTD_FOR_INCLUSION;
 936                         }
 937                         $dom = $this->preprocessToDom( $text, $flag );
 938                         $text = $frame->expand( $dom );
 939                 } else {
 940                         # if $frame is not provided, then use old-style replaceVariables
 941                         $text = $this->replaceVariables( $text );
 942                 }
 943
 944                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 945                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 946
 947                 # Tables need to come after variable replacement for things to work
 948                 # properly; putting them before other transformations should keep
 949                 # exciting things like link expansions from showing up in surprising
 950                 # places.
 951                 $text = $this->doTableStuff( $text );
 952
 953                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 954
 955                 $text = $this->doDoubleUnderscore( $text );
 956
 957                 $text = $this->doHeadings( $text );
 958                 if ( $this->mOptions->getUseDynamicDates() ) {
 959                         $df = DateFormatter::getInstance();
 960                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 961                 }
 962                 $text = $this->doAllQuotes( $text );
 963                 $text = $this->replaceInternalLinks( $text );
 964                 $text = $this->replaceExternalLinks( $text );
 965
 966                 # replaceInternalLinks may sometimes leave behind
 967                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 968                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
 969
 970                 $text = $this->doMagicLinks( $text );
 971                 $text = $this->formatHeadings( $text, $origText, $isMain );
 972
 973                 wfProfileOut( __METHOD__ );
 974                 return $text;
 975         }
 976
 977         /**
 978          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 979          * magic external links.
 980          *
 981          * DML
 982          * @private
 983          */
 984         function doMagicLinks( $text ) {
 985                 wfProfileIn( __METHOD__ );
 986                 $prots = $this->mUrlProtocols;
 987                 $urlChar = self::EXT_LINK_URL_CLASS;
 988                 $text = preg_replace_callback(
 989                         '!(?:                           # Start cases
 990                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
 991                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 992                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 993                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 994                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 995                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 996                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 997                                     [0-9Xx]                 # check digit
 998                                     \b)
 999                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1000                 wfProfileOut( __METHOD__ );
1001                 return $text;
1002         }
1003
1004         function magicLinkCallback( $m ) {
1005                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1006                         # Skip anchor
1007                         return $m[0];
1008                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1009                         # Skip HTML element
1010                         return $m[0];
1011                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1012                         # Free external link
1013                         return $this->makeFreeExternalLink( $m[0] );
1014                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1015                         # RFC or PMID
1016                         $CssClass = '';
1017                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1018                                 $keyword = 'RFC';
1019                                 $urlmsg = 'rfcurl';
1020                                 $CssClass = 'mw-magiclink-rfc';
1021                                 $id = $m[4];
1022                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1023                                 $keyword = 'PMID';
1024                                 $urlmsg = 'pubmedurl';
1025                                 $CssClass = 'mw-magiclink-pmid';
1026                                 $id = $m[4];
1027                         } else {
1028                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1029                                         substr( $m[0], 0, 20 ) . '"' );
1030                         }
1031                         $url = wfMsg( $urlmsg, $id);
1032                         $sk = $this->mOptions->getSkin();
1033                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1034                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1035                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1036                         # ISBN
1037                         $isbn = $m[5];
1038                         $num = strtr( $isbn, array(
1039                                 '-' => '',
1040                                 ' ' => '',
1041                                 'x' => 'X',
1042                         ));
1043                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1044                         return'<a href="' .
1045                                 $titleObj->escapeLocalUrl() .
1046                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1047                 } else {
1048                         return $m[0];
1049                 }
1050         }
1051
1052         /**
1053          * Make a free external link, given a user-supplied URL
1054          * @return HTML
1055          * @private
1056          */
1057         function makeFreeExternalLink( $url ) {
1058                 global $wgContLang;
1059                 wfProfileIn( __METHOD__ );
1060
1061                 $sk = $this->mOptions->getSkin();
1062                 $trail = '';
1063
1064                 # The characters '<' and '>' (which were escaped by
1065                 # removeHTMLtags()) should not be included in
1066                 # URLs, per RFC 2396.
1067                 $m2 = array();
1068                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1069                         $trail = substr( $url, $m2[0][1] ) . $trail;
1070                         $url = substr( $url, 0, $m2[0][1] );
1071                 }
1072
1073                 # Move trailing punctuation to $trail
1074                 $sep = ',;\.:!?';
1075                 # If there is no left bracket, then consider right brackets fair game too
1076                 if ( strpos( $url, '(' ) === false ) {
1077                         $sep .= ')';
1078                 }
1079
1080                 $numSepChars = strspn( strrev( $url ), $sep );
1081                 if ( $numSepChars ) {
1082                         $trail = substr( $url, -$numSepChars ) . $trail;
1083                         $url = substr( $url, 0, -$numSepChars );
1084                 }
1085
1086                 $url = Sanitizer::cleanUrl( $url );
1087
1088                 # Is this an external image?
1089                 $text = $this->maybeMakeExternalImage( $url );
1090                 if ( $text === false ) {
1091                         # Not an image, make a link
1092                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1093                                 $this->getExternalLinkAttribs( $url ) );
1094                         # Register it in the output object...
1095                         # Replace unnecessary URL escape codes with their equivalent characters
1096                         $pasteurized = self::replaceUnusualEscapes( $url );
1097                         $this->mOutput->addExternalLink( $pasteurized );
1098                 }
1099                 wfProfileOut( __METHOD__ );
1100                 return $text . $trail;
1101         }
1102
1103
1104         /**
1105          * Parse headers and return html
1106          *
1107          * @private
1108          */
1109         function doHeadings( $text ) {
1110                 wfProfileIn( __METHOD__ );
1111                 for ( $i = 6; $i >= 1; --$i ) {
1112                         $h = str_repeat( '=', $i );
1113                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1114                           "<h$i>\\1</h$i>", $text );
1115                 }
1116                 wfProfileOut( __METHOD__ );
1117                 return $text;
1118         }
1119
1120         /**
1121          * Replace single quotes with HTML markup
1122          * @private
1123          * @return string the altered text
1124          */
1125         function doAllQuotes( $text ) {
1126                 wfProfileIn( __METHOD__ );
1127                 $outtext = '';
1128                 $lines = StringUtils::explode( "\n", $text );
1129                 foreach ( $lines as $line ) {
1130                         $outtext .= $this->doQuotes( $line ) . "\n";
1131                 }
1132                 $outtext = substr( $outtext, 0,-1 );
1133                 wfProfileOut( __METHOD__ );
1134                 return $outtext;
1135         }
1136
1137         /**
1138          * Helper function for doAllQuotes()
1139          */
1140         public function doQuotes( $text ) {
1141                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1142                 if ( count( $arr ) == 1 ) {
1143                         return $text;
1144                 } else {
1145                         # First, do some preliminary work. This may shift some apostrophes from
1146                         # being mark-up to being text. It also counts the number of occurrences
1147                         # of bold and italics mark-ups.
1148                         $i = 0;
1149                         $numbold = 0;
1150                         $numitalics = 0;
1151                         foreach ( $arr as $r ) {
1152                                 if ( ( $i % 2 ) == 1 ) {
1153                                         # If there are ever four apostrophes, assume the first is supposed to
1154                                         # be text, and the remaining three constitute mark-up for bold text.
1155                                         if ( strlen( $arr[$i] ) == 4 ) {
1156                                                 $arr[$i-1] .= "'";
1157                                                 $arr[$i] = "'''";
1158                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1159                                                 # If there are more than 5 apostrophes in a row, assume they're all
1160                                                 # text except for the last 5.
1161                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1162                                                 $arr[$i] = "'''''";
1163                                         }
1164                                         # Count the number of occurrences of bold and italics mark-ups.
1165                                         # We are not counting sequences of five apostrophes.
1166                                         if ( strlen( $arr[$i] ) == 2 ) {
1167                                                 $numitalics++;
1168                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1169                                                 $numbold++;
1170                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1171                                                 $numitalics++;
1172                                                 $numbold++;
1173                                         }
1174                                 }
1175                                 $i++;
1176                         }
1177
1178                         # If there is an odd number of both bold and italics, it is likely
1179                         # that one of the bold ones was meant to be an apostrophe followed
1180                         # by italics. Which one we cannot know for certain, but it is more
1181                         # likely to be one that has a single-letter word before it.
1182                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1183                                 $i = 0;
1184                                 $firstsingleletterword = -1;
1185                                 $firstmultiletterword = -1;
1186                                 $firstspace = -1;
1187                                 foreach ( $arr as $r ) {
1188                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1189                                                 $x1 = substr( $arr[$i-1], -1 );
1190                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1191                                                 if ( $x1 === ' ' ) {
1192                                                         if ( $firstspace == -1 ) {
1193                                                                 $firstspace = $i;
1194                                                         }
1195                                                 } elseif ( $x2 === ' ') {
1196                                                         if ( $firstsingleletterword == -1 ) {
1197                                                                 $firstsingleletterword = $i;
1198                                                         }
1199                                                 } else {
1200                                                         if ( $firstmultiletterword == -1 ) {
1201                                                                 $firstmultiletterword = $i;
1202                                                         }
1203                                                 }
1204                                         }
1205                                         $i++;
1206                                 }
1207
1208                                 # If there is a single-letter word, use it!
1209                                 if ( $firstsingleletterword > -1 ) {
1210                                         $arr[$firstsingleletterword] = "''";
1211                                         $arr[$firstsingleletterword-1] .= "'";
1212                                 } elseif ( $firstmultiletterword > -1 ) {
1213                                         # If not, but there's a multi-letter word, use that one.
1214                                         $arr[$firstmultiletterword] = "''";
1215                                         $arr[$firstmultiletterword-1] .= "'";
1216                                 } elseif ( $firstspace > -1 ) {
1217                                         # ... otherwise use the first one that has neither.
1218                                         # (notice that it is possible for all three to be -1 if, for example,
1219                                         # there is only one pentuple-apostrophe in the line)
1220                                         $arr[$firstspace] = "''";
1221                                         $arr[$firstspace-1] .= "'";
1222                                 }
1223                         }
1224
1225                         # Now let's actually convert our apostrophic mush to HTML!
1226                         $output = '';
1227                         $buffer = '';
1228                         $state = '';
1229                         $i = 0;
1230                         foreach ( $arr as $r ) {
1231                                 if ( ( $i % 2 ) == 0 ) {
1232                                         if ( $state === 'both' ) {
1233                                                 $buffer .= $r;
1234                                         } else {
1235                                                 $output .= $r;
1236                                         }
1237                                 } else {
1238                                         if ( strlen( $r ) == 2 ) {
1239                                                 if ( $state === 'i' ) {
1240                                                         $output .= '</i>'; $state = '';
1241                                                 } elseif ( $state === 'bi' ) {
1242                                                         $output .= '</i>'; $state = 'b';
1243                                                 } elseif ( $state === 'ib' ) {
1244                                                         $output .= '</b></i><b>'; $state = 'b';
1245                                                 } elseif ( $state === 'both' ) {
1246                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1247                                                 } else { # $state can be 'b' or ''
1248                                                         $output .= '<i>'; $state .= 'i';
1249                                                 }
1250                                         } elseif ( strlen( $r ) == 3 ) {
1251                                                 if ( $state === 'b' ) {
1252                                                         $output .= '</b>'; $state = '';
1253                                                 } elseif ( $state === 'bi' ) {
1254                                                         $output .= '</i></b><i>'; $state = 'i';
1255                                                 } elseif ( $state === 'ib' ) {
1256                                                         $output .= '</b>'; $state = 'i';
1257                                                 } elseif ( $state === 'both' ) {
1258                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1259                                                 } else { # $state can be 'i' or ''
1260                                                         $output .= '<b>'; $state .= 'b';
1261                                                 }
1262                                         } elseif ( strlen( $r ) == 5 ) {
1263                                                 if ( $state === 'b' ) {
1264                                                         $output .= '</b><i>'; $state = 'i';
1265                                                 } elseif ( $state === 'i' ) {
1266                                                         $output .= '</i><b>'; $state = 'b';
1267                                                 } elseif ( $state === 'bi' ) {
1268                                                         $output .= '</i></b>'; $state = '';
1269                                                 } elseif ( $state === 'ib' ) {
1270                                                         $output .= '</b></i>'; $state = '';
1271                                                 } elseif ( $state === 'both' ) {
1272                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1273                                                 } else { # ($state == '')
1274                                                         $buffer = ''; $state = 'both';
1275                                                 }
1276                                         }
1277                                 }
1278                                 $i++;
1279                         }
1280                         # Now close all remaining tags.  Notice that the order is important.
1281                         if ( $state === 'b' || $state === 'ib' ) {
1282                                 $output .= '</b>';
1283                         }
1284                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1285                                 $output .= '</i>';
1286                         }
1287                         if ( $state === 'bi' ) {
1288                                 $output .= '</b>';
1289                         }
1290                         # There might be lonely ''''', so make sure we have a buffer
1291                         if ( $state === 'both' && $buffer ) {
1292                                 $output .= '<b><i>'.$buffer.'</i></b>';
1293                         }
1294                         return $output;
1295                 }
1296         }
1297
1298         /**
1299          * Replace external links (REL)
1300          *
1301          * Note: this is all very hackish and the order of execution matters a lot.
1302          * Make sure to run maintenance/parserTests.php if you change this code.
1303          *
1304          * @private
1305          */
1306         function replaceExternalLinks( $text ) {
1307                 global $wgContLang;
1308                 wfProfileIn( __METHOD__ );
1309
1310                 $sk = $this->mOptions->getSkin();
1311
1312                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1313                 $s = array_shift( $bits );
1314
1315                 $i = 0;
1316                 while ( $i<count( $bits ) ) {
1317                         $url = $bits[$i++];
1318                         $protocol = $bits[$i++];
1319                         $text = $bits[$i++];
1320                         $trail = $bits[$i++];
1321
1322                         # The characters '<' and '>' (which were escaped by
1323                         # removeHTMLtags()) should not be included in
1324                         # URLs, per RFC 2396.
1325                         $m2 = array();
1326                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1327                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1328                                 $url = substr( $url, 0, $m2[0][1] );
1329                         }
1330
1331                         # If the link text is an image URL, replace it with an <img> tag
1332                         # This happened by accident in the original parser, but some people used it extensively
1333                         $img = $this->maybeMakeExternalImage( $text );
1334                         if ( $img !== false ) {
1335                                 $text = $img;
1336                         }
1337
1338                         $dtrail = '';
1339
1340                         # Set linktype for CSS - if URL==text, link is essentially free
1341                         $linktype = ( $text === $url ) ? 'free' : 'text';
1342
1343                         # No link text, e.g. [http://domain.tld/some.link]
1344                         if ( $text == '' ) {
1345                                 # Autonumber if allowed. See bug #5918
1346                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1347                                         $langObj = $this->getFunctionLang();
1348                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1349                                         $linktype = 'autonumber';
1350                                 } else {
1351                                         # Otherwise just use the URL
1352                                         $text = htmlspecialchars( $url );
1353                                         $linktype = 'free';
1354                                 }
1355                         } else {
1356                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1357                                 # Check for trail
1358                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1359                         }
1360
1361                         $text = $wgContLang->markNoConversion( $text );
1362
1363                         $url = Sanitizer::cleanUrl( $url );
1364
1365                         # Use the encoded URL
1366                         # This means that users can paste URLs directly into the text
1367                         # Funny characters like &ouml; aren't valid in URLs anyway
1368                         # This was changed in August 2004
1369                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1370                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1371
1372                         # Register link in the output object.
1373                         # Replace unnecessary URL escape codes with the referenced character
1374                         # This prevents spammers from hiding links from the filters
1375                         $pasteurized = self::replaceUnusualEscapes( $url );
1376                         $this->mOutput->addExternalLink( $pasteurized );
1377                 }
1378
1379                 wfProfileOut( __METHOD__ );
1380                 return $s;
1381         }
1382
1383         /**
1384          * Get an associative array of additional HTML attributes appropriate for a
1385          * particular external link.  This currently may include rel => nofollow
1386          * (depending on configuration, namespace, and the URL's domain) and/or a
1387          * target attribute (depending on configuration).
1388          *
1389          * @param string $url Optional URL, to extract the domain from for rel =>
1390          *   nofollow if appropriate
1391          * @return array Associative array of HTML attributes
1392          */
1393         function getExternalLinkAttribs( $url = false ) {
1394                 $attribs = array();
1395                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1396                 $ns = $this->mTitle->getNamespace();
1397                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1398                         $attribs['rel'] = 'nofollow';
1399
1400                         global $wgNoFollowDomainExceptions;
1401                         if ( $wgNoFollowDomainExceptions ) {
1402                                 $bits = wfParseUrl( $url );
1403                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1404                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1405                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1406                                                         unset( $attribs['rel'] );
1407                                                         break;
1408                                                 }
1409                                         }
1410                                 }
1411                         }
1412                 }
1413                 if ( $this->mOptions->getExternalLinkTarget() ) {
1414                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1415                 }
1416                 return $attribs;
1417         }
1418
1419
1420         /**
1421          * Replace unusual URL escape codes with their equivalent characters
1422          * @param string
1423          * @return string
1424          * @static
1425          * @todo  This can merge genuinely required bits in the path or query string,
1426          *        breaking legit URLs. A proper fix would treat the various parts of
1427          *        the URL differently; as a workaround, just use the output for
1428          *        statistical records, not for actual linking/output.
1429          */
1430         static function replaceUnusualEscapes( $url ) {
1431                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1432                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1433         }
1434
1435         /**
1436          * Callback function used in replaceUnusualEscapes().
