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 = '/\[(\b(' . wfUrlProtocols() . ')'.
 132                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/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;
 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                 # The position of the convert() call should not be changed. it
 348                 # assumes that the links are all replaced and the only thing left
 349                 # is the <nowiki> mark.
 350                 if ( !( $wgDisableLangConversion
 351                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 352                                 || $this->mTitle->isTalkPage()
 353                                 || $this->mTitle->isConversionTable() ) ) {
 354                         $text = $wgContLang->convert( $text );
 355                 }
 356
 357                 # A title may have been set in a conversion rule.
 358                 # Note that if a user tries to set a title in a conversion
 359                 # rule but content conversion was not done, then the parser
 360                 # won't pick it up.  This is probably expected behavior.
 361                 if ( $wgContLang->getConvRuleTitle() ) {
 362                         $this->mOutput->setTitleText( $wgContLang->getConvRuleTitle() );
 363                 }
 364
 365                 $text = $this->mStripState->unstripNoWiki( $text );
 366
 367                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 368
 369 //!JF Move to its own function
 370
 371                 $uniq_prefix = $this->mUniqPrefix;
 372                 $matches = array();
 373                 $elements = array_keys( $this->mTransparentTagHooks );
 374                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 375
 376                 foreach ( $matches as $marker => $data ) {
 377                         list( $element, $content, $params, $tag ) = $data;
 378                         $tagName = strtolower( $element );
 379                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 380                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 381                         } else {
 382                                 $output = $tag;
 383                         }
 384                         $this->mStripState->general->setPair( $marker, $output );
 385                 }
 386                 $text = $this->mStripState->unstripGeneral( $text );
 387
 388                 $text = Sanitizer::normalizeCharReferences( $text );
 389
 390                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 391                         $text = MWTidy::tidy( $text );
 392                 } else {
 393                         # attempt to sanitize at least some nesting problems
 394                         # (bug #2702 and quite a few others)
 395                         $tidyregs = array(
 396                                 # ''Something [http://www.cool.com cool''] -->
 397                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 398                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 399                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 400                                 # fix up an anchor inside another anchor, only
 401                                 # at least for a single single nested link (bug 3695)
 402                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 403                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 404                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 405                                 # contains a div, so fix it up here; replace
 406                                 # div with escaped text
 407                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 408                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 409                                 # remove empty italic or bold tag pairs, some
 410                                 # introduced by rules above
 411                                 '/<([bi])><\/\\1>/' => '',
 412                         );
 413
 414                         $text = preg_replace(
 415                                 array_keys( $tidyregs ),
 416                                 array_values( $tidyregs ),
 417                                 $text );
 418                 }
 419                 global $wgExpensiveParserFunctionLimit;
 420                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 421                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 422                 }
 423
 424                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 425
 426                 # Information on include size limits, for the benefit of users who try to skirt them
 427                 if ( $this->mOptions->getEnableLimitReport() ) {
 428                         $max = $this->mOptions->getMaxIncludeSize();
 429                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 430                         $limitReport =
 431                                 "NewPP limit report\n" .
 432                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 433                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 434                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 435                                 $PFreport;
 436                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 437                         $text .= "\n<!-- \n$limitReport-->\n";
 438                 }
 439                 $this->mOutput->setText( $text );
 440
 441                 $this->mRevisionId = $oldRevisionId;
 442                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 443                 wfProfileOut( $fname );
 444                 wfProfileOut( __METHOD__ );
 445
 446                 return $this->mOutput;
 447         }
 448
 449         /**
 450          * Recursive parser entry point that can be called from an extension tag
 451          * hook.
 452          *
 453          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 454          *
 455          * @param $text String: text extension wants to have parsed
 456          * @param PPFrame $frame: The frame to use for expanding any template variables
 457          */
 458         function recursiveTagParse( $text, $frame=false ) {
 459                 wfProfileIn( __METHOD__ );
 460                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 461                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 462                 $text = $this->internalParse( $text, false, $frame );
 463                 wfProfileOut( __METHOD__ );
 464                 return $text;
 465         }
 466
 467         /**
 468          * Expand templates and variables in the text, producing valid, static wikitext.
 469          * Also removes comments.
 470          */
 471         function preprocess( $text, $title, $options, $revid = null ) {
 472                 wfProfileIn( __METHOD__ );
 473                 $this->clearState();
 474                 $this->setOutputType( self::OT_PREPROCESS );
 475                 $this->mOptions = $options;
 476                 $this->setTitle( $title );
 477                 if ( $revid !== null ) {
 478                         $this->mRevisionId = $revid;
 479                 }
 480                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 481                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 482                 $text = $this->replaceVariables( $text );
 483                 $text = $this->mStripState->unstripBoth( $text );
 484                 wfProfileOut( __METHOD__ );
 485                 return $text;
 486         }
 487
 488         /**
 489          * Process the wikitext for the ?preload= feature. (bug 5210)
 490          *
 491          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 492          * comments, templates, arguments, tags hooks and parser functions are untouched.
 493          */
 494         public function getPreloadText( $text, $title, $options ) {
 495                 # Parser (re)initialisation
 496                 $this->clearState();
 497                 $this->setOutputType( self::OT_PLAIN );
 498                 $this->mOptions = $options;
 499                 $this->setTitle( $title );
 500
 501                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 502                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 503                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 504         }
 505
 506         /**
 507          * Get a random string
 508          *
 509          * @private
 510          * @static
 511          */
 512         function getRandomString() {
 513                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 514         }
 515
 516         function &getTitle() { return $this->mTitle; }
 517         function getOptions() { return $this->mOptions; }
 518         function getRevisionId() { return $this->mRevisionId; }
 519         function getOutput() { return $this->mOutput; }
 520         function nextLinkID() { return $this->mLinkID++; }
 521
 522         function getFunctionLang() {
 523                 global $wgLang, $wgContLang;
 524
 525                 $target = $this->mOptions->getTargetLanguage();
 526                 if ( $target !== null ) {
 527                         return $target;
 528                 } else {
 529                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 530                 }
 531         }
 532
 533         /**
 534          * Get a preprocessor object
 535          */
 536         function getPreprocessor() {
 537                 if ( !isset( $this->mPreprocessor ) ) {
 538                         $class = $this->mPreprocessorClass;
 539                         $this->mPreprocessor = new $class( $this );
 540                 }
 541                 return $this->mPreprocessor;
 542         }
 543
 544         /**
 545          * Replaces all occurrences of HTML-style comments and the given tags
 546          * in the text with a random marker and returns the next text. The output
 547          * parameter $matches will be an associative array filled with data in
 548          * the form:
 549          *   'UNIQ-xxxxx' => array(
 550          *     'element',
 551          *     'tag content',
 552          *     array( 'param' => 'x' ),
 553          *     '<element param="x">tag content</element>' ) )
 554          *
 555          * @param $elements list of element names. Comments are always extracted.
 556          * @param $text Source text string.
 557          * @param $uniq_prefix
 558          *
 559          * @public
 560          * @static
 561          */
 562         function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 563                 static $n = 1;
 564                 $stripped = '';
 565                 $matches = array();
 566
 567                 $taglist = implode( '|', $elements );
 568                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 569
 570                 while ( $text != '' ) {
 571                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 572                         $stripped .= $p[0];
 573                         if ( count( $p ) < 5 ) {
 574                                 break;
 575                         }
 576                         if ( count( $p ) > 5 ) {
 577                                 # comment
 578                                 $element    = $p[4];
 579                                 $attributes = '';
 580                                 $close      = '';
 581                                 $inside     = $p[5];
 582                         } else {
 583                                 # tag
 584                                 $element    = $p[1];
 585                                 $attributes = $p[2];
 586                                 $close      = $p[3];
 587                                 $inside     = $p[4];
 588                         }
 589
 590                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 591                         $stripped .= $marker;
 592
 593                         if ( $close === '/>' ) {
 594                                 # Empty element tag, <tag />
 595                                 $content = null;
 596                                 $text = $inside;
 597                                 $tail = null;
 598                         } else {
 599                                 if ( $element === '!--' ) {
 600                                         $end = '/(-->)/';
 601                                 } else {
 602                                         $end = "/(<\\/$element\\s*>)/i";
 603                                 }
 604                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 605                                 $content = $q[0];
 606                                 if ( count( $q ) < 3 ) {
 607                                         # No end tag -- let it run out to the end of the text.
 608                                         $tail = '';
 609                                         $text = '';
 610                                 } else {
 611                                         $tail = $q[1];
 612                                         $text = $q[2];
 613                                 }
 614                         }
 615
 616                         $matches[$marker] = array( $element,
 617                                 $content,
 618                                 Sanitizer::decodeTagAttributes( $attributes ),
 619                                 "<$element$attributes$close$content$tail" );
 620                 }
 621                 return $stripped;
 622         }
 623
 624         /**
 625          * Get a list of strippable XML-like elements
 626          */
 627         function getStripList() {
 628                 return $this->mStripList;
 629         }
 630
 631         /**
 632          * @deprecated use replaceVariables
 633          */
 634         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 635                 return $text;
 636         }
 637
 638         /**
 639          * Restores pre, math, and other extensions removed by strip()
 640          *
 641          * always call unstripNoWiki() after this one
 642          * @private
 643          * @deprecated use $this->mStripState->unstrip()
 644          */
 645         function unstrip( $text, $state ) {
 646                 return $state->unstripGeneral( $text );
 647         }
 648
 649         /**
 650          * Always call this after unstrip() to preserve the order
 651          *
 652          * @private
 653          * @deprecated use $this->mStripState->unstrip()
 654          */
 655         function unstripNoWiki( $text, $state ) {
 656                 return $state->unstripNoWiki( $text );
 657         }
 658
 659         /**
 660          * @deprecated use $this->mStripState->unstripBoth()
 661          */
 662         function unstripForHTML( $text ) {
 663                 return $this->mStripState->unstripBoth( $text );
 664         }
 665
 666         /**
 667          * Add an item to the strip state
 668          * Returns the unique tag which must be inserted into the stripped text
 669          * The tag will be replaced with the original text in unstrip()
 670          *
 671          * @private
 672          */
 673         function insertStripItem( $text ) {
 674                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 675                 $this->mMarkerIndex++;
 676                 $this->mStripState->general->setPair( $rnd, $text );
 677                 return $rnd;
 678         }
 679
 680         /**
 681          * Interface with html tidy
 682          * @deprecated Use MWTidy::tidy()
 683          */
 684         public static function tidy( $text ) {
 685                 wfDeprecated( __METHOD__ );
 686                 return MWTidy::tidy( $text );
 687         }
 688
 689         /**
 690          * parse the wiki syntax used to render tables
 691          *
 692          * @private
 693          */
 694         function doTableStuff( $text ) {
 695                 wfProfileIn( __METHOD__ );
 696
 697                 $lines = StringUtils::explode( "\n", $text );
 698                 $out = '';
 699                 $td_history = array(); # Is currently a td tag open?
 700                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 701                 $tr_history = array(); # Is currently a tr tag open?
 702                 $tr_attributes = array(); # history of tr attributes
 703                 $has_opened_tr = array(); # Did this table open a <tr> element?
 704                 $indent_level = 0; # indent level of the table
 705
 706                 foreach ( $lines as $outLine ) {
 707                         $line = trim( $outLine );
 708
 709                         if ( $line == '' ) { # empty line, go to next line
 710                                 $out .= $outLine."\n";
 711                                 continue;
 712                         }
 713                         $first_character = $line[0];
 714                         $matches = array();
 715
 716                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 717                                 # First check if we are starting a new table
 718                                 $indent_level = strlen( $matches[1] );
 719
 720                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 721                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 722
 723                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 724                                 array_push( $td_history , false );
 725                                 array_push( $last_tag_history , '' );
 726                                 array_push( $tr_history , false );
 727                                 array_push( $tr_attributes , '' );
 728                                 array_push( $has_opened_tr , false );
 729                         } elseif ( count( $td_history ) == 0 ) {
 730                                 # Don't do any of the following
 731                                 $out .= $outLine."\n";
 732                                 continue;
 733                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 734                                 # We are ending a table
 735                                 $line = '</table>' . substr( $line , 2 );
 736                                 $last_tag = array_pop( $last_tag_history );
 737
 738                                 if ( !array_pop( $has_opened_tr ) ) {
 739                                         $line = "<tr><td></td></tr>{$line}";
 740                                 }
 741
 742                                 if ( array_pop( $tr_history ) ) {
 743                                         $line = "</tr>{$line}";
 744                                 }
 745
 746                                 if ( array_pop( $td_history ) ) {
 747                                         $line = "</{$last_tag}>{$line}";
 748                                 }
 749                                 array_pop( $tr_attributes );
 750                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 751                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 752                                 # Now we have a table row
 753                                 $line = preg_replace( '#^\|-+#', '', $line );
 754
 755                                 # Whats after the tag is now only attributes
 756                                 $attributes = $this->mStripState->unstripBoth( $line );
 757                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 758                                 array_pop( $tr_attributes );
 759                                 array_push( $tr_attributes, $attributes );
 760
 761                                 $line = '';
 762                                 $last_tag = array_pop( $last_tag_history );
 763                                 array_pop( $has_opened_tr );
 764                                 array_push( $has_opened_tr , true );
 765
 766                                 if ( array_pop( $tr_history ) ) {
 767                                         $line = '</tr>';
 768                                 }
 769
 770                                 if ( array_pop( $td_history ) ) {
 771                                         $line = "</{$last_tag}>{$line}";
 772                                 }
 773
 774                                 $outLine = $line;
 775                                 array_push( $tr_history , false );
 776                                 array_push( $td_history , false );
 777                                 array_push( $last_tag_history , '' );
 778                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 779                                 # This might be cell elements, td, th or captions
 780                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 781                                         $first_character = '+';
 782                                         $line = substr( $line , 1 );
 783                                 }
 784
 785                                 $line = substr( $line , 1 );
 786
 787                                 if ( $first_character === '!' ) {
 788                                         $line = str_replace( '!!' , '||' , $line );
 789                                 }
 790
 791                                 # Split up multiple cells on the same line.
 792                                 # FIXME : This can result in improper nesting of tags processed
 793                                 # by earlier parser steps, but should avoid splitting up eg
 794                                 # attribute values containing literal "||".
 795                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 796
 797                                 $outLine = '';
 798
 799                                 # Loop through each table cell
 800                                 foreach ( $cells as $cell ) {
 801                                         $previous = '';
 802                                         if ( $first_character !== '+' ) {
 803                                                 $tr_after = array_pop( $tr_attributes );
 804                                                 if ( !array_pop( $tr_history ) ) {
 805                                                         $previous = "<tr{$tr_after}>\n";
 806                                                 }
 807                                                 array_push( $tr_history , true );
 808                                                 array_push( $tr_attributes , '' );
 809                                                 array_pop( $has_opened_tr );
 810                                                 array_push( $has_opened_tr , true );
 811                                         }
 812
 813                                         $last_tag = array_pop( $last_tag_history );
 814
 815                                         if ( array_pop( $td_history ) ) {
 816                                                 $previous = "</{$last_tag}>{$previous}";
 817                                         }
 818
 819                                         if ( $first_character === '|' ) {
 820                                                 $last_tag = 'td';
 821                                         } elseif ( $first_character === '!' ) {
 822                                                 $last_tag = 'th';
 823                                         } elseif ( $first_character === '+' ) {
 824                                                 $last_tag = 'caption';
 825                                         } else {
 826                                                 $last_tag = '';
 827                                         }
 828
 829                                         array_push( $last_tag_history , $last_tag );
 830
 831                                         # A cell could contain both parameters and data
 832                                         $cell_data = explode( '|' , $cell , 2 );
 833
 834                                         # Bug 553: Note that a '|' inside an invalid link should not
 835                                         # be mistaken as delimiting cell parameters
 836                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 837                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 838                                         } elseif ( count( $cell_data ) == 1 ) {
 839                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 840                                         } else {
 841                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 842                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 843                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 844                                         }
 845
 846                                         $outLine .= $cell;
 847                                         array_push( $td_history , true );
 848                                 }
 849                         }
 850                         $out .= $outLine . "\n";
 851                 }
 852
 853                 # Closing open td, tr && table
 854                 while ( count( $td_history ) > 0 ) {
 855                         if ( array_pop( $td_history ) ) {
 856                                 $out .= "</td>\n";
 857                         }
 858                         if ( array_pop( $tr_history ) ) {
 859                                 $out .= "</tr>\n";
 860                         }
 861                         if ( !array_pop( $has_opened_tr ) ) {
 862                                 $out .= "<tr><td></td></tr>\n" ;
 863                         }
 864
 865                         $out .= "</table>\n";
 866                 }
 867
 868                 # Remove trailing line-ending (b/c)
 869                 if ( substr( $out, -1 ) === "\n" ) {
 870                         $out = substr( $out, 0, -1 );
 871                 }
 872
 873                 # special case: don't return empty table
 874                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 875                         $out = '';
 876                 }
 877
 878                 wfProfileOut( __METHOD__ );
 879
 880                 return $out;
 881         }
 882
 883         /**
 884          * Helper function for parse() that transforms wiki markup into
 885          * HTML. Only called for $mOutputType == self::OT_HTML.
 886          *
 887          * @private
 888          */
 889         function internalParse( $text, $isMain = true, $frame=false ) {
 890                 wfProfileIn( __METHOD__ );
 891
 892                 $origText = $text;
 893
 894                 # Hook to suspend the parser in this state
 895                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 896                         wfProfileOut( __METHOD__ );
 897                         return $text ;
 898                 }
 899
 900                 # if $frame is provided, then use $frame for replacing any variables
 901                 if ( $frame ) {
 902                         # use frame depth to infer how include/noinclude tags should be handled
 903                         # depth=0 means this is the top-level document; otherwise it's an included document
 904                         if ( !$frame->depth ) {
 905                                 $flag = 0;
 906                         } else {
 907                                 $flag = Parser::PTD_FOR_INCLUSION;
 908                         $dom = $this->preprocessToDom( $text, $flag );
 909                         $text = $frame->expand( $dom );
 910                         }
 911                 } else {
 912                         # if $frame is not provided, then use old-style replaceVariables
 913                         $text = $this->replaceVariables( $text );
 914                 }
 915
 916                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 917                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 918
 919                 # Tables need to come after variable replacement for things to work
 920                 # properly; putting them before other transformations should keep
 921                 # exciting things like link expansions from showing up in surprising
 922                 # places.
 923                 $text = $this->doTableStuff( $text );
 924
 925                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 926
 927                 $text = $this->doDoubleUnderscore( $text );
 928
 929                 $text = $this->doHeadings( $text );
 930                 if ( $this->mOptions->getUseDynamicDates() ) {
 931                         $df = DateFormatter::getInstance();
 932                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 933                 }
 934                 $text = $this->doAllQuotes( $text );
 935                 $text = $this->replaceInternalLinks( $text );
 936                 $text = $this->replaceExternalLinks( $text );
 937
 938                 # replaceInternalLinks may sometimes leave behind
 939                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 940                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
 941
 942                 $text = $this->doMagicLinks( $text );
 943                 $text = $this->formatHeadings( $text, $origText, $isMain );
 944
 945                 wfProfileOut( __METHOD__ );
 946                 return $text;
 947         }
 948
 949         /**
 950          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 951          * magic external links.
