includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *     produces HTML output
  21  * preSaveTransform().
  22  *     produces altered wiki markup.
  23  * preprocess()
  24  *     removes HTML comments and expands templates
  25  * cleanSig()
  26  *     Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *     Extracts sections from an article for section editing
  29  * getPreloadText()
  30  *     Removes <noinclude> sections, and <includeonly> tags.
  31  *
  32  * Globals used:
  33  *    objects:   $wgLang, $wgContLang
  34  *
  35  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  36  *
  37  * settings:
  38  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  39  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  40  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  41  *  $wgMaxArticleSize*
  42  *
  43  *  * only within ParserOptions
  44  * </pre>
  45  *
  46  * @ingroup Parser
  47  */
  48 class Parser {
  49         /**
  50          * Update this version number when the ParserOutput format
  51          * changes in an incompatible way, so the parser cache
  52          * can automatically discard old data.
  53          */
  54         const VERSION = '1.6.4';
  55
  56         # Flags for Parser::setFunctionHook
  57         # Also available as global constants from Defines.php
  58         const SFH_NO_HASH = 1;
  59         const SFH_OBJECT_ARGS = 2;
  60
  61         # Constants needed for external link processing
  62         # Everything except bracket, space, or control characters
  63         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  64         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  65                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  66
  67         # State constants for the definition list colon extraction
  68         const COLON_STATE_TEXT = 0;
  69         const COLON_STATE_TAG = 1;
  70         const COLON_STATE_TAGSTART = 2;
  71         const COLON_STATE_CLOSETAG = 3;
  72         const COLON_STATE_TAGSLASH = 4;
  73         const COLON_STATE_COMMENT = 5;
  74         const COLON_STATE_COMMENTDASH = 6;
  75         const COLON_STATE_COMMENTDASHDASH = 7;
  76
  77         # Flags for preprocessToDom
  78         const PTD_FOR_INCLUSION = 1;
  79
  80         # Allowed values for $this->mOutputType
  81         # Parameter to startExternalParse().
  82         const OT_HTML = 1; # like parse()
  83         const OT_WIKI = 2; # like preSaveTransform()
  84         const OT_PREPROCESS = 3; # like preprocess()
  85         const OT_MSG = 3;
  86         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  87
  88         # Marker Suffix needs to be accessible staticly.
  89         const MARKER_SUFFIX = "-QINU\x7f";
  90
  91         # Persistent:
  92         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  93         var $mSubstWords, $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex;
  94         var $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList;
  95         var $mVarCache, $mConf, $mFunctionTagHooks;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mLinkHolders, $mLinkID;
 102         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 103         var $mTplExpandCache; # empty-frame expansion cache
 104         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 105         var $mExpensiveFunctionCount; # number of expensive parser function calls
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions;      # ParserOptions object
 110         var $mTitle;        # Title context, used for self-link rendering and similar things
 111         var $mOutputType;   # Output type, one of the OT_xxx constants
 112         var $ot;            # Shortcut alias, see setOutputType()
 113         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 114         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 115         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 116
 117         /**
 118          * Constructor
 119          *
 120          * @public
 121          */
 122         function __construct( $conf = array() ) {
 123                 $this->mConf = $conf;
 124                 $this->mTagHooks = array();
 125                 $this->mTransparentTagHooks = array();
 126                 $this->mFunctionHooks = array();
 127                 $this->mFunctionTagHooks = array();
 128                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 129                 $this->mDefaultStripList = $this->mStripList = array();
 130                 $this->mUrlProtocols = wfUrlProtocols();
 131                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 132                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 133                 $this->mVarCache = array();
 134                 if ( isset( $conf['preprocessorClass'] ) ) {
 135                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 136                 } elseif ( extension_loaded( 'domxml' ) ) {
 137                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 138                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 139                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 140                 } elseif ( extension_loaded( 'dom' ) ) {
 141                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 142                 } else {
 143                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 144                 }
 145                 $this->mMarkerIndex = 0;
 146                 $this->mFirstCall = true;
 147         }
 148
 149         /**
 150          * Reduce memory usage to reduce the impact of circular references
 151          */
 152         function __destruct() {
 153                 if ( isset( $this->mLinkHolders ) ) {
 154                         $this->mLinkHolders->__destruct();
 155                 }
 156                 foreach ( $this as $name => $value ) {
 157                         unset( $this->$name );
 158                 }
 159         }
 160
 161         /**
 162          * Do various kinds of initialisation on the first call of the parser
 163          */
 164         function firstCallInit() {
 165                 if ( !$this->mFirstCall ) {
 166                         return;
 167                 }
 168                 $this->mFirstCall = false;
 169
 170                 wfProfileIn( __METHOD__ );
 171
 172                 CoreParserFunctions::register( $this );
 173                 CoreTagHooks::register( $this );
 174                 $this->initialiseVariables();
 175
 176                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 177                 wfProfileOut( __METHOD__ );
 178         }
 179
 180         /**
 181          * Clear Parser state
 182          *
 183          * @private
 184          */
 185         function clearState() {
 186                 wfProfileIn( __METHOD__ );
 187                 if ( $this->mFirstCall ) {
 188                         $this->firstCallInit();
 189                 }
 190                 $this->mOutput = new ParserOutput;
 191                 $this->mAutonumber = 0;
 192                 $this->mLastSection = '';
 193                 $this->mDTopen = false;
 194                 $this->mIncludeCount = array();
 195                 $this->mStripState = new StripState;
 196                 $this->mArgStack = false;
 197                 $this->mInPre = false;
 198                 $this->mLinkHolders = new LinkHolderArray( $this );
 199                 $this->mLinkID = 0;
 200                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 201                 $this->mVarCache = array();
 202
 203                 /**
 204                  * Prefix for temporary replacement strings for the multipass parser.
 205                  * \x07 should never appear in input as it's disallowed in XML.
 206                  * Using it at the front also gives us a little extra robustness
 207                  * since it shouldn't match when butted up against identifier-like
 208                  * string constructs.
 209                  *
 210                  * Must not consist of all title characters, or else it will change
 211                  * the behaviour of <nowiki> in a link.
 212                  */
 213                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 214                 # Changed to \x7f to allow XML double-parsing -- TS
 215                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 216
 217
 218                 # Clear these on every parse, bug 4549
 219                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 220
 221                 $this->mShowToc = true;
 222                 $this->mForceTocPosition = false;
 223                 $this->mIncludeSizes = array(
 224                         'post-expand' => 0,
 225                         'arg' => 0,
 226                 );
 227                 $this->mPPNodeCount = 0;
 228                 $this->mDefaultSort = false;
 229                 $this->mHeadings = array();
 230                 $this->mDoubleUnderscores = array();
 231                 $this->mExpensiveFunctionCount = 0;
 232
 233                 # Fix cloning
 234                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 235                         $this->mPreprocessor = null;
 236                 }
 237
 238                 wfRunHooks( 'ParserClearState', array( &$this ) );
 239                 wfProfileOut( __METHOD__ );
 240         }
 241
 242         /**
 243          * Convert wikitext to HTML
 244          * Do not call this function recursively.