1437          * Replaces unusual URL escape codes with their equivalent character
1438          * @static
1439          * @private
1440          */
1441         private static function replaceUnusualEscapesCallback( $matches ) {
1442                 $char = urldecode( $matches[0] );
1443                 $ord = ord( $char );
1444                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1445                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1446                         # No, shouldn't be escaped
1447                         return $char;
1448                 } else {
1449                         # Yes, leave it escaped
1450                         return $matches[0];
1451                 }
1452         }
1453
1454         /**
1455          * make an image if it's allowed, either through the global
1456          * option, through the exception, or through the on-wiki whitelist
1457          * @private
1458          */
1459         function maybeMakeExternalImage( $url ) {
1460                 $sk = $this->mOptions->getSkin();
1461                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1462                 $imagesexception = !empty( $imagesfrom );
1463                 $text = false;
1464                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1465                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1466                         $imagematch = false;
1467                         foreach ( $imagesfrom as $match ) {
1468                                 if ( strpos( $url, $match ) === 0 ) {
1469                                         $imagematch = true;
1470                                         break;
1471                                 }
1472                         }
1473                 } elseif ( $imagesexception ) {
1474                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1475                 } else {
1476                         $imagematch = false;
1477                 }
1478                 if ( $this->mOptions->getAllowExternalImages()
1479                      || ( $imagesexception && $imagematch ) ) {
1480                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1481                                 # Image found
1482                                 $text = $sk->makeExternalImage( $url );
1483                         }
1484                 }
1485                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1486                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1487                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1488                         foreach ( $whitelist as $entry ) {
1489                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1490                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1491                                         continue;
1492                                 }
1493                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1494                                         # Image matches a whitelist entry
1495                                         $text = $sk->makeExternalImage( $url );
1496                                         break;
1497                                 }
1498                         }
1499                 }
1500                 return $text;
1501         }
1502
1503         /**
1504          * Process [[ ]] wikilinks
1505          * @return processed text
1506          *
1507          * @private
1508          */
1509         function replaceInternalLinks( $s ) {
1510                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1511                 return $s;
1512         }
1513
1514         /**
1515          * Process [[ ]] wikilinks (RIL)
1516          * @return LinkHolderArray
1517          *
1518          * @private
1519          */
1520         function replaceInternalLinks2( &$s ) {
1521                 global $wgContLang;
1522
1523                 wfProfileIn( __METHOD__ );
1524
1525                 wfProfileIn( __METHOD__.'-setup' );
1526                 static $tc = FALSE, $e1, $e1_img;
1527                 # the % is needed to support urlencoded titles as well
1528                 if ( !$tc ) {
1529                         $tc = Title::legalChars() . '#%';
1530                         # Match a link having the form [[namespace:link|alternate]]trail
1531                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1532                         # Match cases where there is no "]]", which might still be images
1533                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1534                 }
1535
1536                 $sk = $this->mOptions->getSkin();
1537                 $holders = new LinkHolderArray( $this );
1538
1539                 # split the entire text string on occurences of [[
1540                 $a = StringUtils::explode( '[[', ' ' . $s );
1541                 # get the first element (all text up to first [[), and remove the space we added
1542                 $s = $a->current();
1543                 $a->next();
1544                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1545                 $s = substr( $s, 1 );
1546
1547                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1548                 $e2 = null;
1549                 if ( $useLinkPrefixExtension ) {
1550                         # Match the end of a line for a word that's not followed by whitespace,
1551                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1552                         $e2 = wfMsgForContent( 'linkprefix' );
1553                 }
1554
1555                 if ( is_null( $this->mTitle ) ) {
1556                         wfProfileOut( __METHOD__.'-setup' );
1557                         wfProfileOut( __METHOD__ );
1558                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1559                 }
1560                 $nottalk = !$this->mTitle->isTalkPage();
1561
1562                 if ( $useLinkPrefixExtension ) {
1563                         $m = array();
1564                         if ( preg_match( $e2, $s, $m ) ) {
1565                                 $first_prefix = $m[2];
1566                         } else {
1567                                 $first_prefix = false;
1568                         }
1569                 } else {
1570                         $prefix = '';
1571                 }
1572
1573                 if ( $wgContLang->hasVariants() ) {
1574                         $selflink = $wgContLang->convertLinkToAllVariants( $this->mTitle->getPrefixedText() );
1575                 } else {
1576                         $selflink = array( $this->mTitle->getPrefixedText() );
1577                 }
1578                 $useSubpages = $this->areSubpagesAllowed();
1579                 wfProfileOut( __METHOD__.'-setup' );
1580
1581                 # Loop for each link
1582                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1583                         # Check for excessive memory usage
1584                         if ( $holders->isBig() ) {
1585                                 # Too big
1586                                 # Do the existence check, replace the link holders and clear the array
1587                                 $holders->replace( $s );
1588                                 $holders->clear();
1589                         }
1590
1591                         if ( $useLinkPrefixExtension ) {
1592                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1593                                 if ( preg_match( $e2, $s, $m ) ) {
1594                                         $prefix = $m[2];
1595                                         $s = $m[1];
1596                                 } else {
1597                                         $prefix='';
1598                                 }
1599                                 # first link
1600                                 if ( $first_prefix ) {
1601                                         $prefix = $first_prefix;
1602                                         $first_prefix = false;
1603                                 }
1604                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1605                         }
1606
1607                         $might_be_img = false;
1608
1609                         wfProfileIn( __METHOD__."-e1" );
1610                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1611                                 $text = $m[2];
1612                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1613                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1614                                 # the real problem is with the $e1 regex
1615                                 # See bug 1300.
1616                                 #
1617                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1618                                 # and no image is in sight. See bug 2095.
1619                                 #
1620                                 if ( $text !== '' &&
1621                                         substr( $m[3], 0, 1 ) === ']' &&
1622                                         strpos( $text, '[' ) !== false
1623                                 )
1624                                 {
1625                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1626                                         $m[3] = substr( $m[3], 1 );
1627                                 }
1628                                 # fix up urlencoded title texts
1629                                 if ( strpos( $m[1], '%' ) !== false ) {
1630                                         # Should anchors '#' also be rejected?
1631                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode( $m[1] ) );
1632                                 }
1633                                 $trail = $m[3];
1634                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1635                                 $might_be_img = true;
1636                                 $text = $m[2];
1637                                 if ( strpos( $m[1], '%' ) !== false ) {
1638                                         $m[1] = urldecode( $m[1] );
1639                                 }
1640                                 $trail = "";
1641                         } else { # Invalid form; output directly
1642                                 $s .= $prefix . '[[' . $line ;
1643                                 wfProfileOut( __METHOD__."-e1" );
1644                                 continue;
1645                         }
1646                         wfProfileOut( __METHOD__."-e1" );
1647                         wfProfileIn( __METHOD__."-misc" );
1648
1649                         # Don't allow internal links to pages containing
1650                         # PROTO: where PROTO is a valid URL protocol; these
1651                         # should be external links.
1652                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1653                                 $s .= $prefix . '[[' . $line ;
1654                                 wfProfileOut( __METHOD__."-misc" );
1655                                 continue;
1656                         }
1657
1658                         # Make subpage if necessary
1659                         if ( $useSubpages ) {
1660                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1661                         } else {
1662                                 $link = $m[1];
1663                         }
1664
1665                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1666                         if ( !$noforce ) {
1667                                 # Strip off leading ':'
1668                                 $link = substr( $link, 1 );
1669                         }
1670
1671                         wfProfileOut( __METHOD__."-misc" );
1672                         wfProfileIn( __METHOD__."-title" );
1673                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1674                         if ( $nt === null ) {
1675                                 $s .= $prefix . '[[' . $line;
1676                                 wfProfileOut( __METHOD__."-title" );
1677                                 continue;
1678                         }
1679
1680                         $ns = $nt->getNamespace();
1681                         $iw = $nt->getInterWiki();
1682                         wfProfileOut( __METHOD__."-title" );
1683
1684                         if ( $might_be_img ) { # if this is actually an invalid link
1685                                 wfProfileIn( __METHOD__."-might_be_img" );
1686                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1687                                         $found = false;
1688                                         while ( true ) {
1689                                                 # look at the next 'line' to see if we can close it there
1690                                                 $a->next();
1691                                                 $next_line = $a->current();
1692                                                 if ( $next_line === false || $next_line === null ) {
1693                                                         break;
1694                                                 }
1695                                                 $m = explode( ']]', $next_line, 3 );
1696                                                 if ( count( $m ) == 3 ) {
1697                                                         # the first ]] closes the inner link, the second the image
1698                                                         $found = true;
1699                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1700                                                         $trail = $m[2];
1701                                                         break;
1702                                                 } elseif ( count( $m ) == 2 ) {
1703                                                         # if there's exactly one ]] that's fine, we'll keep looking
1704                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1705                                                 } else {
1706                                                         # if $next_line is invalid too, we need look no further
1707                                                         $text .= '[[' . $next_line;
1708                                                         break;
1709                                                 }
1710                                         }
1711                                         if ( !$found ) {
1712                                                 # we couldn't find the end of this imageLink, so output it raw
1713                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1714                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1715                                                 $s .= "{$prefix}[[$link|$text";
1716                                                 # note: no $trail, because without an end, there *is* no trail
1717                                                 wfProfileOut( __METHOD__."-might_be_img" );
1718                                                 continue;
1719                                         }
1720                                 } else { # it's not an image, so output it raw
1721                                         $s .= "{$prefix}[[$link|$text";
1722                                         # note: no $trail, because without an end, there *is* no trail
1723                                         wfProfileOut( __METHOD__."-might_be_img" );
1724                                         continue;
1725                                 }
1726                                 wfProfileOut( __METHOD__."-might_be_img" );
1727                         }
1728
1729                         $wasblank = ( $text  == '' );
1730                         if ( $wasblank ) {
1731                                 $text = $link;
1732                         }
1733
1734                         # Link not escaped by : , create the various objects
1735                         if ( $noforce ) {
1736
1737                                 # Interwikis
1738                                 wfProfileIn( __METHOD__."-interwiki" );
1739                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1740                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1741                                         $s = rtrim( $s . $prefix );
1742                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1743                                         wfProfileOut( __METHOD__."-interwiki" );
1744                                         continue;
1745                                 }
1746                                 wfProfileOut( __METHOD__."-interwiki" );
1747
1748                                 if ( $ns == NS_FILE ) {
1749                                         wfProfileIn( __METHOD__."-image" );
1750                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1751                                                 if ( $wasblank ) {
1752                                                         # if no parameters were passed, $text
1753                                                         # becomes something like "File:Foo.png",
1754                                                         # which we don't want to pass on to the
1755                                                         # image generator
1756                                                         $text = '';
1757                                                 } else {
1758                                                         # recursively parse links inside the image caption
1759                                                         # actually, this will parse them in any other parameters, too,
1760                                                         # but it might be hard to fix that, and it doesn't matter ATM
1761                                                         $text = $this->replaceExternalLinks( $text );
1762                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1763                                                 }
1764                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1765                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1766                                         } else {
1767                                                 $s .= $prefix . $trail;
1768                                         }
1769                                         $this->mOutput->addImage( $nt->getDBkey() );
1770                                         wfProfileOut( __METHOD__."-image" );
1771                                         continue;
1772
1773                                 }
1774
1775                                 if ( $ns == NS_CATEGORY ) {
1776                                         wfProfileIn( __METHOD__."-category" );
1777                                         $s = rtrim( $s . "\n" ); # bug 87
1778
1779                                         if ( $wasblank ) {
1780                                                 $sortkey = $this->getDefaultSort();
1781                                         } else {
1782                                                 $sortkey = $text;
1783                                         }
1784                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1785                                         $sortkey = str_replace( "\n", '', $sortkey );
1786                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1787                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1788
1789                                         /**
1790                                          * Strip the whitespace Category links produce, see bug 87
1791                                          * @todo We might want to use trim($tmp, "\n") here.
1792                                          */
1793                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '': $prefix . $trail;
1794
1795                                         wfProfileOut( __METHOD__."-category" );
1796                                         continue;
1797                                 }
1798                         }
1799
1800                         # Self-link checking
1801                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1802                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1803                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1804                                         continue;
1805                                 }
1806                         }
1807
1808                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1809                         # FIXME: Should do batch file existence checks, see comment below
1810                         if ( $ns == NS_MEDIA ) {
1811                                 wfProfileIn( __METHOD__."-media" );
1812                                 # Give extensions a chance to select the file revision for us
1813                                 $skip = $time = false;
1814                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1815                                 if ( $skip ) {
1816                                         $link = $sk->link( $nt );
1817                                 } else {
1818                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1819                                 }
1820                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1821                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1822                                 $this->mOutput->addImage( $nt->getDBkey() );
1823                                 wfProfileOut( __METHOD__."-media" );
1824                                 continue;
1825                         }
1826
1827                         wfProfileIn( __METHOD__."-always_known" );
1828                         # Some titles, such as valid special pages or files in foreign repos, should
1829                         # be shown as bluelinks even though they're not included in the page table
1830                         #
1831                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1832                         # batch file existence checks for NS_FILE and NS_MEDIA
1833                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1834                                 $this->mOutput->addLink( $nt );
1835                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1836                         } else {
1837                                 # Links will be added to the output link list after checking
1838                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1839                         }
1840                         wfProfileOut( __METHOD__."-always_known" );
1841                 }
1842                 wfProfileOut( __METHOD__ );
1843                 return $holders;
1844         }
1845
1846         /**
1847          * Make a link placeholder. The text returned can be later resolved to a real link with
1848          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1849          * parsing of interwiki links, and secondly to allow all existence checks and
1850          * article length checks (for stub links) to be bundled into a single query.
1851          *
1852          * @deprecated
1853          */
1854         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1855                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1856         }
1857
1858         /**
1859          * Render a forced-blue link inline; protect against double expansion of
1860          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1861          * Since this little disaster has to split off the trail text to avoid
1862          * breaking URLs in the following text without breaking trails on the
1863          * wiki links, it's been made into a horrible function.
1864          *
1865          * @param Title $nt
1866          * @param string $text
1867          * @param string $query
1868          * @param string $trail
1869          * @param string $prefix
1870          * @return string HTML-wikitext mix oh yuck
1871          */
1872         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1873                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1874                 $sk = $this->mOptions->getSkin();
1875                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1876                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1877                 return $this->armorLinks( $link ) . $trail;
1878         }
1879
1880         /**
1881          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1882          * going to go through further parsing steps before inline URL expansion.
1883          *
1884          * Not needed quite as much as it used to be since free links are a bit
1885          * more sensible these days. But bracketed links are still an issue.
1886          *
1887          * @param string more-or-less HTML
1888          * @return string less-or-more HTML with NOPARSE bits
1889          */
1890         function armorLinks( $text ) {
1891                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1892                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1893         }
1894
1895         /**
1896          * Return true if subpage links should be expanded on this page.
1897          * @return bool
1898          */
1899         function areSubpagesAllowed() {
1900                 # Some namespaces don't allow subpages
1901                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1902         }
1903
1904         /**
1905          * Handle link to subpage if necessary
1906          * @param string $target the source of the link
1907          * @param string &$text the link text, modified as necessary
1908          * @return string the full name of the link
1909          * @private
1910          */
1911         function maybeDoSubpageLink( $target, &$text ) {
1912                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1913         }
1914
1915         /**#@+
1916          * Used by doBlockLevels()
1917          * @private
1918          */
1919         function closeParagraph() {
1920                 $result = '';
1921                 if ( $this->mLastSection != '' ) {
1922                         $result = '</' . $this->mLastSection  . ">\n";
1923                 }
1924                 $this->mInPre = false;
1925                 $this->mLastSection = '';
1926                 return $result;
1927         }
1928         /**
1929          * getCommon() returns the length of the longest common substring
1930          * of both arguments, starting at the beginning of both.
1931          * @private
1932          */
1933         function getCommon( $st1, $st2 ) {
1934                 $fl = strlen( $st1 );
1935                 $shorter = strlen( $st2 );
1936                 if ( $fl < $shorter ) {
1937                         $shorter = $fl;
1938                 }
1939
1940                 for ( $i = 0; $i < $shorter; ++$i ) {
1941                         if ( $st1{$i} != $st2{$i} ) {
1942                                 break;
1943                         }
1944                 }
1945                 return $i;
1946         }
1947         /**
1948          * These next three functions open, continue, and close the list
1949          * element appropriate to the prefix character passed into them.
1950          * @private
1951          */
1952         function openList( $char ) {
1953                 $result = $this->closeParagraph();
1954
1955                 if ( '*' === $char ) {
1956                         $result .= '<ul><li>';
1957                 } elseif ( '#' === $char ) {
1958                         $result .= '<ol><li>';
1959                 } elseif ( ':' === $char ) {
1960                         $result .= '<dl><dd>';
1961                 } elseif ( ';' === $char ) {
1962                         $result .= '<dl><dt>';
1963                         $this->mDTopen = true;
1964                 } else {
1965                         $result = '<!-- ERR 1 -->';
1966                 }
1967
1968                 return $result;
1969         }
1970
1971         /**
1972          * TODO: document
1973          * @param $char String
1974          * @private
1975          */
1976         function nextItem( $char ) {
1977                 if ( '*' === $char || '#' === $char ) {
1978                         return '</li><li>';
1979                 } elseif ( ':' === $char || ';' === $char ) {
1980                         $close = '</dd>';
1981                         if ( $this->mDTopen ) {
1982                                 $close = '</dt>';
1983                         }
1984                         if ( ';' === $char ) {
1985                                 $this->mDTopen = true;
1986                                 return $close . '<dt>';
1987                         } else {
1988                                 $this->mDTopen = false;
1989                                 return $close . '<dd>';
1990                         }
1991                 }
1992                 return '<!-- ERR 2 -->';
1993         }
1994
1995         /**
1996          * TODO: document
1997          * @param $char String
1998          * @private
1999          */
2000         function closeList( $char ) {
2001                 if ( '*' === $char ) {
2002                         $text = '</li></ul>';
2003                 } elseif ( '#' === $char ) {
2004                         $text = '</li></ol>';
2005                 } elseif ( ':' === $char ) {
2006                         if ( $this->mDTopen ) {
2007                                 $this->mDTopen = false;
2008                                 $text = '</dt></dl>';
2009                         } else {
2010                                 $text = '</dd></dl>';
2011                         }
2012                 } else {
2013                         return '<!-- ERR 3 -->';
2014                 }
2015                 return $text."\n";
2016         }
2017         /**#@-*/
2018
2019         /**
2020          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2021          *
2022          * @param $linestart bool whether or not this is at the start of a line.