 952          *
 953          * DML
 954          * @private
 955          */
 956         function doMagicLinks( $text ) {
 957                 wfProfileIn( __METHOD__ );
 958                 $prots = $this->mUrlProtocols;
 959                 $urlChar = self::EXT_LINK_URL_CLASS;
 960                 $text = preg_replace_callback(
 961                         '!(?:                           # Start cases
 962                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
 963                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 964                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 965                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 966                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 967                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 968                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 969                                     [0-9Xx]                 # check digit
 970                                     \b)
 971                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 972                 wfProfileOut( __METHOD__ );
 973                 return $text;
 974         }
 975
 976         function magicLinkCallback( $m ) {
 977                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 978                         # Skip anchor
 979                         return $m[0];
 980                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 981                         # Skip HTML element
 982                         return $m[0];
 983                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 984                         # Free external link
 985                         return $this->makeFreeExternalLink( $m[0] );
 986                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 987                         # RFC or PMID
 988                         $CssClass = '';
 989                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 990                                 $keyword = 'RFC';
 991                                 $urlmsg = 'rfcurl';
 992                                 $CssClass = 'mw-magiclink-rfc';
 993                                 $id = $m[4];
 994                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
 995                                 $keyword = 'PMID';
 996                                 $urlmsg = 'pubmedurl';
 997                                 $CssClass = 'mw-magiclink-pmid';
 998                                 $id = $m[4];
 999                         } else {
1000                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1001                                         substr( $m[0], 0, 20 ) . '"' );
1002                         }
1003                         $url = wfMsg( $urlmsg, $id);
1004                         $sk = $this->mOptions->getSkin();
1005                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1006                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1007                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1008                         # ISBN
1009                         $isbn = $m[5];
1010                         $num = strtr( $isbn, array(
1011                                 '-' => '',
1012                                 ' ' => '',
1013                                 'x' => 'X',
1014                         ));
1015                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1016                         return'<a href="' .
1017                                 $titleObj->escapeLocalUrl() .
1018                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1019                 } else {
1020                         return $m[0];
1021                 }
1022         }
1023
1024         /**
1025          * Make a free external link, given a user-supplied URL
1026          * @return HTML
1027          * @private
1028          */
1029         function makeFreeExternalLink( $url ) {
1030                 global $wgContLang;
1031                 wfProfileIn( __METHOD__ );
1032
1033                 $sk = $this->mOptions->getSkin();
1034                 $trail = '';
1035
1036                 # The characters '<' and '>' (which were escaped by
1037                 # removeHTMLtags()) should not be included in
1038                 # URLs, per RFC 2396.
1039                 $m2 = array();
1040                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1041                         $trail = substr( $url, $m2[0][1] ) . $trail;
1042                         $url = substr( $url, 0, $m2[0][1] );
1043                 }
1044
1045                 # Move trailing punctuation to $trail
1046                 $sep = ',;\.:!?';
1047                 # If there is no left bracket, then consider right brackets fair game too
1048                 if ( strpos( $url, '(' ) === false ) {
1049                         $sep .= ')';
1050                 }
1051
1052                 $numSepChars = strspn( strrev( $url ), $sep );
1053                 if ( $numSepChars ) {
1054                         $trail = substr( $url, -$numSepChars ) . $trail;
1055                         $url = substr( $url, 0, -$numSepChars );
1056                 }
1057
1058                 $url = Sanitizer::cleanUrl( $url );
1059
1060                 # Is this an external image?
1061                 $text = $this->maybeMakeExternalImage( $url );
1062                 if ( $text === false ) {
1063                         # Not an image, make a link
1064                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1065                                 $this->getExternalLinkAttribs( $url ) );
1066                         # Register it in the output object...
1067                         # Replace unnecessary URL escape codes with their equivalent characters
1068                         $pasteurized = self::replaceUnusualEscapes( $url );
1069                         $this->mOutput->addExternalLink( $pasteurized );
1070                 }
1071                 wfProfileOut( __METHOD__ );
1072                 return $text . $trail;
1073         }
1074
1075
1076         /**
1077          * Parse headers and return html
1078          *
1079          * @private
1080          */
1081         function doHeadings( $text ) {
1082                 wfProfileIn( __METHOD__ );
1083                 for ( $i = 6; $i >= 1; --$i ) {
1084                         $h = str_repeat( '=', $i );
1085                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1086                           "<h$i>\\1</h$i>", $text );
1087                 }
1088                 wfProfileOut( __METHOD__ );
1089                 return $text;
1090         }
1091
1092         /**
1093          * Replace single quotes with HTML markup
1094          * @private
1095          * @return string the altered text
1096          */
1097         function doAllQuotes( $text ) {
1098                 wfProfileIn( __METHOD__ );
1099                 $outtext = '';
1100                 $lines = StringUtils::explode( "\n", $text );
1101                 foreach ( $lines as $line ) {
1102                         $outtext .= $this->doQuotes( $line ) . "\n";
1103                 }
1104                 $outtext = substr( $outtext, 0,-1 );
1105                 wfProfileOut( __METHOD__ );
1106                 return $outtext;
1107         }
1108
1109         /**
1110          * Helper function for doAllQuotes()
1111          */
1112         public function doQuotes( $text ) {
1113                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1114                 if ( count( $arr ) == 1 ) {
1115                         return $text;
1116                 } else {
1117                         # First, do some preliminary work. This may shift some apostrophes from
1118                         # being mark-up to being text. It also counts the number of occurrences
1119                         # of bold and italics mark-ups.
1120                         $i = 0;
1121                         $numbold = 0;
1122                         $numitalics = 0;
1123                         foreach ( $arr as $r ) {
1124                                 if ( ( $i % 2 ) == 1 ) {
1125                                         # If there are ever four apostrophes, assume the first is supposed to
1126                                         # be text, and the remaining three constitute mark-up for bold text.
1127                                         if ( strlen( $arr[$i] ) == 4 ) {
1128                                                 $arr[$i-1] .= "'";
1129                                                 $arr[$i] = "'''";
1130                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1131                                                 # If there are more than 5 apostrophes in a row, assume they're all
1132                                                 # text except for the last 5.
1133                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1134                                                 $arr[$i] = "'''''";
1135                                         }
1136                                         # Count the number of occurrences of bold and italics mark-ups.
1137                                         # We are not counting sequences of five apostrophes.
1138                                         if ( strlen( $arr[$i] ) == 2 ) {
1139                                                 $numitalics++;
1140                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1141                                                 $numbold++;
1142                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1143                                                 $numitalics++;
1144                                                 $numbold++;
1145                                         }
1146                                 }
1147                                 $i++;
1148                         }
1149
1150                         # If there is an odd number of both bold and italics, it is likely
1151                         # that one of the bold ones was meant to be an apostrophe followed
1152                         # by italics. Which one we cannot know for certain, but it is more
1153                         # likely to be one that has a single-letter word before it.
1154                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1155                                 $i = 0;
1156                                 $firstsingleletterword = -1;
1157                                 $firstmultiletterword = -1;
1158                                 $firstspace = -1;
1159                                 foreach ( $arr as $r ) {
1160                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1161                                                 $x1 = substr( $arr[$i-1], -1 );
1162                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1163                                                 if ( $x1 === ' ' ) {
1164                                                         if ( $firstspace == -1 ) {
1165                                                                 $firstspace = $i;
1166                                                         }
1167                                                 } elseif ( $x2 === ' ') {
1168                                                         if ( $firstsingleletterword == -1 ) {
1169                                                                 $firstsingleletterword = $i;
1170                                                         }
1171                                                 } else {
1172                                                         if ( $firstmultiletterword == -1 ) {
1173                                                                 $firstmultiletterword = $i;
1174                                                         }
1175                                                 }
1176                                         }
1177                                         $i++;
1178                                 }
1179
1180                                 # If there is a single-letter word, use it!
1181                                 if ( $firstsingleletterword > -1 ) {
1182                                         $arr[$firstsingleletterword] = "''";
1183                                         $arr[$firstsingleletterword-1] .= "'";
1184                                 } elseif ( $firstmultiletterword > -1 ) {
1185                                         # If not, but there's a multi-letter word, use that one.
1186                                         $arr[$firstmultiletterword] = "''";
1187                                         $arr[$firstmultiletterword-1] .= "'";
1188                                 } elseif ( $firstspace > -1 ) {
1189                                         # ... otherwise use the first one that has neither.
1190                                         # (notice that it is possible for all three to be -1 if, for example,
1191                                         # there is only one pentuple-apostrophe in the line)
1192                                         $arr[$firstspace] = "''";
1193                                         $arr[$firstspace-1] .= "'";
1194                                 }
1195                         }
1196
1197                         # Now let's actually convert our apostrophic mush to HTML!
1198                         $output = '';
1199                         $buffer = '';
1200                         $state = '';
1201                         $i = 0;
1202                         foreach ( $arr as $r ) {
1203                                 if ( ( $i % 2 ) == 0 ) {
1204                                         if ( $state === 'both' ) {
1205                                                 $buffer .= $r;
1206                                         } else {
1207                                                 $output .= $r;
1208                                         }
1209                                 } else {
1210                                         if ( strlen( $r ) == 2 ) {
1211                                                 if ( $state === 'i' ) {
1212                                                         $output .= '</i>'; $state = '';
1213                                                 } elseif ( $state === 'bi' ) {
1214                                                         $output .= '</i>'; $state = 'b';
1215                                                 } elseif ( $state === 'ib' ) {
1216                                                         $output .= '</b></i><b>'; $state = 'b';
1217                                                 } elseif ( $state === 'both' ) {
1218                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1219                                                 } else { # $state can be 'b' or ''
1220                                                         $output .= '<i>'; $state .= 'i';
1221                                                 }
1222                                         } elseif ( strlen( $r ) == 3 ) {
1223                                                 if ( $state === 'b' ) {
1224                                                         $output .= '</b>'; $state = '';
1225                                                 } elseif ( $state === 'bi' ) {
1226                                                         $output .= '</i></b><i>'; $state = 'i';
1227                                                 } elseif ( $state === 'ib' ) {
1228                                                         $output .= '</b>'; $state = 'i';
1229                                                 } elseif ( $state === 'both' ) {
1230                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1231                                                 } else { # $state can be 'i' or ''
1232                                                         $output .= '<b>'; $state .= 'b';
1233                                                 }
1234                                         } elseif ( strlen( $r ) == 5 ) {
1235                                                 if ( $state === 'b' ) {
1236                                                         $output .= '</b><i>'; $state = 'i';
1237                                                 } elseif ( $state === 'i' ) {
1238                                                         $output .= '</i><b>'; $state = 'b';
1239                                                 } elseif ( $state === 'bi' ) {
1240                                                         $output .= '</i></b>'; $state = '';
1241                                                 } elseif ( $state === 'ib' ) {
1242                                                         $output .= '</b></i>'; $state = '';
1243                                                 } elseif ( $state === 'both' ) {
1244                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1245                                                 } else { # ($state == '')
1246                                                         $buffer = ''; $state = 'both';
1247                                                 }
1248                                         }
1249                                 }
1250                                 $i++;
1251                         }
1252                         # Now close all remaining tags.  Notice that the order is important.
1253                         if ( $state === 'b' || $state === 'ib' ) {
1254                                 $output .= '</b>';
1255                         }
1256                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1257                                 $output .= '</i>';
1258                         }
1259                         if ( $state === 'bi' ) {
1260                                 $output .= '</b>';
1261                         }
1262                         # There might be lonely ''''', so make sure we have a buffer
1263                         if ( $state === 'both' && $buffer ) {
1264                                 $output .= '<b><i>'.$buffer.'</i></b>';
1265                         }
1266                         return $output;
1267                 }
1268         }
1269
1270         /**
1271          * Replace external links (REL)
1272          *
1273          * Note: this is all very hackish and the order of execution matters a lot.
1274          * Make sure to run maintenance/parserTests.php if you change this code.
1275          *
1276          * @private
1277          */
1278         function replaceExternalLinks( $text ) {
1279                 global $wgContLang;
1280                 wfProfileIn( __METHOD__ );
1281
1282                 $sk = $this->mOptions->getSkin();
1283
1284                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1285                 $s = array_shift( $bits );
1286
1287                 $i = 0;
1288                 while ( $i<count( $bits ) ) {
1289                         $url = $bits[$i++];
1290                         $protocol = $bits[$i++];
1291                         $text = $bits[$i++];
1292                         $trail = $bits[$i++];
1293
1294                         # The characters '<' and '>' (which were escaped by
1295                         # removeHTMLtags()) should not be included in
1296                         # URLs, per RFC 2396.
1297                         $m2 = array();
1298                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1299                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1300                                 $url = substr( $url, 0, $m2[0][1] );
1301                         }
1302
1303                         # If the link text is an image URL, replace it with an <img> tag
1304                         # This happened by accident in the original parser, but some people used it extensively
1305                         $img = $this->maybeMakeExternalImage( $text );
1306                         if ( $img !== false ) {
1307                                 $text = $img;
1308                         }
1309
1310                         $dtrail = '';
1311
1312                         # Set linktype for CSS - if URL==text, link is essentially free
1313                         $linktype = ( $text === $url ) ? 'free' : 'text';
1314
1315                         # No link text, e.g. [http://domain.tld/some.link]
1316                         if ( $text == '' ) {
1317                                 # Autonumber if allowed. See bug #5918
1318                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1319                                         $langObj = $this->getFunctionLang();
1320                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1321                                         $linktype = 'autonumber';
1322                                 } else {
1323                                         # Otherwise just use the URL
1324                                         $text = htmlspecialchars( $url );
1325                                         $linktype = 'free';
1326                                 }
1327                         } else {
1328                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1329                                 # Check for trail
1330                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1331                         }
1332
1333                         $text = $wgContLang->markNoConversion( $text );
1334
1335                         $url = Sanitizer::cleanUrl( $url );
1336
1337                         # Use the encoded URL
1338                         # This means that users can paste URLs directly into the text
1339                         # Funny characters like &ouml; aren't valid in URLs anyway
1340                         # This was changed in August 2004
1341                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1342                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1343
1344                         # Register link in the output object.
1345                         # Replace unnecessary URL escape codes with the referenced character
1346                         # This prevents spammers from hiding links from the filters
1347                         $pasteurized = self::replaceUnusualEscapes( $url );
1348                         $this->mOutput->addExternalLink( $pasteurized );
1349                 }
1350
1351                 wfProfileOut( __METHOD__ );
1352                 return $s;
1353         }
1354
1355         /**
1356          * Get an associative array of additional HTML attributes appropriate for a
1357          * particular external link.  This currently may include rel => nofollow
1358          * (depending on configuration, namespace, and the URL's domain) and/or a
1359          * target attribute (depending on configuration).
1360          *
1361          * @param string $url Optional URL, to extract the domain from for rel =>
1362          *   nofollow if appropriate
1363          * @return array Associative array of HTML attributes
1364          */
1365         function getExternalLinkAttribs( $url = false ) {
1366                 $attribs = array();
1367                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1368                 $ns = $this->mTitle->getNamespace();
1369                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1370                         $attribs['rel'] = 'nofollow';
1371
1372                         global $wgNoFollowDomainExceptions;
1373                         if ( $wgNoFollowDomainExceptions ) {
1374                                 $bits = wfParseUrl( $url );
1375                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1376                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1377                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1378                                                         unset( $attribs['rel'] );
1379                                                         break;
1380                                                 }
1381                                         }
1382                                 }
1383                         }
1384                 }
1385                 if ( $this->mOptions->getExternalLinkTarget() ) {
1386                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1387                 }
1388                 return $attribs;
1389         }
1390
1391
1392         /**
1393          * Replace unusual URL escape codes with their equivalent characters
1394          * @param string
1395          * @return string
1396          * @static
1397          * @todo  This can merge genuinely required bits in the path or query string,
1398          *        breaking legit URLs. A proper fix would treat the various parts of
1399          *        the URL differently; as a workaround, just use the output for
1400          *        statistical records, not for actual linking/output.
1401          */
1402         static function replaceUnusualEscapes( $url ) {
1403                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1404                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1405         }
1406
1407         /**
1408          * Callback function used in replaceUnusualEscapes().
1409          * Replaces unusual URL escape codes with their equivalent character
1410          * @static
1411          * @private
1412          */
1413         private static function replaceUnusualEscapesCallback( $matches ) {
1414                 $char = urldecode( $matches[0] );
1415                 $ord = ord( $char );
1416                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1417                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1418                         # No, shouldn't be escaped
1419                         return $char;
1420                 } else {
1421                         # Yes, leave it escaped
1422                         return $matches[0];
1423                 }
1424         }
1425
1426         /**
1427          * make an image if it's allowed, either through the global
1428          * option, through the exception, or through the on-wiki whitelist
1429          * @private
1430          */
1431         function maybeMakeExternalImage( $url ) {
1432                 $sk = $this->mOptions->getSkin();
1433                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1434                 $imagesexception = !empty( $imagesfrom );
1435                 $text = false;
1436                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1437                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1438                         $imagematch = false;
1439                         foreach ( $imagesfrom as $match ) {
1440                                 if ( strpos( $url, $match ) === 0 ) {
1441                                         $imagematch = true;
1442                                         break;
1443                                 }
1444                         }
1445                 } elseif ( $imagesexception ) {
1446                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1447                 } else {
1448                         $imagematch = false;
1449                 }
1450                 if ( $this->mOptions->getAllowExternalImages()
1451                      || ( $imagesexception && $imagematch ) ) {
1452                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1453                                 # Image found
1454                                 $text = $sk->makeExternalImage( $url );
1455                         }
1456                 }
1457                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1458                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1459                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1460                         foreach ( $whitelist as $entry ) {
1461                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1462                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1463                                         continue;
1464                                 }
1465                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1466                                         # Image matches a whitelist entry
1467                                         $text = $sk->makeExternalImage( $url );
1468                                         break;
1469                                 }
1470                         }
1471                 }
1472                 return $text;
1473         }
1474
1475         /**
1476          * Process [[ ]] wikilinks
1477          * @return processed text
1478          *
1479          * @private
1480          */
1481         function replaceInternalLinks( $s ) {
1482                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1483                 return $s;
1484         }
1485
1486         /**
1487          * Process [[ ]] wikilinks (RIL)
1488          * @return LinkHolderArray
1489          *
1490          * @private
1491          */
1492         function replaceInternalLinks2( &$s ) {
1493                 global $wgContLang;
1494
1495                 wfProfileIn( __METHOD__ );
1496
1497                 wfProfileIn( __METHOD__.'-setup' );
1498                 static $tc = FALSE, $e1, $e1_img;
1499                 # the % is needed to support urlencoded titles as well
1500                 if ( !$tc ) {
1501                         $tc = Title::legalChars() . '#%';
1502                         # Match a link having the form [[namespace:link|alternate]]trail
1503                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1504                         # Match cases where there is no "]]", which might still be images
1505                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1506                 }
1507
1508                 $sk = $this->mOptions->getSkin();
1509                 $holders = new LinkHolderArray( $this );
1510
1511                 # split the entire text string on occurences of [[
1512                 $a = StringUtils::explode( '[[', ' ' . $s );
1513                 # get the first element (all text up to first [[), and remove the space we added
1514                 $s = $a->current();
1515                 $a->next();
1516                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1517                 $s = substr( $s, 1 );
1518
1519                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1520                 $e2 = null;
1521                 if ( $useLinkPrefixExtension ) {
1522                         # Match the end of a line for a word that's not followed by whitespace,
1523                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1524                         $e2 = wfMsgForContent( 'linkprefix' );
1525                 }
1526
1527                 if ( is_null( $this->mTitle ) ) {
1528                         wfProfileOut( __METHOD__.'-setup' );
1529                         wfProfileOut( __METHOD__ );
1530                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1531                 }
1532                 $nottalk = !$this->mTitle->isTalkPage();
1533
1534                 if ( $useLinkPrefixExtension ) {
1535                         $m = array();
1536                         if ( preg_match( $e2, $s, $m ) ) {
1537                                 $first_prefix = $m[2];
1538                         } else {
1539                                 $first_prefix = false;
1540                         }
1541                 } else {
1542                         $prefix = '';
1543                 }
1544
1545                 if ( $wgContLang->hasVariants() ) {
1546                         $selflink = $wgContLang->convertLinkToAllVariants( $this->mTitle->getPrefixedText() );
1547                 } else {
1548                         $selflink = array( $this->mTitle->getPrefixedText() );
1549                 }
1550                 $useSubpages = $this->areSubpagesAllowed();
1551                 wfProfileOut( __METHOD__.'-setup' );
1552
1553                 # Loop for each link
1554                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1555                         # Check for excessive memory usage
1556                         if ( $holders->isBig() ) {
1557                                 # Too big
1558                                 # Do the existence check, replace the link holders and clear the array
1559                                 $holders->replace( $s );
1560                                 $holders->clear();
1561                         }
1562
1563                         if ( $useLinkPrefixExtension ) {
1564                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1565                                 if ( preg_match( $e2, $s, $m ) ) {
1566                                         $prefix = $m[2];
1567                                         $s = $m[1];
1568                                 } else {
1569                                         $prefix='';
1570                                 }
1571                                 # first link
1572                                 if ( $first_prefix ) {
1573                                         $prefix = $first_prefix;
1574                                         $first_prefix = false;
1575                                 }
1576                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1577                         }
1578
1579                         $might_be_img = false;
1580
1581                         wfProfileIn( __METHOD__."-e1" );
1582                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1583                                 $text = $m[2];
1584                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1585                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1586                                 # the real problem is with the $e1 regex
1587                                 # See bug 1300.