 245          *
 246          * @param $text String: text we want to parse
 247          * @param $title A title object
 248          * @param $options ParserOptions
 249          * @param $linestart boolean
 250          * @param $clearState boolean
 251          * @param $revid Int: number to pass in {{REVISIONID}}
 252          * @return ParserOutput a ParserOutput
 253          */
 254         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 255                 /**
 256                  * First pass--just handle <nowiki> sections, pass the rest off
 257                  * to internalParse() which does all the real work.
 258                  */
 259
 260                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 261                 $fname = __METHOD__.'-' . wfGetCaller();
 262                 wfProfileIn( __METHOD__ );
 263                 wfProfileIn( $fname );
 264
 265                 if ( $clearState ) {
 266                         $this->clearState();
 267                 }
 268
 269                 $this->mOptions = $options;
 270                 $this->setTitle( $title ); # Page title has to be set for the pre-processor
 271
 272                 $oldRevisionId = $this->mRevisionId;
 273                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 274                 if ( $revid !== null ) {
 275                         $this->mRevisionId = $revid;
 276                         $this->mRevisionTimestamp = null;
 277                 }
 278                 $this->setOutputType( self::OT_HTML );
 279                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 280                 # No more strip!
 281                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 282                 $text = $this->internalParse( $text );
 283
 284                 $text = $this->mStripState->unstripGeneral( $text );
 285
 286                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 287                 $fixtags = array(
 288                         # french spaces, last one Guillemet-left
 289                         # only if there is something before the space
 290                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 291                         # french spaces, Guillemet-right
 292                         '/(\\302\\253) /' => '\\1&#160;',
 293                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 294                 );
 295                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 296
 297                 $text = $this->doBlockLevels( $text, $linestart );
 298
 299                 $this->replaceLinkHolders( $text );
 300
 301                 /**
 302                  * The page doesn't get language converted if
 303                  * a) It's disabled
 304                  * b) Content isn't converted
 305                  * c) It's a conversion table
 306                  */
 307                 if ( !( $wgDisableLangConversion
 308                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 309                                 || $this->mTitle->isConversionTable() ) ) {
 310
 311                         # The position of the convert() call should not be changed. it
 312                         # assumes that the links are all replaced and the only thing left
 313                         # is the <nowiki> mark.
 314
 315                         $text = $wgContLang->convert( $text );
 316                 }
 317
 318                 /**
 319                  * A page get its title converted except:
 320                  * a) Language conversion is globally disabled
 321                  * b) Title convert is globally disabled
 322                  * c) The page is a redirect page
 323                  * d) User request with a "linkconvert" set to "no"
 324                  * e) A "nocontentconvert" magic word has been set
 325                  * f) A "notitleconvert" magic word has been set
 326                  * g) User sets "noconvertlink" in his/her preference
 327                  *
 328                  * Note that if a user tries to set a title in a conversion
 329                  * rule but content conversion was not done, then the parser
 330                  * won't pick it up.  This is probably expected behavior.
 331                  */
 332                 if ( !( $wgDisableLangConversion
 333                                 || $wgDisableTitleConversion
 334                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 335                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 336                                 || $this->mOutput->getDisplayTitle() !== false ) )
 337                 {
 338                         $convruletitle = $wgContLang->getConvRuleTitle();
 339                         if ( $convruletitle ) {
 340                                 $this->mOutput->setTitleText( $convruletitle );
 341                         } else {
 342                                 $titleText = $wgContLang->convertTitle( $title );
 343                                 $this->mOutput->setTitleText( $titleText );
 344                         }
 345                 }
 346
 347                 $text = $this->mStripState->unstripNoWiki( $text );
 348
 349                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 350
 351 //!JF Move to its own function
 352
 353                 $uniq_prefix = $this->mUniqPrefix;
 354                 $matches = array();
 355                 $elements = array_keys( $this->mTransparentTagHooks );
 356                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 357
 358                 foreach ( $matches as $marker => $data ) {
 359                         list( $element, $content, $params, $tag ) = $data;
 360                         $tagName = strtolower( $element );
 361                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 362                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 363                         } else {
 364                                 $output = $tag;
 365                         }
 366                         $this->mStripState->general->setPair( $marker, $output );
 367                 }
 368                 $text = $this->mStripState->unstripGeneral( $text );
 369
 370                 $text = Sanitizer::normalizeCharReferences( $text );
 371
 372                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 373                         $text = MWTidy::tidy( $text );
 374                 } else {
 375                         # attempt to sanitize at least some nesting problems
 376                         # (bug #2702 and quite a few others)
 377                         $tidyregs = array(
 378                                 # ''Something [http://www.cool.com cool''] -->
 379                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 380                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 381                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 382                                 # fix up an anchor inside another anchor, only
 383                                 # at least for a single single nested link (bug 3695)
 384                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 385                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 386                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 387                                 # contains a div, so fix it up here; replace
 388                                 # div with escaped text
 389                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 390                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 391                                 # remove empty italic or bold tag pairs, some
 392                                 # introduced by rules above
 393                                 '/<([bi])><\/\\1>/' => '',
 394                         );
 395
 396                         $text = preg_replace(
 397                                 array_keys( $tidyregs ),
 398                                 array_values( $tidyregs ),
 399                                 $text );
 400                 }
 401                 global $wgExpensiveParserFunctionLimit;
 402                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 403                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 404                 }
 405
 406                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 407
 408                 # Information on include size limits, for the benefit of users who try to skirt them
 409                 if ( $this->mOptions->getEnableLimitReport() ) {
 410                         $max = $this->mOptions->getMaxIncludeSize();
 411                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 412                         $limitReport =
 413                                 "NewPP limit report\n" .
 414                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 415                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 416                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 417                                 $PFreport;
 418                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 419                         $text .= "\n<!-- \n$limitReport-->\n";
 420                 }
 421                 $this->mOutput->setText( $text );
 422
 423                 $this->mRevisionId = $oldRevisionId;
 424                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 425                 wfProfileOut( $fname );
 426                 wfProfileOut( __METHOD__ );
 427
 428                 return $this->mOutput;
 429         }
 430
 431         /**
 432          * Recursive parser entry point that can be called from an extension tag
 433          * hook.