2023          * @private
2024          * @return string the lists rendered as HTML
2025          */
2026         function doBlockLevels( $text, $linestart ) {
2027                 wfProfileIn( __METHOD__ );
2028
2029                 # Parsing through the text line by line.  The main thing
2030                 # happening here is handling of block-level elements p, pre,
2031                 # and making lists from lines starting with * # : etc.
2032                 #
2033                 $textLines = StringUtils::explode( "\n", $text );
2034
2035                 $lastPrefix = $output = '';
2036                 $this->mDTopen = $inBlockElem = false;
2037                 $prefixLength = 0;
2038                 $paragraphStack = false;
2039
2040                 foreach ( $textLines as $oLine ) {
2041                         # Fix up $linestart
2042                         if ( !$linestart ) {
2043                                 $output .= $oLine;
2044                                 $linestart = true;
2045                                 continue;
2046                         }
2047                         # * = ul
2048                         # # = ol
2049                         # ; = dt
2050                         # : = dd
2051
2052                         $lastPrefixLength = strlen( $lastPrefix );
2053                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2054                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2055                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2056                         if ( !$this->mInPre ) {
2057                                 # Multiple prefixes may abut each other for nested lists.
2058                                 $prefixLength = strspn( $oLine, '*#:;' );
2059                                 $prefix = substr( $oLine, 0, $prefixLength );
2060
2061                                 # eh?
2062                                 # ; and : are both from definition-lists, so they're equivalent
2063                                 #  for the purposes of determining whether or not we need to open/close
2064                                 #  elements.
2065                                 $prefix2 = str_replace( ';', ':', $prefix );
2066                                 $t = substr( $oLine, $prefixLength );
2067                                 $this->mInPre = (bool)$preOpenMatch;
2068                         } else {
2069                                 # Don't interpret any other prefixes in preformatted text
2070                                 $prefixLength = 0;
2071                                 $prefix = $prefix2 = '';
2072                                 $t = $oLine;
2073                         }
2074
2075                         # List generation
2076                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2077                                 # Same as the last item, so no need to deal with nesting or opening stuff
2078                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2079                                 $paragraphStack = false;
2080
2081                                 if ( substr( $prefix, -1 ) === ';') {
2082                                         # The one nasty exception: definition lists work like this:
2083                                         # ; title : definition text
2084                                         # So we check for : in the remainder text to split up the
2085                                         # title and definition, without b0rking links.
2086                                         $term = $t2 = '';
2087                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2088                                                 $t = $t2;
2089                                                 $output .= $term . $this->nextItem( ':' );
2090                                         }
2091                                 }
2092                         } elseif ( $prefixLength || $lastPrefixLength ) {
2093                                 # We need to open or close prefixes, or both.
2094
2095                                 # Either open or close a level...
2096                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2097                                 $paragraphStack = false;
2098
2099                                 # Close all the prefixes which aren't shared.
2100                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2101                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2102                                         --$lastPrefixLength;
2103                                 }
2104
2105                                 # Continue the current prefix if appropriate.
2106                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2107                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2108                                 }
2109
2110                                 # Open prefixes where appropriate.
2111                                 while ( $prefixLength > $commonPrefixLength ) {
2112                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2113                                         $output .= $this->openList( $char );
2114
2115                                         if ( ';' === $char ) {
2116                                                 # FIXME: This is dupe of code above
2117                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2118                                                         $t = $t2;
2119                                                         $output .= $term . $this->nextItem( ':' );
2120                                                 }
2121                                         }
2122                                         ++$commonPrefixLength;
2123                                 }
2124                                 $lastPrefix = $prefix2;
2125                         }
2126
2127                         # If we have no prefixes, go to paragraph mode.
2128                         if ( 0 == $prefixLength ) {
2129                                 wfProfileIn( __METHOD__."-paragraph" );
2130                                 # No prefix (not in list)--go to paragraph mode
2131                                 # XXX: use a stack for nestable elements like span, table and div
2132                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2133                                 $closematch = preg_match(
2134                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2135                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2136                                 if ( $openmatch or $closematch ) {
2137                                         $paragraphStack = false;
2138                                         # TODO bug 5718: paragraph closed
2139                                         $output .= $this->closeParagraph();
2140                                         if ( $preOpenMatch and !$preCloseMatch ) {
2141                                                 $this->mInPre = true;
2142                                         }
2143                                         if ( $closematch ) {
2144                                                 $inBlockElem = false;
2145                                         } else {
2146                                                 $inBlockElem = true;
2147                                         }
2148                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2149                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2150                                                 # pre
2151                                                 if ( $this->mLastSection !== 'pre' ) {
2152                                                         $paragraphStack = false;
2153                                                         $output .= $this->closeParagraph().'<pre>';
2154                                                         $this->mLastSection = 'pre';
2155                                                 }
2156                                                 $t = substr( $t, 1 );
2157                                         } else {
2158                                                 # paragraph
2159                                                 if ( trim( $t ) == '' ) {
2160                                                         if ( $paragraphStack ) {
2161                                                                 $output .= $paragraphStack.'<br />';
2162                                                                 $paragraphStack = false;
2163                                                                 $this->mLastSection = 'p';
2164                                                         } else {
2165                                                                 if ( $this->mLastSection !== 'p' ) {
2166                                                                         $output .= $this->closeParagraph();
2167                                                                         $this->mLastSection = '';
2168                                                                         $paragraphStack = '<p>';
2169                                                                 } else {
2170                                                                         $paragraphStack = '</p><p>';
2171                                                                 }
2172                                                         }
2173                                                 } else {
2174                                                         if ( $paragraphStack ) {
2175                                                                 $output .= $paragraphStack;
2176                                                                 $paragraphStack = false;
2177                                                                 $this->mLastSection = 'p';
2178                                                         } elseif ( $this->mLastSection !== 'p' ) {
2179                                                                 $output .= $this->closeParagraph().'<p>';
2180                                                                 $this->mLastSection = 'p';
2181                                                         }
2182                                                 }
2183                                         }
2184                                 }
2185                                 wfProfileOut( __METHOD__."-paragraph" );
2186                         }
2187                         # somewhere above we forget to get out of pre block (bug 785)
2188                         if ( $preCloseMatch && $this->mInPre ) {
2189                                 $this->mInPre = false;
2190                         }
2191                         if ( $paragraphStack === false ) {
2192                                 $output .= $t."\n";
2193                         }
2194                 }
2195                 while ( $prefixLength ) {
2196                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2197                         --$prefixLength;
2198                 }
2199                 if ( $this->mLastSection != '' ) {
2200                         $output .= '</' . $this->mLastSection . '>';
2201                         $this->mLastSection = '';
2202                 }
2203
2204                 wfProfileOut( __METHOD__ );
2205                 return $output;
2206         }
2207
2208         /**
2209          * Split up a string on ':', ignoring any occurences inside tags
2210          * to prevent illegal overlapping.
2211          * @param string $str the string to split
2212          * @param string &$before set to everything before the ':'
2213          * @param string &$after set to everything after the ':'
2214          * return string the position of the ':', or false if none found
2215          */
2216         function findColonNoLinks( $str, &$before, &$after ) {
2217                 wfProfileIn( __METHOD__ );
2218
2219                 $pos = strpos( $str, ':' );
2220                 if ( $pos === false ) {
2221                         # Nothing to find!
2222                         wfProfileOut( __METHOD__ );
2223                         return false;
2224                 }
2225
2226                 $lt = strpos( $str, '<' );
2227                 if ( $lt === false || $lt > $pos ) {
2228                         # Easy; no tag nesting to worry about
2229                         $before = substr( $str, 0, $pos );
2230                         $after = substr( $str, $pos+1 );
2231                         wfProfileOut( __METHOD__ );
2232                         return $pos;
2233                 }
2234
2235                 # Ugly state machine to walk through avoiding tags.
2236                 $state = self::COLON_STATE_TEXT;
2237                 $stack = 0;
2238                 $len = strlen( $str );
2239                 for( $i = 0; $i < $len; $i++ ) {
2240                         $c = $str{$i};
2241
2242                         switch( $state ) {
2243                         # (Using the number is a performance hack for common cases)
2244                         case 0: # self::COLON_STATE_TEXT:
2245                                 switch( $c ) {
2246                                 case "<":
2247                                         # Could be either a <start> tag or an </end> tag
2248                                         $state = self::COLON_STATE_TAGSTART;
2249                                         break;
2250                                 case ":":
2251                                         if ( $stack == 0 ) {
2252                                                 # We found it!
2253                                                 $before = substr( $str, 0, $i );
2254                                                 $after = substr( $str, $i + 1 );
2255                                                 wfProfileOut( __METHOD__ );
2256                                                 return $i;
2257                                         }
2258                                         # Embedded in a tag; don't break it.
2259                                         break;
2260                                 default:
2261                                         # Skip ahead looking for something interesting
2262                                         $colon = strpos( $str, ':', $i );
2263                                         if ( $colon === false ) {
2264                                                 # Nothing else interesting
2265                                                 wfProfileOut( __METHOD__ );
2266                                                 return false;
2267                                         }
2268                                         $lt = strpos( $str, '<', $i );
2269                                         if ( $stack === 0 ) {
2270                                                 if ( $lt === false || $colon < $lt ) {
2271                                                         # We found it!
2272                                                         $before = substr( $str, 0, $colon );
2273                                                         $after = substr( $str, $colon + 1 );
2274                                                         wfProfileOut( __METHOD__ );
2275                                                         return $i;
2276                                                 }
2277                                         }
2278                                         if ( $lt === false ) {
2279                                                 # Nothing else interesting to find; abort!
2280                                                 # We're nested, but there's no close tags left. Abort!
2281                                                 break 2;
2282                                         }
2283                                         # Skip ahead to next tag start
2284                                         $i = $lt;
2285                                         $state = self::COLON_STATE_TAGSTART;
2286                                 }
2287                                 break;
2288                         case 1: # self::COLON_STATE_TAG:
2289                                 # In a <tag>
2290                                 switch( $c ) {
2291                                 case ">":
2292                                         $stack++;
2293                                         $state = self::COLON_STATE_TEXT;
2294                                         break;
2295                                 case "/":
2296                                         # Slash may be followed by >?
2297                                         $state = self::COLON_STATE_TAGSLASH;
2298                                         break;
2299                                 default:
2300                                         # ignore
2301                                 }
2302                                 break;
2303                         case 2: # self::COLON_STATE_TAGSTART:
2304                                 switch( $c ) {
2305                                 case "/":
2306                                         $state = self::COLON_STATE_CLOSETAG;
2307                                         break;
2308                                 case "!":
2309                                         $state = self::COLON_STATE_COMMENT;
2310                                         break;
2311                                 case ">":
2312                                         # Illegal early close? This shouldn't happen D:
2313                                         $state = self::COLON_STATE_TEXT;
2314                                         break;
2315                                 default:
2316                                         $state = self::COLON_STATE_TAG;
2317                                 }
2318                                 break;
2319                         case 3: # self::COLON_STATE_CLOSETAG:
2320                                 # In a </tag>
2321                                 if ( $c === ">" ) {
2322                                         $stack--;
2323                                         if ( $stack < 0 ) {
2324                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2325                                                 wfProfileOut( __METHOD__ );
2326                                                 return false;
2327                                         }
2328                                         $state = self::COLON_STATE_TEXT;
2329                                 }
2330                                 break;
2331                         case self::COLON_STATE_TAGSLASH:
2332                                 if ( $c === ">" ) {
2333                                         # Yes, a self-closed tag <blah/>
2334                                         $state = self::COLON_STATE_TEXT;
2335                                 } else {
2336                                         # Probably we're jumping the gun, and this is an attribute
2337                                         $state = self::COLON_STATE_TAG;
2338                                 }
2339                                 break;
2340                         case 5: # self::COLON_STATE_COMMENT:
2341                                 if ( $c === "-" ) {
2342                                         $state = self::COLON_STATE_COMMENTDASH;
2343                                 }
2344                                 break;
2345                         case self::COLON_STATE_COMMENTDASH:
2346                                 if ( $c === "-" ) {
2347                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2348                                 } else {
2349                                         $state = self::COLON_STATE_COMMENT;
2350                                 }
2351                                 break;
2352                         case self::COLON_STATE_COMMENTDASHDASH:
2353                                 if ( $c === ">" ) {
2354                                         $state = self::COLON_STATE_TEXT;
2355                                 } else {
2356                                         $state = self::COLON_STATE_COMMENT;
2357                                 }
2358                                 break;
2359                         default:
2360                                 throw new MWException( "State machine error in " . __METHOD__ );
2361                         }
2362                 }
2363                 if ( $stack > 0 ) {
2364                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2365                         return false;
2366                 }
2367                 wfProfileOut( __METHOD__ );
2368                 return false;
2369         }
2370
2371         /**
2372          * Return value of a magic variable (like PAGENAME)
2373          *
2374          * @private
2375          */
2376         function getVariableValue( $index, $frame=false ) {
2377                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2378                 global $wgScriptPath, $wgStylePath;
2379
2380                 /**
2381                  * Some of these require message or data lookups and can be
2382                  * expensive to check many times.
2383                  */
2384                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2385                         if ( isset( $this->mVarCache[$index] ) ) {
2386                                 return $this->mVarCache[$index];
2387                         }
2388                 }
2389
2390                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2391                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2392
2393                 # Use the time zone
2394                 global $wgLocaltimezone;
2395                 if ( isset( $wgLocaltimezone ) ) {
2396                         $oldtz = date_default_timezone_get();
2397                         date_default_timezone_set( $wgLocaltimezone );
2398                 }
2399
2400                 $localTimestamp = date( 'YmdHis', $ts );
2401                 $localMonth = date( 'm', $ts );
2402                 $localMonth1 = date( 'n', $ts );
2403                 $localMonthName = date( 'n', $ts );
2404                 $localDay = date( 'j', $ts );
2405                 $localDay2 = date( 'd', $ts );
2406                 $localDayOfWeek = date( 'w', $ts );
2407                 $localWeek = date( 'W', $ts );
2408                 $localYear = date( 'Y', $ts );
2409                 $localHour = date( 'H', $ts );
2410                 if ( isset( $wgLocaltimezone ) ) {
2411                         date_default_timezone_set( $oldtz );
2412                 }
2413
2414                 switch ( $index ) {
2415                         case 'currentmonth':
2416                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2417                                 break;
2418                         case 'currentmonth1':
2419                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2420                                 break;
2421                         case 'currentmonthname':
2422                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2423                                 break;
2424                         case 'currentmonthnamegen':
2425                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2426                                 break;
2427                         case 'currentmonthabbrev':
2428                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2429                                 break;
2430                         case 'currentday':
2431                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2432                                 break;
2433                         case 'currentday2':
2434                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2435                                 break;
2436                         case 'localmonth':
2437                                 $value = $wgContLang->formatNum( $localMonth );
2438                                 break;
2439                         case 'localmonth1':
2440                                 $value = $wgContLang->formatNum( $localMonth1 );
2441                                 break;
2442                         case 'localmonthname':
2443                                 $value = $wgContLang->getMonthName( $localMonthName );
2444                                 break;
2445                         case 'localmonthnamegen':
2446                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2447                                 break;
2448                         case 'localmonthabbrev':
2449                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2450                                 break;
2451                         case 'localday':
2452                                 $value = $wgContLang->formatNum( $localDay );
2453                                 break;
2454                         case 'localday2':
2455                                 $value = $wgContLang->formatNum( $localDay2 );
2456                                 break;
2457                         case 'pagename':
2458                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2459                                 break;
2460                         case 'pagenamee':
2461                                 $value = $this->mTitle->getPartialURL();
2462                                 break;
2463                         case 'fullpagename':
2464                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2465                                 break;
2466                         case 'fullpagenamee':
2467                                 $value = $this->mTitle->getPrefixedURL();
2468                                 break;
2469                         case 'subpagename':
2470                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2471                                 break;
2472                         case 'subpagenamee':
2473                                 $value = $this->mTitle->getSubpageUrlForm();
2474                                 break;
2475                         case 'basepagename':
2476                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2477                                 break;
2478                         case 'basepagenamee':
2479                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2480                                 break;
2481                         case 'talkpagename':
2482                                 if ( $this->mTitle->canTalk() ) {
2483                                         $talkPage = $this->mTitle->getTalkPage();
2484                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2485                                 } else {
2486                                         $value = '';
2487                                 }
2488                                 break;
2489                         case 'talkpagenamee':
2490                                 if ( $this->mTitle->canTalk() ) {
2491                                         $talkPage = $this->mTitle->getTalkPage();
2492                                         $value = $talkPage->getPrefixedUrl();
2493                                 } else {
2494                                         $value = '';
2495                                 }
2496                                 break;
2497                         case 'subjectpagename':
2498                                 $subjPage = $this->mTitle->getSubjectPage();
2499                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2500                                 break;
2501                         case 'subjectpagenamee':
2502                                 $subjPage = $this->mTitle->getSubjectPage();
2503                                 $value = $subjPage->getPrefixedUrl();
2504                                 break;
2505                         case 'revisionid':
2506                                 # Let the edit saving system know we should parse the page
2507                                 # *after* a revision ID has been assigned.
2508                                 $this->mOutput->setFlag( 'vary-revision' );
2509                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2510                                 $value = $this->mRevisionId;
2511                                 break;
2512                         case 'revisionday':
2513                                 # Let the edit saving system know we should parse the page
2514                                 # *after* a revision ID has been assigned. This is for null edits.
2515                                 $this->mOutput->setFlag( 'vary-revision' );
2516                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2517                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2518                                 break;
2519                         case 'revisionday2':
2520                                 # Let the edit saving system know we should parse the page
2521                                 # *after* a revision ID has been assigned. This is for null edits.
2522                                 $this->mOutput->setFlag( 'vary-revision' );
2523                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2524                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2525                                 break;
2526                         case 'revisionmonth':
2527                                 # Let the edit saving system know we should parse the page
2528                                 # *after* a revision ID has been assigned. This is for null edits.