1588                                 #
1589                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1590                                 # and no image is in sight. See bug 2095.
1591                                 #
1592                                 if ( $text !== '' &&
1593                                         substr( $m[3], 0, 1 ) === ']' &&
1594                                         strpos( $text, '[' ) !== false
1595                                 )
1596                                 {
1597                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1598                                         $m[3] = substr( $m[3], 1 );
1599                                 }
1600                                 # fix up urlencoded title texts
1601                                 if ( strpos( $m[1], '%' ) !== false ) {
1602                                         # Should anchors '#' also be rejected?
1603                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode( $m[1] ) );
1604                                 }
1605                                 $trail = $m[3];
1606                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1607                                 $might_be_img = true;
1608                                 $text = $m[2];
1609                                 if ( strpos( $m[1], '%' ) !== false ) {
1610                                         $m[1] = urldecode( $m[1] );
1611                                 }
1612                                 $trail = "";
1613                         } else { # Invalid form; output directly
1614                                 $s .= $prefix . '[[' . $line ;
1615                                 wfProfileOut( __METHOD__."-e1" );
1616                                 continue;
1617                         }
1618                         wfProfileOut( __METHOD__."-e1" );
1619                         wfProfileIn( __METHOD__."-misc" );
1620
1621                         # Don't allow internal links to pages containing
1622                         # PROTO: where PROTO is a valid URL protocol; these
1623                         # should be external links.
1624                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1625                                 $s .= $prefix . '[[' . $line ;
1626                                 wfProfileOut( __METHOD__."-misc" );
1627                                 continue;
1628                         }
1629
1630                         # Make subpage if necessary
1631                         if ( $useSubpages ) {
1632                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1633                         } else {
1634                                 $link = $m[1];
1635                         }
1636
1637                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1638                         if ( !$noforce ) {
1639                                 # Strip off leading ':'
1640                                 $link = substr( $link, 1 );
1641                         }
1642
1643                         wfProfileOut( __METHOD__."-misc" );
1644                         wfProfileIn( __METHOD__."-title" );
1645                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1646                         if ( $nt === null ) {
1647                                 $s .= $prefix . '[[' . $line;
1648                                 wfProfileOut( __METHOD__."-title" );
1649                                 continue;
1650                         }
1651
1652                         $ns = $nt->getNamespace();
1653                         $iw = $nt->getInterWiki();
1654                         wfProfileOut( __METHOD__."-title" );
1655
1656                         if ( $might_be_img ) { # if this is actually an invalid link
1657                                 wfProfileIn( __METHOD__."-might_be_img" );
1658                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1659                                         $found = false;
1660                                         while ( true ) {
1661                                                 # look at the next 'line' to see if we can close it there
1662                                                 $a->next();
1663                                                 $next_line = $a->current();
1664                                                 if ( $next_line === false || $next_line === null ) {
1665                                                         break;
1666                                                 }
1667                                                 $m = explode( ']]', $next_line, 3 );
1668                                                 if ( count( $m ) == 3 ) {
1669                                                         # the first ]] closes the inner link, the second the image
1670                                                         $found = true;
1671                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1672                                                         $trail = $m[2];
1673                                                         break;
1674                                                 } elseif ( count( $m ) == 2 ) {
1675                                                         # if there's exactly one ]] that's fine, we'll keep looking
1676                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1677                                                 } else {
1678                                                         # if $next_line is invalid too, we need look no further
1679                                                         $text .= '[[' . $next_line;
1680                                                         break;
1681                                                 }
1682                                         }
1683                                         if ( !$found ) {
1684                                                 # we couldn't find the end of this imageLink, so output it raw
1685                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1686                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1687                                                 $s .= "{$prefix}[[$link|$text";
1688                                                 # note: no $trail, because without an end, there *is* no trail
1689                                                 wfProfileOut( __METHOD__."-might_be_img" );
1690                                                 continue;
1691                                         }
1692                                 } else { # it's not an image, so output it raw
1693                                         $s .= "{$prefix}[[$link|$text";
1694                                         # note: no $trail, because without an end, there *is* no trail
1695                                         wfProfileOut( __METHOD__."-might_be_img" );
1696                                         continue;
1697                                 }
1698                                 wfProfileOut( __METHOD__."-might_be_img" );
1699                         }
1700
1701                         $wasblank = ( $text  == '' );
1702                         if ( $wasblank ) {
1703                                 $text = $link;
1704                         }
1705
1706                         # Link not escaped by : , create the various objects
1707                         if ( $noforce ) {
1708
1709                                 # Interwikis
1710                                 wfProfileIn( __METHOD__."-interwiki" );
1711                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1712                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1713                                         $s = rtrim( $s . $prefix );
1714                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1715                                         wfProfileOut( __METHOD__."-interwiki" );
1716                                         continue;
1717                                 }
1718                                 wfProfileOut( __METHOD__."-interwiki" );
1719
1720                                 if ( $ns == NS_FILE ) {
1721                                         wfProfileIn( __METHOD__."-image" );
1722                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1723                                                 if ( $wasblank ) {
1724                                                         # if no parameters were passed, $text
1725                                                         # becomes something like "File:Foo.png",
1726                                                         # which we don't want to pass on to the
1727                                                         # image generator
1728                                                         $text = '';
1729                                                 } else {
1730                                                         # recursively parse links inside the image caption
1731                                                         # actually, this will parse them in any other parameters, too,
1732                                                         # but it might be hard to fix that, and it doesn't matter ATM
1733                                                         $text = $this->replaceExternalLinks( $text );
1734                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1735                                                 }
1736                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1737                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1738                                         } else {
1739                                                 $s .= $prefix . $trail;
1740                                         }
1741                                         $this->mOutput->addImage( $nt->getDBkey() );
1742                                         wfProfileOut( __METHOD__."-image" );
1743                                         continue;
1744
1745                                 }
1746
1747                                 if ( $ns == NS_CATEGORY ) {
1748                                         wfProfileIn( __METHOD__."-category" );
1749                                         $s = rtrim( $s . "\n" ); # bug 87
1750
1751                                         if ( $wasblank ) {
1752                                                 $sortkey = $this->getDefaultSort();
1753                                         } else {
1754                                                 $sortkey = $text;
1755                                         }
1756                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1757                                         $sortkey = str_replace( "\n", '', $sortkey );
1758                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1759                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1760
1761                                         /**
1762                                          * Strip the whitespace Category links produce, see bug 87
1763                                          * @todo We might want to use trim($tmp, "\n") here.
1764                                          */
1765                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '': $prefix . $trail;
1766
1767                                         wfProfileOut( __METHOD__."-category" );
1768                                         continue;
1769                                 }
1770                         }
1771
1772                         # Self-link checking
1773                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1774                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1775                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1776                                         continue;
1777                                 }
1778                         }
1779
1780                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1781                         # FIXME: Should do batch file existence checks, see comment below
1782                         if ( $ns == NS_MEDIA ) {
1783                                 wfProfileIn( __METHOD__."-media" );
1784                                 # Give extensions a chance to select the file revision for us
1785                                 $skip = $time = false;
1786                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1787                                 if ( $skip ) {
1788                                         $link = $sk->link( $nt );
1789                                 } else {
1790                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1791                                 }
1792                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1793                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1794                                 $this->mOutput->addImage( $nt->getDBkey() );
1795                                 wfProfileOut( __METHOD__."-media" );
1796                                 continue;
1797                         }
1798
1799                         wfProfileIn( __METHOD__."-always_known" );
1800                         # Some titles, such as valid special pages or files in foreign repos, should
1801                         # be shown as bluelinks even though they're not included in the page table
1802                         #
1803                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1804                         # batch file existence checks for NS_FILE and NS_MEDIA
1805                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1806                                 $this->mOutput->addLink( $nt );
1807                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1808                         } else {
1809                                 # Links will be added to the output link list after checking
1810                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1811                         }
1812                         wfProfileOut( __METHOD__."-always_known" );
1813                 }
1814                 wfProfileOut( __METHOD__ );
1815                 return $holders;
1816         }
1817
1818         /**
1819          * Make a link placeholder. The text returned can be later resolved to a real link with
1820          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1821          * parsing of interwiki links, and secondly to allow all existence checks and
1822          * article length checks (for stub links) to be bundled into a single query.
1823          *
1824          * @deprecated
1825          */
1826         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1827                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1828         }
1829
1830         /**
1831          * Render a forced-blue link inline; protect against double expansion of
1832          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1833          * Since this little disaster has to split off the trail text to avoid
1834          * breaking URLs in the following text without breaking trails on the
1835          * wiki links, it's been made into a horrible function.
1836          *
1837          * @param Title $nt
1838          * @param string $text
1839          * @param string $query
1840          * @param string $trail
1841          * @param string $prefix
1842          * @return string HTML-wikitext mix oh yuck
1843          */
1844         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1845                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1846                 $sk = $this->mOptions->getSkin();
1847                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1848                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1849                 return $this->armorLinks( $link ) . $trail;
1850         }
1851
1852         /**
1853          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1854          * going to go through further parsing steps before inline URL expansion.
1855          *
1856          * Not needed quite as much as it used to be since free links are a bit
1857          * more sensible these days. But bracketed links are still an issue.
1858          *
1859          * @param string more-or-less HTML
1860          * @return string less-or-more HTML with NOPARSE bits
1861          */
1862         function armorLinks( $text ) {
1863                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1864                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1865         }
1866
1867         /**
1868          * Return true if subpage links should be expanded on this page.
1869          * @return bool
1870          */
1871         function areSubpagesAllowed() {
1872                 # Some namespaces don't allow subpages
1873                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1874         }
1875
1876         /**
1877          * Handle link to subpage if necessary
1878          * @param string $target the source of the link
1879          * @param string &$text the link text, modified as necessary
1880          * @return string the full name of the link
1881          * @private
1882          */
1883         function maybeDoSubpageLink( $target, &$text ) {
1884                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1885         }
1886
1887         /**#@+
1888          * Used by doBlockLevels()
1889          * @private
1890          */
1891         function closeParagraph() {
1892                 $result = '';
1893                 if ( $this->mLastSection != '' ) {
1894                         $result = '</' . $this->mLastSection  . ">\n";
1895                 }
1896                 $this->mInPre = false;
1897                 $this->mLastSection = '';
1898                 return $result;
1899         }
1900         /**
1901          * getCommon() returns the length of the longest common substring
1902          * of both arguments, starting at the beginning of both.
1903          * @private
1904          */
1905         function getCommon( $st1, $st2 ) {
1906                 $fl = strlen( $st1 );
1907                 $shorter = strlen( $st2 );
1908                 if ( $fl < $shorter ) {
1909                         $shorter = $fl;
1910                 }
1911
1912                 for ( $i = 0; $i < $shorter; ++$i ) {
1913                         if ( $st1{$i} != $st2{$i} ) {
1914                                 break;
1915                         }
1916                 }
1917                 return $i;
1918         }
1919         /**
1920          * These next three functions open, continue, and close the list
1921          * element appropriate to the prefix character passed into them.
1922          * @private
1923          */
1924         function openList( $char ) {
1925                 $result = $this->closeParagraph();
1926
1927                 if ( '*' === $char ) {
1928                         $result .= '<ul><li>';
1929                 } elseif ( '#' === $char ) {
1930                         $result .= '<ol><li>';
1931                 } elseif ( ':' === $char ) {
1932                         $result .= '<dl><dd>';
1933                 } elseif ( ';' === $char ) {
1934                         $result .= '<dl><dt>';
1935                         $this->mDTopen = true;
1936                 } else {
1937                         $result = '<!-- ERR 1 -->';
1938                 }
1939
1940                 return $result;
1941         }
1942
1943         /**
1944          * TODO: document
1945          * @param $char String
1946          * @private
1947          */
1948         function nextItem( $char ) {
1949                 if ( '*' === $char || '#' === $char ) {
1950                         return '</li><li>';
1951                 } elseif ( ':' === $char || ';' === $char ) {
1952                         $close = '</dd>';
1953                         if ( $this->mDTopen ) {
1954                                 $close = '</dt>';
1955                         }
1956                         if ( ';' === $char ) {
1957                                 $this->mDTopen = true;
1958                                 return $close . '<dt>';
1959                         } else {
1960                                 $this->mDTopen = false;
1961                                 return $close . '<dd>';
1962                         }
1963                 }
1964                 return '<!-- ERR 2 -->';
1965         }
1966
1967         /**
1968          * TODO: document
1969          * @param $char String
1970          * @private
1971          */
1972         function closeList( $char ) {
1973                 if ( '*' === $char ) {
1974                         $text = '</li></ul>';
1975                 } elseif ( '#' === $char ) {
1976                         $text = '</li></ol>';
1977                 } elseif ( ':' === $char ) {
1978                         if ( $this->mDTopen ) {
1979                                 $this->mDTopen = false;
1980                                 $text = '</dt></dl>';
1981                         } else {
1982                                 $text = '</dd></dl>';
1983                         }
1984                 } else {
1985                         return '<!-- ERR 3 -->';
1986                 }
1987                 return $text."\n";
1988         }
1989         /**#@-*/
1990
1991         /**
1992          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1993          *
1994          * @param $linestart bool whether or not this is at the start of a line.
1995          * @private
1996          * @return string the lists rendered as HTML
1997          */
1998         function doBlockLevels( $text, $linestart ) {
1999                 wfProfileIn( __METHOD__ );
2000
2001                 # Parsing through the text line by line.  The main thing
2002                 # happening here is handling of block-level elements p, pre,
2003                 # and making lists from lines starting with * # : etc.
2004                 #
2005                 $textLines = StringUtils::explode( "\n", $text );
2006
2007                 $lastPrefix = $output = '';
2008                 $this->mDTopen = $inBlockElem = false;
2009                 $prefixLength = 0;
2010                 $paragraphStack = false;
2011
2012                 foreach ( $textLines as $oLine ) {
2013                         # Fix up $linestart
2014                         if ( !$linestart ) {
2015                                 $output .= $oLine;
2016                                 $linestart = true;
2017                                 continue;
2018                         }
2019                         # * = ul
2020                         # # = ol
2021                         # ; = dt
2022                         # : = dd
2023
2024                         $lastPrefixLength = strlen( $lastPrefix );
2025                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2026                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2027                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2028                         if ( !$this->mInPre ) {
2029                                 # Multiple prefixes may abut each other for nested lists.
2030                                 $prefixLength = strspn( $oLine, '*#:;' );
2031                                 $prefix = substr( $oLine, 0, $prefixLength );
2032
2033                                 # eh?
2034                                 # ; and : are both from definition-lists, so they're equivalent
2035                                 #  for the purposes of determining whether or not we need to open/close
2036                                 #  elements.
2037                                 $prefix2 = str_replace( ';', ':', $prefix );
2038                                 $t = substr( $oLine, $prefixLength );
2039                                 $this->mInPre = (bool)$preOpenMatch;
2040                         } else {
2041                                 # Don't interpret any other prefixes in preformatted text
2042                                 $prefixLength = 0;
2043                                 $prefix = $prefix2 = '';
2044                                 $t = $oLine;
2045                         }
2046
2047                         # List generation
2048                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2049                                 # Same as the last item, so no need to deal with nesting or opening stuff
2050                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2051                                 $paragraphStack = false;
2052
2053                                 if ( substr( $prefix, -1 ) === ';') {
2054                                         # The one nasty exception: definition lists work like this:
2055                                         # ; title : definition text
2056                                         # So we check for : in the remainder text to split up the
2057                                         # title and definition, without b0rking links.
2058                                         $term = $t2 = '';
2059                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2060                                                 $t = $t2;
2061                                                 $output .= $term . $this->nextItem( ':' );
2062                                         }
2063                                 }
2064                         } elseif ( $prefixLength || $lastPrefixLength ) {
2065                                 # We need to open or close prefixes, or both.
2066
2067                                 # Either open or close a level...
2068                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2069                                 $paragraphStack = false;
2070
2071                                 # Close all the prefixes which aren't shared.
2072                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2073                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2074                                         --$lastPrefixLength;
2075                                 }
2076
2077                                 # Continue the current prefix if appropriate.
2078                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2079                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2080                                 }
2081
2082                                 # Open prefixes where appropriate.
2083                                 while ( $prefixLength > $commonPrefixLength ) {
2084                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2085                                         $output .= $this->openList( $char );
2086
2087                                         if ( ';' === $char ) {
2088                                                 # FIXME: This is dupe of code above
2089                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2090                                                         $t = $t2;
2091                                                         $output .= $term . $this->nextItem( ':' );
2092                                                 }
2093                                         }
2094                                         ++$commonPrefixLength;
2095                                 }
2096                                 $lastPrefix = $prefix2;
2097                         }
2098
2099                         # If we have no prefixes, go to paragraph mode.
2100                         if ( 0 == $prefixLength ) {
2101                                 wfProfileIn( __METHOD__."-paragraph" );
2102                                 # No prefix (not in list)--go to paragraph mode
2103                                 # XXX: use a stack for nestable elements like span, table and div
2104                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2105                                 $closematch = preg_match(
2106                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2107                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2108                                 if ( $openmatch or $closematch ) {
2109                                         $paragraphStack = false;
2110                                         # TODO bug 5718: paragraph closed
2111                                         $output .= $this->closeParagraph();
2112                                         if ( $preOpenMatch and !$preCloseMatch ) {
2113                                                 $this->mInPre = true;
2114                                         }
2115                                         if ( $closematch ) {
2116                                                 $inBlockElem = false;
2117                                         } else {
2118                                                 $inBlockElem = true;
2119                                         }
2120                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2121                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2122                                                 # pre
2123                                                 if ( $this->mLastSection !== 'pre' ) {
2124                                                         $paragraphStack = false;
2125                                                         $output .= $this->closeParagraph().'<pre>';
2126                                                         $this->mLastSection = 'pre';
2127                                                 }
2128                                                 $t = substr( $t, 1 );
2129                                         } else {
2130                                                 # paragraph
2131                                                 if ( trim( $t ) == '' ) {
2132                                                         if ( $paragraphStack ) {
2133                                                                 $output .= $paragraphStack.'<br />';
2134                                                                 $paragraphStack = false;
2135                                                                 $this->mLastSection = 'p';
2136                                                         } else {
2137                                                                 if ( $this->mLastSection !== 'p' ) {
2138                                                                         $output .= $this->closeParagraph();
2139                                                                         $this->mLastSection = '';
2140                                                                         $paragraphStack = '<p>';
2141                                                                 } else {
2142                                                                         $paragraphStack = '</p><p>';
2143                                                                 }
2144                                                         }
2145                                                 } else {
2146                                                         if ( $paragraphStack ) {
2147                                                                 $output .= $paragraphStack;
2148                                                                 $paragraphStack = false;
2149                                                                 $this->mLastSection = 'p';
2150                                                         } elseif ( $this->mLastSection !== 'p' ) {
2151                                                                 $output .= $this->closeParagraph().'<p>';
2152                                                                 $this->mLastSection = 'p';
2153                                                         }
2154                                                 }
2155                                         }
2156                                 }
2157                                 wfProfileOut( __METHOD__."-paragraph" );
2158                         }
2159                         # somewhere above we forget to get out of pre block (bug 785)
2160                         if ( $preCloseMatch && $this->mInPre ) {
2161                                 $this->mInPre = false;
2162                         }
2163                         if ( $paragraphStack === false ) {
2164                                 $output .= $t."\n";
2165                         }
2166                 }
2167                 while ( $prefixLength ) {
2168                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2169                         --$prefixLength;
2170                 }
2171                 if ( $this->mLastSection != '' ) {
2172                         $output .= '</' . $this->mLastSection . '>';
2173                         $this->mLastSection = '';
2174                 }
2175
2176                 wfProfileOut( __METHOD__ );
2177                 return $output;
2178         }
2179
2180         /**
2181          * Split up a string on ':', ignoring any occurences inside tags
2182          * to prevent illegal overlapping.