 434          *
 435          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 436          *
 437          * @param $text String: text extension wants to have parsed
 438          * @param $frame PPFrame: The frame to use for expanding any template variables
 439          */
 440         function recursiveTagParse( $text, $frame=false ) {
 441                 wfProfileIn( __METHOD__ );
 442                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 443                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 444                 $text = $this->internalParse( $text, false, $frame );
 445                 wfProfileOut( __METHOD__ );
 446                 return $text;
 447         }
 448
 449         /**
 450          * Expand templates and variables in the text, producing valid, static wikitext.
 451          * Also removes comments.
 452          */
 453         function preprocess( $text, $title, $options, $revid = null ) {
 454                 wfProfileIn( __METHOD__ );
 455                 $this->clearState();
 456                 $this->setOutputType( self::OT_PREPROCESS );
 457                 $this->mOptions = $options;
 458                 $this->setTitle( $title );
 459                 if ( $revid !== null ) {
 460                         $this->mRevisionId = $revid;
 461                 }
 462                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 463                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 464                 $text = $this->replaceVariables( $text );
 465                 $text = $this->mStripState->unstripBoth( $text );
 466                 wfProfileOut( __METHOD__ );
 467                 return $text;
 468         }
 469
 470         /**
 471          * Process the wikitext for the ?preload= feature. (bug 5210)
 472          *
 473          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 474          * comments, templates, arguments, tags hooks and parser functions are untouched.
 475          */
 476         public function getPreloadText( $text, $title, $options ) {
 477                 # Parser (re)initialisation
 478                 $this->clearState();
 479                 $this->setOutputType( self::OT_PLAIN );
 480                 $this->mOptions = $options;
 481                 $this->setTitle( $title );
 482
 483                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 484                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 485                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 486         }
 487
 488         /**
 489          * Get a random string
 490          *
 491          * @private
 492          * @static
 493          */
 494         static function getRandomString() {
 495                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 496         }
 497
 498         /**
 499          * Accessor for mUniqPrefix.
 500          *
 501          * @return String
 502          */
 503         public function uniqPrefix() {
 504                 if ( !isset( $this->mUniqPrefix ) ) {
 505                         # @todo Fixme: this is probably *horribly wrong*
 506                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 507                         # if this is called for a parser cache hit, the parser may not
 508                         # have ever been initialized in the first place.
 509                         # Not really sure what the heck is supposed to be going on here.
 510                         return '';
 511                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 512                 }
 513                 return $this->mUniqPrefix;
 514         }
 515
 516         /**
 517          * Set the context title
 518          */
 519         function setTitle( $t ) {
 520                 if ( !$t || $t instanceof FakeTitle ) {
 521                         $t = Title::newFromText( 'NO TITLE' );
 522                 }
 523
 524                 if ( strval( $t->getFragment() ) !== '' ) {
 525                         # Strip the fragment to avoid various odd effects
 526                         $this->mTitle = clone $t;
 527                         $this->mTitle->setFragment( '' );
 528                 } else {
 529                         $this->mTitle = $t;
 530                 }
 531         }
 532
 533         /**
 534          * Accessor for the Title object
 535          *
 536          * @return Title object
 537          */
 538         function &getTitle() {
 539                 return $this->mTitle;
 540         }
 541
 542         /**
 543          * Accessor/mutator for the Title object
 544          *
 545          * @param $x New Title object or null to just get the current one
 546          * @return Title object
 547          */
 548         function Title( $x = null ) {
 549                 return wfSetVar( $this->mTitle, $x );
 550         }
 551
 552         /**
 553          * Set the output type
 554          *
 555          * @param $ot Integer: new value
 556          */
 557         function setOutputType( $ot ) {
 558                 $this->mOutputType = $ot;
 559                 # Shortcut alias
 560                 $this->ot = array(
 561                         'html' => $ot == self::OT_HTML,
 562                         'wiki' => $ot == self::OT_WIKI,
 563                         'pre' => $ot == self::OT_PREPROCESS,
 564                         'plain' => $ot == self::OT_PLAIN,
 565                 );
 566         }
 567
 568         /**
 569          * Accessor/mutator for the output type
 570          *
 571          * @param $x New value or null to just get the current one
 572          * @return Integer
 573          */
 574         function OutputType( $x = null ) {
 575                 return wfSetVar( $this->mOutputType, $x );
 576         }
 577
 578         /**
 579          * Get the ParserOutput object
 580          *
 581          * @return ParserOutput object
 582          */
 583         function getOutput() {
 584                 return $this->mOutput;
 585         }
 586
 587         /**
 588          * Get the ParserOptions object
 589          *
 590          * @return ParserOptions object
 591          */
 592         function getOptions() {
 593                 return $this->mOptions;
 594         }
 595
 596         /**
 597          * Accessor/mutator for the ParserOptions object
 598          *
 599          * @param $x New value or null to just get the current one
 600          * @return Current ParserOptions object
 601          */
 602         function Options( $x = null ) {
 603                 return wfSetVar( $this->mOptions, $x );
 604         }
 605
 606         function nextLinkID() {
 607                 return $this->mLinkID++;
 608         }
 609
 610         function getFunctionLang() {
 611                 global $wgLang, $wgContLang;
 612
 613                 $target = $this->mOptions->getTargetLanguage();
 614                 if ( $target !== null ) {
 615                         return $target;
 616                 } else {
 617                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 618                 }
 619         }
 620
 621         /**
 622          * Get a preprocessor object
 623          *
 624          * @return Preprocessor instance
 625          */
 626         function getPreprocessor() {
 627                 if ( !isset( $this->mPreprocessor ) ) {
 628                         $class = $this->mPreprocessorClass;
 629                         $this->mPreprocessor = new $class( $this );
 630                 }
 631                 return $this->mPreprocessor;
 632         }
 633
 634         /**
 635          * Replaces all occurrences of HTML-style comments and the given tags
 636          * in the text with a random marker and returns the next text. The output
 637          * parameter $matches will be an associative array filled with data in
 638          * the form:
 639          *   'UNIQ-xxxxx' => array(
 640          *     'element',
 641          *     'tag content',
 642          *     array( 'param' => 'x' ),
 643          *     '<element param="x">tag content</element>' ) )
 644          *
 645          * @param $elements list of element names. Comments are always extracted.
 646          * @param $text Source text string.