2529                                 $this->mOutput->setFlag( 'vary-revision' );
2530                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2531                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2532                                 break;
2533                         case 'revisionmonth1':
2534                                 # Let the edit saving system know we should parse the page
2535                                 # *after* a revision ID has been assigned. This is for null edits.
2536                                 $this->mOutput->setFlag( 'vary-revision' );
2537                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2538                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2539                                 break;
2540                         case 'revisionyear':
2541                                 # Let the edit saving system know we should parse the page
2542                                 # *after* a revision ID has been assigned. This is for null edits.
2543                                 $this->mOutput->setFlag( 'vary-revision' );
2544                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2545                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2546                                 break;
2547                         case 'revisiontimestamp':
2548                                 # Let the edit saving system know we should parse the page
2549                                 # *after* a revision ID has been assigned. This is for null edits.
2550                                 $this->mOutput->setFlag( 'vary-revision' );
2551                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2552                                 $value = $this->getRevisionTimestamp();
2553                                 break;
2554                         case 'revisionuser':
2555                                 # Let the edit saving system know we should parse the page
2556                                 # *after* a revision ID has been assigned. This is for null edits.
2557                                 $this->mOutput->setFlag( 'vary-revision' );
2558                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2559                                 $value = $this->getRevisionUser();
2560                                 break;
2561                         case 'namespace':
2562                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2563                                 break;
2564                         case 'namespacee':
2565                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2566                                 break;
2567                         case 'talkspace':
2568                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2569                                 break;
2570                         case 'talkspacee':
2571                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2572                                 break;
2573                         case 'subjectspace':
2574                                 $value = $this->mTitle->getSubjectNsText();
2575                                 break;
2576                         case 'subjectspacee':
2577                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2578                                 break;
2579                         case 'currentdayname':
2580                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2581                                 break;
2582                         case 'currentyear':
2583                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2584                                 break;
2585                         case 'currenttime':
2586                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2587                                 break;
2588                         case 'currenthour':
2589                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2590                                 break;
2591                         case 'currentweek':
2592                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2593                                 # int to remove the padding
2594                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2595                                 break;
2596                         case 'currentdow':
2597                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2598                                 break;
2599                         case 'localdayname':
2600                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2601                                 break;
2602                         case 'localyear':
2603                                 $value = $wgContLang->formatNum( $localYear, true );
2604                                 break;
2605                         case 'localtime':
2606                                 $value = $wgContLang->time( $localTimestamp, false, false );
2607                                 break;
2608                         case 'localhour':
2609                                 $value = $wgContLang->formatNum( $localHour, true );
2610                                 break;
2611                         case 'localweek':
2612                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2613                                 # int to remove the padding
2614                                 $value = $wgContLang->formatNum( (int)$localWeek );
2615                                 break;
2616                         case 'localdow':
2617                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2618                                 break;
2619                         case 'numberofarticles':
2620                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2621                                 break;
2622                         case 'numberoffiles':
2623                                 $value = $wgContLang->formatNum( SiteStats::images() );
2624                                 break;
2625                         case 'numberofusers':
2626                                 $value = $wgContLang->formatNum( SiteStats::users() );
2627                                 break;
2628                         case 'numberofactiveusers':
2629                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2630                                 break;
2631                         case 'numberofpages':
2632                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2633                                 break;
2634                         case 'numberofadmins':
2635                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2636                                 break;
2637                         case 'numberofedits':
2638                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2639                                 break;
2640                         case 'numberofviews':
2641                                 $value = $wgContLang->formatNum( SiteStats::views() );
2642                                 break;
2643                         case 'currenttimestamp':
2644                                 $value = wfTimestamp( TS_MW, $ts );
2645                                 break;
2646                         case 'localtimestamp':
2647                                 $value = $localTimestamp;
2648                                 break;
2649                         case 'currentversion':
2650                                 $value = SpecialVersion::getVersion();
2651                                 break;
2652                         case 'sitename':
2653                                 return $wgSitename;
2654                         case 'server':
2655                                 return $wgServer;
2656                         case 'servername':
2657                                 return $wgServerName;
2658                         case 'scriptpath':
2659                                 return $wgScriptPath;
2660                         case 'stylepath':
2661                                 return $wgStylePath;
2662                         case 'directionmark':
2663                                 return $wgContLang->getDirMark();
2664                         case 'contentlanguage':
2665                                 global $wgContLanguageCode;
2666                                 return $wgContLanguageCode;
2667                         default:
2668                                 $ret = null;
2669                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2670                                         return $ret;
2671                                 } else {
2672                                         return null;
2673                                 }
2674                 }
2675
2676                 if ( $index )
2677                         $this->mVarCache[$index] = $value;
2678
2679                 return $value;
2680         }
2681
2682         /**
2683          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2684          *
2685          * @private
2686          */
2687         function initialiseVariables() {
2688                 wfProfileIn( __METHOD__ );
2689                 $variableIDs = MagicWord::getVariableIDs();
2690                 $substIDs = MagicWord::getSubstIDs();
2691
2692                 $this->mVariables = new MagicWordArray( $variableIDs );
2693                 $this->mSubstWords = new MagicWordArray( $substIDs );
2694                 wfProfileOut( __METHOD__ );
2695         }
2696
2697         /**
2698          * Preprocess some wikitext and return the document tree.
2699          * This is the ghost of replace_variables().
2700          *
2701          * @param string $text The text to parse
2702          * @param integer flags Bitwise combination of:
2703          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2704          *                                     included. Default is to assume a direct page view.
2705          *
2706          * The generated DOM tree must depend only on the input text and the flags.
2707          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2708          *
2709          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2710          * change in the DOM tree for a given text, must be passed through the section identifier
2711          * in the section edit link and thus back to extractSections().
2712          *
2713          * The output of this function is currently only cached in process memory, but a persistent
2714          * cache may be implemented at a later date which takes further advantage of these strict
2715          * dependency requirements.
2716          *
2717          * @private
2718          */
2719         function preprocessToDom( $text, $flags = 0 ) {
2720                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2721                 return $dom;
2722         }
2723
2724         /**
2725          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2726          */
2727         public static function splitWhitespace( $s ) {
2728                 $ltrimmed = ltrim( $s );
2729                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2730                 $trimmed = rtrim( $ltrimmed );
2731                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2732                 if ( $diff > 0 ) {
2733                         $w2 = substr( $ltrimmed, -$diff );
2734                 } else {
2735                         $w2 = '';
2736                 }
2737                 return array( $w1, $trimmed, $w2 );
2738         }
2739
2740         /**
2741          * Replace magic variables, templates, and template arguments
2742          * with the appropriate text. Templates are substituted recursively,
2743          * taking care to avoid infinite loops.
2744          *
2745          * Note that the substitution depends on value of $mOutputType:
2746          *  self::OT_WIKI: only {{subst:}} templates
2747          *  self::OT_PREPROCESS: templates but not extension tags
2748          *  self::OT_HTML: all templates and extension tags
2749          *
2750          * @param string $tex The text to transform
2751          * @param PPFrame $frame Object describing the arguments passed to the template.
2752          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2753          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2754          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2755          * @private
2756          */
2757         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2758                 # Is there any text? Also, Prevent too big inclusions!
2759                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2760                         return $text;
2761                 }
2762                 wfProfileIn( __METHOD__ );
2763
2764                 if ( $frame === false ) {
2765                         $frame = $this->getPreprocessor()->newFrame();
2766                 } elseif ( !( $frame instanceof PPFrame ) ) {
2767                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2768                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2769                 }
2770
2771                 $dom = $this->preprocessToDom( $text );
2772                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2773                 $text = $frame->expand( $dom, $flags );
2774
2775                 wfProfileOut( __METHOD__ );
2776                 return $text;
2777         }
2778
2779         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2780         static function createAssocArgs( $args ) {
2781                 $assocArgs = array();
2782                 $index = 1;
2783                 foreach ( $args as $arg ) {
2784                         $eqpos = strpos( $arg, '=' );
2785                         if ( $eqpos === false ) {
2786                                 $assocArgs[$index++] = $arg;
2787                         } else {
2788                                 $name = trim( substr( $arg, 0, $eqpos ) );
2789                                 $value = trim( substr( $arg, $eqpos+1 ) );
2790                                 if ( $value === false ) {
2791                                         $value = '';
2792                                 }
2793                                 if ( $name !== false ) {
2794                                         $assocArgs[$name] = $value;
2795                                 }
2796                         }
2797                 }
2798
2799                 return $assocArgs;
2800         }
2801
2802         /**
2803          * Warn the user when a parser limitation is reached
2804          * Will warn at most once the user per limitation type
2805          *
2806          * @param string $limitationType, should be one of:
2807          *   'expensive-parserfunction' (corresponding messages:
2808          *       'expensive-parserfunction-warning',
2809          *       'expensive-parserfunction-category')
2810          *   'post-expand-template-argument' (corresponding messages:
2811          *       'post-expand-template-argument-warning',
2812          *       'post-expand-template-argument-category')
2813          *   'post-expand-template-inclusion' (corresponding messages:
2814          *       'post-expand-template-inclusion-warning',
2815          *       'post-expand-template-inclusion-category')
2816          * @params int $current, $max When an explicit limit has been
2817          *       exceeded, provide the values (optional)
2818          */
2819         function limitationWarn( $limitationType, $current=null, $max=null) {
2820                 # does no harm if $current and $max are present but are unnecessary for the message
2821                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2822                 $this->mOutput->addWarning( $warning );
2823                 $this->addTrackingCategory( "$limitationType-category" );
2824         }
2825
2826         /**
2827          * Return the text of a template, after recursively
2828          * replacing any variables or templates within the template.
2829          *
2830          * @param array $piece The parts of the template
2831          *  $piece['title']: the title, i.e. the part before the |
2832          *  $piece['parts']: the parameter array
2833          *  $piece['lineStart']: whether the brace was at the start of a line
2834          * @param PPFrame The current frame, contains template arguments
2835          * @return string the text of the template
2836          * @private
2837          */
2838         function braceSubstitution( $piece, $frame ) {
2839                 global $wgContLang, $wgNonincludableNamespaces;
2840                 wfProfileIn( __METHOD__ );
2841                 wfProfileIn( __METHOD__.'-setup' );
2842
2843                 # Flags
2844                 $found = false;             # $text has been filled
2845                 $nowiki = false;            # wiki markup in $text should be escaped
2846                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2847                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2848                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2849                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2850
2851                 # Title object, where $text came from
2852                 $title = null;
2853
2854                 # $part1 is the bit before the first |, and must contain only title characters.
2855                 # Various prefixes will be stripped from it later.
2856                 $titleWithSpaces = $frame->expand( $piece['title'] );
2857                 $part1 = trim( $titleWithSpaces );
2858                 $titleText = false;
2859
2860                 # Original title text preserved for various purposes
2861                 $originalTitle = $part1;
2862
2863                 # $args is a list of argument nodes, starting from index 0, not including $part1
2864                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2865                 wfProfileOut( __METHOD__.'-setup' );
2866
2867                 # SUBST
2868                 wfProfileIn( __METHOD__.'-modifiers' );
2869                 if ( !$found ) {
2870
2871                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2872
2873                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2874                         # Decide whether to expand template or keep wikitext as-is.
2875                         if ( $this->ot['wiki'] ) {
2876                                 if ( $substMatch === false ) {
2877                                         $literal = true;  # literal when in PST with no prefix
2878                                 } else {
2879                                         $literal = false; # expand when in PST with subst: or safesubst:
2880                                 }
2881                         } else {
2882                                 if ( $substMatch == 'subst' ) {
2883                                         $literal = true;  # literal when not in PST with plain subst:
2884                                 } else {
2885                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2886                                 }
2887                         }
2888                         if ( $literal ) {
2889                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2890                                 $isLocalObj = true;
2891                                 $found = true;
2892                         }
2893                 }
2894
2895                 # Variables
2896                 if ( !$found && $args->getLength() == 0 ) {
2897                         $id = $this->mVariables->matchStartToEnd( $part1 );
2898                         if ( $id !== false ) {
2899                                 $text = $this->getVariableValue( $id, $frame );
2900                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
2901                                         $this->mOutput->mContainsOldMagic = true;
2902                                 }
2903                                 $found = true;
2904                         }
2905                 }
2906
2907                 # MSG, MSGNW and RAW
2908                 if ( !$found ) {
2909                         # Check for MSGNW:
2910                         $mwMsgnw = MagicWord::get( 'msgnw' );
2911                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2912                                 $nowiki = true;
2913                         } else {
2914                                 # Remove obsolete MSG:
2915                                 $mwMsg = MagicWord::get( 'msg' );
2916                                 $mwMsg->matchStartAndRemove( $part1 );
2917                         }
2918
2919                         # Check for RAW:
2920                         $mwRaw = MagicWord::get( 'raw' );
2921                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2922                                 $forceRawInterwiki = true;
2923                         }
2924                 }
2925                 wfProfileOut( __METHOD__.'-modifiers' );
2926
2927                 # Parser functions
2928                 if ( !$found ) {
2929                         wfProfileIn( __METHOD__ . '-pfunc' );
2930
2931                         $colonPos = strpos( $part1, ':' );
2932                         if ( $colonPos !== false ) {
2933                                 # Case sensitive functions
2934                                 $function = substr( $part1, 0, $colonPos );
2935                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2936                                         $function = $this->mFunctionSynonyms[1][$function];
2937                                 } else {
2938                                         # Case insensitive functions
2939                                         $function = $wgContLang->lc( $function );
2940                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2941                                                 $function = $this->mFunctionSynonyms[0][$function];
2942                                         } else {
2943                                                 $function = false;
2944                                         }
2945                                 }
2946                                 if ( $function ) {
2947                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2948                                         $initialArgs = array( &$this );
2949                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2950                                         if ( $flags & SFH_OBJECT_ARGS ) {
2951                                                 # Add a frame parameter, and pass the arguments as an array
2952                                                 $allArgs = $initialArgs;
2953                                                 $allArgs[] = $frame;
2954                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2955                                                         $funcArgs[] = $args->item( $i );
2956                                                 }
2957                                                 $allArgs[] = $funcArgs;
2958                                         } else {
2959                                                 # Convert arguments to plain text
2960                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2961                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2962                                                 }
2963                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2964                                         }
2965
2966                                         # Workaround for PHP bug 35229 and similar
2967                                         if ( !is_callable( $callback ) ) {
2968                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2969                                                 wfProfileOut( __METHOD__ );
2970                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2971                                         }
2972                                         $result = call_user_func_array( $callback, $allArgs );
2973                                         $found = true;
2974                                         $noparse = true;
2975                                         $preprocessFlags = 0;
2976
2977                                         if ( is_array( $result ) ) {
2978                                                 if ( isset( $result[0] ) ) {
2979                                                         $text = $result[0];
2980                                                         unset( $result[0] );
2981                                                 }
2982
2983                                                 # Extract flags into the local scope
2984                                                 # This allows callers to set flags such as nowiki, found, etc.
2985                                                 extract( $result );
2986                                         } else {
2987                                                 $text = $result;
2988                                         }
2989                                         if ( !$noparse ) {
2990                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2991                                                 $isChildObj = true;
2992                                         }
2993                                 }
2994                         }
2995                         wfProfileOut( __METHOD__ . '-pfunc' );
2996                 }
2997
2998                 # Finish mangling title and then check for loops.
2999                 # Set $title to a Title object and $titleText to the PDBK
3000                 if ( !$found ) {
3001                         $ns = NS_TEMPLATE;
3002                         # Split the title into page and subpage
3003                         $subpage = '';
3004                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3005                         if ( $subpage !== '' ) {
3006                                 $ns = $this->mTitle->getNamespace();
3007                         }
3008                         $title = Title::newFromText( $part1, $ns );
3009                         if ( $title ) {
3010                                 $titleText = $title->getPrefixedText();
3011                                 # Check for language variants if the template is not found
3012                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3013                                         $wgContLang->findVariantLink( $part1, $title, true );
3014                                 }
3015                                 # Do recursion depth check
3016                                 $limit = $this->mOptions->getMaxTemplateDepth();
3017                                 if ( $frame->depth >= $limit ) {
3018                                         $found = true;
3019                                         $text = '<span class="error">'
3020                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3021                                                 . '</span>';
3022                                 }
3023                         }
3024                 }
3025
3026                 # Load from database
3027                 if ( !$found && $title ) {
3028                         wfProfileIn( __METHOD__ . '-loadtpl' );
3029                         if ( !$title->isExternal() ) {
3030                                 if ( $title->getNamespace() == NS_SPECIAL
3031                                         && $this->mOptions->getAllowSpecialInclusion()
3032                                         && $this->ot['html'] )
3033                                 {
3034                                         $text = SpecialPage::capturePath( $title );
3035                                         if ( is_string( $text ) ) {
3036                                                 $found = true;
3037                                                 $isHTML = true;
3038                                                 $this->disableCache();
3039                                         }
3040                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3041                                         $found = false; # access denied
3042                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3043                                 } else {
3044                                         list( $text, $title ) = $this->getTemplateDom( $title );
3045                                         if ( $text !== false ) {
3046                                                 $found = true;
3047                                                 $isChildObj = true;
3048                                         }
3049                                 }
3050
3051                                 # If the title is valid but undisplayable, make a link to it
3052                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3053                                         $text = "[[:$titleText]]";
3054                                         $found = true;
3055                                 }
3056                         } elseif ( $title->isTrans() ) {
3057                                 # Interwiki transclusion
3058                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3059                                         $text = $this->interwikiTransclude( $title, 'render' );
3060                                         $isHTML = true;
3061                                 } else {
3062                                         $text = $this->interwikiTransclude( $title, 'raw' );
3063                                         # Preprocess it like a template
3064                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3065                                         $isChildObj = true;
3066                                 }
3067                                 $found = true;
3068                         }
3069
3070                         # Do infinite loop check
3071                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3072                         if ( !$frame->loopCheck( $title ) ) {
3073                                 $found = true;
3074                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3075                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3076                         }
3077                         wfProfileOut( __METHOD__ . '-loadtpl' );
3078                 }
3079
3080                 # If we haven't found text to substitute by now, we're done
3081                 # Recover the source wikitext and return it
3082                 if ( !$found ) {
3083                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3084                         wfProfileOut( __METHOD__ );
3085                         return array( 'object' => $text );
3086                 }
3087
3088                 # Expand DOM-style return values in a child frame
3089                 if ( $isChildObj ) {
3090                         # Clean up argument array
3091                         $newFrame = $frame->newChild( $args, $title );
3092
3093                         if ( $nowiki ) {
3094                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3095                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3096                                 # Expansion is eligible for the empty-frame cache
3097                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3098                                         $text = $this->mTplExpandCache[$titleText];
3099                                 } else {
3100                                         $text = $newFrame->expand( $text );
3101                                         $this->mTplExpandCache[$titleText] = $text;
3102                                 }
3103                         } else {
3104                                 # Uncached expansion
3105                                 $text = $newFrame->expand( $text );
3106                         }
3107                 }
3108                 if ( $isLocalObj && $nowiki ) {
3109                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3110                         $isLocalObj = false;
3111                 }
3112
3113                 # Replace raw HTML by a placeholder
3114                 # Add a blank line preceding, to prevent it from mucking up
3115                 # immediately preceding headings
3116                 if ( $isHTML ) {
3117                         $text = "\n\n" . $this->insertStripItem( $text );
3118                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3119                         # Escape nowiki-style return values
3120                         $text = wfEscapeWikiText( $text );
3121                 } elseif ( is_string( $text )
3122                         && !$piece['lineStart']
3123                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3124                 {
3125                         # Bug 529: if the template begins with a table or block-level
3126                         # element, it should be treated as beginning a new line.