2183          * @param string $str the string to split
2184          * @param string &$before set to everything before the ':'
2185          * @param string &$after set to everything after the ':'
2186          * return string the position of the ':', or false if none found
2187          */
2188         function findColonNoLinks( $str, &$before, &$after ) {
2189                 wfProfileIn( __METHOD__ );
2190
2191                 $pos = strpos( $str, ':' );
2192                 if ( $pos === false ) {
2193                         # Nothing to find!
2194                         wfProfileOut( __METHOD__ );
2195                         return false;
2196                 }
2197
2198                 $lt = strpos( $str, '<' );
2199                 if ( $lt === false || $lt > $pos ) {
2200                         # Easy; no tag nesting to worry about
2201                         $before = substr( $str, 0, $pos );
2202                         $after = substr( $str, $pos+1 );
2203                         wfProfileOut( __METHOD__ );
2204                         return $pos;
2205                 }
2206
2207                 # Ugly state machine to walk through avoiding tags.
2208                 $state = self::COLON_STATE_TEXT;
2209                 $stack = 0;
2210                 $len = strlen( $str );
2211                 for( $i = 0; $i < $len; $i++ ) {
2212                         $c = $str{$i};
2213
2214                         switch( $state ) {
2215                         # (Using the number is a performance hack for common cases)
2216                         case 0: # self::COLON_STATE_TEXT:
2217                                 switch( $c ) {
2218                                 case "<":
2219                                         # Could be either a <start> tag or an </end> tag
2220                                         $state = self::COLON_STATE_TAGSTART;
2221                                         break;
2222                                 case ":":
2223                                         if ( $stack == 0 ) {
2224                                                 # We found it!
2225                                                 $before = substr( $str, 0, $i );
2226                                                 $after = substr( $str, $i + 1 );
2227                                                 wfProfileOut( __METHOD__ );
2228                                                 return $i;
2229                                         }
2230                                         # Embedded in a tag; don't break it.
2231                                         break;
2232                                 default:
2233                                         # Skip ahead looking for something interesting
2234                                         $colon = strpos( $str, ':', $i );
2235                                         if ( $colon === false ) {
2236                                                 # Nothing else interesting
2237                                                 wfProfileOut( __METHOD__ );
2238                                                 return false;
2239                                         }
2240                                         $lt = strpos( $str, '<', $i );
2241                                         if ( $stack === 0 ) {
2242                                                 if ( $lt === false || $colon < $lt ) {
2243                                                         # We found it!
2244                                                         $before = substr( $str, 0, $colon );
2245                                                         $after = substr( $str, $colon + 1 );
2246                                                         wfProfileOut( __METHOD__ );
2247                                                         return $i;
2248                                                 }
2249                                         }
2250                                         if ( $lt === false ) {
2251                                                 # Nothing else interesting to find; abort!
2252                                                 # We're nested, but there's no close tags left. Abort!
2253                                                 break 2;
2254                                         }
2255                                         # Skip ahead to next tag start
2256                                         $i = $lt;
2257                                         $state = self::COLON_STATE_TAGSTART;
2258                                 }
2259                                 break;
2260                         case 1: # self::COLON_STATE_TAG:
2261                                 # In a <tag>
2262                                 switch( $c ) {
2263                                 case ">":
2264                                         $stack++;
2265                                         $state = self::COLON_STATE_TEXT;
2266                                         break;
2267                                 case "/":
2268                                         # Slash may be followed by >?
2269                                         $state = self::COLON_STATE_TAGSLASH;
2270                                         break;
2271                                 default:
2272                                         # ignore
2273                                 }
2274                                 break;
2275                         case 2: # self::COLON_STATE_TAGSTART:
2276                                 switch( $c ) {
2277                                 case "/":
2278                                         $state = self::COLON_STATE_CLOSETAG;
2279                                         break;
2280                                 case "!":
2281                                         $state = self::COLON_STATE_COMMENT;
2282                                         break;
2283                                 case ">":
2284                                         # Illegal early close? This shouldn't happen D:
2285                                         $state = self::COLON_STATE_TEXT;
2286                                         break;
2287                                 default:
2288                                         $state = self::COLON_STATE_TAG;
2289                                 }
2290                                 break;
2291                         case 3: # self::COLON_STATE_CLOSETAG:
2292                                 # In a </tag>
2293                                 if ( $c === ">" ) {
2294                                         $stack--;
2295                                         if ( $stack < 0 ) {
2296                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2297                                                 wfProfileOut( __METHOD__ );
2298                                                 return false;
2299                                         }
2300                                         $state = self::COLON_STATE_TEXT;
2301                                 }
2302                                 break;
2303                         case self::COLON_STATE_TAGSLASH:
2304                                 if ( $c === ">" ) {
2305                                         # Yes, a self-closed tag <blah/>
2306                                         $state = self::COLON_STATE_TEXT;
2307                                 } else {
2308                                         # Probably we're jumping the gun, and this is an attribute
2309                                         $state = self::COLON_STATE_TAG;
2310                                 }
2311                                 break;
2312                         case 5: # self::COLON_STATE_COMMENT:
2313                                 if ( $c === "-" ) {
2314                                         $state = self::COLON_STATE_COMMENTDASH;
2315                                 }
2316                                 break;
2317                         case self::COLON_STATE_COMMENTDASH:
2318                                 if ( $c === "-" ) {
2319                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2320                                 } else {
2321                                         $state = self::COLON_STATE_COMMENT;
2322                                 }
2323                                 break;
2324                         case self::COLON_STATE_COMMENTDASHDASH:
2325                                 if ( $c === ">" ) {
2326                                         $state = self::COLON_STATE_TEXT;
2327                                 } else {
2328                                         $state = self::COLON_STATE_COMMENT;
2329                                 }
2330                                 break;
2331                         default:
2332                                 throw new MWException( "State machine error in " . __METHOD__ );
2333                         }
2334                 }
2335                 if ( $stack > 0 ) {
2336                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2337                         return false;
2338                 }
2339                 wfProfileOut( __METHOD__ );
2340                 return false;
2341         }
2342
2343         /**
2344          * Return value of a magic variable (like PAGENAME)
2345          *
2346          * @private
2347          */
2348         function getVariableValue( $index, $frame=false ) {
2349                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2350                 global $wgScriptPath, $wgStylePath;
2351
2352                 /**
2353                  * Some of these require message or data lookups and can be
2354                  * expensive to check many times.
2355                  */
2356                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2357                         if ( isset( $this->mVarCache[$index] ) ) {
2358                                 return $this->mVarCache[$index];
2359                         }
2360                 }
2361
2362                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2363                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2364
2365                 # Use the time zone
2366                 global $wgLocaltimezone;
2367                 if ( isset( $wgLocaltimezone ) ) {
2368                         $oldtz = date_default_timezone_get();
2369                         date_default_timezone_set( $wgLocaltimezone );
2370                 }
2371
2372                 $localTimestamp = date( 'YmdHis', $ts );
2373                 $localMonth = date( 'm', $ts );
2374                 $localMonth1 = date( 'n', $ts );
2375                 $localMonthName = date( 'n', $ts );
2376                 $localDay = date( 'j', $ts );
2377                 $localDay2 = date( 'd', $ts );
2378                 $localDayOfWeek = date( 'w', $ts );
2379                 $localWeek = date( 'W', $ts );
2380                 $localYear = date( 'Y', $ts );
2381                 $localHour = date( 'H', $ts );
2382                 if ( isset( $wgLocaltimezone ) ) {
2383                         date_default_timezone_set( $oldtz );
2384                 }
2385
2386                 switch ( $index ) {
2387                         case 'currentmonth':
2388                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2389                                 break;
2390                         case 'currentmonth1':
2391                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2392                                 break;
2393                         case 'currentmonthname':
2394                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2395                                 break;
2396                         case 'currentmonthnamegen':
2397                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2398                                 break;
2399                         case 'currentmonthabbrev':
2400                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2401                                 break;
2402                         case 'currentday':
2403                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2404                                 break;
2405                         case 'currentday2':
2406                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2407                                 break;
2408                         case 'localmonth':
2409                                 $value = $wgContLang->formatNum( $localMonth );
2410                                 break;
2411                         case 'localmonth1':
2412                                 $value = $wgContLang->formatNum( $localMonth1 );
2413                                 break;
2414                         case 'localmonthname':
2415                                 $value = $wgContLang->getMonthName( $localMonthName );
2416                                 break;
2417                         case 'localmonthnamegen':
2418                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2419                                 break;
2420                         case 'localmonthabbrev':
2421                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2422                                 break;
2423                         case 'localday':
2424                                 $value = $wgContLang->formatNum( $localDay );
2425                                 break;
2426                         case 'localday2':
2427                                 $value = $wgContLang->formatNum( $localDay2 );
2428                                 break;
2429                         case 'pagename':
2430                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2431                                 break;
2432                         case 'pagenamee':
2433                                 $value = $this->mTitle->getPartialURL();
2434                                 break;
2435                         case 'fullpagename':
2436                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2437                                 break;
2438                         case 'fullpagenamee':
2439                                 $value = $this->mTitle->getPrefixedURL();
2440                                 break;
2441                         case 'subpagename':
2442                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2443                                 break;
2444                         case 'subpagenamee':
2445                                 $value = $this->mTitle->getSubpageUrlForm();
2446                                 break;
2447                         case 'basepagename':
2448                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2449                                 break;
2450                         case 'basepagenamee':
2451                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2452                                 break;
2453                         case 'talkpagename':
2454                                 if ( $this->mTitle->canTalk() ) {
2455                                         $talkPage = $this->mTitle->getTalkPage();
2456                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2457                                 } else {
2458                                         $value = '';
2459                                 }
2460                                 break;
2461                         case 'talkpagenamee':
2462                                 if ( $this->mTitle->canTalk() ) {
2463                                         $talkPage = $this->mTitle->getTalkPage();
2464                                         $value = $talkPage->getPrefixedUrl();
2465                                 } else {
2466                                         $value = '';
2467                                 }
2468                                 break;
2469                         case 'subjectpagename':
2470                                 $subjPage = $this->mTitle->getSubjectPage();
2471                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2472                                 break;
2473                         case 'subjectpagenamee':
2474                                 $subjPage = $this->mTitle->getSubjectPage();
2475                                 $value = $subjPage->getPrefixedUrl();
2476                                 break;
2477                         case 'revisionid':
2478                                 # Let the edit saving system know we should parse the page
2479                                 # *after* a revision ID has been assigned.
2480                                 $this->mOutput->setFlag( 'vary-revision' );
2481                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2482                                 $value = $this->mRevisionId;
2483                                 break;
2484                         case 'revisionday':
2485                                 # Let the edit saving system know we should parse the page
2486                                 # *after* a revision ID has been assigned. This is for null edits.
2487                                 $this->mOutput->setFlag( 'vary-revision' );
2488                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2489                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2490                                 break;
2491                         case 'revisionday2':
2492                                 # Let the edit saving system know we should parse the page
2493                                 # *after* a revision ID has been assigned. This is for null edits.
2494                                 $this->mOutput->setFlag( 'vary-revision' );
2495                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2496                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2497                                 break;
2498                         case 'revisionmonth':
2499                                 # Let the edit saving system know we should parse the page
2500                                 # *after* a revision ID has been assigned. This is for null edits.
2501                                 $this->mOutput->setFlag( 'vary-revision' );
2502                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2503                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2504                                 break;
2505                         case 'revisionyear':
2506                                 # Let the edit saving system know we should parse the page
2507                                 # *after* a revision ID has been assigned. This is for null edits.
2508                                 $this->mOutput->setFlag( 'vary-revision' );
2509                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2510                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2511                                 break;
2512                         case 'revisiontimestamp':
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__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2517                                 $value = $this->getRevisionTimestamp();
2518                                 break;
2519                         case 'revisionuser':
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__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2524                                 $value = $this->getRevisionUser();
2525                                 break;
2526                         case 'namespace':
2527                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2528                                 break;
2529                         case 'namespacee':
2530                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2531                                 break;
2532                         case 'talkspace':
2533                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2534                                 break;
2535                         case 'talkspacee':
2536                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2537                                 break;
2538                         case 'subjectspace':
2539                                 $value = $this->mTitle->getSubjectNsText();
2540                                 break;
2541                         case 'subjectspacee':
2542                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2543                                 break;
2544                         case 'currentdayname':
2545                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2546                                 break;
2547                         case 'currentyear':
2548                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2549                                 break;
2550                         case 'currenttime':
2551                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2552                                 break;
2553                         case 'currenthour':
2554                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2555                                 break;
2556                         case 'currentweek':
2557                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2558                                 # int to remove the padding
2559                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2560                                 break;
2561                         case 'currentdow':
2562                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2563                                 break;
2564                         case 'localdayname':
2565                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2566                                 break;
2567                         case 'localyear':
2568                                 $value = $wgContLang->formatNum( $localYear, true );
2569                                 break;
2570                         case 'localtime':
2571                                 $value = $wgContLang->time( $localTimestamp, false, false );
2572                                 break;
2573                         case 'localhour':
2574                                 $value = $wgContLang->formatNum( $localHour, true );
2575                                 break;
2576                         case 'localweek':
2577                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2578                                 # int to remove the padding
2579                                 $value = $wgContLang->formatNum( (int)$localWeek );
2580                                 break;
2581                         case 'localdow':
2582                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2583                                 break;
2584                         case 'numberofarticles':
2585                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2586                                 break;
2587                         case 'numberoffiles':
2588                                 $value = $wgContLang->formatNum( SiteStats::images() );
2589                                 break;
2590                         case 'numberofusers':
2591                                 $value = $wgContLang->formatNum( SiteStats::users() );
2592                                 break;
2593                         case 'numberofactiveusers':
2594                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2595                                 break;
2596                         case 'numberofpages':
2597                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2598                                 break;
2599                         case 'numberofadmins':
2600                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2601                                 break;
2602                         case 'numberofedits':
2603                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2604                                 break;
2605                         case 'numberofviews':
2606                                 $value = $wgContLang->formatNum( SiteStats::views() );
2607                                 break;
2608                         case 'currenttimestamp':
2609                                 $value = wfTimestamp( TS_MW, $ts );
2610                                 break;
2611                         case 'localtimestamp':
2612                                 $value = $localTimestamp;
2613                                 break;
2614                         case 'currentversion':
2615                                 $value = SpecialVersion::getVersion();
2616                                 break;
2617                         case 'sitename':
2618                                 return $wgSitename;
2619                         case 'server':
2620                                 return $wgServer;
2621                         case 'servername':
2622                                 return $wgServerName;
2623                         case 'scriptpath':
2624                                 return $wgScriptPath;
2625                         case 'stylepath':
2626                                 return $wgStylePath;
2627                         case 'directionmark':
2628                                 return $wgContLang->getDirMark();
2629                         case 'contentlanguage':
2630                                 global $wgContLanguageCode;
2631                                 return $wgContLanguageCode;
2632                         default:
2633                                 $ret = null;
2634                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2635                                         return $ret;
2636                                 } else {
2637                                         return null;
2638                                 }
2639                 }
2640
2641                 if ( $index )
2642                         $this->mVarCache[$index] = $value;
2643
2644                 return $value;
2645         }
2646
2647         /**
2648          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2649          *
2650          * @private
2651          */
2652         function initialiseVariables() {
2653                 wfProfileIn( __METHOD__ );
2654                 $variableIDs = MagicWord::getVariableIDs();
2655                 $substIDs = MagicWord::getSubstIDs();
2656
2657                 $this->mVariables = new MagicWordArray( $variableIDs );
2658                 $this->mSubstWords = new MagicWordArray( $substIDs );
2659                 wfProfileOut( __METHOD__ );
2660         }
2661
2662         /**
2663          * Preprocess some wikitext and return the document tree.
2664          * This is the ghost of replace_variables().
2665          *
2666          * @param string $text The text to parse
2667          * @param integer flags Bitwise combination of:
2668          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2669          *                                     included. Default is to assume a direct page view.
2670          *
2671          * The generated DOM tree must depend only on the input text and the flags.
2672          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2673          *
2674          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2675          * change in the DOM tree for a given text, must be passed through the section identifier
2676          * in the section edit link and thus back to extractSections().
2677          *
2678          * The output of this function is currently only cached in process memory, but a persistent
2679          * cache may be implemented at a later date which takes further advantage of these strict
2680          * dependency requirements.
2681          *
2682          * @private
2683          */
2684         function preprocessToDom( $text, $flags = 0 ) {
2685                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2686                 return $dom;
2687         }
2688
2689         /**
2690          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2691          */
2692         public static function splitWhitespace( $s ) {
2693                 $ltrimmed = ltrim( $s );
2694                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2695                 $trimmed = rtrim( $ltrimmed );
2696                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2697                 if ( $diff > 0 ) {
2698                         $w2 = substr( $ltrimmed, -$diff );
2699                 } else {
2700                         $w2 = '';
2701                 }
2702                 return array( $w1, $trimmed, $w2 );
2703         }
2704
2705         /**
2706          * Replace magic variables, templates, and template arguments
2707          * with the appropriate text. Templates are substituted recursively,
2708          * taking care to avoid infinite loops.
2709          *
2710          * Note that the substitution depends on value of $mOutputType:
2711          *  self::OT_WIKI: only {{subst:}} templates
2712          *  self::OT_PREPROCESS: templates but not extension tags
2713          *  self::OT_HTML: all templates and extension tags
2714          *
2715          * @param string $tex The text to transform
2716          * @param PPFrame $frame Object describing the arguments passed to the template.
2717          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2718          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2719          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2720          * @private
2721          */
2722         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2723                 # Is there any text? Also, Prevent too big inclusions!
2724                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2725                         return $text;
2726                 }
2727                 wfProfileIn( __METHOD__ );
2728
2729                 if ( $frame === false ) {
2730                         $frame = $this->getPreprocessor()->newFrame();
2731                 } elseif ( !( $frame instanceof PPFrame ) ) {
2732                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2733                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2734                 }
2735
2736                 $dom = $this->preprocessToDom( $text );
2737                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2738                 $text = $frame->expand( $dom, $flags );
2739
2740                 wfProfileOut( __METHOD__ );
2741                 return $text;
2742         }
2743
2744         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2745         static function createAssocArgs( $args ) {
2746                 $assocArgs = array();
2747                 $index = 1;
2748                 foreach ( $args as $arg ) {
2749                         $eqpos = strpos( $arg, '=' );
2750                         if ( $eqpos === false ) {
2751                                 $assocArgs[$index++] = $arg;
2752                         } else {
2753                                 $name = trim( substr( $arg, 0, $eqpos ) );
2754                                 $value = trim( substr( $arg, $eqpos+1 ) );
2755                                 if ( $value === false ) {
2756                                         $value = '';
2757                                 }
2758                                 if ( $name !== false ) {
2759                                         $assocArgs[$name] = $value;
2760                                 }
2761                         }
2762                 }
2763
2764                 return $assocArgs;
2765         }
2766
2767         /**
2768          * Warn the user when a parser limitation is reached
2769          * Will warn at most once the user per limitation type
2770          *
2771          * @param string $limitationType, should be one of:
2772          *   'expensive-parserfunction' (corresponding messages:
2773          *       'expensive-parserfunction-warning',
2774          *       'expensive-parserfunction-category')
2775          *   'post-expand-template-argument' (corresponding messages:
2776          *       'post-expand-template-argument-warning',
2777          *       'post-expand-template-argument-category')
2778          *   'post-expand-template-inclusion' (corresponding messages:
2779          *       'post-expand-template-inclusion-warning',
2780          *       'post-expand-template-inclusion-category')
2781          * @params int $current, $max When an explicit limit has been
2782          *       exceeded, provide the values (optional)
2783          */
2784         function limitationWarn( $limitationType, $current=null, $max=null) {
2785                 # does no harm if $current and $max are present but are unnecessary for the message
2786                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2787                 $this->mOutput->addWarning( $warning );
2788                 $this->addTrackingCategory( "$limitationType-category" );
2789         }
2790
2791         /**
2792          * Return the text of a template, after recursively
2793          * replacing any variables or templates within the template.