 647          * @param $matches Out parameter, Array: extarcted tags
 648          * @param $uniq_prefix
 649          * @return String: stripped text
 650          *
 651          * @static
 652          */
 653         public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 654                 static $n = 1;
 655                 $stripped = '';
 656                 $matches = array();
 657
 658                 $taglist = implode( '|', $elements );
 659                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 660
 661                 while ( $text != '' ) {
 662                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 663                         $stripped .= $p[0];
 664                         if ( count( $p ) < 5 ) {
 665                                 break;
 666                         }
 667                         if ( count( $p ) > 5 ) {
 668                                 # comment
 669                                 $element    = $p[4];
 670                                 $attributes = '';
 671                                 $close      = '';
 672                                 $inside     = $p[5];
 673                         } else {
 674                                 # tag
 675                                 $element    = $p[1];
 676                                 $attributes = $p[2];
 677                                 $close      = $p[3];
 678                                 $inside     = $p[4];
 679                         }
 680
 681                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 682                         $stripped .= $marker;
 683
 684                         if ( $close === '/>' ) {
 685                                 # Empty element tag, <tag />
 686                                 $content = null;
 687                                 $text = $inside;
 688                                 $tail = null;
 689                         } else {
 690                                 if ( $element === '!--' ) {
 691                                         $end = '/(-->)/';
 692                                 } else {
 693                                         $end = "/(<\\/$element\\s*>)/i";
 694                                 }
 695                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 696                                 $content = $q[0];
 697                                 if ( count( $q ) < 3 ) {
 698                                         # No end tag -- let it run out to the end of the text.
 699                                         $tail = '';
 700                                         $text = '';
 701                                 } else {
 702                                         $tail = $q[1];
 703                                         $text = $q[2];
 704                                 }
 705                         }
 706
 707                         $matches[$marker] = array( $element,
 708                                 $content,
 709                                 Sanitizer::decodeTagAttributes( $attributes ),
 710                                 "<$element$attributes$close$content$tail" );
 711                 }
 712                 return $stripped;
 713         }
 714
 715         /**
 716          * Get a list of strippable XML-like elements
 717          */
 718         function getStripList() {
 719                 return $this->mStripList;
 720         }
 721
 722         /**
 723          * @deprecated use replaceVariables
 724          */
 725         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 726                 return $text;
 727         }
 728
 729         /**
 730          * Restores pre, math, and other extensions removed by strip()
 731          *
 732          * always call unstripNoWiki() after this one
 733          * @private
 734          * @deprecated use $this->mStripState->unstrip()
 735          */
 736         function unstrip( $text, $state ) {
 737                 return $state->unstripGeneral( $text );
 738         }
 739
 740         /**
 741          * Always call this after unstrip() to preserve the order
 742          *
 743          * @private
 744          * @deprecated use $this->mStripState->unstrip()
 745          */
 746         function unstripNoWiki( $text, $state ) {
 747                 return $state->unstripNoWiki( $text );
 748         }
 749
 750         /**
 751          * @deprecated use $this->mStripState->unstripBoth()
 752          */
 753         function unstripForHTML( $text ) {
 754                 return $this->mStripState->unstripBoth( $text );
 755         }
 756
 757         /**
 758          * Add an item to the strip state
 759          * Returns the unique tag which must be inserted into the stripped text
 760          * The tag will be replaced with the original text in unstrip()
 761          *
 762          * @private
 763          */
 764         function insertStripItem( $text ) {
 765                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 766                 $this->mMarkerIndex++;
 767                 $this->mStripState->general->setPair( $rnd, $text );
 768                 return $rnd;
 769         }
 770
 771         /**
 772          * Interface with html tidy
 773          * @deprecated Use MWTidy::tidy()
 774          */
 775         public static function tidy( $text ) {
 776                 wfDeprecated( __METHOD__ );
 777                 return MWTidy::tidy( $text );
 778         }
 779
 780         /**
 781          * parse the wiki syntax used to render tables
 782          *
 783          * @private
 784          */
 785         function doTableStuff( $text ) {
 786                 wfProfileIn( __METHOD__ );
 787
 788                 $lines = StringUtils::explode( "\n", $text );
 789                 $text = null;
 790                 $out = '';
 791                 $td_history = array(); # Is currently a td tag open?
 792                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 793                 $tr_history = array(); # Is currently a tr tag open?
 794                 $tr_attributes = array(); # history of tr attributes
 795                 $has_opened_tr = array(); # Did this table open a <tr> element?
 796                 $indent_level = 0; # indent level of the table
 797
 798                 $table_tag = 'table';
 799                 $tr_tag = 'tr';
 800                 $th_tag = 'th';
 801                 $td_tag = 'td';
 802                 $caption_tag = 'caption';
 803
 804                 $extra_table_attribs = null;
 805                 $extra_tr_attribs = null;
 806                 $extra_td_attribs = null;
 807
 808                 $convert_style = false;
 809
 810                 foreach ( $lines as $outLine ) {
 811                         $line = trim( $outLine );
 812
 813                         if ( $line === '' ) { # empty line, go to next line
 814                                 $out .= $outLine."\n";
 815                                 continue;
 816                         }
 817
 818                         $first_character = $line[0];
 819                         $matches = array();
 820
 821                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 822                                 # First check if we are starting a new table
 823                                 $indent_level = strlen( $matches[1] );
 824
 825                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 826
 827                                 $attr = Sanitizer::decodeTagAttributes( $attributes );
 828
 829                                 $mode = @$attr['mode'];
 830                                 if ( !$mode ) $mode = 'data';
 831
 832                                 if ( $mode == 'grid' || $mode == 'layout' ) {
 833                                         $table_tag = 'div';
 834                                         $tr_tag = 'div';
 835                                         $th_tag = 'div';
 836                                         $td_tag = 'div';
 837                                         $caption_tag = 'div';
 838
 839                                         $extra_table_attribs = array( 'class' => 'grid-table' );
 840                                         $extra_tr_attribs = array( 'class' => 'grid-row' );
 841                                         $extra_td_attribs = array( 'class' => 'grid-cell' );
 842
 843                                         $convert_style = true;
 844                                 }
 845
 846                                 if ($convert_style) $attr['style'] = Sanitizer::styleFromAttributes( $attr );
 847                                 $attr = Sanitizer::validateTagAttributes( $attr, $table_tag );
 848                                 $attributes = Sanitizer::collapseTagAttributes( $attr, $extra_table_attribs );
 849
 850                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<$table_tag{$attributes}>";
 851                                 array_push( $td_history , false );
 852                                 array_push( $last_tag_history , '' );
 853                                 array_push( $tr_history , false );
 854                                 array_push( $tr_attributes , '' );
 855                                 array_push( $has_opened_tr , false );
 856                         } elseif ( count( $td_history ) == 0 ) {
 857                                 # Don't do any of the following
 858                                 $out .= $outLine."\n";
 859                                 continue;