3127                         # This behaviour is somewhat controversial.
3128                         $text = "\n" . $text;
3129                 }
3130
3131                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3132                         # Error, oversize inclusion
3133                         if ( $titleText !== false ) {
3134                                 # Make a working, properly escaped link if possible (bug 23588)
3135                                 $text = "[[:$titleText]]";
3136                         } else {
3137                                 # This will probably not be a working link, but at least it may
3138                                 # provide some hint of where the problem is
3139                                 preg_replace( '/^:/', '', $originalTitle );
3140                                 $text = "[[:$originalTitle]]";
3141                         }
3142                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3143                         $this->limitationWarn( 'post-expand-template-inclusion' );
3144                 }
3145
3146                 if ( $isLocalObj ) {
3147                         $ret = array( 'object' => $text );
3148                 } else {
3149                         $ret = array( 'text' => $text );
3150                 }
3151
3152                 wfProfileOut( __METHOD__ );
3153                 return $ret;
3154         }
3155
3156         /**
3157          * Get the semi-parsed DOM representation of a template with a given title,
3158          * and its redirect destination title. Cached.
3159          */
3160         function getTemplateDom( $title ) {
3161                 $cacheTitle = $title;
3162                 $titleText = $title->getPrefixedDBkey();
3163
3164                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3165                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3166                         $title = Title::makeTitle( $ns, $dbk );
3167                         $titleText = $title->getPrefixedDBkey();
3168                 }
3169                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3170                         return array( $this->mTplDomCache[$titleText], $title );
3171                 }
3172
3173                 # Cache miss, go to the database
3174                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3175
3176                 if ( $text === false ) {
3177                         $this->mTplDomCache[$titleText] = false;
3178                         return array( false, $title );
3179                 }
3180
3181                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3182                 $this->mTplDomCache[ $titleText ] = $dom;
3183
3184                 if ( !$title->equals( $cacheTitle ) ) {
3185                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3186                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3187                 }
3188
3189                 return array( $dom, $title );
3190         }
3191
3192         /**
3193          * Fetch the unparsed text of a template and register a reference to it.
3194          */
3195         function fetchTemplateAndTitle( $title ) {
3196                 $templateCb = $this->mOptions->getTemplateCallback();
3197                 $stuff = call_user_func( $templateCb, $title, $this );
3198                 $text = $stuff['text'];
3199                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3200                 if ( isset( $stuff['deps'] ) ) {
3201                         foreach ( $stuff['deps'] as $dep ) {
3202                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3203                         }
3204                 }
3205                 return array( $text, $finalTitle );
3206         }
3207
3208         function fetchTemplate( $title ) {
3209                 $rv = $this->fetchTemplateAndTitle( $title );
3210                 return $rv[0];
3211         }
3212
3213         /**
3214          * Static function to get a template
3215          * Can be overridden via ParserOptions::setTemplateCallback().
3216          */
3217         static function statelessFetchTemplate( $title, $parser=false ) {
3218                 $text = $skip = false;
3219                 $finalTitle = $title;
3220                 $deps = array();
3221
3222                 # Loop to fetch the article, with up to 1 redirect
3223                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3224                         # Give extensions a chance to select the revision instead
3225                         $id = false; # Assume current
3226                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3227
3228                         if ( $skip ) {
3229                                 $text = false;
3230                                 $deps[] = array(
3231                                         'title' => $title,
3232                                         'page_id' => $title->getArticleID(),
3233                                         'rev_id' => null );
3234                                 break;
3235                         }
3236                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3237                         $rev_id = $rev ? $rev->getId() : 0;
3238                         # If there is no current revision, there is no page
3239                         if ( $id === false && !$rev ) {
3240                                 $linkCache = LinkCache::singleton();
3241                                 $linkCache->addBadLinkObj( $title );
3242                         }
3243
3244                         $deps[] = array(
3245                                 'title' => $title,
3246                                 'page_id' => $title->getArticleID(),
3247                                 'rev_id' => $rev_id );
3248
3249                         if ( $rev ) {
3250                                 $text = $rev->getText();
3251                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3252                                 global $wgContLang;
3253                                 $message = $wgContLang->lcfirst( $title->getText() );
3254                                 $text = wfMsgForContentNoTrans( $message );
3255                                 if ( wfEmptyMsg( $message, $text ) ) {
3256                                         $text = false;
3257                                         break;
3258                                 }
3259                         } else {
3260                                 break;
3261                         }
3262                         if ( $text === false ) {
3263                                 break;
3264                         }
3265                         # Redirect?
3266                         $finalTitle = $title;
3267                         $title = Title::newFromRedirect( $text );
3268                 }
3269                 return array(
3270                         'text' => $text,
3271                         'finalTitle' => $finalTitle,
3272                         'deps' => $deps );
3273         }
3274
3275         /**
3276          * Transclude an interwiki link.
3277          */
3278         function interwikiTransclude( $title, $action ) {
3279                 global $wgEnableScaryTranscluding;
3280
3281                 if ( !$wgEnableScaryTranscluding ) {
3282                         return wfMsg('scarytranscludedisabled');
3283                 }
3284
3285                 $url = $title->getFullUrl( "action=$action" );
3286
3287                 if ( strlen( $url ) > 255 ) {
3288                         return wfMsg( 'scarytranscludetoolong' );
3289                 }
3290                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3291         }
3292
3293         function fetchScaryTemplateMaybeFromCache( $url ) {
3294                 global $wgTranscludeCacheExpiry;
3295                 $dbr = wfGetDB( DB_SLAVE );
3296                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3297                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3298                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3299                 if ( $obj ) {
3300                         return $obj->tc_contents;
3301                 }
3302
3303                 $text = Http::get( $url );
3304                 if ( !$text ) {
3305                         return wfMsg( 'scarytranscludefailed', $url );
3306                 }
3307
3308                 $dbw = wfGetDB( DB_MASTER );
3309                 $dbw->replace( 'transcache', array('tc_url'), array(
3310                         'tc_url' => $url,
3311                         'tc_time' => $dbw->timestamp( time() ),
3312                         'tc_contents' => $text)
3313                 );
3314                 return $text;
3315         }
3316
3317
3318         /**
3319          * Triple brace replacement -- used for template arguments
3320          * @private
3321          */
3322         function argSubstitution( $piece, $frame ) {
3323                 wfProfileIn( __METHOD__ );
3324
3325                 $error = false;
3326                 $parts = $piece['parts'];
3327                 $nameWithSpaces = $frame->expand( $piece['title'] );
3328                 $argName = trim( $nameWithSpaces );
3329                 $object = false;
3330                 $text = $frame->getArgument( $argName );
3331                 if (  $text === false && $parts->getLength() > 0
3332                   && (
3333                     $this->ot['html']
3334                     || $this->ot['pre']
3335                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3336                   )
3337                 ) {
3338                         # No match in frame, use the supplied default
3339                         $object = $parts->item( 0 )->getChildren();
3340                 }
3341                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3342                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3343                         $this->limitationWarn( 'post-expand-template-argument' );
3344                 }
3345
3346                 if ( $text === false && $object === false ) {
3347                         # No match anywhere
3348                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3349                 }
3350                 if ( $error !== false ) {
3351                         $text .= $error;
3352                 }
3353                 if ( $object !== false ) {
3354                         $ret = array( 'object' => $object );
3355                 } else {
3356                         $ret = array( 'text' => $text );
3357                 }
3358
3359                 wfProfileOut( __METHOD__ );
3360                 return $ret;
3361         }
3362
3363         /**
3364          * Return the text to be used for a given extension tag.
3365          * This is the ghost of strip().
3366          *
3367          * @param array $params Associative array of parameters:
3368          *     name       PPNode for the tag name
3369          *     attr       PPNode for unparsed text where tag attributes are thought to be
3370          *     attributes Optional associative array of parsed attributes
3371          *     inner      Contents of extension element
3372          *     noClose    Original text did not have a close tag
3373          * @param PPFrame $frame
3374          */
3375         function extensionSubstitution( $params, $frame ) {
3376                 global $wgRawHtml, $wgContLang;
3377
3378                 $name = $frame->expand( $params['name'] );
3379                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3380                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3381                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3382
3383                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3384                         ( $this->ot['html'] || $this->ot['pre'] );
3385                 if ( $isFunctionTag ) {
3386                         $markerType = 'none';
3387                 } else {
3388                         $markerType = 'general';
3389                 }
3390                 if ( $this->ot['html'] || $isFunctionTag ) {
3391                         $name = strtolower( $name );
3392                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3393                         if ( isset( $params['attributes'] ) ) {
3394                                 $attributes = $attributes + $params['attributes'];
3395                         }
3396
3397                         if ( isset( $this->mTagHooks[$name] ) ) {
3398                                 # Workaround for PHP bug 35229 and similar
3399                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3400                                         throw new MWException( "Tag hook for $name is not callable\n" );
3401                                 }
3402                                 $output = call_user_func_array( $this->mTagHooks[$name],
3403                                         array( $content, $attributes, $this, $frame ) );
3404                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3405                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3406                                 if ( !is_callable( $callback ) ) {
3407                                         throw new MWException( "Tag hook for $name is not callable\n" );
3408                                 }
3409
3410                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3411                         } else {
3412                                 $output = '<span class="error">Invalid tag extension name: ' .
3413                                         htmlspecialchars( $name ) . '</span>';
3414                         }
3415
3416                         if ( is_array( $output ) ) {
3417                                 # Extract flags to local scope (to override $markerType)
3418                                 $flags = $output;
3419                                 $output = $flags[0];
3420                                 unset( $flags[0] );
3421                                 extract( $flags );
3422                         }
3423                 } else {
3424                         if ( is_null( $attrText ) ) {
3425                                 $attrText = '';
3426                         }
3427                         if ( isset( $params['attributes'] ) ) {
3428                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3429                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3430                                                 htmlspecialchars( $attrValue ) . '"';
3431                                 }
3432                         }
3433                         if ( $content === null ) {
3434                                 $output = "<$name$attrText/>";
3435                         } else {
3436                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3437                                 $output = "<$name$attrText>$content$close";
3438                         }
3439                 }
3440
3441                 if ( $markerType === 'none' ) {
3442                         return $output;
3443                 } elseif ( $markerType === 'nowiki' ) {
3444                         $this->mStripState->nowiki->setPair( $marker, $output );
3445                 } elseif ( $markerType === 'general' ) {
3446                         $this->mStripState->general->setPair( $marker, $output );
3447                 } else {
3448                         throw new MWException( __METHOD__.': invalid marker type' );
3449                 }
3450                 return $marker;
3451         }
3452
3453         /**
3454          * Increment an include size counter
3455          *
3456          * @param string $type The type of expansion
3457          * @param integer $size The size of the text
3458          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3459          */
3460         function incrementIncludeSize( $type, $size ) {
3461                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3462                         return false;
3463                 } else {
3464                         $this->mIncludeSizes[$type] += $size;
3465                         return true;
3466                 }
3467         }
3468
3469         /**
3470          * Increment the expensive function count
3471          *
3472          * @return boolean False if the limit has been exceeded
3473          */
3474         function incrementExpensiveFunctionCount() {
3475                 global $wgExpensiveParserFunctionLimit;
3476                 $this->mExpensiveFunctionCount++;
3477                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3478                         return true;
3479                 }
3480                 return false;
3481         }
3482
3483         /**
3484          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3485          * Fills $this->mDoubleUnderscores, returns the modified text
3486          */
3487         function doDoubleUnderscore( $text ) {
3488                 wfProfileIn( __METHOD__ );
3489
3490                 # The position of __TOC__ needs to be recorded
3491                 $mw = MagicWord::get( 'toc' );
3492                 if ( $mw->match( $text ) ) {
3493                         $this->mShowToc = true;
3494                         $this->mForceTocPosition = true;
3495
3496                         # Set a placeholder. At the end we'll fill it in with the TOC.
3497                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3498
3499                         # Only keep the first one.
3500                         $text = $mw->replace( '', $text );
3501                 }
3502
3503                 # Now match and remove the rest of them
3504                 $mwa = MagicWord::getDoubleUnderscoreArray();
3505                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3506
3507                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3508                         $this->mOutput->mNoGallery = true;
3509                 }
3510                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3511                         $this->mShowToc = false;
3512                 }
3513                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3514                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3515                         $this->addTrackingCategory( 'hidden-category-category' );
3516                 }
3517                 # (bug 8068) Allow control over whether robots index a page.
3518                 #
3519                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3520                 # is not desirable, the last one on the page should win.
3521                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3522                         $this->mOutput->setIndexPolicy( 'noindex' );
3523                         $this->addTrackingCategory( 'noindex-category' );
3524                 }
3525                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3526                         $this->mOutput->setIndexPolicy( 'index' );
3527                         $this->addTrackingCategory( 'index-category' );
3528                 }
3529
3530                 wfProfileOut( __METHOD__ );
3531                 return $text;
3532         }
3533
3534         /**
3535          * Add a tracking category, getting the title from a system message,
3536          * or print a debug message if the title is invalid.
3537          * @param $msg String message key
3538          * @return Bool whether the addition was successful
3539          */
3540         protected function addTrackingCategory( $msg ) {
3541                 $cat = wfMsgForContent( $msg );
3542
3543                 # Allow tracking categories to be disabled by setting them to "-"
3544                 if ( $cat === '-' ) {
3545                         return false;
3546                 }
3547
3548                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3549                 if ( $containerCategory ) {
3550                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3551                         return true;
3552                 } else {
3553                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3554                         return false;
3555                 }
3556         }
3557
3558         /**
3559          * This function accomplishes several tasks:
3560          * 1) Auto-number headings if that option is enabled
3561          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3562          * 3) Add a Table of contents on the top for users who have enabled the option
3563          * 4) Auto-anchor headings
3564          *
3565          * It loops through all headlines, collects the necessary data, then splits up the
3566          * string and re-inserts the newly formatted headlines.
3567          *
3568          * @param string $text
3569          * @param string $origText Original, untouched wikitext
3570          * @param boolean $isMain
3571          * @private
3572          */
3573         function formatHeadings( $text, $origText, $isMain=true ) {
3574                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3575
3576                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3577                 $showEditLink = $this->mOptions->getEditSection();
3578
3579                 # Do not call quickUserCan unless necessary
3580                 if ( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3581                         $showEditLink = 0;
3582                 }
3583
3584                 # Inhibit editsection links if requested in the page
3585                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3586                         $showEditLink = 0;
3587                 }
3588
3589                 # Get all headlines for numbering them and adding funky stuff like [edit]
3590                 # links - this is for later, but we need the number of headlines right now
3591                 $matches = array();
3592                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3593
3594                 # if there are fewer than 4 headlines in the article, do not show TOC
3595                 # unless it's been explicitly enabled.