2794          *
2795          * @param array $piece The parts of the template
2796          *  $piece['title']: the title, i.e. the part before the |
2797          *  $piece['parts']: the parameter array
2798          *  $piece['lineStart']: whether the brace was at the start of a line
2799          * @param PPFrame The current frame, contains template arguments
2800          * @return string the text of the template
2801          * @private
2802          */
2803         function braceSubstitution( $piece, $frame ) {
2804                 global $wgContLang, $wgNonincludableNamespaces;
2805                 wfProfileIn( __METHOD__ );
2806                 wfProfileIn( __METHOD__.'-setup' );
2807
2808                 # Flags
2809                 $found = false;             # $text has been filled
2810                 $nowiki = false;            # wiki markup in $text should be escaped
2811                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2812                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2813                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2814                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2815
2816                 # Title object, where $text came from
2817                 $title = null;
2818
2819                 # $part1 is the bit before the first |, and must contain only title characters.
2820                 # Various prefixes will be stripped from it later.
2821                 $titleWithSpaces = $frame->expand( $piece['title'] );
2822                 $part1 = trim( $titleWithSpaces );
2823                 $titleText = false;
2824
2825                 # Original title text preserved for various purposes
2826                 $originalTitle = $part1;
2827
2828                 # $args is a list of argument nodes, starting from index 0, not including $part1
2829                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2830                 wfProfileOut( __METHOD__.'-setup' );
2831
2832                 # SUBST
2833                 wfProfileIn( __METHOD__.'-modifiers' );
2834                 if ( !$found ) {
2835
2836                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2837
2838                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2839                         # Decide whether to expand template or keep wikitext as-is.
2840                         if ( $this->ot['wiki'] ) {
2841                                 if ( $substMatch === false ) {
2842                                         $literal = true;  # literal when in PST with no prefix
2843                                 } else {
2844                                         $literal = false; # expand when in PST with subst: or safesubst:
2845                                 }
2846                         } else {
2847                                 if ( $substMatch == 'subst' ) {
2848                                         $literal = true;  # literal when not in PST with plain subst:
2849                                 } else {
2850                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2851                                 }
2852                         }
2853                         if ( $literal ) {
2854                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2855                                 $isLocalObj = true;
2856                                 $found = true;
2857                         }
2858                 }
2859
2860                 # Variables
2861                 if ( !$found && $args->getLength() == 0 ) {
2862                         $id = $this->mVariables->matchStartToEnd( $part1 );
2863                         if ( $id !== false ) {
2864                                 $text = $this->getVariableValue( $id, $frame );
2865                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
2866                                         $this->mOutput->mContainsOldMagic = true;
2867                                 }
2868                                 $found = true;
2869                         }
2870                 }
2871
2872                 # MSG, MSGNW and RAW
2873                 if ( !$found ) {
2874                         # Check for MSGNW:
2875                         $mwMsgnw = MagicWord::get( 'msgnw' );
2876                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2877                                 $nowiki = true;
2878                         } else {
2879                                 # Remove obsolete MSG:
2880                                 $mwMsg = MagicWord::get( 'msg' );
2881                                 $mwMsg->matchStartAndRemove( $part1 );
2882                         }
2883
2884                         # Check for RAW:
2885                         $mwRaw = MagicWord::get( 'raw' );
2886                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2887                                 $forceRawInterwiki = true;
2888                         }
2889                 }
2890                 wfProfileOut( __METHOD__.'-modifiers' );
2891
2892                 # Parser functions
2893                 if ( !$found ) {
2894                         wfProfileIn( __METHOD__ . '-pfunc' );
2895
2896                         $colonPos = strpos( $part1, ':' );
2897                         if ( $colonPos !== false ) {
2898                                 # Case sensitive functions
2899                                 $function = substr( $part1, 0, $colonPos );
2900                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2901                                         $function = $this->mFunctionSynonyms[1][$function];
2902                                 } else {
2903                                         # Case insensitive functions
2904                                         $function = $wgContLang->lc( $function );
2905                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2906                                                 $function = $this->mFunctionSynonyms[0][$function];
2907                                         } else {
2908                                                 $function = false;
2909                                         }
2910                                 }
2911                                 if ( $function ) {
2912                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2913                                         $initialArgs = array( &$this );
2914                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2915                                         if ( $flags & SFH_OBJECT_ARGS ) {
2916                                                 # Add a frame parameter, and pass the arguments as an array
2917                                                 $allArgs = $initialArgs;
2918                                                 $allArgs[] = $frame;
2919                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2920                                                         $funcArgs[] = $args->item( $i );
2921                                                 }
2922                                                 $allArgs[] = $funcArgs;
2923                                         } else {
2924                                                 # Convert arguments to plain text
2925                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2926                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2927                                                 }
2928                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2929                                         }
2930
2931                                         # Workaround for PHP bug 35229 and similar
2932                                         if ( !is_callable( $callback ) ) {
2933                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2934                                                 wfProfileOut( __METHOD__ );
2935                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2936                                         }
2937                                         $result = call_user_func_array( $callback, $allArgs );
2938                                         $found = true;
2939                                         $noparse = true;
2940                                         $preprocessFlags = 0;
2941
2942                                         if ( is_array( $result ) ) {
2943                                                 if ( isset( $result[0] ) ) {
2944                                                         $text = $result[0];
2945                                                         unset( $result[0] );
2946                                                 }
2947
2948                                                 # Extract flags into the local scope
2949                                                 # This allows callers to set flags such as nowiki, found, etc.
2950                                                 extract( $result );
2951                                         } else {
2952                                                 $text = $result;
2953                                         }
2954                                         if ( !$noparse ) {
2955                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2956                                                 $isChildObj = true;
2957                                         }
2958                                 }
2959                         }
2960                         wfProfileOut( __METHOD__ . '-pfunc' );
2961                 }
2962
2963                 # Finish mangling title and then check for loops.
2964                 # Set $title to a Title object and $titleText to the PDBK
2965                 if ( !$found ) {
2966                         $ns = NS_TEMPLATE;
2967                         # Split the title into page and subpage
2968                         $subpage = '';
2969                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2970                         if ( $subpage !== '' ) {
2971                                 $ns = $this->mTitle->getNamespace();
2972                         }
2973                         $title = Title::newFromText( $part1, $ns );
2974                         if ( $title ) {
2975                                 $titleText = $title->getPrefixedText();
2976                                 # Check for language variants if the template is not found
2977                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
2978                                         $wgContLang->findVariantLink( $part1, $title, true );
2979                                 }
2980                                 # Do recursion depth check
2981                                 $limit = $this->mOptions->getMaxTemplateDepth();
2982                                 if ( $frame->depth >= $limit ) {
2983                                         $found = true;
2984                                         $text = '<span class="error">'
2985                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
2986                                                 . '</span>';
2987                                 }
2988                         }
2989                 }
2990
2991                 # Load from database
2992                 if ( !$found && $title ) {
2993                         wfProfileIn( __METHOD__ . '-loadtpl' );
2994                         if ( !$title->isExternal() ) {
2995                                 if ( $title->getNamespace() == NS_SPECIAL
2996                                         && $this->mOptions->getAllowSpecialInclusion()
2997                                         && $this->ot['html'] )
2998                                 {
2999                                         $text = SpecialPage::capturePath( $title );
3000                                         if ( is_string( $text ) ) {
3001                                                 $found = true;
3002                                                 $isHTML = true;
3003                                                 $this->disableCache();
3004                                         }
3005                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3006                                         $found = false; # access denied
3007                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3008                                 } else {
3009                                         list( $text, $title ) = $this->getTemplateDom( $title );
3010                                         if ( $text !== false ) {
3011                                                 $found = true;
3012                                                 $isChildObj = true;
3013                                         }
3014                                 }
3015
3016                                 # If the title is valid but undisplayable, make a link to it
3017                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3018                                         $text = "[[:$titleText]]";
3019                                         $found = true;
3020                                 }
3021                         } elseif ( $title->isTrans() ) {
3022                                 # Interwiki transclusion
3023                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3024                                         $text = $this->interwikiTransclude( $title, 'render' );
3025                                         $isHTML = true;
3026                                 } else {
3027                                         $text = $this->interwikiTransclude( $title, 'raw' );
3028                                         # Preprocess it like a template
3029                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3030                                         $isChildObj = true;
3031                                 }
3032                                 $found = true;
3033                         }
3034
3035                         # Do infinite loop check
3036                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3037                         if ( !$frame->loopCheck( $title ) ) {
3038                                 $found = true;
3039                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3040                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3041                         }
3042                         wfProfileOut( __METHOD__ . '-loadtpl' );
3043                 }
3044
3045                 # If we haven't found text to substitute by now, we're done
3046                 # Recover the source wikitext and return it
3047                 if ( !$found ) {
3048                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3049                         wfProfileOut( __METHOD__ );
3050                         return array( 'object' => $text );
3051                 }
3052
3053                 # Expand DOM-style return values in a child frame
3054                 if ( $isChildObj ) {
3055                         # Clean up argument array
3056                         $newFrame = $frame->newChild( $args, $title );
3057
3058                         if ( $nowiki ) {
3059                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3060                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3061                                 # Expansion is eligible for the empty-frame cache
3062                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3063                                         $text = $this->mTplExpandCache[$titleText];
3064                                 } else {
3065                                         $text = $newFrame->expand( $text );
3066                                         $this->mTplExpandCache[$titleText] = $text;
3067                                 }
3068                         } else {
3069                                 # Uncached expansion
3070                                 $text = $newFrame->expand( $text );
3071                         }
3072                 }
3073                 if ( $isLocalObj && $nowiki ) {
3074                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3075                         $isLocalObj = false;
3076                 }
3077
3078                 # Replace raw HTML by a placeholder
3079                 # Add a blank line preceding, to prevent it from mucking up
3080                 # immediately preceding headings
3081                 if ( $isHTML ) {
3082                         $text = "\n\n" . $this->insertStripItem( $text );
3083                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3084                         # Escape nowiki-style return values
3085                         $text = wfEscapeWikiText( $text );
3086                 } elseif ( is_string( $text )
3087                         && !$piece['lineStart']
3088                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3089                 {
3090                         # Bug 529: if the template begins with a table or block-level
3091                         # element, it should be treated as beginning a new line.
3092                         # This behaviour is somewhat controversial.
3093                         $text = "\n" . $text;
3094                 }
3095
3096                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3097                         # Error, oversize inclusion
3098                         $text = "[[$originalTitle]]" .
3099                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3100                         $this->limitationWarn( 'post-expand-template-inclusion' );
3101                 }
3102
3103                 if ( $isLocalObj ) {
3104                         $ret = array( 'object' => $text );
3105                 } else {
3106                         $ret = array( 'text' => $text );
3107                 }
3108
3109                 wfProfileOut( __METHOD__ );
3110                 return $ret;
3111         }
3112
3113         /**
3114          * Get the semi-parsed DOM representation of a template with a given title,
3115          * and its redirect destination title. Cached.
3116          */
3117         function getTemplateDom( $title ) {
3118                 $cacheTitle = $title;
3119                 $titleText = $title->getPrefixedDBkey();
3120
3121                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3122                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3123                         $title = Title::makeTitle( $ns, $dbk );
3124                         $titleText = $title->getPrefixedDBkey();
3125                 }
3126                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3127                         return array( $this->mTplDomCache[$titleText], $title );
3128                 }
3129
3130                 # Cache miss, go to the database
3131                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3132
3133                 if ( $text === false ) {
3134                         $this->mTplDomCache[$titleText] = false;
3135                         return array( false, $title );
3136                 }
3137
3138                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3139                 $this->mTplDomCache[ $titleText ] = $dom;
3140
3141                 if ( !$title->equals( $cacheTitle ) ) {
3142                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3143                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3144                 }
3145
3146                 return array( $dom, $title );
3147         }
3148
3149         /**
3150          * Fetch the unparsed text of a template and register a reference to it.
3151          */
3152         function fetchTemplateAndTitle( $title ) {
3153                 $templateCb = $this->mOptions->getTemplateCallback();
3154                 $stuff = call_user_func( $templateCb, $title, $this );
3155                 $text = $stuff['text'];
3156                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3157                 if ( isset( $stuff['deps'] ) ) {
3158                         foreach ( $stuff['deps'] as $dep ) {
3159                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3160                         }
3161                 }
3162                 return array( $text, $finalTitle );
3163         }
3164
3165         function fetchTemplate( $title ) {
3166                 $rv = $this->fetchTemplateAndTitle( $title );
3167                 return $rv[0];
3168         }
3169
3170         /**
3171          * Static function to get a template
3172          * Can be overridden via ParserOptions::setTemplateCallback().
3173          */
3174         static function statelessFetchTemplate( $title, $parser=false ) {
3175                 $text = $skip = false;
3176                 $finalTitle = $title;
3177                 $deps = array();
3178
3179                 # Loop to fetch the article, with up to 1 redirect
3180                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3181                         # Give extensions a chance to select the revision instead
3182                         $id = false; # Assume current
3183                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3184
3185                         if ( $skip ) {
3186                                 $text = false;
3187                                 $deps[] = array(
3188                                         'title' => $title,
3189                                         'page_id' => $title->getArticleID(),
3190                                         'rev_id' => null );
3191                                 break;
3192                         }
3193                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3194                         $rev_id = $rev ? $rev->getId() : 0;
3195                         # If there is no current revision, there is no page
3196                         if ( $id === false && !$rev ) {
3197                                 $linkCache = LinkCache::singleton();
3198                                 $linkCache->addBadLinkObj( $title );
3199                         }
3200
3201                         $deps[] = array(
3202                                 'title' => $title,
3203                                 'page_id' => $title->getArticleID(),
3204                                 'rev_id' => $rev_id );
3205
3206                         if ( $rev ) {
3207                                 $text = $rev->getText();
3208                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3209                                 global $wgContLang;
3210                                 $message = $wgContLang->lcfirst( $title->getText() );
3211                                 $text = wfMsgForContentNoTrans( $message );
3212                                 if ( wfEmptyMsg( $message, $text ) ) {
3213                                         $text = false;
3214                                         break;
3215                                 }
3216                         } else {
3217                                 break;
3218                         }
3219                         if ( $text === false ) {
3220                                 break;
3221                         }
3222                         # Redirect?
3223                         $finalTitle = $title;
3224                         $title = Title::newFromRedirect( $text );
3225                 }
3226                 return array(
3227                         'text' => $text,
3228                         'finalTitle' => $finalTitle,
3229                         'deps' => $deps );
3230         }
3231
3232         /**
3233          * Transclude an interwiki link.
3234          */
3235         function interwikiTransclude( $title, $action ) {
3236                 global $wgEnableScaryTranscluding;
3237
3238                 if ( !$wgEnableScaryTranscluding ) {
3239                         return wfMsg('scarytranscludedisabled');
3240                 }
3241
3242                 $url = $title->getFullUrl( "action=$action" );
3243
3244                 if ( strlen( $url ) > 255 ) {
3245                         return wfMsg( 'scarytranscludetoolong' );
3246                 }
3247                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3248         }
3249
3250         function fetchScaryTemplateMaybeFromCache( $url ) {
3251                 global $wgTranscludeCacheExpiry;
3252                 $dbr = wfGetDB( DB_SLAVE );
3253                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3254                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3255                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3256                 if ( $obj ) {
3257                         return $obj->tc_contents;
3258                 }
3259
3260                 $text = Http::get( $url );
3261                 if ( !$text ) {
3262                         return wfMsg( 'scarytranscludefailed', $url );
3263                 }
3264
3265                 $dbw = wfGetDB( DB_MASTER );
3266                 $dbw->replace( 'transcache', array('tc_url'), array(
3267                         'tc_url' => $url,
3268                         'tc_time' => $dbw->timestamp( time() ),
3269                         'tc_contents' => $text)
3270                 );
3271                 return $text;
3272         }
3273
3274
3275         /**
3276          * Triple brace replacement -- used for template arguments
3277          * @private
3278          */
3279         function argSubstitution( $piece, $frame ) {
3280                 wfProfileIn( __METHOD__ );
3281
3282                 $error = false;
3283                 $parts = $piece['parts'];
3284                 $nameWithSpaces = $frame->expand( $piece['title'] );
3285                 $argName = trim( $nameWithSpaces );
3286                 $object = false;
3287                 $text = $frame->getArgument( $argName );
3288                 if (  $text === false && $parts->getLength() > 0
3289                   && (
3290                     $this->ot['html']
3291                     || $this->ot['pre']
3292                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3293                   )
3294                 ) {
3295                         # No match in frame, use the supplied default
3296                         $object = $parts->item( 0 )->getChildren();
3297                 }
3298                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3299                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3300                         $this->limitationWarn( 'post-expand-template-argument' );
3301                 }
3302
3303                 if ( $text === false && $object === false ) {
3304                         # No match anywhere
3305                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3306                 }
3307                 if ( $error !== false ) {
3308                         $text .= $error;
3309                 }
3310                 if ( $object !== false ) {
3311                         $ret = array( 'object' => $object );
3312                 } else {
3313                         $ret = array( 'text' => $text );
3314                 }
3315
3316                 wfProfileOut( __METHOD__ );
3317                 return $ret;
3318         }
3319
3320         /**
3321          * Return the text to be used for a given extension tag.
3322          * This is the ghost of strip().
3323          *
3324          * @param array $params Associative array of parameters:
3325          *     name       PPNode for the tag name
3326          *     attr       PPNode for unparsed text where tag attributes are thought to be
3327          *     attributes Optional associative array of parsed attributes
3328          *     inner      Contents of extension element
3329          *     noClose    Original text did not have a close tag
3330          * @param PPFrame $frame
3331          */
3332         function extensionSubstitution( $params, $frame ) {
3333                 global $wgRawHtml, $wgContLang;
3334
3335                 $name = $frame->expand( $params['name'] );
3336                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3337                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3338                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3339
3340                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3341                         ( $this->ot['html'] || $this->ot['pre'] );
3342                 if ( $isFunctionTag ) {
3343                         $markerType = 'none';
3344                 } else {
3345                         $markerType = 'general';
3346                 }
3347                 if ( $this->ot['html'] || $isFunctionTag ) {
3348                         $name = strtolower( $name );
3349                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3350                         if ( isset( $params['attributes'] ) ) {
3351                                 $attributes = $attributes + $params['attributes'];
3352                         }
3353
3354                         if ( isset( $this->mTagHooks[$name] ) ) {
3355                                 # Workaround for PHP bug 35229 and similar
3356                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3357                                         throw new MWException( "Tag hook for $name is not callable\n" );
3358                                 }
3359                                 $output = call_user_func_array( $this->mTagHooks[$name],
3360                                         array( $content, $attributes, $this, $frame ) );
3361                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3362                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3363                                 if ( !is_callable( $callback ) ) {
3364                                         throw new MWException( "Tag hook for $name is not callable\n" );
3365                                 }
3366
3367                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3368                         } else {
3369                                 $output = '<span class="error">Invalid tag extension name: ' .