 860                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 861                                 # We are ending a table
 862                                 $line = "</$table_tag>" . substr( $line , 2 );
 863                                 $last_tag = array_pop( $last_tag_history );
 864
 865                                 if ( !array_pop( $has_opened_tr ) ) {
 866                                         $line = "<$tr_tag><$td_tag></$td_tag></$tr_tag>{$line}";
 867                                 }
 868
 869                                 if ( array_pop( $tr_history ) ) {
 870                                         $line = "</$tr_tag>{$line}";
 871                                 }
 872
 873                                 if ( array_pop( $td_history ) ) {
 874                                         $line = "</{$last_tag}>{$line}";
 875                                 }
 876                                 array_pop( $tr_attributes );
 877                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 878                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 879                                 # Now we have a table row
 880                                 $line = preg_replace( '#^\|-+#', '', $line );
 881
 882                                 # Whats after the tag is now only attributes
 883                                 $attributes = $this->mStripState->unstripBoth( $line );
 884
 885                                 $attr = Sanitizer::decodeTagAttributes( $attributes );
 886                                 if ($convert_style) $attr['style'] = Sanitizer::styleFromAttributes( $attr );
 887                                 $attr = Sanitizer::validateTagAttributes( $attr, $tr_tag );
 888                                 $attributes = Sanitizer::collapseTagAttributes( $attr, $extra_tr_attribs );
 889
 890                                 array_pop( $tr_attributes );
 891                                 array_push( $tr_attributes, $attributes );
 892
 893                                 $line = '';
 894                                 $last_tag = array_pop( $last_tag_history );
 895                                 array_pop( $has_opened_tr );
 896                                 array_push( $has_opened_tr , true );
 897
 898                                 if ( array_pop( $tr_history ) ) {
 899                                         $line = "</$tr_tag>";
 900                                 }
 901
 902                                 if ( array_pop( $td_history ) ) {
 903                                         $line = "</{$last_tag}>{$line}";
 904                                 }
 905
 906                                 $outLine = $line;
 907                                 array_push( $tr_history , false );
 908                                 array_push( $td_history , false );
 909                                 array_push( $last_tag_history , '' );
 910                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 911                                 # This might be cell elements, td, th or captions
 912                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 913                                         $first_character = '+';
 914                                         $line = substr( $line , 1 );
 915                                 }
 916
 917                                 $line = substr( $line , 1 );
 918
 919                                 if ( $first_character === '!' ) {
 920                                         $line = str_replace( '!!' , '||' , $line );
 921                                 }
 922
 923                                 # Split up multiple cells on the same line.
 924                                 # FIXME : This can result in improper nesting of tags processed
 925                                 # by earlier parser steps, but should avoid splitting up eg
 926                                 # attribute values containing literal "||".
 927                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 928
 929                                 $outLine = '';
 930
 931                                 # Loop through each table cell
 932                                 foreach ( $cells as $cell ) {
 933                                         $previous = '';
 934                                         if ( $first_character !== '+' ) {
 935                                                 $tr_after = array_pop( $tr_attributes );
 936                                                 if ( !array_pop( $tr_history ) ) {
 937                                                         $previous = "<$tr_tag{$tr_after}>\n";
 938                                                 }
 939                                                 array_push( $tr_history , true );
 940                                                 array_push( $tr_attributes , '' );
 941                                                 array_pop( $has_opened_tr );
 942                                                 array_push( $has_opened_tr , true );
 943                                         }
 944
 945                                         $last_tag = array_pop( $last_tag_history );
 946
 947                                         if ( array_pop( $td_history ) ) {
 948                                                 $previous = "</{$last_tag}>{$previous}";
 949                                         }
 950
 951                                         if ( $first_character === '|' ) {
 952                                                 $last_tag = $td_tag;
 953                                         } elseif ( $first_character === '!' ) {
 954                                                 $last_tag = $th_tag;
 955                                         } elseif ( $first_character === '+' ) {
 956                                                 $last_tag = $caption_tag;
 957                                         } else {
 958                                                 $last_tag = '';
 959                                         }
 960
 961                                         array_push( $last_tag_history , $last_tag );
 962
 963                                         # A cell could contain both parameters and data
 964                                         $cell_data = explode( '|' , $cell , 2 );
 965
 966                                         $attributes = '';
 967
 968                                         # Bug 553: Note that a '|' inside an invalid link should not
 969                                         # be mistaken as delimiting cell parameters
 970                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 971                                                 if ($extra_td_attribs) $attributes = Sanitizer::collapseTagAttributes( $extra_td_attribs );
 972                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell}";
 973                                         } elseif ( count( $cell_data ) == 1 ) {
 974                                                 if ($extra_td_attribs) $attributes = Sanitizer::collapseTagAttributes( $extra_td_attribs );
 975                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[0]}";
 976                                         } else {
 977                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 978
 979                                                 $attr = Sanitizer::decodeTagAttributes( $attributes );
 980                                                 if ($convert_style) $attr['style'] = Sanitizer::styleFromAttributes( $attr );
 981                                                 $attr = Sanitizer::validateTagAttributes( $attr, $last_tag );
 982                                                 $attributes = Sanitizer::collapseTagAttributes( $attr, $extra_td_attribs );
 983
 984                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 985                                         }
 986
 987                                         $outLine .= $cell;
 988                                         array_push( $td_history , true );
 989                                 }
 990                         }
 991                         $out .= $outLine . "\n";
 992                 }
 993
 994                 # Closing open td, tr && table
 995                 while ( count( $td_history ) > 0 ) {
 996                         if ( array_pop( $td_history ) ) {
 997                                 $out .= "</$td_tag>\n";
 998                         }
 999                         if ( array_pop( $tr_history ) ) {
1000                                 $out .= "</$tr_tag>\n";
1001                         }
1002                         if ( !array_pop( $has_opened_tr ) ) {
1003                                 $out .= "<$tr_tag><$td_tag></$td_tag></$tr_tag>\n" ;
1004                         }
1005
1006                         $out .= "</$table_tag>\n";
1007                 }
1008
1009                 # Remove trailing line-ending (b/c)
1010                 if ( substr( $out, -1 ) === "\n" ) {
1011                         $out = substr( $out, 0, -1 );
1012                 }
1013
1014                 # special case: don't return empty table
1015                 if ( $out === "<$table_tag>\n<$tr_tag><$td_tag></$td_tag></$tr_tag>\n</$table_tag>" ) {
1016                         $out = '';
1017                 }
1018
1019                 wfProfileOut( __METHOD__ );
1020
1021                 return $out;
1022         }
1023
1024         /**
1025          * Helper function for parse() that transforms wiki markup into
1026          * HTML. Only called for $mOutputType == self::OT_HTML.