3596                 $enoughToc = $this->mShowToc &&
3597                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3598
3599                 # Allow user to stipulate that a page should have a "new section"
3600                 # link added via __NEWSECTIONLINK__
3601                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3602                         $this->mOutput->setNewSection( true );
3603                 }
3604
3605                 # Allow user to remove the "new section"
3606                 # link via __NONEWSECTIONLINK__
3607                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3608                         $this->mOutput->hideNewSection( true );
3609                 }
3610
3611                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3612                 # override above conditions and always show TOC above first header
3613                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3614                         $this->mShowToc = true;
3615                         $enoughToc = true;
3616                 }
3617
3618                 # We need this to perform operations on the HTML
3619                 $sk = $this->mOptions->getSkin();
3620
3621                 # headline counter
3622                 $headlineCount = 0;
3623                 $numVisible = 0;
3624
3625                 # Ugh .. the TOC should have neat indentation levels which can be
3626                 # passed to the skin functions. These are determined here
3627                 $toc = '';
3628                 $full = '';
3629                 $head = array();
3630                 $sublevelCount = array();
3631                 $levelCount = array();
3632                 $toclevel = 0;
3633                 $level = 0;
3634                 $prevlevel = 0;
3635                 $toclevel = 0;
3636                 $prevtoclevel = 0;
3637                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3638                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3639                 $oldType = $this->mOutputType;
3640                 $this->setOutputType( self::OT_WIKI );
3641                 $frame = $this->getPreprocessor()->newFrame();
3642                 $root = $this->preprocessToDom( $origText );
3643                 $node = $root->getFirstChild();
3644                 $byteOffset = 0;
3645                 $tocraw = array();
3646
3647                 foreach ( $matches[3] as $headline ) {
3648                         $isTemplate = false;
3649                         $titleText = false;
3650                         $sectionIndex = false;
3651                         $numbering = '';
3652                         $markerMatches = array();
3653                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3654                                 $serial = $markerMatches[1];
3655                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3656                                 $isTemplate = ( $titleText != $baseTitleText );
3657                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3658                         }
3659
3660                         if ( $toclevel ) {
3661                                 $prevlevel = $level;
3662                                 $prevtoclevel = $toclevel;
3663                         }
3664                         $level = $matches[1][$headlineCount];
3665
3666                         if ( $level > $prevlevel ) {
3667                                 # Increase TOC level
3668                                 $toclevel++;
3669                                 $sublevelCount[$toclevel] = 0;
3670                                 if ( $toclevel<$wgMaxTocLevel ) {
3671                                         $prevtoclevel = $toclevel;
3672                                         $toc .= $sk->tocIndent();
3673                                         $numVisible++;
3674                                 }
3675                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3676                                 # Decrease TOC level, find level to jump to
3677
3678                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3679                                         if ( $levelCount[$i] == $level ) {
3680                                                 # Found last matching level
3681                                                 $toclevel = $i;
3682                                                 break;
3683                                         } elseif ( $levelCount[$i] < $level ) {
3684                                                 # Found first matching level below current level
3685                                                 $toclevel = $i + 1;
3686                                                 break;
3687                                         }
3688                                 }
3689                                 if ( $i == 0 ) {
3690                                         $toclevel = 1;
3691                                 }
3692                                 if ( $toclevel<$wgMaxTocLevel ) {
3693                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3694                                                 # Unindent only if the previous toc level was shown :p
3695                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3696                                                 $prevtoclevel = $toclevel;
3697                                         } else {
3698                                                 $toc .= $sk->tocLineEnd();
3699                                         }
3700                                 }
3701                         } else {
3702                                 # No change in level, end TOC line
3703                                 if ( $toclevel<$wgMaxTocLevel ) {
3704                                         $toc .= $sk->tocLineEnd();
3705                                 }
3706                         }
3707
3708                         $levelCount[$toclevel] = $level;
3709
3710                         # count number of headlines for each level
3711                         @$sublevelCount[$toclevel]++;
3712                         $dot = 0;
3713                         for( $i = 1; $i <= $toclevel; $i++ ) {
3714                                 if ( !empty( $sublevelCount[$i] ) ) {
3715                                         if ( $dot ) {
3716                                                 $numbering .= '.';
3717                                         }
3718                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3719                                         $dot = 1;
3720                                 }
3721                         }
3722
3723                         # The safe header is a version of the header text safe to use for links
3724                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3725                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3726
3727                         # Remove link placeholders by the link text.
3728                         #     <!--LINK number-->
3729                         # turns into
3730                         #     link text with suffix
3731                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3732
3733                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3734                         $tocline = preg_replace(
3735                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3736                                 array( '',                          '<$1>' ),
3737                                 $safeHeadline
3738                         );
3739                         $tocline = trim( $tocline );
3740
3741                         # For the anchor, strip out HTML-y stuff period
3742                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3743                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3744                         $safeHeadline = trim( $safeHeadline );
3745
3746                         # Save headline for section edit hint before it's escaped
3747                         $headlineHint = $safeHeadline;
3748
3749                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3750                                 # For reverse compatibility, provide an id that's
3751                                 # HTML4-compatible, like we used to.
3752                                 #
3753                                 # It may be worth noting, academically, that it's possible for
3754                                 # the legacy anchor to conflict with a non-legacy headline
3755                                 # anchor on the page.  In this case likely the "correct" thing
3756                                 # would be to either drop the legacy anchors or make sure
3757                                 # they're numbered first.  However, this would require people
3758                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3759                                 # manually, so let's not bother worrying about it.
3760                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3761                                         array( 'noninitial', 'legacy' ) );
3762                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3763
3764                                 if ( $legacyHeadline == $safeHeadline ) {
3765                                         # No reason to have both (in fact, we can't)
3766                                         $legacyHeadline = false;
3767                                 }
3768                         } else {
3769                                 $legacyHeadline = false;
3770                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3771                                         'noninitial' );
3772                         }
3773
3774                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3775                         # Does this apply to Unicode characters?  Because we aren't
3776                         # handling those here.
3777                         $arrayKey = strtolower( $safeHeadline );
3778                         if ( $legacyHeadline === false ) {
3779                                 $legacyArrayKey = false;
3780                         } else {
3781                                 $legacyArrayKey = strtolower( $legacyHeadline );
3782                         }
3783
3784                         # count how many in assoc. array so we can track dupes in anchors
3785                         if ( isset( $refers[$arrayKey] ) ) {
3786                                 $refers[$arrayKey]++;
3787                         } else {
3788                                 $refers[$arrayKey] = 1;
3789                         }
3790                         if ( isset( $refers[$legacyArrayKey] ) ) {
3791                                 $refers[$legacyArrayKey]++;
3792                         } else {
3793                                 $refers[$legacyArrayKey] = 1;
3794                         }
3795
3796                         # Don't number the heading if it is the only one (looks silly)
3797                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3798                                 # the two are different if the line contains a link
3799                                 $headline = $numbering . ' ' . $headline;
3800                         }
3801
3802                         # Create the anchor for linking from the TOC to the section
3803                         $anchor = $safeHeadline;
3804                         $legacyAnchor = $legacyHeadline;
3805                         if ( $refers[$arrayKey] > 1 ) {
3806                                 $anchor .= '_' . $refers[$arrayKey];
3807                         }
3808                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3809                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3810                         }
3811                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3812                                 $toc .= $sk->tocLine( $anchor, $tocline,
3813                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3814                         }
3815
3816                         # Add the section to the section tree
3817                         # Find the DOM node for this header
3818                         while ( $node && !$isTemplate ) {
3819                                 if ( $node->getName() === 'h' ) {
3820                                         $bits = $node->splitHeading();
3821                                         if ( $bits['i'] == $sectionIndex )
3822                                                 break;
3823                                 }
3824                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3825                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3826                                 $node = $node->getNextSibling();
3827                         }
3828                         $tocraw[] = array(
3829                                 'toclevel' => $toclevel,
3830                                 'level' => $level,
3831                                 'line' => $tocline,
3832                                 'number' => $numbering,
3833                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3834                                 'fromtitle' => $titleText,
3835                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3836                                 'anchor' => $anchor,
3837                         );
3838
3839                         # give headline the correct <h#> tag
3840                         if ( $showEditLink && $sectionIndex !== false ) {
3841                                 if ( $isTemplate ) {
3842                                         # Put a T flag in the section identifier, to indicate to extractSections()
3843                                         # that sections inside <includeonly> should be counted.
3844                                         $editlink = $sk->doEditSectionLink( Title::newFromText( $titleText ), "T-$sectionIndex" );
3845                                 } else {
3846                                         $editlink = $sk->doEditSectionLink( $this->mTitle, $sectionIndex, $headlineHint );
3847                                 }
3848                         } else {
3849                                 $editlink = '';
3850                         }
3851                         $head[$headlineCount] = $sk->makeHeadline( $level,
3852                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3853                                 $editlink, $legacyAnchor );
3854
3855                         $headlineCount++;
3856                 }
3857
3858                 $this->setOutputType( $oldType );
3859
3860                 # Never ever show TOC if no headers
3861                 if ( $numVisible < 1 ) {
3862                         $enoughToc = false;
3863                 }
3864
3865                 if ( $enoughToc ) {
3866                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3867                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3868                         }
3869                         $toc = $sk->tocList( $toc );
3870                         $this->mOutput->setTOCHTML( $toc );
3871                 }
3872
3873                 if ( $isMain ) {
3874                         $this->mOutput->setSections( $tocraw );
3875                 }
3876
3877                 # split up and insert constructed headlines
3878
3879                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3880                 $i = 0;
3881
3882                 foreach ( $blocks as $block ) {
3883                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3884                                 # This is the [edit] link that appears for the top block of text when
3885                                 # section editing is enabled
3886
3887                                 # Disabled because it broke block formatting
3888                                 # For example, a bullet point in the top line
3889                                 # $full .= $sk->editSectionLink(0);
3890                         }
3891                         $full .= $block;
3892                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3893                                 # Top anchor now in skin
3894                                 $full = $full.$toc;
3895                         }
3896
3897                         if ( !empty( $head[$i] ) ) {
3898                                 $full .= $head[$i];
3899                         }
3900                         $i++;
3901                 }
3902                 if ( $this->mForceTocPosition ) {
3903                         return str_replace( '<!--MWTOC-->', $toc, $full );
3904                 } else {
3905                         return $full;
3906                 }
3907         }
3908
3909         /**
3910          * Merge $tree2 into $tree1 by replacing the section with index
3911          * $section in $tree1 and its descendants with the sections in $tree2.
3912          * Note that in the returned section tree, only the 'index' and
3913          * 'byteoffset' fields are guaranteed to be correct.
3914          * @param $tree1 array Section tree from ParserOutput::getSectons()
3915         *  @param $tree2 array Section tree
3916          * @param $section int Section index
3917          * @param $title Title Title both section trees come from
3918          * @param $len2 int Length of the original wikitext for $tree2
3919          * @return array Merged section tree
3920          */
3921         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3922                 global $wgContLang;
3923                 $newTree = array();
3924                 $targetLevel = false;
3925                 $merged = false;
3926                 $lastLevel = 1;
3927                 $nextIndex = 1;
3928                 $numbering = array( 0 );
3929                 $titletext = $title->getPrefixedDBkey();
3930                 foreach ( $tree1 as $s ) {
3931                         if ( $targetLevel !== false ) {
3932                                 if ( $s['level'] <= $targetLevel ) {
3933                                         # We've skipped enough
3934                                         $targetLevel = false;
3935                                 } else {
3936                                         continue;
3937                                 }
3938                         }
3939                         if ( $s['index'] != $section ||
3940                                         $s['fromtitle'] != $titletext ) {
3941                                 self::incrementNumbering( $numbering,
3942                                         $s['toclevel'], $lastLevel );
3943
3944                                 # Rewrite index, byteoffset and number
3945                                 if ( $s['fromtitle'] == $titletext ) {
3946                                         $s['index'] = $nextIndex++;
3947                                         if ( $merged ) {
3948                                                 $s['byteoffset'] += $len2;
3949                                         }
3950                                 }
3951                                 $s['number']  = implode( '.', array_map(
3952                                         array( $wgContLang, 'formatnum' ),
3953                                         $numbering ) );
3954                                 $lastLevel = $s['toclevel'];
3955                                 $newTree[] = $s;
3956                         } else {
3957                                 # We're at $section
3958                                 # Insert sections from $tree2 here
3959                                 foreach ( $tree2 as $s2 ) {
3960                                         # Rewrite the fields in $s2
3961                                         # before inserting it
3962                                         $s2['toclevel'] += $s['toclevel'] - 1;
3963                                         $s2['level'] += $s['level'] - 1;
3964                                         $s2['index'] = $nextIndex++;
3965                                         $s2['byteoffset'] += $s['byteoffset'];
3966
3967                                         self::incrementNumbering( $numbering,
3968                                                 $s2['toclevel'], $lastLevel );
3969                                         $s2['number']  = implode( '.', array_map(
3970                                                 array( $wgContLang, 'formatnum' ),
3971                                                 $numbering ) );
3972                                         $lastLevel = $s2['toclevel'];
3973                                         $newTree[] = $s2;
3974                                 }
3975                                 # Skip all descendants of $section in $tree1
3976                                 $targetLevel = $s['level'];
3977                                 $merged = true;
3978                         }
3979                 }
3980                 return $newTree;
3981         }
3982
3983         /**
3984          * Increment a section number. Helper function for mergeSectionTrees()
3985          * @param $number array Array representing a section number
3986          * @param $level int Current TOC level (depth)
3987          * @param $lastLevel int Level of previous TOC entry
3988          */
3989         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3990                 if ( $level > $lastLevel ) {
3991                         $number[$level - 1] = 1;
3992                 } elseif ( $level < $lastLevel ) {
3993                         foreach ( $number as $key => $unused )
3994                                 if ( $key >= $level ) {
3995                                         unset( $number[$key] );
3996                                 }
3997                         $number[$level - 1]++;
3998                 } else {
3999                         $number[$level - 1]++;
4000                 }
4001         }
4002
4003         /**
4004          * Transform wiki markup when saving a page by doing \r\n -> \n
4005          * conversion, substitting signatures, {{subst:}} templates, etc.
4006          *
4007          * @param string $text the text to transform
4008          * @param Title &$title the Title object for the current article
4009          * @param User $user the User object describing the current user
4010          * @param ParserOptions $options parsing options
4011          * @param bool $clearState whether to clear the parser state first
4012          * @return string the altered wiki markup
4013          * @public
4014          */
4015         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
4016                 $this->mOptions = $options;
4017                 $this->setTitle( $title );
4018                 $this->setOutputType( self::OT_WIKI );
4019
4020                 if ( $clearState ) {
4021                         $this->clearState();
4022                 }
4023
4024                 $pairs = array(
4025                         "\r\n" => "\n",
4026                 );
4027                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4028                 $text = $this->pstPass2( $text, $user );
4029                 $text = $this->mStripState->unstripBoth( $text );
4030                 return $text;
4031         }
4032
4033         /**
4034          * Pre-save transform helper function
4035          * @private
4036          */
4037         function pstPass2( $text, $user ) {
4038                 global $wgContLang, $wgLocaltimezone;
4039
4040                 # Note: This is the timestamp saved as hardcoded wikitext to
4041                 # the database, we use $wgContLang here in order to give
4042                 # everyone the same signature and use the default one rather
4043                 # than the one selected in each user's preferences.
4044                 # (see also bug 12815)
4045                 $ts = $this->mOptions->getTimestamp();
4046                 if ( isset( $wgLocaltimezone ) ) {
4047                         $tz = $wgLocaltimezone;
4048                 } else {
4049                         $tz = date_default_timezone_get();
4050                 }
4051
4052                 $unixts = wfTimestamp( TS_UNIX, $ts );
4053                 $oldtz = date_default_timezone_get();
4054                 date_default_timezone_set( $tz );
4055                 $ts = date( 'YmdHis', $unixts );
4056                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4057
4058                 # Allow translation of timezones trough wiki. date() can return
4059                 # whatever crap the system uses, localised or not, so we cannot
4060                 # ship premade translations.
4061                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4062                 $value = wfMsgForContent( $key );
4063                 if ( !wfEmptyMsg( $key, $value ) ) {
4064                         $tzMsg = $value;
4065                 }
4066
4067                 date_default_timezone_set( $oldtz );
4068
4069                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4070
4071                 # Variable replacement
4072                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4073                 $text = $this->replaceVariables( $text );
4074
4075                 # Signatures
4076                 $sigText = $this->getUserSig( $user );
4077                 $text = strtr( $text, array(
4078                         '~~~~~' => $d,
4079                         '~~~~' => "$sigText $d",
4080                         '~~~' => $sigText
4081                 ) );
4082
4083                 # Context links: [[|name]] and [[name (context)|]]
4084                 global $wgLegalTitleChars;
4085                 $tc = "[$wgLegalTitleChars]";
4086                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4087
4088                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4089                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4090                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4091                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4092
4093                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4094                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4095                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4096                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4097
4098                 $t = $this->mTitle->getText();
4099                 $m = array();
4100                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4101                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4102                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4103                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4104                 } else {
4105                         # if there's no context, don't bother duplicating the title
4106                         $text = preg_replace( $p2, '[[\\1]]', $text );
4107                 }
4108
4109                 # Trim trailing whitespace
4110                 $text = rtrim( $text );
4111
4112                 return $text;
4113         }
4114
4115         /**
4116          * Fetch the user's signature text, if any, and normalize to
4117          * validated, ready-to-insert wikitext.
4118          * If you have pre-fetched the nickname or the fancySig option, you can
4119          * specify them here to save a database query.
4120          *
4121          * @param User $user
4122          * @return string
4123          */
4124         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4125                 global $wgMaxSigChars;
4126
4127                 $username = $user->getName();
4128
4129                 # If not given, retrieve from the user object.
4130                 if ( $nickname === false )
4131                         $nickname = $user->getOption( 'nickname' );
4132
4133                 if ( is_null( $fancySig ) ) {
4134                         $fancySig = $user->getBoolOption( 'fancysig' );
4135                 }
4136
4137                 $nickname = $nickname == null ? $username : $nickname;
4138
4139                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4140                         $nickname = $username;
4141                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4142                 } elseif ( $fancySig !== false ) {
4143                         # Sig. might contain markup; validate this
4144                         if ( $this->validateSig( $nickname ) !== false ) {
4145                                 # Validated; clean up (if needed) and return it
4146                                 return $this->cleanSig( $nickname, true );
4147                         } else {
4148                                 # Failed to validate; fall back to the default
4149                                 $nickname = $username;
4150                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4151                         }
4152                 }
4153
4154                 # Make sure nickname doesnt get a sig in a sig
4155                 $nickname = $this->cleanSigInSig( $nickname );
4156
4157                 # If we're still here, make it a link to the user page
4158                 $userText = wfEscapeWikiText( $username );
4159                 $nickText = wfEscapeWikiText( $nickname );
4160                 if ( $user->isAnon() )  {
4161                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4162                 } else {
4163                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4164                 }
4165         }
4166
4167         /**
4168          * Check that the user's signature contains no bad XML
4169          *
4170          * @param string $text
4171          * @return mixed An expanded string, or false if invalid.