3370                                         htmlspecialchars( $name ) . '</span>';
3371                         }
3372
3373                         if ( is_array( $output ) ) {
3374                                 # Extract flags to local scope (to override $markerType)
3375                                 $flags = $output;
3376                                 $output = $flags[0];
3377                                 unset( $flags[0] );
3378                                 extract( $flags );
3379                         }
3380                 } else {
3381                         if ( is_null( $attrText ) ) {
3382                                 $attrText = '';
3383                         }
3384                         if ( isset( $params['attributes'] ) ) {
3385                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3386                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3387                                                 htmlspecialchars( $attrValue ) . '"';
3388                                 }
3389                         }
3390                         if ( $content === null ) {
3391                                 $output = "<$name$attrText/>";
3392                         } else {
3393                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3394                                 $output = "<$name$attrText>$content$close";
3395                         }
3396                 }
3397
3398                 if ( $markerType === 'none' ) {
3399                         return $output;
3400                 } elseif ( $markerType === 'nowiki' ) {
3401                         $this->mStripState->nowiki->setPair( $marker, $output );
3402                 } elseif ( $markerType === 'general' ) {
3403                         $this->mStripState->general->setPair( $marker, $output );
3404                 } else {
3405                         throw new MWException( __METHOD__.': invalid marker type' );
3406                 }
3407                 return $marker;
3408         }
3409
3410         /**
3411          * Increment an include size counter
3412          *
3413          * @param string $type The type of expansion
3414          * @param integer $size The size of the text
3415          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3416          */
3417         function incrementIncludeSize( $type, $size ) {
3418                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3419                         return false;
3420                 } else {
3421                         $this->mIncludeSizes[$type] += $size;
3422                         return true;
3423                 }
3424         }
3425
3426         /**
3427          * Increment the expensive function count
3428          *
3429          * @return boolean False if the limit has been exceeded
3430          */
3431         function incrementExpensiveFunctionCount() {
3432                 global $wgExpensiveParserFunctionLimit;
3433                 $this->mExpensiveFunctionCount++;
3434                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3435                         return true;
3436                 }
3437                 return false;
3438         }
3439
3440         /**
3441          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3442          * Fills $this->mDoubleUnderscores, returns the modified text
3443          */
3444         function doDoubleUnderscore( $text ) {
3445                 wfProfileIn( __METHOD__ );
3446
3447                 # The position of __TOC__ needs to be recorded
3448                 $mw = MagicWord::get( 'toc' );
3449                 if ( $mw->match( $text ) ) {
3450                         $this->mShowToc = true;
3451                         $this->mForceTocPosition = true;
3452
3453                         # Set a placeholder. At the end we'll fill it in with the TOC.
3454                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3455
3456                         # Only keep the first one.
3457                         $text = $mw->replace( '', $text );
3458                 }
3459
3460                 # Now match and remove the rest of them
3461                 $mwa = MagicWord::getDoubleUnderscoreArray();
3462                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3463
3464                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3465                         $this->mOutput->mNoGallery = true;
3466                 }
3467                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3468                         $this->mShowToc = false;
3469                 }
3470                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3471                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3472                         $this->addTrackingCategory( 'hidden-category-category' );
3473                 }
3474                 # (bug 8068) Allow control over whether robots index a page.
3475                 #
3476                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3477                 # is not desirable, the last one on the page should win.
3478                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3479                         $this->mOutput->setIndexPolicy( 'noindex' );
3480                         $this->addTrackingCategory( 'noindex-category' );
3481                 }
3482                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3483                         $this->mOutput->setIndexPolicy( 'index' );
3484                         $this->addTrackingCategory( 'index-category' );
3485                 }
3486
3487                 wfProfileOut( __METHOD__ );
3488                 return $text;
3489         }
3490
3491         /**
3492          * Add a tracking category, getting the title from a system message,
3493          * or print a debug message if the title is invalid.
3494          * @param $msg String message key
3495          * @return Bool whether the addition was successful
3496          */
3497         protected function addTrackingCategory( $msg ) {
3498                 $cat = wfMsgForContent( $msg );
3499
3500                 # Allow tracking categories to be disabled by setting them to "-"
3501                 if ( $cat === '-' ) {
3502                         return false;
3503                 }
3504
3505                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3506                 if ( $containerCategory ) {
3507                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3508                         return true;
3509                 } else {
3510                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3511                         return false;
3512                 }
3513         }
3514
3515         /**
3516          * This function accomplishes several tasks:
3517          * 1) Auto-number headings if that option is enabled
3518          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3519          * 3) Add a Table of contents on the top for users who have enabled the option
3520          * 4) Auto-anchor headings
3521          *
3522          * It loops through all headlines, collects the necessary data, then splits up the
3523          * string and re-inserts the newly formatted headlines.
3524          *
3525          * @param string $text
3526          * @param string $origText Original, untouched wikitext
3527          * @param boolean $isMain
3528          * @private
3529          */
3530         function formatHeadings( $text, $origText, $isMain=true ) {
3531                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3532
3533                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3534                 $showEditLink = $this->mOptions->getEditSection();
3535
3536                 # Do not call quickUserCan unless necessary
3537                 if ( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3538                         $showEditLink = 0;
3539                 }
3540
3541                 # Inhibit editsection links if requested in the page
3542                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3543                         $showEditLink = 0;
3544                 }
3545
3546                 # Get all headlines for numbering them and adding funky stuff like [edit]
3547                 # links - this is for later, but we need the number of headlines right now
3548                 $matches = array();
3549                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3550
3551                 # if there are fewer than 4 headlines in the article, do not show TOC
3552                 # unless it's been explicitly enabled.
3553                 $enoughToc = $this->mShowToc &&
3554                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3555
3556                 # Allow user to stipulate that a page should have a "new section"
3557                 # link added via __NEWSECTIONLINK__
3558                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3559                         $this->mOutput->setNewSection( true );
3560                 }
3561
3562                 # Allow user to remove the "new section"
3563                 # link via __NONEWSECTIONLINK__
3564                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3565                         $this->mOutput->hideNewSection( true );
3566                 }
3567
3568                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3569                 # override above conditions and always show TOC above first header
3570                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3571                         $this->mShowToc = true;
3572                         $enoughToc = true;
3573                 }
3574
3575                 # We need this to perform operations on the HTML
3576                 $sk = $this->mOptions->getSkin();
3577
3578                 # headline counter
3579                 $headlineCount = 0;
3580                 $numVisible = 0;
3581
3582                 # Ugh .. the TOC should have neat indentation levels which can be
3583                 # passed to the skin functions. These are determined here
3584                 $toc = '';
3585                 $full = '';
3586                 $head = array();
3587                 $sublevelCount = array();
3588                 $levelCount = array();
3589                 $toclevel = 0;
3590                 $level = 0;
3591                 $prevlevel = 0;
3592                 $toclevel = 0;
3593                 $prevtoclevel = 0;
3594                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3595                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3596                 $oldType = $this->mOutputType;
3597                 $this->setOutputType( self::OT_WIKI );
3598                 $frame = $this->getPreprocessor()->newFrame();
3599                 $root = $this->preprocessToDom( $origText );
3600                 $node = $root->getFirstChild();
3601                 $byteOffset = 0;
3602                 $tocraw = array();
3603
3604                 foreach ( $matches[3] as $headline ) {
3605                         $isTemplate = false;
3606                         $titleText = false;
3607                         $sectionIndex = false;
3608                         $numbering = '';
3609                         $markerMatches = array();
3610                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3611                                 $serial = $markerMatches[1];
3612                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3613                                 $isTemplate = ( $titleText != $baseTitleText );
3614                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3615                         }
3616
3617                         if ( $toclevel ) {
3618                                 $prevlevel = $level;
3619                                 $prevtoclevel = $toclevel;
3620                         }
3621                         $level = $matches[1][$headlineCount];
3622
3623                         if ( $level > $prevlevel ) {
3624                                 # Increase TOC level
3625                                 $toclevel++;
3626                                 $sublevelCount[$toclevel] = 0;
3627                                 if ( $toclevel<$wgMaxTocLevel ) {
3628                                         $prevtoclevel = $toclevel;
3629                                         $toc .= $sk->tocIndent();
3630                                         $numVisible++;
3631                                 }
3632                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3633                                 # Decrease TOC level, find level to jump to
3634
3635                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3636                                         if ( $levelCount[$i] == $level ) {
3637                                                 # Found last matching level
3638                                                 $toclevel = $i;
3639                                                 break;
3640                                         } elseif ( $levelCount[$i] < $level ) {
3641                                                 # Found first matching level below current level
3642                                                 $toclevel = $i + 1;
3643                                                 break;
3644                                         }
3645                                 }
3646                                 if ( $i == 0 ) {
3647                                         $toclevel = 1;
3648                                 }
3649                                 if ( $toclevel<$wgMaxTocLevel ) {
3650                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3651                                                 # Unindent only if the previous toc level was shown :p
3652                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3653                                                 $prevtoclevel = $toclevel;
3654                                         } else {
3655                                                 $toc .= $sk->tocLineEnd();
3656                                         }
3657                                 }
3658                         } else {
3659                                 # No change in level, end TOC line
3660                                 if ( $toclevel<$wgMaxTocLevel ) {
3661                                         $toc .= $sk->tocLineEnd();
3662                                 }
3663                         }
3664
3665                         $levelCount[$toclevel] = $level;
3666
3667                         # count number of headlines for each level
3668                         @$sublevelCount[$toclevel]++;
3669                         $dot = 0;
3670                         for( $i = 1; $i <= $toclevel; $i++ ) {
3671                                 if ( !empty( $sublevelCount[$i] ) ) {
3672                                         if ( $dot ) {
3673                                                 $numbering .= '.';
3674                                         }
3675                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3676                                         $dot = 1;
3677                                 }
3678                         }
3679
3680                         # The safe header is a version of the header text safe to use for links
3681                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3682                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3683
3684                         # Remove link placeholders by the link text.
3685                         #     <!--LINK number-->
3686                         # turns into
3687                         #     link text with suffix
3688                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3689
3690                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3691                         $tocline = preg_replace(
3692                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3693                                 array( '',                          '<$1>' ),
3694                                 $safeHeadline
3695                         );
3696                         $tocline = trim( $tocline );
3697
3698                         # For the anchor, strip out HTML-y stuff period
3699                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3700                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3701                         $safeHeadline = trim( $safeHeadline );
3702
3703                         # Save headline for section edit hint before it's escaped
3704                         $headlineHint = $safeHeadline;
3705
3706                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3707                                 # For reverse compatibility, provide an id that's
3708                                 # HTML4-compatible, like we used to.
3709                                 #
3710                                 # It may be worth noting, academically, that it's possible for
3711                                 # the legacy anchor to conflict with a non-legacy headline
3712                                 # anchor on the page.  In this case likely the "correct" thing
3713                                 # would be to either drop the legacy anchors or make sure
3714                                 # they're numbered first.  However, this would require people
3715                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3716                                 # manually, so let's not bother worrying about it.
3717                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3718                                         array( 'noninitial', 'legacy' ) );
3719                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3720
3721                                 if ( $legacyHeadline == $safeHeadline ) {
3722                                         # No reason to have both (in fact, we can't)
3723                                         $legacyHeadline = false;
3724                                 }
3725                         } else {
3726                                 $legacyHeadline = false;
3727                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3728                                         'noninitial' );
3729                         }
3730
3731                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3732                         # Does this apply to Unicode characters?  Because we aren't
3733                         # handling those here.
3734                         $arrayKey = strtolower( $safeHeadline );
3735                         if ( $legacyHeadline === false ) {
3736                                 $legacyArrayKey = false;
3737                         } else {
3738                                 $legacyArrayKey = strtolower( $legacyHeadline );
3739                         }
3740
3741                         # count how many in assoc. array so we can track dupes in anchors
3742                         if ( isset( $refers[$arrayKey] ) ) {
3743                                 $refers[$arrayKey]++;
3744                         } else {
3745                                 $refers[$arrayKey] = 1;
3746                         }
3747                         if ( isset( $refers[$legacyArrayKey] ) ) {
3748                                 $refers[$legacyArrayKey]++;
3749                         } else {
3750                                 $refers[$legacyArrayKey] = 1;
3751                         }
3752
3753                         # Don't number the heading if it is the only one (looks silly)
3754                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3755                                 # the two are different if the line contains a link
3756                                 $headline = $numbering . ' ' . $headline;
3757                         }
3758
3759                         # Create the anchor for linking from the TOC to the section
3760                         $anchor = $safeHeadline;
3761                         $legacyAnchor = $legacyHeadline;
3762                         if ( $refers[$arrayKey] > 1 ) {
3763                                 $anchor .= '_' . $refers[$arrayKey];
3764                         }
3765                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3766                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3767                         }
3768                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3769                                 $toc .= $sk->tocLine( $anchor, $tocline,
3770                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3771                         }
3772
3773                         # Add the section to the section tree
3774                         # Find the DOM node for this header
3775                         while ( $node && !$isTemplate ) {
3776                                 if ( $node->getName() === 'h' ) {
3777                                         $bits = $node->splitHeading();
3778                                         if ( $bits['i'] == $sectionIndex )
3779                                                 break;
3780                                 }
3781                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3782                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3783                                 $node = $node->getNextSibling();
3784                         }
3785                         $tocraw[] = array(
3786                                 'toclevel' => $toclevel,
3787                                 'level' => $level,
3788                                 'line' => $tocline,
3789                                 'number' => $numbering,
3790                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3791                                 'fromtitle' => $titleText,
3792                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3793                                 'anchor' => $anchor,
3794                         );
3795
3796                         # give headline the correct <h#> tag
3797                         if ( $showEditLink && $sectionIndex !== false ) {
3798                                 if ( $isTemplate ) {
3799                                         # Put a T flag in the section identifier, to indicate to extractSections()
3800                                         # that sections inside <includeonly> should be counted.
3801                                         $editlink = $sk->doEditSectionLink( Title::newFromText( $titleText ), "T-$sectionIndex" );
3802                                 } else {
3803                                         $editlink = $sk->doEditSectionLink( $this->mTitle, $sectionIndex, $headlineHint );
3804                                 }
3805                         } else {
3806                                 $editlink = '';
3807                         }
3808                         $head[$headlineCount] = $sk->makeHeadline( $level,
3809                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3810                                 $editlink, $legacyAnchor );
3811
3812                         $headlineCount++;
3813                 }
3814
3815                 $this->setOutputType( $oldType );
3816
3817                 # Never ever show TOC if no headers
3818                 if ( $numVisible < 1 ) {
3819                         $enoughToc = false;
3820                 }
3821
3822                 if ( $enoughToc ) {
3823                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3824                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3825                         }
3826                         $toc = $sk->tocList( $toc );
3827                         $this->mOutput->setTOCHTML( $toc );
3828                 }
3829
3830                 if ( $isMain ) {
3831                         $this->mOutput->setSections( $tocraw );
3832                 }
3833
3834                 # split up and insert constructed headlines
3835
3836                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3837                 $i = 0;
3838
3839                 foreach ( $blocks as $block ) {
3840                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3841                                 # This is the [edit] link that appears for the top block of text when
3842                                 # section editing is enabled
3843
3844                                 # Disabled because it broke block formatting
3845                                 # For example, a bullet point in the top line
3846                                 # $full .= $sk->editSectionLink(0);
3847                         }
3848                         $full .= $block;
3849                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3850                                 # Top anchor now in skin
3851                                 $full = $full.$toc;
3852                         }
3853
3854                         if ( !empty( $head[$i] ) ) {
3855                                 $full .= $head[$i];
3856                         }
3857                         $i++;
3858                 }
3859                 if ( $this->mForceTocPosition ) {
3860                         return str_replace( '<!--MWTOC-->', $toc, $full );
3861                 } else {
3862                         return $full;
3863                 }
3864         }
3865
3866         /**
3867          * Merge $tree2 into $tree1 by replacing the section with index
3868          * $section in $tree1 and its descendants with the sections in $tree2.
3869          * Note that in the returned section tree, only the 'index' and
3870          * 'byteoffset' fields are guaranteed to be correct.
3871          * @param $tree1 array Section tree from ParserOutput::getSectons()
3872         *  @param $tree2 array Section tree
3873          * @param $section int Section index
3874          * @param $title Title Title both section trees come from
3875          * @param $len2 int Length of the original wikitext for $tree2
3876          * @return array Merged section tree
3877          */
3878         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3879                 global $wgContLang;
3880                 $newTree = array();
3881                 $targetLevel = false;
3882                 $merged = false;
3883                 $lastLevel = 1;
3884                 $nextIndex = 1;
3885                 $numbering = array( 0 );
3886                 $titletext = $title->getPrefixedDBkey();
3887                 foreach ( $tree1 as $s ) {
3888                         if ( $targetLevel !== false ) {
3889                                 if ( $s['level'] <= $targetLevel ) {
3890                                         # We've skipped enough
3891                                         $targetLevel = false;
3892                                 } else {
3893                                         continue;
3894                                 }
3895                         }
3896                         if ( $s['index'] != $section ||
3897                                         $s['fromtitle'] != $titletext ) {
3898                                 self::incrementNumbering( $numbering,
3899                                         $s['toclevel'], $lastLevel );
3900
3901                                 # Rewrite index, byteoffset and number
3902                                 if ( $s['fromtitle'] == $titletext ) {
3903                                         $s['index'] = $nextIndex++;
3904                                         if ( $merged ) {
3905                                                 $s['byteoffset'] += $len2;
3906                                         }
3907                                 }
3908                                 $s['number']  = implode( '.', array_map(
3909                                         array( $wgContLang, 'formatnum' ),
3910                                         $numbering ) );
3911                                 $lastLevel = $s['toclevel'];
3912                                 $newTree[] = $s;
3913                         } else {
3914                                 # We're at $section
3915                                 # Insert sections from $tree2 here
3916                                 foreach ( $tree2 as $s2 ) {
3917                                         # Rewrite the fields in $s2
3918                                         # before inserting it
3919                                         $s2['toclevel'] += $s['toclevel'] - 1;
3920                                         $s2['level'] += $s['level'] - 1;
3921                                         $s2['index'] = $nextIndex++;
3922                                         $s2['byteoffset'] += $s['byteoffset'];
3923
3924                                         self::incrementNumbering( $numbering,
3925                                                 $s2['toclevel'], $lastLevel );
3926                                         $s2['number']  = implode( '.', array_map(
3927                                                 array( $wgContLang, 'formatnum' ),
3928                                                 $numbering ) );
3929                                         $lastLevel = $s2['toclevel'];
3930                                         $newTree[] = $s2;
3931                                 }
3932                                 # Skip all descendants of $section in $tree1
3933                                 $targetLevel = $s['level'];
3934                                 $merged = true;
3935                         }
3936                 }
3937                 return $newTree;
3938         }
3939
3940         /**
3941          * Increment a section number. Helper function for mergeSectionTrees()
3942          * @param $number array Array representing a section number
3943          * @param $level int Current TOC level (depth)
3944          * @param $lastLevel int Level of previous TOC entry
3945          */
3946         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3947                 if ( $level > $lastLevel ) {
3948                         $number[$level - 1] = 1;
3949                 } elseif ( $level < $lastLevel ) {
3950                         foreach ( $number as $key => $unused )
3951                                 if ( $key >= $level ) {
3952                                         unset( $number[$key] );
3953                                 }
3954                         $number[$level - 1]++;
3955                 } else {
3956                         $number[$level - 1]++;
3957                 }
3958         }
3959
3960         /**
3961          * Transform wiki markup when saving a page by doing \r\n -> \n
3962          * conversion, substitting signatures, {{subst:}} templates, etc.
3963          *
3964          * @param string $text the text to transform
3965          * @param Title &$title the Title object for the current article
3966          * @param User $user the User object describing the current user
3967          * @param ParserOptions $options parsing options
3968          * @param bool $clearState whether to clear the parser state first
3969          * @return string the altered wiki markup
3970          * @public
3971          */
3972         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3973                 $this->mOptions = $options;
3974                 $this->setTitle( $title );
3975                 $this->setOutputType( self::OT_WIKI );
3976
3977                 if ( $clearState ) {
3978                         $this->clearState();
3979                 }
3980
3981                 $pairs = array(
3982                         "\r\n" => "\n",
3983                 );
3984                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3985                 $text = $this->pstPass2( $text, $user );
3986                 $text = $this->mStripState->unstripBoth( $text );
3987                 return $text;
3988         }
3989
3990         /**
3991          * Pre-save transform helper function
3992          * @private
3993          */
3994         function pstPass2( $text, $user ) {
3995                 global $wgContLang, $wgLocaltimezone;
3996
3997                 # Note: This is the timestamp saved as hardcoded wikitext to
3998                 # the database, we use $wgContLang here in order to give
3999                 # everyone the same signature and use the default one rather
4000                 # than the one selected in each user's preferences.