1027          *
1028          * @private
1029          */
1030         function internalParse( $text, $isMain = true, $frame=false ) {
1031                 wfProfileIn( __METHOD__ );
1032
1033                 $origText = $text;
1034
1035                 # Hook to suspend the parser in this state
1036                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1037                         wfProfileOut( __METHOD__ );
1038                         return $text ;
1039                 }
1040
1041                 # if $frame is provided, then use $frame for replacing any variables
1042                 if ( $frame ) {
1043                         # use frame depth to infer how include/noinclude tags should be handled
1044                         # depth=0 means this is the top-level document; otherwise it's an included document
1045                         if ( !$frame->depth ) {
1046                                 $flag = 0;
1047                         } else {
1048                                 $flag = Parser::PTD_FOR_INCLUSION;
1049                         }
1050                         $dom = $this->preprocessToDom( $text, $flag );
1051                         $text = $frame->expand( $dom );
1052                 } else {
1053                         # if $frame is not provided, then use old-style replaceVariables
1054                         $text = $this->replaceVariables( $text );
1055                 }
1056
1057                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1058                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1059
1060                 # Tables need to come after variable replacement for things to work
1061                 # properly; putting them before other transformations should keep
1062                 # exciting things like link expansions from showing up in surprising
1063                 # places.
1064                 $text = $this->doTableStuff( $text );
1065
1066                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1067
1068                 $text = $this->doDoubleUnderscore( $text );
1069
1070                 $text = $this->doHeadings( $text );
1071                 if ( $this->mOptions->getUseDynamicDates() ) {
1072                         $df = DateFormatter::getInstance();
1073                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1074                 }
1075                 $text = $this->replaceInternalLinks( $text );
1076                 $text = $this->doAllQuotes( $text );
1077                 $text = $this->replaceExternalLinks( $text );
1078
1079                 # replaceInternalLinks may sometimes leave behind
1080                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1081                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1082
1083                 $text = $this->doMagicLinks( $text );
1084                 $text = $this->formatHeadings( $text, $origText, $isMain );
1085
1086                 wfProfileOut( __METHOD__ );
1087                 return $text;
1088         }
1089
1090         /**
1091          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1092          * magic external links.
1093          *
1094          * DML
1095          * @private
1096          */
1097         function doMagicLinks( $text ) {
1098                 wfProfileIn( __METHOD__ );
1099                 $prots = $this->mUrlProtocols;
1100                 $urlChar = self::EXT_LINK_URL_CLASS;
1101                 $text = preg_replace_callback(
1102                         '!(?:                           # Start cases
1103                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1104                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1105                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1106                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1107                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1108                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1109                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1110                                     [0-9Xx]                 # check digit
1111                                     \b)
1112                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1113                 wfProfileOut( __METHOD__ );
1114                 return $text;
1115         }
1116
1117         function magicLinkCallback( $m ) {
1118                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1119                         # Skip anchor
1120                         return $m[0];
1121                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1122                         # Skip HTML element
1123                         return $m[0];
1124                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1125                         # Free external link
1126                         return $this->makeFreeExternalLink( $m[0] );
1127                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1128                         # RFC or PMID
1129                         $CssClass = '';
1130                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1131                                 $keyword = 'RFC';
1132                                 $urlmsg = 'rfcurl';
1133                                 $CssClass = 'mw-magiclink-rfc';
1134                                 $id = $m[4];
1135                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1136                                 $keyword = 'PMID';
1137                                 $urlmsg = 'pubmedurl';
1138                                 $CssClass = 'mw-magiclink-pmid';
1139                                 $id = $m[4];
1140                         } else {
1141                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1142                                         substr( $m[0], 0, 20 ) . '"' );
1143                         }
1144                         $url = wfMsg( $urlmsg, $id);
1145                         $sk = $this->mOptions->getSkin();
1146                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1147                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1148                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1149                         # ISBN
1150                         $isbn = $m[5];
1151                         $num = strtr( $isbn, array(
1152                                 '-' => '',
1153                                 ' ' => '',
1154                                 'x' => 'X',
1155                         ));
1156                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1157                         return'<a href="' .
1158                                 $titleObj->escapeLocalUrl() .
1159                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1160                 } else {
1161                         return $m[0];
1162                 }
1163         }
1164
1165         /**
1166          * Make a free external link, given a user-supplied URL
1167          * @return HTML
1168          * @private
1169          */
1170         function makeFreeExternalLink( $url ) {
1171                 global $wgContLang;
1172                 wfProfileIn( __METHOD__ );
1173
1174                 $sk = $this->mOptions->getSkin();
1175                 $trail = '';
1176
1177                 # The characters '<' and '>' (which were escaped by
1178                 # removeHTMLtags()) should not be included in
1179                 # URLs, per RFC 2396.
1180                 $m2 = array();
1181                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1182                         $trail = substr( $url, $m2[0][1] ) . $trail;
1183                         $url = substr( $url, 0, $m2[0][1] );
1184                 }
1185
1186                 # Move trailing punctuation to $trail
1187                 $sep = ',;\.:!?';
1188                 # If there is no left bracket, then consider right brackets fair game too
1189                 if ( strpos( $url, '(' ) === false ) {
1190                         $sep .= ')';
1191                 }
1192
1193                 $numSepChars = strspn( strrev( $url ), $sep );
1194                 if ( $numSepChars ) {
1195                         $trail = substr( $url, -$numSepChars ) . $trail;
1196                         $url = substr( $url, 0, -$numSepChars );
1197                 }
1198
1199                 $url = Sanitizer::cleanUrl( $url );
1200
1201                 # Is this an external image?
1202                 $text = $this->maybeMakeExternalImage( $url );
1203                 if ( $text === false ) {
1204                         # Not an image, make a link
1205                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1206                                 $this->getExternalLinkAttribs( $url ) );
1207                         # Register it in the output object...
1208                         # Replace unnecessary URL escape codes with their equivalent characters
1209                         $pasteurized = self::replaceUnusualEscapes( $url );
1210                         $this->mOutput->addExternalLink( $pasteurized );
1211                 }
1212                 wfProfileOut( __METHOD__ );
1213                 return $text . $trail;
1214         }
1215
1216
1217         /**
1218          * Parse headers and return html
1219          *
1220          * @private
1221          */
1222         function doHeadings( $text ) {
1223                 wfProfileIn( __METHOD__ );
1224                 for ( $i = 6; $i >= 1; --$i ) {
1225                         $h = str_repeat( '=', $i );
1226                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1227                           "<h$i>\\1</h$i>", $text );
1228                 }
1229                 wfProfileOut( __METHOD__ );
1230                 return $text;
1231         }
1232
1233         /**
1234          * Replace single quotes with HTML markup
1235          * @private
1236          * @return string the altered text
1237          */
1238         function doAllQuotes( $text ) {
1239                 wfProfileIn( __METHOD__ );
1240                 $outtext = '';
1241                 $lines = StringUtils::explode( "\n", $text );
1242                 foreach ( $lines as $line ) {
1243                         $outtext .= $this->doQuotes( $line ) . "\n";
1244                 }
1245                 $outtext = substr( $outtext, 0,-1 );
1246                 wfProfileOut( __METHOD__ );
1247                 return $outtext;
1248         }
1249
1250         /**
1251          * Helper function for doAllQuotes()
1252          */
1253         public function doQuotes( $text ) {
1254                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1255                 if ( count( $arr ) == 1 ) {
1256                         return $text;
1257                 } else {
1258                         # First, do some preliminary work. This may shift some apostrophes from
1259                         # being mark-up to being text. It also counts the number of occurrences
1260                         # of bold and italics mark-ups.