4172          */
4173         function validateSig( $text ) {
4174                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4175         }
4176
4177         /**
4178          * Clean up signature text
4179          *
4180          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4181          * 2) Substitute all transclusions
4182          *
4183          * @param string $text
4184          * @param $parsing Whether we're cleaning (preferences save) or parsing
4185          * @return string Signature text
4186          */
4187         function cleanSig( $text, $parsing = false ) {
4188                 if ( !$parsing ) {
4189                         global $wgTitle;
4190                         $this->clearState();
4191                         $this->setTitle( $wgTitle );
4192                         $this->mOptions = new ParserOptions;
4193                         $this->setOutputType = self::OT_PREPROCESS;
4194                 }
4195
4196                 # Option to disable this feature
4197                 if ( !$this->mOptions->getCleanSignatures() ) {
4198                         return $text;
4199                 }
4200
4201                 # FIXME: regex doesn't respect extension tags or nowiki
4202                 #  => Move this logic to braceSubstitution()
4203                 $substWord = MagicWord::get( 'subst' );
4204                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4205                 $substText = '{{' . $substWord->getSynonym( 0 );
4206
4207                 $text = preg_replace( $substRegex, $substText, $text );
4208                 $text = $this->cleanSigInSig( $text );
4209                 $dom = $this->preprocessToDom( $text );
4210                 $frame = $this->getPreprocessor()->newFrame();
4211                 $text = $frame->expand( $dom );
4212
4213                 if ( !$parsing ) {
4214                         $text = $this->mStripState->unstripBoth( $text );
4215                 }
4216
4217                 return $text;
4218         }
4219
4220         /**
4221          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4222          * @param string $text
4223          * @return string Signature text with /~{3,5}/ removed
4224          */
4225         function cleanSigInSig( $text ) {
4226                 $text = preg_replace( '/~{3,5}/', '', $text );
4227                 return $text;
4228         }
4229
4230         /**
4231          * Set up some variables which are usually set up in parse()
4232          * so that an external function can call some class members with confidence
4233          * @public
4234          */
4235         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4236                 $this->setTitle( $title );
4237                 $this->mOptions = $options;
4238                 $this->setOutputType( $outputType );
4239                 if ( $clearState ) {
4240                         $this->clearState();
4241                 }
4242         }
4243
4244         /**
4245          * Wrapper for preprocess()
4246          *
4247          * @param string $text the text to preprocess
4248          * @param ParserOptions $options  options
4249          * @return string
4250          * @public
4251          */
4252         function transformMsg( $text, $options ) {
4253                 global $wgTitle;
4254                 static $executing = false;
4255
4256                 # Guard against infinite recursion
4257                 if ( $executing ) {
4258                         return $text;
4259                 }
4260                 $executing = true;
4261
4262                 wfProfileIn( __METHOD__ );
4263                 $text = $this->preprocess( $text, $wgTitle, $options );
4264
4265                 $executing = false;
4266                 wfProfileOut( __METHOD__ );
4267                 return $text;
4268         }
4269
4270         /**
4271          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4272          * The callback should have the following form:
4273          *    function myParserHook( $text, $params, $parser ) { ... }
4274          *
4275          * Transform and return $text. Use $parser for any required context, e.g. use
4276          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4277          *
4278          * @public
4279          *
4280          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4281          * @param mixed $callback The callback function (and object) to use for the tag
4282          *
4283          * @return The old value of the mTagHooks array associated with the hook
4284          */
4285         function setHook( $tag, $callback ) {
4286                 $tag = strtolower( $tag );
4287                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4288                 $this->mTagHooks[$tag] = $callback;
4289                 if ( !in_array( $tag, $this->mStripList ) ) {
4290                         $this->mStripList[] = $tag;
4291                 }
4292
4293                 return $oldVal;
4294         }
4295
4296         function setTransparentTagHook( $tag, $callback ) {
4297                 $tag = strtolower( $tag );
4298                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4299                 $this->mTransparentTagHooks[$tag] = $callback;
4300
4301                 return $oldVal;
4302         }
4303
4304         /**
4305          * Remove all tag hooks
4306          */
4307         function clearTagHooks() {
4308                 $this->mTagHooks = array();
4309                 $this->mStripList = $this->mDefaultStripList;
4310         }
4311
4312         /**
4313          * Create a function, e.g. {{sum:1|2|3}}
4314          * The callback function should have the form:
4315          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4316          *
4317          * Or with SFH_OBJECT_ARGS:
4318          *    function myParserFunction( $parser, $frame, $args ) { ... }
4319          *
4320          * The callback may either return the text result of the function, or an array with the text
4321          * in element 0, and a number of flags in the other elements. The names of the flags are
4322          * specified in the keys. Valid flags are:
4323          *   found                     The text returned is valid, stop processing the template. This
4324          *                             is on by default.
4325          *   nowiki                    Wiki markup in the return value should be escaped
4326          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4327          *
4328          * @public
4329          *
4330          * @param string $id The magic word ID
4331          * @param mixed $callback The callback function (and object) to use
4332          * @param integer $flags a combination of the following flags:
4333          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4334          *
4335          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4336          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4337          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4338          *     the arguments, and to control the way they are expanded.
4339          *
4340          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4341          *     arguments, for instance:
4342          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4343          *
4344          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4345          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4346          *     working if/when this is changed.
4347          *
4348          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4349          *     expansion.
4350          *
4351          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4352          *     about the methods available in PPFrame and PPNode.
4353          *
4354          * @return The old callback function for this name, if any
4355          */
4356         function setFunctionHook( $id, $callback, $flags = 0 ) {
4357                 global $wgContLang;
4358
4359                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4360                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4361
4362                 # Add to function cache
4363                 $mw = MagicWord::get( $id );
4364                 if ( !$mw )
4365                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4366
4367                 $synonyms = $mw->getSynonyms();
4368                 $sensitive = intval( $mw->isCaseSensitive() );
4369
4370                 foreach ( $synonyms as $syn ) {
4371                         # Case
4372                         if ( !$sensitive ) {
4373                                 $syn = $wgContLang->lc( $syn );
4374                         }
4375                         # Add leading hash
4376                         if ( !( $flags & SFH_NO_HASH ) ) {
4377                                 $syn = '#' . $syn;
4378                         }
4379                         # Remove trailing colon
4380                         if ( substr( $syn, -1, 1 ) === ':' ) {
4381                                 $syn = substr( $syn, 0, -1 );
4382                         }
4383                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4384                 }
4385                 return $oldVal;
4386         }
4387
4388         /**
4389          * Get all registered function hook identifiers
4390          *
4391          * @return array
4392          */
4393         function getFunctionHooks() {
4394                 return array_keys( $this->mFunctionHooks );
4395         }
4396
4397         /**
4398          * Create a tag function, e.g. <test>some stuff</test>.
4399          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4400          * Unlike parser functions, their content is not preprocessed.
4401          */
4402         function setFunctionTagHook( $tag, $callback, $flags ) {
4403                 $tag = strtolower( $tag );
4404                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4405                         $this->mFunctionTagHooks[$tag] : null;
4406                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4407
4408                 if ( !in_array( $tag, $this->mStripList ) ) {
4409                         $this->mStripList[] = $tag;
4410                 }
4411
4412                 return $old;
4413         }
4414
4415         /**
4416          * FIXME: update documentation. makeLinkObj() is deprecated.
4417          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4418          * Placeholders created in Skin::makeLinkObj()
4419          * Returns an array of link CSS classes, indexed by PDBK.
4420          */
4421         function replaceLinkHolders( &$text, $options = 0 ) {
4422                 return $this->mLinkHolders->replace( $text );
4423         }
4424
4425         /**
4426          * Replace <!--LINK--> link placeholders with plain text of links
4427          * (not HTML-formatted).
4428          * @param string $text
4429          * @return string
4430          */
4431         function replaceLinkHoldersText( $text ) {
4432                 return $this->mLinkHolders->replaceText( $text );
4433         }
4434
4435         /**
4436          * Renders an image gallery from a text with one line per image.
4437          * text labels may be given by using |-style alternative text. E.g.
4438          *   Image:one.jpg|The number "1"
4439          *   Image:tree.jpg|A tree
4440          * given as text will return the HTML of a gallery with two images,
4441          * labeled 'The number "1"' and
4442          * 'A tree'.
4443          */
4444         function renderImageGallery( $text, $params ) {
4445                 $ig = new ImageGallery();
4446                 $ig->setContextTitle( $this->mTitle );
4447                 $ig->setShowBytes( false );
4448                 $ig->setShowFilename( false );
4449                 $ig->setParser( $this );
4450                 $ig->setHideBadImages();
4451                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4452                 $ig->useSkin( $this->mOptions->getSkin() );
4453                 $ig->mRevisionId = $this->mRevisionId;
4454
4455                 if ( isset( $params['showfilename'] ) ) {
4456                         $ig->setShowFilename( true );
4457                 } else {
4458                         $ig->setShowFilename( false );
4459                 }
4460                 if ( isset( $params['caption'] ) ) {
4461                         $caption = $params['caption'];
4462                         $caption = htmlspecialchars( $caption );
4463                         $caption = $this->replaceInternalLinks( $caption );
4464                         $ig->setCaptionHtml( $caption );
4465                 }
4466                 if ( isset( $params['perrow'] ) ) {
4467                         $ig->setPerRow( $params['perrow'] );
4468                 }
4469                 if ( isset( $params['widths'] ) ) {
4470                         $ig->setWidths( $params['widths'] );
4471                 }
4472                 if ( isset( $params['heights'] ) ) {
4473                         $ig->setHeights( $params['heights'] );
4474                 }
4475
4476                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4477
4478                 $lines = StringUtils::explode( "\n", $text );
4479                 foreach ( $lines as $line ) {
4480                         # match lines like these:
4481                         # Image:someimage.jpg|This is some image
4482                         $matches = array();
4483                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4484                         # Skip empty lines
4485                         if ( count( $matches ) == 0 ) {
4486                                 continue;
4487                         }
4488
4489                         if ( strpos( $matches[0], '%' ) !== false ) {
4490                                 $matches[1] = urldecode( $matches[1] );
4491                         }
4492                         $tp = Title::newFromText( $matches[1] );
4493                         $nt =& $tp;
4494                         if ( is_null( $nt ) ) {
4495                                 # Bogus title. Ignore these so we don't bomb out later.
4496                                 continue;
4497                         }
4498                         if ( isset( $matches[3] ) ) {
4499                                 $label = $matches[3];
4500                         } else {
4501                                 $label = '';
4502                         }
4503
4504                         $html = $this->recursiveTagParse( trim( $label ) );
4505
4506                         $ig->add( $nt, $html );
4507
4508                         # Only add real images (bug #5586)
4509                         if ( $nt->getNamespace() == NS_FILE ) {
4510                                 $this->mOutput->addImage( $nt->getDBkey() );
4511                         }
4512                 }
4513                 return $ig->toHTML();
4514         }
4515
4516         function getImageParams( $handler ) {
4517                 if ( $handler ) {
4518                         $handlerClass = get_class( $handler );
4519                 } else {
4520                         $handlerClass = '';
4521                 }
4522                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4523                         # Initialise static lists
4524                         static $internalParamNames = array(
4525                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4526                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4527                                         'bottom', 'text-bottom' ),
4528                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4529                                         'upright', 'border', 'link', 'alt' ),
4530                         );
4531                         static $internalParamMap;
4532                         if ( !$internalParamMap ) {
4533                                 $internalParamMap = array();
4534                                 foreach ( $internalParamNames as $type => $names ) {
4535                                         foreach ( $names as $name ) {
4536                                                 $magicName = str_replace( '-', '_', "img_$name" );
4537                                                 $internalParamMap[$magicName] = array( $type, $name );
4538                                         }
4539                                 }
4540                         }
4541
4542                         # Add handler params
4543                         $paramMap = $internalParamMap;
4544                         if ( $handler ) {
4545                                 $handlerParamMap = $handler->getParamMap();
4546                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4547                                         $paramMap[$magic] = array( 'handler', $paramName );
4548                                 }
4549                         }
4550                         $this->mImageParams[$handlerClass] = $paramMap;
4551                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4552                 }
4553                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4554         }
4555
4556         /**
4557          * Parse image options text and use it to make an image
4558          * @param Title $title
4559          * @param string $options
4560          * @param LinkHolderArray $holders
4561          */
4562         function makeImage( $title, $options, $holders = false ) {
4563                 # Check if the options text is of the form "options|alt text"
4564                 # Options are:
4565                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4566                 #  * left       no resizing, just left align. label is used for alt= only
4567                 #  * right      same, but right aligned
4568                 #  * none       same, but not aligned
4569                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4570                 #  * center     center the image
4571                 #  * frame      Keep original image size, no magnify-button.
4572                 #  * framed     Same as "frame"
4573                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4574                 #  * upright    reduce width for upright images, rounded to full __0 px
4575                 #  * border     draw a 1px border around the image
4576                 #  * alt        Text for HTML alt attribute (defaults to empty)
4577                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4578                 # vertical-align values (no % or length right now):
4579                 #  * baseline
4580                 #  * sub
4581                 #  * super
4582                 #  * top
4583                 #  * text-top
4584                 #  * middle
4585                 #  * bottom
4586                 #  * text-bottom
4587
4588                 $parts = StringUtils::explode( "|", $options );
4589                 $sk = $this->mOptions->getSkin();
4590
4591                 # Give extensions a chance to select the file revision for us
4592                 $skip = $time = $descQuery = false;
4593                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4594
4595                 if ( $skip ) {
4596                         return $sk->link( $title );
4597                 }
4598
4599                 # Get the file
4600                 $imagename = $title->getDBkey();
4601                 $file = wfFindFile( $title, array( 'time' => $time ) );
4602                 # Get parameter map
4603                 $handler = $file ? $file->getHandler() : false;
4604
4605                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4606
4607                 # Process the input parameters
4608                 $caption = '';
4609                 $params = array( 'frame' => array(), 'handler' => array(),
4610                         'horizAlign' => array(), 'vertAlign' => array() );
4611                 foreach ( $parts as $part ) {
4612                         $part = trim( $part );
4613                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4614                         $validated = false;
4615                         if ( isset( $paramMap[$magicName] ) ) {
4616                                 list( $type, $paramName ) = $paramMap[$magicName];
4617
4618                                 # Special case; width and height come in one variable together
4619                                 if ( $type === 'handler' && $paramName === 'width' ) {
4620                                         $m = array();
4621                                         # (bug 13500) In both cases (width/height and width only),
4622                                         # permit trailing "px" for backward compatibility.
4623                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4624                                                 $width = intval( $m[1] );
4625                                                 $height = intval( $m[2] );
4626                                                 if ( $handler->validateParam( 'width', $width ) ) {
4627                                                         $params[$type]['width'] = $width;
4628                                                         $validated = true;
4629                                                 }
4630                                                 if ( $handler->validateParam( 'height', $height ) ) {
4631                                                         $params[$type]['height'] = $height;
4632                                                         $validated = true;
4633                                                 }
4634                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4635                                                 $width = intval( $value );
4636                                                 if ( $handler->validateParam( 'width', $width ) ) {
4637                                                         $params[$type]['width'] = $width;
4638                                                         $validated = true;
4639                                                 }
4640                                         } # else no validation -- bug 13436
4641                                 } else {
4642                                         if ( $type === 'handler' ) {
4643                                                 # Validate handler parameter
4644                                                 $validated = $handler->validateParam( $paramName, $value );
4645                                         } else {
4646                                                 # Validate internal parameters
4647                                                 switch( $paramName ) {
4648                                                 case 'manualthumb':
4649                                                 case 'alt':
4650                                                         # @todo Fixme: possibly check validity here for
4651                                                         # manualthumb? downstream behavior seems odd with
4652                                                         # missing manual thumbs.
4653                                                         $validated = true;
4654                                                         $value = $this->stripAltText( $value, $holders );
4655                                                         break;
4656                                                 case 'link':
4657                                                         $chars = self::EXT_LINK_URL_CLASS;
4658                                                         $prots = $this->mUrlProtocols;
4659                                                         if ( $value === '' ) {
4660                                                                 $paramName = 'no-link';
4661                                                                 $value = true;
4662                                                                 $validated = true;
4663                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4664                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4665                                                                         $paramName = 'link-url';
4666                                                                         $this->mOutput->addExternalLink( $value );
4667                                                                         $validated = true;
4668                                                                 }
4669                                                         } else {
4670                                                                 $linkTitle = Title::newFromText( $value );
4671                                                                 if ( $linkTitle ) {
4672                                                                         $paramName = 'link-title';
4673                                                                         $value = $linkTitle;
4674                                                                         $this->mOutput->addLink( $linkTitle );
4675                                                                         $validated = true;
4676                                                                 }
4677                                                         }
4678                                                         break;
4679                                                 default:
4680                                                         # Most other things appear to be empty or numeric...
4681                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4682                                                 }
4683                                         }
4684
4685                                         if ( $validated ) {
4686                                                 $params[$type][$paramName] = $value;
4687                                         }
4688                                 }
4689                         }
4690                         if ( !$validated ) {
4691                                 $caption = $part;
4692                         }
4693                 }
4694
4695                 # Process alignment parameters
4696                 if ( $params['horizAlign'] ) {
4697                         $params['frame']['align'] = key( $params['horizAlign'] );
4698                 }
4699                 if ( $params['vertAlign'] ) {
4700                         $params['frame']['valign'] = key( $params['vertAlign'] );
4701                 }
4702
4703                 $params['frame']['caption'] = $caption;
4704
4705                 # Will the image be presented in a frame, with the caption below?
4706                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4707                                  isset( $params['frame']['framed'] ) ||
4708                                  isset( $params['frame']['thumbnail'] ) ||
4709                                  isset( $params['frame']['manualthumb'] );
4710
4711                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4712                 # came to also set the caption, ordinary text after the image -- which
4713                 # makes no sense, because that just repeats the text multiple times in
4714                 # screen readers.  It *also* came to set the title attribute.
4715                 #
4716                 # Now that we have an alt attribute, we should not set the alt text to
4717                 # equal the caption: that's worse than useless, it just repeats the
4718                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4719                 # use the unnamed parameter for alt text as well, just for the time be-
4720                 # ing, if the unnamed param is set and the alt param is not.
4721                 #
4722                 # For the future, we need to figure out if we want to tweak this more,
4723                 # e.g., introducing a title= parameter for the title; ignoring the un-
4724                 # named parameter entirely for images without a caption; adding an ex-
4725                 # plicit caption= parameter and preserving the old magic unnamed para-
4726                 # meter for BC; ...