4001                 # (see also bug 12815)
4002                 $ts = $this->mOptions->getTimestamp();
4003                 if ( isset( $wgLocaltimezone ) ) {
4004                         $tz = $wgLocaltimezone;
4005                 } else {
4006                         $tz = date_default_timezone_get();
4007                 }
4008
4009                 $unixts = wfTimestamp( TS_UNIX, $ts );
4010                 $oldtz = date_default_timezone_get();
4011                 date_default_timezone_set( $tz );
4012                 $ts = date( 'YmdHis', $unixts );
4013                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4014
4015                 # Allow translation of timezones trough wiki. date() can return
4016                 # whatever crap the system uses, localised or not, so we cannot
4017                 # ship premade translations.
4018                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4019                 $value = wfMsgForContent( $key );
4020                 if ( !wfEmptyMsg( $key, $value ) ) {
4021                         $tzMsg = $value;
4022                 }
4023
4024                 date_default_timezone_set( $oldtz );
4025
4026                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4027
4028                 # Variable replacement
4029                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4030                 $text = $this->replaceVariables( $text );
4031
4032                 # Signatures
4033                 $sigText = $this->getUserSig( $user );
4034                 $text = strtr( $text, array(
4035                         '~~~~~' => $d,
4036                         '~~~~' => "$sigText $d",
4037                         '~~~' => $sigText
4038                 ) );
4039
4040                 # Context links: [[|name]] and [[name (context)|]]
4041                 global $wgLegalTitleChars;
4042                 $tc = "[$wgLegalTitleChars]";
4043                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4044
4045                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4046                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4047                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4048                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4049
4050                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4051                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4052                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4053                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4054
4055                 $t = $this->mTitle->getText();
4056                 $m = array();
4057                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4058                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4059                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4060                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4061                 } else {
4062                         # if there's no context, don't bother duplicating the title
4063                         $text = preg_replace( $p2, '[[\\1]]', $text );
4064                 }
4065
4066                 # Trim trailing whitespace
4067                 $text = rtrim( $text );
4068
4069                 return $text;
4070         }
4071
4072         /**
4073          * Fetch the user's signature text, if any, and normalize to
4074          * validated, ready-to-insert wikitext.
4075          * If you have pre-fetched the nickname or the fancySig option, you can
4076          * specify them here to save a database query.
4077          *
4078          * @param User $user
4079          * @return string
4080          */
4081         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4082                 global $wgMaxSigChars;
4083
4084                 $username = $user->getName();
4085
4086                 # If not given, retrieve from the user object.
4087                 if ( $nickname === false )
4088                         $nickname = $user->getOption( 'nickname' );
4089
4090                 if ( is_null( $fancySig ) ) {
4091                         $fancySig = $user->getBoolOption( 'fancysig' );
4092                 }
4093
4094                 $nickname = $nickname == null ? $username : $nickname;
4095
4096                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4097                         $nickname = $username;
4098                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4099                 } elseif ( $fancySig !== false ) {
4100                         # Sig. might contain markup; validate this
4101                         if ( $this->validateSig( $nickname ) !== false ) {
4102                                 # Validated; clean up (if needed) and return it
4103                                 return $this->cleanSig( $nickname, true );
4104                         } else {
4105                                 # Failed to validate; fall back to the default
4106                                 $nickname = $username;
4107                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4108                         }
4109                 }
4110
4111                 # Make sure nickname doesnt get a sig in a sig
4112                 $nickname = $this->cleanSigInSig( $nickname );
4113
4114                 # If we're still here, make it a link to the user page
4115                 $userText = wfEscapeWikiText( $username );
4116                 $nickText = wfEscapeWikiText( $nickname );
4117                 if ( $user->isAnon() )  {
4118                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4119                 } else {
4120                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4121                 }
4122         }
4123
4124         /**
4125          * Check that the user's signature contains no bad XML
4126          *
4127          * @param string $text
4128          * @return mixed An expanded string, or false if invalid.
4129          */
4130         function validateSig( $text ) {
4131                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4132         }
4133
4134         /**
4135          * Clean up signature text
4136          *
4137          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4138          * 2) Substitute all transclusions
4139          *
4140          * @param string $text
4141          * @param $parsing Whether we're cleaning (preferences save) or parsing
4142          * @return string Signature text
4143          */
4144         function cleanSig( $text, $parsing = false ) {
4145                 if ( !$parsing ) {
4146                         global $wgTitle;
4147                         $this->clearState();
4148                         $this->setTitle( $wgTitle );
4149                         $this->mOptions = new ParserOptions;
4150                         $this->setOutputType = self::OT_PREPROCESS;
4151                 }
4152
4153                 # Option to disable this feature
4154                 if ( !$this->mOptions->getCleanSignatures() ) {
4155                         return $text;
4156                 }
4157
4158                 # FIXME: regex doesn't respect extension tags or nowiki
4159                 #  => Move this logic to braceSubstitution()
4160                 $substWord = MagicWord::get( 'subst' );
4161                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4162                 $substText = '{{' . $substWord->getSynonym( 0 );
4163
4164                 $text = preg_replace( $substRegex, $substText, $text );
4165                 $text = $this->cleanSigInSig( $text );
4166                 $dom = $this->preprocessToDom( $text );
4167                 $frame = $this->getPreprocessor()->newFrame();
4168                 $text = $frame->expand( $dom );
4169
4170                 if ( !$parsing ) {
4171                         $text = $this->mStripState->unstripBoth( $text );
4172                 }
4173
4174                 return $text;
4175         }
4176
4177         /**
4178          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4179          * @param string $text
4180          * @return string Signature text with /~{3,5}/ removed
4181          */
4182         function cleanSigInSig( $text ) {
4183                 $text = preg_replace( '/~{3,5}/', '', $text );
4184                 return $text;
4185         }
4186
4187         /**
4188          * Set up some variables which are usually set up in parse()
4189          * so that an external function can call some class members with confidence
4190          * @public
4191          */
4192         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4193                 $this->setTitle( $title );
4194                 $this->mOptions = $options;
4195                 $this->setOutputType( $outputType );
4196                 if ( $clearState ) {
4197                         $this->clearState();
4198                 }
4199         }
4200
4201         /**
4202          * Wrapper for preprocess()
4203          *
4204          * @param string $text the text to preprocess
4205          * @param ParserOptions $options  options
4206          * @return string
4207          * @public
4208          */
4209         function transformMsg( $text, $options ) {
4210                 global $wgTitle;
4211                 static $executing = false;
4212
4213                 # Guard against infinite recursion
4214                 if ( $executing ) {
4215                         return $text;
4216                 }
4217                 $executing = true;
4218
4219                 wfProfileIn( __METHOD__ );
4220                 $text = $this->preprocess( $text, $wgTitle, $options );
4221
4222                 $executing = false;
4223                 wfProfileOut( __METHOD__ );
4224                 return $text;
4225         }
4226
4227         /**
4228          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4229          * The callback should have the following form:
4230          *    function myParserHook( $text, $params, $parser ) { ... }
4231          *
4232          * Transform and return $text. Use $parser for any required context, e.g. use
4233          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4234          *
4235          * @public
4236          *
4237          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4238          * @param mixed $callback The callback function (and object) to use for the tag
4239          *
4240          * @return The old value of the mTagHooks array associated with the hook
4241          */
4242         function setHook( $tag, $callback ) {
4243                 $tag = strtolower( $tag );
4244                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4245                 $this->mTagHooks[$tag] = $callback;
4246                 if ( !in_array( $tag, $this->mStripList ) ) {
4247                         $this->mStripList[] = $tag;
4248                 }
4249
4250                 return $oldVal;
4251         }
4252
4253         function setTransparentTagHook( $tag, $callback ) {
4254                 $tag = strtolower( $tag );
4255                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4256                 $this->mTransparentTagHooks[$tag] = $callback;
4257
4258                 return $oldVal;
4259         }
4260
4261         /**
4262          * Remove all tag hooks
4263          */
4264         function clearTagHooks() {
4265                 $this->mTagHooks = array();
4266                 $this->mStripList = $this->mDefaultStripList;
4267         }
4268
4269         /**
4270          * Create a function, e.g. {{sum:1|2|3}}
4271          * The callback function should have the form:
4272          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4273          *
4274          * Or with SFH_OBJECT_ARGS:
4275          *    function myParserFunction( $parser, $frame, $args ) { ... }
4276          *
4277          * The callback may either return the text result of the function, or an array with the text
4278          * in element 0, and a number of flags in the other elements. The names of the flags are
4279          * specified in the keys. Valid flags are:
4280          *   found                     The text returned is valid, stop processing the template. This
4281          *                             is on by default.
4282          *   nowiki                    Wiki markup in the return value should be escaped
4283          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4284          *
4285          * @public
4286          *
4287          * @param string $id The magic word ID
4288          * @param mixed $callback The callback function (and object) to use
4289          * @param integer $flags a combination of the following flags:
4290          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4291          *
4292          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4293          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4294          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4295          *     the arguments, and to control the way they are expanded.
4296          *
4297          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4298          *     arguments, for instance:
4299          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4300          *
4301          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4302          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4303          *     working if/when this is changed.
4304          *
4305          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4306          *     expansion.
4307          *
4308          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4309          *     about the methods available in PPFrame and PPNode.
4310          *
4311          * @return The old callback function for this name, if any
4312          */
4313         function setFunctionHook( $id, $callback, $flags = 0 ) {
4314                 global $wgContLang;
4315
4316                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4317                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4318
4319                 # Add to function cache
4320                 $mw = MagicWord::get( $id );
4321                 if ( !$mw )
4322                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4323
4324                 $synonyms = $mw->getSynonyms();
4325                 $sensitive = intval( $mw->isCaseSensitive() );
4326
4327                 foreach ( $synonyms as $syn ) {
4328                         # Case
4329                         if ( !$sensitive ) {
4330                                 $syn = $wgContLang->lc( $syn );
4331                         }
4332                         # Add leading hash
4333                         if ( !( $flags & SFH_NO_HASH ) ) {
4334                                 $syn = '#' . $syn;
4335                         }
4336                         # Remove trailing colon
4337                         if ( substr( $syn, -1, 1 ) === ':' ) {
4338                                 $syn = substr( $syn, 0, -1 );
4339                         }
4340                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4341                 }
4342                 return $oldVal;
4343         }
4344
4345         /**
4346          * Get all registered function hook identifiers
4347          *
4348          * @return array
4349          */
4350         function getFunctionHooks() {
4351                 return array_keys( $this->mFunctionHooks );
4352         }
4353
4354         /**
4355          * Create a tag function, e.g. <test>some stuff</test>.
4356          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4357          * Unlike parser functions, their content is not preprocessed.
4358          */
4359         function setFunctionTagHook( $tag, $callback, $flags ) {
4360                 $tag = strtolower( $tag );
4361                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4362                         $this->mFunctionTagHooks[$tag] : null;
4363                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4364
4365                 if ( !in_array( $tag, $this->mStripList ) ) {
4366                         $this->mStripList[] = $tag;
4367                 }
4368
4369                 return $old;
4370         }
4371
4372         /**
4373          * FIXME: update documentation. makeLinkObj() is deprecated.
4374          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4375          * Placeholders created in Skin::makeLinkObj()
4376          * Returns an array of link CSS classes, indexed by PDBK.
4377          */
4378         function replaceLinkHolders( &$text, $options = 0 ) {
4379                 return $this->mLinkHolders->replace( $text );
4380         }
4381
4382         /**
4383          * Replace <!--LINK--> link placeholders with plain text of links
4384          * (not HTML-formatted).
4385          * @param string $text
4386          * @return string
4387          */
4388         function replaceLinkHoldersText( $text ) {
4389                 return $this->mLinkHolders->replaceText( $text );
4390         }
4391
4392         /**
4393          * Renders an image gallery from a text with one line per image.
4394          * text labels may be given by using |-style alternative text. E.g.
4395          *   Image:one.jpg|The number "1"
4396          *   Image:tree.jpg|A tree
4397          * given as text will return the HTML of a gallery with two images,
4398          * labeled 'The number "1"' and
4399          * 'A tree'.
4400          */
4401         function renderImageGallery( $text, $params ) {
4402                 $ig = new ImageGallery();
4403                 $ig->setContextTitle( $this->mTitle );
4404                 $ig->setShowBytes( false );
4405                 $ig->setShowFilename( false );
4406                 $ig->setParser( $this );
4407                 $ig->setHideBadImages();
4408                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4409                 $ig->useSkin( $this->mOptions->getSkin() );
4410                 $ig->mRevisionId = $this->mRevisionId;
4411
4412                 if ( isset( $params['showfilename'] ) ) {
4413                         $ig->setShowFilename( true );
4414                 } else {
4415                         $ig->setShowFilename( false );
4416                 }
4417                 if ( isset( $params['caption'] ) ) {
4418                         $caption = $params['caption'];
4419                         $caption = htmlspecialchars( $caption );
4420                         $caption = $this->replaceInternalLinks( $caption );
4421                         $ig->setCaptionHtml( $caption );
4422                 }
4423                 if ( isset( $params['perrow'] ) ) {
4424                         $ig->setPerRow( $params['perrow'] );
4425                 }
4426                 if ( isset( $params['widths'] ) ) {
4427                         $ig->setWidths( $params['widths'] );
4428                 }
4429                 if ( isset( $params['heights'] ) ) {
4430                         $ig->setHeights( $params['heights'] );
4431                 }
4432
4433                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4434
4435                 $lines = StringUtils::explode( "\n", $text );
4436                 foreach ( $lines as $line ) {
4437                         # match lines like these:
4438                         # Image:someimage.jpg|This is some image
4439                         $matches = array();
4440                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4441                         # Skip empty lines
4442                         if ( count( $matches ) == 0 ) {
4443                                 continue;
4444                         }
4445
4446                         if ( strpos( $matches[0], '%' ) !== false ) {
4447                                 $matches[1] = urldecode( $matches[1] );
4448                         }
4449                         $tp = Title::newFromText( $matches[1] );
4450                         $nt =& $tp;
4451                         if ( is_null( $nt ) ) {
4452                                 # Bogus title. Ignore these so we don't bomb out later.
4453                                 continue;
4454                         }
4455                         if ( isset( $matches[3] ) ) {
4456                                 $label = $matches[3];
4457                         } else {
4458                                 $label = '';
4459                         }
4460
4461                         $html = $this->recursiveTagParse( trim( $label ) );
4462
4463                         $ig->add( $nt, $html );
4464
4465                         # Only add real images (bug #5586)
4466                         if ( $nt->getNamespace() == NS_FILE ) {
4467                                 $this->mOutput->addImage( $nt->getDBkey() );
4468                         }
4469                 }
4470                 return $ig->toHTML();
4471         }
4472
4473         function getImageParams( $handler ) {
4474                 if ( $handler ) {
4475                         $handlerClass = get_class( $handler );
4476                 } else {
4477                         $handlerClass = '';
4478                 }
4479                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4480                         # Initialise static lists
4481                         static $internalParamNames = array(
4482                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4483                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4484                                         'bottom', 'text-bottom' ),
4485                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4486                                         'upright', 'border', 'link', 'alt' ),
4487                         );
4488                         static $internalParamMap;
4489                         if ( !$internalParamMap ) {
4490                                 $internalParamMap = array();
4491                                 foreach ( $internalParamNames as $type => $names ) {
4492                                         foreach ( $names as $name ) {
4493                                                 $magicName = str_replace( '-', '_', "img_$name" );
4494                                                 $internalParamMap[$magicName] = array( $type, $name );
4495                                         }
4496                                 }
4497                         }
4498
4499                         # Add handler params
4500                         $paramMap = $internalParamMap;
4501                         if ( $handler ) {
4502                                 $handlerParamMap = $handler->getParamMap();
4503                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4504                                         $paramMap[$magic] = array( 'handler', $paramName );
4505                                 }
4506                         }
4507                         $this->mImageParams[$handlerClass] = $paramMap;
4508                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4509                 }
4510                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4511         }
4512
4513         /**
4514          * Parse image options text and use it to make an image
4515          * @param Title $title
4516          * @param string $options
4517          * @param LinkHolderArray $holders
4518          */
4519         function makeImage( $title, $options, $holders = false ) {
4520                 # Check if the options text is of the form "options|alt text"
4521                 # Options are:
4522                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4523                 #  * left       no resizing, just left align. label is used for alt= only
4524                 #  * right      same, but right aligned
4525                 #  * none       same, but not aligned
4526                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4527                 #  * center     center the image
4528                 #  * frame      Keep original image size, no magnify-button.
4529                 #  * framed     Same as "frame"
4530                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4531                 #  * upright    reduce width for upright images, rounded to full __0 px
4532                 #  * border     draw a 1px border around the image
4533                 #  * alt        Text for HTML alt attribute (defaults to empty)
4534                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4535                 # vertical-align values (no % or length right now):
4536                 #  * baseline
4537                 #  * sub
4538                 #  * super
4539                 #  * top
4540                 #  * text-top
4541                 #  * middle
4542                 #  * bottom
4543                 #  * text-bottom
4544
4545                 $parts = StringUtils::explode( "|", $options );
4546                 $sk = $this->mOptions->getSkin();
4547
4548                 # Give extensions a chance to select the file revision for us
4549                 $skip = $time = $descQuery = false;
4550                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4551
4552                 if ( $skip ) {
4553                         return $sk->link( $title );
4554                 }
4555
4556                 # Get the file
4557                 $imagename = $title->getDBkey();
4558                 $file = wfFindFile( $title, array( 'time' => $time ) );
4559                 # Get parameter map
4560                 $handler = $file ? $file->getHandler() : false;
4561
4562                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4563
4564                 # Process the input parameters
4565                 $caption = '';
4566                 $params = array( 'frame' => array(), 'handler' => array(),
4567                         'horizAlign' => array(), 'vertAlign' => array() );
4568                 foreach ( $parts as $part ) {
4569                         $part = trim( $part );
4570                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4571                         $validated = false;
4572                         if ( isset( $paramMap[$magicName] ) ) {
4573                                 list( $type, $paramName ) = $paramMap[$magicName];
4574
4575                                 # Special case; width and height come in one variable together
4576                                 if ( $type === 'handler' && $paramName === 'width' ) {
4577                                         $m = array();
4578                                         # (bug 13500) In both cases (width/height and width only),
4579                                         # permit trailing "px" for backward compatibility.
4580                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4581                                                 $width = intval( $m[1] );
4582                                                 $height = intval( $m[2] );
4583                                                 if ( $handler->validateParam( 'width', $width ) ) {
4584                                                         $params[$type]['width'] = $width;
4585                                                         $validated = true;
4586                                                 }
4587                                                 if ( $handler->validateParam( 'height', $height ) ) {
4588                                                         $params[$type]['height'] = $height;
4589                                                         $validated = true;
4590                                                 }
4591                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4592                                                 $width = intval( $value );
4593                                                 if ( $handler->validateParam( 'width', $width ) ) {
4594                                                         $params[$type]['width'] = $width;
4595                                                         $validated = true;
4596                                                 }
4597                                         } # else no validation -- bug 13436
4598                                 } else {
4599                                         if ( $type === 'handler' ) {
4600                                                 # Validate handler parameter
4601                                                 $validated = $handler->validateParam( $paramName, $value );
4602                                         } else {
4603                                                 # Validate internal parameters
4604                                                 switch( $paramName ) {
4605                                                 case 'manualthumb':
4606                                                 case 'alt':
4607                                                         # @todo Fixme: possibly check validity here for
4608                                                         # manualthumb? downstream behavior seems odd with
4609                                                         # missing manual thumbs.
4610                                                         $validated = true;
4611                                                         $value = $this->stripAltText( $value, $holders );
4612                                                         break;
4613                                                 case 'link':
4614                                                         $chars = self::EXT_LINK_URL_CLASS;
4615                                                         $prots = $this->mUrlProtocols;
4616                                                         if ( $value === '' ) {
4617                                                                 $paramName = 'no-link';
4618                                                                 $value = true;
4619                                                                 $validated = true;
4620                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4621                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4622                                                                         $paramName = 'link-url';
4623                                                                         $this->mOutput->addExternalLink( $value );
4624                                                                         $validated = true;
4625                                                                 }
4626                                                         } else {
4627                                                                 $linkTitle = Title::newFromText( $value );
4628                                                                 if ( $linkTitle ) {
4629                                                                         $paramName = 'link-title';
4630                                                                         $value = $linkTitle;
4631                                                                         $this->mOutput->addLink( $linkTitle );
4632                                                                         $validated = true;
4633                                                                 }
4634                                                         }
4635                                                         break;
4636                                                 default:
4637                                                         # Most other things appear to be empty or numeric...