1261                         $i = 0;
1262                         $numbold = 0;
1263                         $numitalics = 0;
1264                         foreach ( $arr as $r ) {
1265                                 if ( ( $i % 2 ) == 1 ) {
1266                                         # If there are ever four apostrophes, assume the first is supposed to
1267                                         # be text, and the remaining three constitute mark-up for bold text.
1268                                         if ( strlen( $arr[$i] ) == 4 ) {
1269                                                 $arr[$i-1] .= "'";
1270                                                 $arr[$i] = "'''";
1271                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1272                                                 # If there are more than 5 apostrophes in a row, assume they're all
1273                                                 # text except for the last 5.
1274                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1275                                                 $arr[$i] = "'''''";
1276                                         }
1277                                         # Count the number of occurrences of bold and italics mark-ups.
1278                                         # We are not counting sequences of five apostrophes.
1279                                         if ( strlen( $arr[$i] ) == 2 ) {
1280                                                 $numitalics++;
1281                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1282                                                 $numbold++;
1283                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1284                                                 $numitalics++;
1285                                                 $numbold++;
1286                                         }
1287                                 }
1288                                 $i++;
1289                         }
1290
1291                         # If there is an odd number of both bold and italics, it is likely
1292                         # that one of the bold ones was meant to be an apostrophe followed
1293                         # by italics. Which one we cannot know for certain, but it is more
1294                         # likely to be one that has a single-letter word before it.
1295                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1296                                 $i = 0;
1297                                 $firstsingleletterword = -1;
1298                                 $firstmultiletterword = -1;
1299                                 $firstspace = -1;
1300                                 foreach ( $arr as $r ) {
1301                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1302                                                 $x1 = substr( $arr[$i-1], -1 );
1303                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1304                                                 if ( $x1 === ' ' ) {
1305                                                         if ( $firstspace == -1 ) {
1306                                                                 $firstspace = $i;
1307                                                         }
1308                                                 } elseif ( $x2 === ' ') {
1309                                                         if ( $firstsingleletterword == -1 ) {
1310                                                                 $firstsingleletterword = $i;
1311                                                         }
1312                                                 } else {
1313                                                         if ( $firstmultiletterword == -1 ) {
1314                                                                 $firstmultiletterword = $i;
1315                                                         }
1316                                                 }
1317                                         }
1318                                         $i++;
1319                                 }
1320
1321                                 # If there is a single-letter word, use it!
1322                                 if ( $firstsingleletterword > -1 ) {
1323                                         $arr[$firstsingleletterword] = "''";
1324                                         $arr[$firstsingleletterword-1] .= "'";
1325                                 } elseif ( $firstmultiletterword > -1 ) {
1326                                         # If not, but there's a multi-letter word, use that one.
1327                                         $arr[$firstmultiletterword] = "''";
1328                                         $arr[$firstmultiletterword-1] .= "'";
1329                                 } elseif ( $firstspace > -1 ) {
1330                                         # ... otherwise use the first one that has neither.
1331                                         # (notice that it is possible for all three to be -1 if, for example,
1332                                         # there is only one pentuple-apostrophe in the line)
1333                                         $arr[$firstspace] = "''";
1334                                         $arr[$firstspace-1] .= "'";
1335                                 }
1336                         }
1337
1338                         # Now let's actually convert our apostrophic mush to HTML!
1339                         $output = '';
1340                         $buffer = '';
1341                         $state = '';
1342                         $i = 0;
1343                         foreach ( $arr as $r ) {
1344                                 if ( ( $i % 2 ) == 0 ) {
1345                                         if ( $state === 'both' ) {
1346                                                 $buffer .= $r;
1347                                         } else {
1348                                                 $output .= $r;
1349                                         }
1350                                 } else {
1351                                         if ( strlen( $r ) == 2 ) {
1352                                                 if ( $state === 'i' ) {
1353                                                         $output .= '</i>'; $state = '';
1354                                                 } elseif ( $state === 'bi' ) {
1355                                                         $output .= '</i>'; $state = 'b';
1356                                                 } elseif ( $state === 'ib' ) {
1357                                                         $output .= '</b></i><b>'; $state = 'b';
1358                                                 } elseif ( $state === 'both' ) {
1359                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1360                                                 } else { # $state can be 'b' or ''
1361                                                         $output .= '<i>'; $state .= 'i';
1362                                                 }
1363                                         } elseif ( strlen( $r ) == 3 ) {
1364                                                 if ( $state === 'b' ) {
1365                                                         $output .= '</b>'; $state = '';
1366                                                 } elseif ( $state === 'bi' ) {
1367                                                         $output .= '</i></b><i>'; $state = 'i';
1368                                                 } elseif ( $state === 'ib' ) {
1369                                                         $output .= '</b>'; $state = 'i';
1370                                                 } elseif ( $state === 'both' ) {
1371                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1372                                                 } else { # $state can be 'i' or ''
1373                                                         $output .= '<b>'; $state .= 'b';
1374                                                 }
1375                                         } elseif ( strlen( $r ) == 5 ) {
1376                                                 if ( $state === 'b' ) {
1377                                                         $output .= '</b><i>'; $state = 'i';
1378                                                 } elseif ( $state === 'i' ) {
1379                                                         $output .= '</i><b>'; $state = 'b';
1380                                                 } elseif ( $state === 'bi' ) {
1381                                                         $output .= '</i></b>'; $state = '';
1382                                                 } elseif ( $state === 'ib' ) {
1383                                                         $output .= '</b></i>'; $state = '';
1384                                                 } elseif ( $state === 'both' ) {
1385                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1386                                                 } else { # ($state == '')
1387                                                         $buffer = ''; $state = 'both';
1388                                                 }
1389                                         }
1390                                 }
1391                                 $i++;
1392                         }
1393                         # Now close all remaining tags.  Notice that the order is important.