4727                 if ( $imageIsFramed ) { # Framed image
4728                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4729                                 # No caption or alt text, add the filename as the alt text so
4730                                 # that screen readers at least get some description of the image
4731                                 $params['frame']['alt'] = $title->getText();
4732                         }
4733                         # Do not set $params['frame']['title'] because tooltips don't make sense
4734                         # for framed images
4735                 } else { # Inline image
4736                         if ( !isset( $params['frame']['alt'] ) ) {
4737                                 # No alt text, use the "caption" for the alt text
4738                                 if ( $caption !== '') {
4739                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4740                                 } else {
4741                                         # No caption, fall back to using the filename for the
4742                                         # alt text
4743                                         $params['frame']['alt'] = $title->getText();
4744                                 }
4745                         }
4746                         # Use the "caption" for the tooltip text
4747                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4748                 }
4749
4750                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4751
4752                 # Linker does the rest
4753                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4754
4755                 # Give the handler a chance to modify the parser object
4756                 if ( $handler ) {
4757                         $handler->parserTransformHook( $this, $file );
4758                 }
4759
4760                 return $ret;
4761         }
4762
4763         protected function stripAltText( $caption, $holders ) {
4764                 # Strip bad stuff out of the title (tooltip).  We can't just use
4765                 # replaceLinkHoldersText() here, because if this function is called
4766                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4767                 if ( $holders ) {
4768                         $tooltip = $holders->replaceText( $caption );
4769                 } else {
4770                         $tooltip = $this->replaceLinkHoldersText( $caption );
4771                 }
4772
4773                 # make sure there are no placeholders in thumbnail attributes
4774                 # that are later expanded to html- so expand them now and
4775                 # remove the tags
4776                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4777                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4778
4779                 return $tooltip;
4780         }
4781
4782         /**
4783          * Set a flag in the output object indicating that the content is dynamic and
4784          * shouldn't be cached.
4785          */
4786         function disableCache() {
4787                 wfDebug( "Parser output marked as uncacheable.\n" );
4788                 $this->mOutput->mCacheTime = -1;
4789         }
4790
4791         /**#@+
4792          * Callback from the Sanitizer for expanding items found in HTML attribute
4793          * values, so they can be safely tested and escaped.
4794          * @param string $text
4795          * @param PPFrame $frame
4796          * @return string
4797          * @private
4798          */
4799         function attributeStripCallback( &$text, $frame = false ) {
4800                 $text = $this->replaceVariables( $text, $frame );
4801                 $text = $this->mStripState->unstripBoth( $text );
4802                 return $text;
4803         }
4804
4805         /**#@-*/
4806
4807         /**#@+
4808          * Accessor/mutator
4809          */
4810         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4811         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4812         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4813         /**#@-*/
4814
4815         /**#@+
4816          * Accessor
4817          */
4818         function getTags() {
4819                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4820         }
4821         /**#@-*/
4822
4823
4824         /**
4825          * Break wikitext input into sections, and either pull or replace
4826          * some particular section's text.
4827          *
4828          * External callers should use the getSection and replaceSection methods.
4829          *
4830          * @param string $text Page wikitext
4831          * @param string $section A section identifier string of the form:
4832          *   <flag1> - <flag2> - ... - <section number>
4833          *
4834          * Currently the only recognised flag is "T", which means the target section number
4835          * was derived during a template inclusion parse, in other words this is a template
4836          * section edit link. If no flags are given, it was an ordinary section edit link.
4837          * This flag is required to avoid a section numbering mismatch when a section is
4838          * enclosed by <includeonly> (bug 6563).
4839          *
4840          * The section number 0 pulls the text before the first heading; other numbers will
4841          * pull the given section along with its lower-level subsections. If the section is
4842          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4843          *
4844          * @param string $mode One of "get" or "replace"
4845          * @param string $newText Replacement text for section data.
4846          * @return string for "get", the extracted section text.
4847          *                for "replace", the whole page with the section replaced.
4848          */
4849         private function extractSections( $text, $section, $mode, $newText='' ) {
4850                 global $wgTitle;
4851                 $this->clearState();
4852                 $this->setTitle( $wgTitle ); # not generally used but removes an ugly failure mode
4853                 $this->mOptions = new ParserOptions;
4854                 $this->setOutputType( self::OT_PLAIN );
4855                 $outText = '';
4856                 $frame = $this->getPreprocessor()->newFrame();
4857
4858                 # Process section extraction flags
4859                 $flags = 0;
4860                 $sectionParts = explode( '-', $section );
4861                 $sectionIndex = array_pop( $sectionParts );
4862                 foreach ( $sectionParts as $part ) {
4863                         if ( $part === 'T' ) {
4864                                 $flags |= self::PTD_FOR_INCLUSION;
4865                         }
4866                 }
4867                 # Preprocess the text
4868                 $root = $this->preprocessToDom( $text, $flags );
4869
4870                 # <h> nodes indicate section breaks
4871                 # They can only occur at the top level, so we can find them by iterating the root's children
4872                 $node = $root->getFirstChild();
4873
4874                 # Find the target section
4875                 if ( $sectionIndex == 0 ) {
4876                         # Section zero doesn't nest, level=big
4877                         $targetLevel = 1000;
4878                 } else {
4879             while ( $node ) {
4880                 if ( $node->getName() === 'h' ) {
4881                     $bits = $node->splitHeading();
4882                                         if ( $bits['i'] == $sectionIndex ) {
4883                                         $targetLevel = $bits['level'];
4884                                                 break;
4885                                         }
4886                                 }
4887                                 if ( $mode === 'replace' ) {
4888                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4889                                 }
4890                                 $node = $node->getNextSibling();
4891                         }
4892                 }
4893
4894                 if ( !$node ) {
4895                         # Not found
4896                         if ( $mode === 'get' ) {
4897                                 return $newText;
4898                         } else {
4899                                 return $text;
4900                         }
4901                 }
4902
4903                 # Find the end of the section, including nested sections
4904                 do {
4905                         if ( $node->getName() === 'h' ) {
4906                                 $bits = $node->splitHeading();
4907                                 $curLevel = $bits['level'];
4908                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4909                                         break;
4910                                 }
4911                         }
4912                         if ( $mode === 'get' ) {
4913                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4914                         }
4915                         $node = $node->getNextSibling();
4916                 } while ( $node );
4917
4918                 # Write out the remainder (in replace mode only)
4919                 if ( $mode === 'replace' ) {
4920                         # Output the replacement text
4921                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4922                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4923                         # Only add trailing whitespace if there is newText
4924                         if ( $newText != "" ) {
4925                                 $outText .= $newText . "\n\n";
4926                         }
4927
4928                         while ( $node ) {
4929                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4930                                 $node = $node->getNextSibling();
4931                         }
4932                 }
4933
4934                 if ( is_string( $outText ) ) {
4935                         # Re-insert stripped tags
4936                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4937                 }
4938
4939                 return $outText;
4940         }
4941
4942         /**
4943          * This function returns the text of a section, specified by a number ($section).
4944          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4945          * the first section before any such heading (section 0).
4946          *
4947          * If a section contains subsections, these are also returned.
4948          *
4949          * @param string $text text to look in
4950          * @param string $section section identifier
4951          * @param string $deftext default to return if section is not found
4952          * @return string text of the requested section
4953          */
4954         public function getSection( $text, $section, $deftext='' ) {
4955                 return $this->extractSections( $text, $section, "get", $deftext );
4956         }
4957
4958         public function replaceSection( $oldtext, $section, $text ) {
4959                 return $this->extractSections( $oldtext, $section, "replace", $text );
4960         }
4961
4962         /**
4963          * Get the timestamp associated with the current revision, adjusted for
4964          * the default server-local timestamp
4965          */
4966         function getRevisionTimestamp() {
4967                 if ( is_null( $this->mRevisionTimestamp ) ) {
4968                         wfProfileIn( __METHOD__ );
4969                         global $wgContLang;
4970                         $dbr = wfGetDB( DB_SLAVE );
4971                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4972                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4973
4974                         # Normalize timestamp to internal MW format for timezone processing.
4975                         # This has the added side-effect of replacing a null value with
4976                         # the current time, which gives us more sensible behavior for
4977                         # previews.
4978                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4979
4980                         # The cryptic '' timezone parameter tells to use the site-default
4981                         # timezone offset instead of the user settings.
4982                         #
4983                         # Since this value will be saved into the parser cache, served
4984                         # to other users, and potentially even used inside links and such,
4985                         # it needs to be consistent for all visitors.
4986                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4987
4988                         wfProfileOut( __METHOD__ );
4989                 }
4990                 return $this->mRevisionTimestamp;
4991         }
4992
4993         /**
4994          * Get the name of the user that edited the last revision
4995          */
4996         function getRevisionUser() {
4997                 # if this template is subst: the revision id will be blank,
4998                 # so just use the current user's name
4999                 if ( $this->mRevisionId ) {
5000                         $revision = Revision::newFromId( $this->mRevisionId );
5001                         $revuser = $revision->getUserText();
5002                 } else {
5003                         global $wgUser;
5004                         $revuser = $wgUser->getName();
5005                 }
5006                 return $revuser;
5007         }
5008
5009         /**
5010          * Mutator for $mDefaultSort
5011          *
5012          * @param $sort New value
5013          */
5014         public function setDefaultSort( $sort ) {
5015                 $this->mDefaultSort = $sort;
5016         }
5017
5018         /**
5019          * Accessor for $mDefaultSort
5020          * Will use the title/prefixed title if none is set
5021          *
5022          * @return string
5023          */
5024         public function getDefaultSort() {
5025                 global $wgCategoryPrefixedDefaultSortkey;
5026                 if ( $this->mDefaultSort !== false ) {
5027                         return $this->mDefaultSort;
5028                 } elseif ( $this->mTitle->getNamespace() == NS_CATEGORY ||
5029                         !$wgCategoryPrefixedDefaultSortkey )
5030                 {
5031                         return $this->mTitle->getText();
5032                 } else {
5033                         return $this->mTitle->getPrefixedText();
5034                 }
5035         }
5036
5037         /**
5038          * Accessor for $mDefaultSort
5039          * Unlike getDefaultSort(), will return false if none is set
5040          *
5041          * @return string or false
5042          */
5043         public function getCustomDefaultSort() {
5044                 return $this->mDefaultSort;
5045         }
5046
5047         /**
5048          * Try to guess the section anchor name based on a wikitext fragment
5049          * presumably extracted from a heading, for example "Header" from
5050          * "== Header ==".
5051          */
5052         public function guessSectionNameFromWikiText( $text ) {
5053                 # Strip out wikitext links(they break the anchor)
5054                 $text = $this->stripSectionName( $text );
5055                 $headline = Sanitizer::decodeCharReferences( $text );
5056                 # strip out HTML
5057                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5058                 $headline = trim( $headline );
5059                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5060                 $replacearray = array(
5061                         '%3A' => ':',
5062                         '%' => '.'
5063                 );
5064                 return str_replace(
5065                         array_keys( $replacearray ),
5066                         array_values( $replacearray ),
5067                         $sectionanchor );
5068         }
5069
5070         /**
5071          * Strips a text string of wikitext for use in a section anchor
5072          *
5073          * Accepts a text string and then removes all wikitext from the
5074          * string and leaves only the resultant text (i.e. the result of
5075          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5076          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5077          * to create valid section anchors by mimicing the output of the
5078          * parser when headings are parsed.
5079          *
5080          * @param $text string Text string to be stripped of wikitext
5081          * for use in a Section anchor
5082          * @return Filtered text string
5083          */
5084         public function stripSectionName( $text ) {
5085                 # Strip internal link markup
5086                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5087                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5088
5089                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5090                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5091                 # on how many empty links there are on the page - need to figure that out.
5092                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5093
5094                 # Parse wikitext quotes (italics & bold)
5095                 $text = $this->doQuotes( $text );
5096
5097                 # Strip HTML tags
5098                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5099                 return $text;
5100         }
5101
5102         function srvus( $text ) {
5103                 return $this->testSrvus( $text, $this->mOutputType );
5104         }
5105
5106         /**
5107          * strip/replaceVariables/unstrip for preprocessor regression testing
5108          */
5109         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5110                 $this->clearState();
5111                 if ( !$title instanceof Title ) {
5112                         $title = Title::newFromText( $title );
5113                 }
5114                 $this->mTitle = $title;
5115                 $this->mOptions = $options;
5116                 $this->setOutputType( $outputType );
5117                 $text = $this->replaceVariables( $text );
5118                 $text = $this->mStripState->unstripBoth( $text );
5119                 $text = Sanitizer::removeHTMLtags( $text );
5120                 return $text;
5121         }
5122
5123         function testPst( $text, $title, $options ) {
5124                 global $wgUser;
5125                 if ( !$title instanceof Title ) {
5126                         $title = Title::newFromText( $title );
5127                 }
5128                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5129         }
5130
5131         function testPreprocess( $text, $title, $options ) {
5132                 if ( !$title instanceof Title ) {
5133                         $title = Title::newFromText( $title );
5134                 }
5135                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5136         }
5137
5138         function markerSkipCallback( $s, $callback ) {
5139                 $i = 0;
5140                 $out = '';
5141                 while ( $i < strlen( $s ) ) {
5142                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5143                         if ( $markerStart === false ) {
5144                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5145                                 break;
5146                         } else {
5147                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5148                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5149                                 if ( $markerEnd === false ) {
5150                                         $out .= substr( $s, $markerStart );
5151                                         break;
5152                                 } else {
5153                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5154                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5155                                         $i = $markerEnd;
5156                                 }
5157                         }
5158                 }
5159                 return $out;
5160         }
5161
5162         function serialiseHalfParsedText( $text ) {
5163                 $data = array();
5164                 $data['text'] = $text;
5165
5166                 # First, find all strip markers, and store their
5167                 #  data in an array.
5168                 $stripState = new StripState;
5169                 $pos = 0;
5170                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5171                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5172                 {
5173                         $end_pos += strlen( self::MARKER_SUFFIX );
5174                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5175
5176                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5177                                 $replaceArray = $stripState->general;
5178                                 $stripText = $this->mStripState->general->data[$marker];
5179                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5180                                 $replaceArray = $stripState->nowiki;
5181                                 $stripText = $this->mStripState->nowiki->data[$marker];
5182                         } else {
5183                                 throw new MWException( "Hanging strip marker: '$marker'." );
5184                         }
5185
5186                         $replaceArray->setPair( $marker, $stripText );
5187                         $pos = $end_pos;
5188                 }
5189                 $data['stripstate'] = $stripState;
5190
5191                 # Now, find all of our links, and store THEIR
5192                 #  data in an array! :)
5193                 $links = array( 'internal' => array(), 'interwiki' => array() );
5194                 $pos = 0;
5195
5196                 # Internal links
5197                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5198                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5199
5200                         $ns = trim( $ns );
5201                         if ( empty( $links['internal'][$ns] ) ) {
5202                                 $links['internal'][$ns] = array();
5203                         }
5204
5205                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5206                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5207                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5208                 }
5209
5210                 $pos = 0;
5211
5212                 # Interwiki links
5213                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5214                         $data = substr( $text, $start_pos );
5215                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5216                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5217                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5218                 }
5219
5220                 $data['linkholder'] = $links;
5221
5222                 return $data;
5223         }
5224
5225         /**
5226          * TODO: document
5227          * @param $data Array
5228          * @param $intPrefix String unique identifying prefix
5229          * @return String
5230          */
5231         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5232                 if ( !$intPrefix ) {
5233                         $intPrefix = $this->getRandomString();
5234                 }
5235
5236                 # First, extract the strip state.
5237                 $stripState = $data['stripstate'];
5238                 $this->mStripState->general->merge( $stripState->general );
5239                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5240
5241                 # Now, extract the text, and renumber links
5242                 $text = $data['text'];
5243                 $links = $data['linkholder'];
5244
5245                 # Internal...
5246                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5247                         foreach ( $nsLinks as $key => $entry ) {
5248                                 $newKey = $intPrefix . '-' . $key;
5249                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5250
5251                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5252                         }
5253                 }
5254
5255                 # Interwiki...
5256                 foreach ( $links['interwiki'] as $key => $entry ) {
5257                         $newKey = "$intPrefix-$key";
5258                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5259
5260                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5261                 }
5262
5263                 # Should be good to go.
5264                 return $text;
5265         }
5266 }
5267
5268 /**
5269  * @todo document, briefly.
5270  * @ingroup Parser
5271  */
5272 class StripState {
5273         var $general, $nowiki;
5274
5275         function __construct() {
5276                 $this->general = new ReplacementArray;
5277                 $this->nowiki = new ReplacementArray;
5278         }
5279
5280         function unstripGeneral( $text ) {
5281                 wfProfileIn( __METHOD__ );
5282                 do {
5283                         $oldText = $text;
5284                         $text = $this->general->replace( $text );
5285                 } while ( $text !== $oldText );
5286                 wfProfileOut( __METHOD__ );
5287                 return $text;
5288         }
5289
5290         function unstripNoWiki( $text ) {
5291                 wfProfileIn( __METHOD__ );
5292                 do {
5293                         $oldText = $text;
5294                         $text = $this->nowiki->replace( $text );
5295                 } while ( $text !== $oldText );
5296                 wfProfileOut( __METHOD__ );
5297                 return $text;
5298         }
5299
5300         function unstripBoth( $text ) {
5301                 wfProfileIn( __METHOD__ );
5302                 do {
5303                         $oldText = $text;
5304                         $text = $this->general->replace( $text );
5305                         $text = $this->nowiki->replace( $text );
5306                 } while ( $text !== $oldText );
5307                 wfProfileOut( __METHOD__ );
5308                 return $text;
5309         }
5310 }
5311
5312 /**
5313  * @todo document, briefly.
5314  * @ingroup Parser
5315  */
5316 class OnlyIncludeReplacer {
5317         var $output = '';
5318
5319         function replace( $matches ) {
5320                 if ( substr( $matches[1], -1 ) === "\n" ) {
5321                         $this->output .= substr( $matches[1], 0, -1 );
5322                 } else {
5323                         $this->output .= $matches[1];
5324                 }
5325         }
5326 }