4638                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4639                                                 }
4640                                         }
4641
4642                                         if ( $validated ) {
4643                                                 $params[$type][$paramName] = $value;
4644                                         }
4645                                 }
4646                         }
4647                         if ( !$validated ) {
4648                                 $caption = $part;
4649                         }
4650                 }
4651
4652                 # Process alignment parameters
4653                 if ( $params['horizAlign'] ) {
4654                         $params['frame']['align'] = key( $params['horizAlign'] );
4655                 }
4656                 if ( $params['vertAlign'] ) {
4657                         $params['frame']['valign'] = key( $params['vertAlign'] );
4658                 }
4659
4660                 $params['frame']['caption'] = $caption;
4661
4662                 # Will the image be presented in a frame, with the caption below?
4663                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4664                                  isset( $params['frame']['framed'] ) ||
4665                                  isset( $params['frame']['thumbnail'] ) ||
4666                                  isset( $params['frame']['manualthumb'] );
4667
4668                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4669                 # came to also set the caption, ordinary text after the image -- which
4670                 # makes no sense, because that just repeats the text multiple times in
4671                 # screen readers.  It *also* came to set the title attribute.
4672                 #
4673                 # Now that we have an alt attribute, we should not set the alt text to
4674                 # equal the caption: that's worse than useless, it just repeats the
4675                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4676                 # use the unnamed parameter for alt text as well, just for the time be-
4677                 # ing, if the unnamed param is set and the alt param is not.
4678                 #
4679                 # For the future, we need to figure out if we want to tweak this more,
4680                 # e.g., introducing a title= parameter for the title; ignoring the un-
4681                 # named parameter entirely for images without a caption; adding an ex-
4682                 # plicit caption= parameter and preserving the old magic unnamed para-
4683                 # meter for BC; ...
4684                 if ( $imageIsFramed ) { # Framed image
4685                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4686                                 # No caption or alt text, add the filename as the alt text so
4687                                 # that screen readers at least get some description of the image
4688                                 $params['frame']['alt'] = $title->getText();
4689                         }
4690                         # Do not set $params['frame']['title'] because tooltips don't make sense
4691                         # for framed images
4692                 } else { # Inline image
4693                         if ( !isset( $params['frame']['alt'] ) ) {
4694                                 # No alt text, use the "caption" for the alt text
4695                                 if ( $caption !== '') {
4696                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4697                                 } else {
4698                                         # No caption, fall back to using the filename for the
4699                                         # alt text
4700                                         $params['frame']['alt'] = $title->getText();
4701                                 }
4702                         }
4703                         # Use the "caption" for the tooltip text
4704                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4705                 }
4706
4707                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4708
4709                 # Linker does the rest
4710                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4711
4712                 # Give the handler a chance to modify the parser object
4713                 if ( $handler ) {
4714                         $handler->parserTransformHook( $this, $file );
4715                 }
4716
4717                 return $ret;
4718         }
4719
4720         protected function stripAltText( $caption, $holders ) {
4721                 # Strip bad stuff out of the title (tooltip).  We can't just use
4722                 # replaceLinkHoldersText() here, because if this function is called
4723                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4724                 if ( $holders ) {
4725                         $tooltip = $holders->replaceText( $caption );
4726                 } else {
4727                         $tooltip = $this->replaceLinkHoldersText( $caption );
4728                 }
4729
4730                 # make sure there are no placeholders in thumbnail attributes
4731                 # that are later expanded to html- so expand them now and
4732                 # remove the tags
4733                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4734                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4735
4736                 return $tooltip;
4737         }
4738
4739         /**
4740          * Set a flag in the output object indicating that the content is dynamic and
4741          * shouldn't be cached.
4742          */
4743         function disableCache() {
4744                 wfDebug( "Parser output marked as uncacheable.\n" );
4745                 $this->mOutput->mCacheTime = -1;
4746         }
4747
4748         /**#@+
4749          * Callback from the Sanitizer for expanding items found in HTML attribute
4750          * values, so they can be safely tested and escaped.
4751          * @param string $text
4752          * @param PPFrame $frame
4753          * @return string
4754          * @private
4755          */
4756         function attributeStripCallback( &$text, $frame = false ) {
4757                 $text = $this->replaceVariables( $text, $frame );
4758                 $text = $this->mStripState->unstripBoth( $text );
4759                 return $text;
4760         }
4761
4762         /**#@-*/
4763
4764         /**#@+
4765          * Accessor/mutator
4766          */
4767         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4768         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4769         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4770         /**#@-*/
4771
4772         /**#@+
4773          * Accessor
4774          */
4775         function getTags() {
4776                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4777         }
4778         /**#@-*/
4779
4780
4781         /**
4782          * Break wikitext input into sections, and either pull or replace
4783          * some particular section's text.
4784          *
4785          * External callers should use the getSection and replaceSection methods.
4786          *
4787          * @param string $text Page wikitext
4788          * @param string $section A section identifier string of the form:
4789          *   <flag1> - <flag2> - ... - <section number>
4790          *
4791          * Currently the only recognised flag is "T", which means the target section number
4792          * was derived during a template inclusion parse, in other words this is a template
4793          * section edit link. If no flags are given, it was an ordinary section edit link.
4794          * This flag is required to avoid a section numbering mismatch when a section is
4795          * enclosed by <includeonly> (bug 6563).
4796          *
4797          * The section number 0 pulls the text before the first heading; other numbers will
4798          * pull the given section along with its lower-level subsections. If the section is
4799          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4800          *
4801          * @param string $mode One of "get" or "replace"
4802          * @param string $newText Replacement text for section data.
4803          * @return string for "get", the extracted section text.
4804          *                for "replace", the whole page with the section replaced.
4805          */
4806         private function extractSections( $text, $section, $mode, $newText='' ) {
4807                 global $wgTitle;
4808                 $this->clearState();
4809                 $this->setTitle( $wgTitle ); # not generally used but removes an ugly failure mode
4810                 $this->mOptions = new ParserOptions;
4811                 $this->setOutputType( self::OT_PLAIN );
4812                 $outText = '';
4813                 $frame = $this->getPreprocessor()->newFrame();
4814
4815                 # Process section extraction flags
4816                 $flags = 0;
4817                 $sectionParts = explode( '-', $section );
4818                 $sectionIndex = array_pop( $sectionParts );
4819                 foreach ( $sectionParts as $part ) {
4820                         if ( $part === 'T' ) {
4821                                 $flags |= self::PTD_FOR_INCLUSION;
4822                         }
4823                 }
4824                 # Preprocess the text
4825                 $root = $this->preprocessToDom( $text, $flags );
4826
4827                 # <h> nodes indicate section breaks
4828                 # They can only occur at the top level, so we can find them by iterating the root's children
4829                 $node = $root->getFirstChild();
4830
4831                 # Find the target section
4832                 if ( $sectionIndex == 0 ) {
4833                         # Section zero doesn't nest, level=big
4834                         $targetLevel = 1000;
4835                 } else {
4836             while ( $node ) {
4837                 if ( $node->getName() === 'h' ) {
4838                     $bits = $node->splitHeading();
4839                                         if ( $bits['i'] == $sectionIndex ) {
4840                                         $targetLevel = $bits['level'];
4841                                                 break;
4842                                         }
4843                                 }
4844                                 if ( $mode === 'replace' ) {
4845                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4846                                 }
4847                                 $node = $node->getNextSibling();
4848                         }
4849                 }
4850
4851                 if ( !$node ) {
4852                         # Not found
4853                         if ( $mode === 'get' ) {
4854                                 return $newText;
4855                         } else {
4856                                 return $text;
4857                         }
4858                 }
4859
4860                 # Find the end of the section, including nested sections
4861                 do {
4862                         if ( $node->getName() === 'h' ) {
4863                                 $bits = $node->splitHeading();
4864                                 $curLevel = $bits['level'];
4865                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4866                                         break;
4867                                 }
4868                         }
4869                         if ( $mode === 'get' ) {
4870                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4871                         }
4872                         $node = $node->getNextSibling();
4873                 } while ( $node );
4874
4875                 # Write out the remainder (in replace mode only)
4876                 if ( $mode === 'replace' ) {
4877                         # Output the replacement text
4878                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4879                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4880                         # Only add trailing whitespace if there is newText
4881                         if ( $newText != "" ) {
4882                                 $outText .= $newText . "\n\n";
4883                         }
4884
4885                         while ( $node ) {
4886                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4887                                 $node = $node->getNextSibling();
4888                         }
4889                 }
4890
4891                 if ( is_string( $outText ) ) {
4892                         # Re-insert stripped tags
4893                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4894                 }
4895
4896                 return $outText;
4897         }
4898
4899         /**
4900          * This function returns the text of a section, specified by a number ($section).
4901          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4902          * the first section before any such heading (section 0).
4903          *
4904          * If a section contains subsections, these are also returned.
4905          *
4906          * @param string $text text to look in
4907          * @param string $section section identifier
4908          * @param string $deftext default to return if section is not found
4909          * @return string text of the requested section
4910          */
4911         public function getSection( $text, $section, $deftext='' ) {
4912                 return $this->extractSections( $text, $section, "get", $deftext );
4913         }
4914
4915         public function replaceSection( $oldtext, $section, $text ) {
4916                 return $this->extractSections( $oldtext, $section, "replace", $text );
4917         }
4918
4919         /**
4920          * Get the timestamp associated with the current revision, adjusted for
4921          * the default server-local timestamp
4922          */
4923         function getRevisionTimestamp() {
4924                 if ( is_null( $this->mRevisionTimestamp ) ) {
4925                         wfProfileIn( __METHOD__ );
4926                         global $wgContLang;
4927                         $dbr = wfGetDB( DB_SLAVE );
4928                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4929                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4930
4931                         # Normalize timestamp to internal MW format for timezone processing.
4932                         # This has the added side-effect of replacing a null value with
4933                         # the current time, which gives us more sensible behavior for
4934                         # previews.
4935                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4936
4937                         # The cryptic '' timezone parameter tells to use the site-default
4938                         # timezone offset instead of the user settings.
4939                         #
4940                         # Since this value will be saved into the parser cache, served
4941                         # to other users, and potentially even used inside links and such,
4942                         # it needs to be consistent for all visitors.
4943                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4944
4945                         wfProfileOut( __METHOD__ );
4946                 }
4947                 return $this->mRevisionTimestamp;
4948         }
4949
4950         /**
4951          * Get the name of the user that edited the last revision
4952          */
4953         function getRevisionUser() {
4954                 # if this template is subst: the revision id will be blank,
4955                 # so just use the current user's name
4956                 if ( $this->mRevisionId ) {
4957                         $revision = Revision::newFromId( $this->mRevisionId );
4958                         $revuser = $revision->getUserText();
4959                 } else {
4960                         global $wgUser;
4961                         $revuser = $wgUser->getName();
4962                 }
4963                 return $revuser;
4964         }
4965
4966         /**
4967          * Mutator for $mDefaultSort
4968          *
4969          * @param $sort New value
4970          */
4971         public function setDefaultSort( $sort ) {
4972                 $this->mDefaultSort = $sort;
4973         }
4974
4975         /**
4976          * Accessor for $mDefaultSort
4977          * Will use the title/prefixed title if none is set
4978          *
4979          * @return string
4980          */
4981         public function getDefaultSort() {
4982                 global $wgCategoryPrefixedDefaultSortkey;
4983                 if ( $this->mDefaultSort !== false ) {
4984                         return $this->mDefaultSort;
4985                 } elseif ( $this->mTitle->getNamespace() == NS_CATEGORY ||
4986                         !$wgCategoryPrefixedDefaultSortkey )
4987                 {
4988                         return $this->mTitle->getText();
4989                 } else {
4990                         return $this->mTitle->getPrefixedText();
4991                 }
4992         }
4993
4994         /**
4995          * Accessor for $mDefaultSort
4996          * Unlike getDefaultSort(), will return false if none is set
4997          *
4998          * @return string or false
4999          */
5000         public function getCustomDefaultSort() {
5001                 return $this->mDefaultSort;
5002         }
5003
5004         /**
5005          * Try to guess the section anchor name based on a wikitext fragment
5006          * presumably extracted from a heading, for example "Header" from
5007          * "== Header ==".
5008          */
5009         public function guessSectionNameFromWikiText( $text ) {
5010                 # Strip out wikitext links(they break the anchor)
5011                 $text = $this->stripSectionName( $text );
5012                 $headline = Sanitizer::decodeCharReferences( $text );
5013                 # strip out HTML
5014                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5015                 $headline = trim( $headline );
5016                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5017                 $replacearray = array(
5018                         '%3A' => ':',
5019                         '%' => '.'
5020                 );
5021                 return str_replace(
5022                         array_keys( $replacearray ),
5023                         array_values( $replacearray ),
5024                         $sectionanchor );
5025         }
5026
5027         /**
5028          * Strips a text string of wikitext for use in a section anchor
5029          *
5030          * Accepts a text string and then removes all wikitext from the
5031          * string and leaves only the resultant text (i.e. the result of
5032          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5033          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5034          * to create valid section anchors by mimicing the output of the
5035          * parser when headings are parsed.
5036          *
5037          * @param $text string Text string to be stripped of wikitext
5038          * for use in a Section anchor
5039          * @return Filtered text string
5040          */
5041         public function stripSectionName( $text ) {
5042                 # Strip internal link markup
5043                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5044                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5045
5046                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5047                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5048                 # on how many empty links there are on the page - need to figure that out.
5049                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5050
5051                 # Parse wikitext quotes (italics & bold)
5052                 $text = $this->doQuotes( $text );
5053
5054                 # Strip HTML tags
5055                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5056                 return $text;
5057         }
5058
5059         function srvus( $text ) {
5060                 return $this->testSrvus( $text, $this->mOutputType );
5061         }
5062
5063         /**
5064          * strip/replaceVariables/unstrip for preprocessor regression testing
5065          */
5066         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5067                 $this->clearState();
5068                 if ( !$title instanceof Title ) {
5069                         $title = Title::newFromText( $title );
5070                 }
5071                 $this->mTitle = $title;
5072                 $this->mOptions = $options;
5073                 $this->setOutputType( $outputType );
5074                 $text = $this->replaceVariables( $text );
5075                 $text = $this->mStripState->unstripBoth( $text );
5076                 $text = Sanitizer::removeHTMLtags( $text );
5077                 return $text;
5078         }
5079
5080         function testPst( $text, $title, $options ) {
5081                 global $wgUser;
5082                 if ( !$title instanceof Title ) {
5083                         $title = Title::newFromText( $title );
5084                 }
5085                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5086         }
5087
5088         function testPreprocess( $text, $title, $options ) {
5089                 if ( !$title instanceof Title ) {
5090                         $title = Title::newFromText( $title );
5091                 }
5092                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5093         }
5094
5095         function markerSkipCallback( $s, $callback ) {
5096                 $i = 0;
5097                 $out = '';
5098                 while ( $i < strlen( $s ) ) {
5099                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5100                         if ( $markerStart === false ) {
5101                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5102                                 break;
5103                         } else {
5104                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5105                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5106                                 if ( $markerEnd === false ) {
5107                                         $out .= substr( $s, $markerStart );
5108                                         break;
5109                                 } else {
5110                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5111                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5112                                         $i = $markerEnd;
5113                                 }
5114                         }
5115                 }
5116                 return $out;
5117         }
5118
5119         function serialiseHalfParsedText( $text ) {
5120                 $data = array();
5121                 $data['text'] = $text;
5122
5123                 # First, find all strip markers, and store their
5124                 #  data in an array.
5125                 $stripState = new StripState;
5126                 $pos = 0;
5127                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5128                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5129                 {
5130                         $end_pos += strlen( self::MARKER_SUFFIX );
5131                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5132
5133                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5134                                 $replaceArray = $stripState->general;
5135                                 $stripText = $this->mStripState->general->data[$marker];
5136                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5137                                 $replaceArray = $stripState->nowiki;
5138                                 $stripText = $this->mStripState->nowiki->data[$marker];
5139                         } else {
5140                                 throw new MWException( "Hanging strip marker: '$marker'." );
5141                         }
5142
5143                         $replaceArray->setPair( $marker, $stripText );
5144                         $pos = $end_pos;
5145                 }
5146                 $data['stripstate'] = $stripState;
5147
5148                 # Now, find all of our links, and store THEIR
5149                 #  data in an array! :)
5150                 $links = array( 'internal' => array(), 'interwiki' => array() );
5151                 $pos = 0;
5152
5153                 # Internal links
5154                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5155                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5156
5157                         $ns = trim( $ns );
5158                         if ( empty( $links['internal'][$ns] ) ) {
5159                                 $links['internal'][$ns] = array();
5160                         }
5161
5162                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5163                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5164                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5165                 }
5166
5167                 $pos = 0;
5168
5169                 # Interwiki links
5170                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5171                         $data = substr( $text, $start_pos );
5172                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5173                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5174                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5175                 }
5176
5177                 $data['linkholder'] = $links;
5178
5179                 return $data;
5180         }
5181
5182         /**
5183          * TODO: document
5184          * @param $data Array
5185          * @param $intPrefix String unique identifying prefix
5186          * @return String
5187          */
5188         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5189                 if ( !$intPrefix ) {
5190                         $intPrefix = $this->getRandomString();
5191                 }
5192
5193                 # First, extract the strip state.
5194                 $stripState = $data['stripstate'];
5195                 $this->mStripState->general->merge( $stripState->general );
5196                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5197
5198                 # Now, extract the text, and renumber links
5199                 $text = $data['text'];
5200                 $links = $data['linkholder'];
5201
5202                 # Internal...
5203                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5204                         foreach ( $nsLinks as $key => $entry ) {
5205                                 $newKey = $intPrefix . '-' . $key;
5206                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5207
5208                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5209                         }
5210                 }
5211
5212                 # Interwiki...
5213                 foreach ( $links['interwiki'] as $key => $entry ) {
5214                         $newKey = "$intPrefix-$key";
5215                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5216
5217                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5218                 }
5219
5220                 # Should be good to go.
5221                 return $text;
5222         }
5223 }
5224
5225 /**
5226  * @todo document, briefly.
5227  * @ingroup Parser
5228  */
5229 class StripState {
5230         var $general, $nowiki;
5231
5232         function __construct() {
5233                 $this->general = new ReplacementArray;
5234                 $this->nowiki = new ReplacementArray;
5235         }
5236
5237         function unstripGeneral( $text ) {
5238                 wfProfileIn( __METHOD__ );
5239                 do {
5240                         $oldText = $text;
5241                         $text = $this->general->replace( $text );
5242                 } while ( $text !== $oldText );
5243                 wfProfileOut( __METHOD__ );
5244                 return $text;
5245         }
5246
5247         function unstripNoWiki( $text ) {
5248                 wfProfileIn( __METHOD__ );
5249                 do {
5250                         $oldText = $text;
5251                         $text = $this->nowiki->replace( $text );
5252                 } while ( $text !== $oldText );
5253                 wfProfileOut( __METHOD__ );
5254                 return $text;
5255         }
5256
5257         function unstripBoth( $text ) {
5258                 wfProfileIn( __METHOD__ );
5259                 do {
5260                         $oldText = $text;
5261                         $text = $this->general->replace( $text );
5262                         $text = $this->nowiki->replace( $text );
5263                 } while ( $text !== $oldText );
5264                 wfProfileOut( __METHOD__ );
5265                 return $text;
5266         }
5267 }
5268
5269 /**
5270  * @todo document, briefly.
5271  * @ingroup Parser
5272  */
5273 class OnlyIncludeReplacer {
5274         var $output = '';
5275
5276         function replace( $matches ) {
5277                 if ( substr( $matches[1], -1 ) === "\n" ) {
5278                         $this->output .= substr( $matches[1], 0, -1 );
5279                 } else {
5280                         $this->output .= $matches[1];
5281                 }
5282         }
5283 }