1394                         if ( $state === 'b' || $state === 'ib' ) {
1395                                 $output .= '</b>';
1396                         }
1397                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1398                                 $output .= '</i>';
1399                         }
1400                         if ( $state === 'bi' ) {
1401                                 $output .= '</b>';
1402                         }
1403                         # There might be lonely ''''', so make sure we have a buffer
1404                         if ( $state === 'both' && $buffer ) {
1405                                 $output .= '<b><i>'.$buffer.'</i></b>';
1406                         }
1407                         return $output;
1408                 }
1409         }
1410
1411         /**
1412          * Replace external links (REL)
1413          *
1414          * Note: this is all very hackish and the order of execution matters a lot.
1415          * Make sure to run maintenance/parserTests.php if you change this code.
1416          *
1417          * @private
1418          */
1419         function replaceExternalLinks( $text ) {
1420                 global $wgContLang;
1421                 wfProfileIn( __METHOD__ );
1422
1423                 $sk = $this->mOptions->getSkin();
1424
1425                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1426                 $s = array_shift( $bits );
1427
1428                 $i = 0;
1429                 while ( $i<count( $bits ) ) {
1430                         $url = $bits[$i++];
1431                         $protocol = $bits[$i++];
1432                         $text = $bits[$i++];
1433                         $trail = $bits[$i++];
1434
1435                         # The characters '<' and '>' (which were escaped by
1436                         # removeHTMLtags()) should not be included in
1437                         # URLs, per RFC 2396.
1438                         $m2 = array();
1439                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1440                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1441                                 $url = substr( $url, 0, $m2[0][1] );
1442                         }
1443
1444                         # If the link text is an image URL, replace it with an <img> tag
1445                         # This happened by accident in the original parser, but some people used it extensively
1446                         $img = $this->maybeMakeExternalImage( $text );
1447                         if ( $img !== false ) {
1448                                 $text = $img;
1449                         }
1450
1451                         $dtrail = '';
1452
1453                         # Set linktype for CSS - if URL==text, link is essentially free
1454                         $linktype = ( $text === $url ) ? 'free' : 'text';
1455
1456                         # No link text, e.g. [http://domain.tld/some.link]
1457                         if ( $text == '' ) {
1458                                 # Autonumber if allowed. See bug #5918
1459                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1460                                         $langObj = $this->getFunctionLang();
1461                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1462                                         $linktype = 'autonumber';
1463                                 } else {
1464                                         # Otherwise just use the URL
1465                                         $text = htmlspecialchars( $url );
1466                                         $linktype = 'free';
1467                                 }
1468                         } else {
1469                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1470                                 # Check for trail
1471                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1472                         }
1473
1474                         $text = $wgContLang->markNoConversion( $text );
1475
1476                         $url = Sanitizer::cleanUrl( $url );
1477
1478                         # Use the encoded URL
1479                         # This means that users can paste URLs directly into the text
1480                         # Funny characters like ö aren't valid in URLs anyway
1481                         # This was changed in August 2004
1482                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1483                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1484
1485                         # Register link in the output object.
1486                         # Replace unnecessary URL escape codes with the referenced character
1487                         # This prevents spammers from hiding links from the filters
1488                         $pasteurized = self::replaceUnusualEscapes( $url );
1489                         $this->mOutput->addExternalLink( $pasteurized );
1490                 }
1491
1492                 wfProfileOut( __METHOD__ );
1493                 return $s;
1494         }
1495
1496         /**
1497          * Get an associative array of additional HTML attributes appropriate for a
1498          * particular external link.  This currently may include rel => nofollow
1499          * (depending on configuration, namespace, and the URL's domain) and/or a
1500          * target attribute (depending on configuration).
1501          *
1502          * @param $url String: optional URL, to extract the domain from for rel =>
1503          *   nofollow if appropriate
1504          * @return Array: associative array of HTML attributes
1505          */
1506         function getExternalLinkAttribs( $url = false ) {
1507                 $attribs = array();
1508                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1509                 $ns = $this->mTitle->getNamespace();
1510                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1511                         $attribs['rel'] = 'nofollow';
1512
1513                         global $wgNoFollowDomainExceptions;
1514                         if ( $wgNoFollowDomainExceptions ) {
1515                                 $bits = wfParseUrl( $url );
1516                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1517                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1518                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1519                                                         unset( $attribs['rel'] );
1520                                                         break;
1521                                                 }
1522                                         }
1523                                 }
1524                         }
1525                 }
1526                 if ( $this->mOptions->getExternalLinkTarget() ) {
1527                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1528                 }
1529                 return $attribs;
1530         }
1531
1532
1533         /**
1534          * Replace unusual URL escape codes with their equivalent characters
1535          *
1536          * @param $url String
1537          * @return String
1538          *
1539          * @todo  This can merge genuinely required bits in the path or query string,
1540          *        breaking legit URLs. A proper fix would treat the various parts of
1541          *        the URL differently; as a workaround, just use the output for
1542          *        statistical records, not for actual linking/output.
1543          */
1544         static function replaceUnusualEscapes( $url ) {
1545                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1546                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1547         }
1548
1549         /**
1550          * Callback function used in replaceUnusualEscapes().
1551          * Replaces unusual URL escape codes with their equivalent character
1552          */
1553         private static function replaceUnusualEscapesCallback( $matches ) {
1554                 $char = urldecode( $matches[0] );
1555                 $ord = ord( $char );
1556                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1557                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1558                         # No, shouldn't be escaped
1559                         return $char;
1560                 } else {
1561                         # Yes, leave it escaped
1562                         return $matches[0];
1563                 }
1564         }
1565
1566         /**
1567          * make an image if it's allowed, either through the global
1568          * option, through the exception, or through the on-wiki whitelist
1569          * @private
1570          */
1571         function maybeMakeExternalImage( $url ) {
1572                 $sk = $this->mOptions->getSkin();
1573                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1574                 $imagesexception = !empty( $imagesfrom );
1575                 $text = false;
1576                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1577                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1578                         $imagematch = false;
1579                         foreach ( $imagesfrom as $match ) {
1580                                 if ( strpos( $url, $match ) === 0 ) {
1581                                         $imagematch = true;
1582                                         break;
1583                                 }
1584                         }
1585                 } elseif ( $imagesexception ) {
1586                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1587                 } else {
1588                         $imagematch = false;
1589                 }
1590                 if ( $this->mOptions->getAllowExternalImages()
1591                      || ( $imagesexception && $imagematch ) ) {
1592                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1593                                 # Image found
1594                                 $text = $sk->makeExternalImage( $url );
1595                         }
1596                 }
1597                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1598                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1599                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1600                         foreach ( $whitelist as $entry ) {
1601                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1602                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1603                                         continue;
1604                                 }
1605                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1606                                         # Image matches a whitelist entry
1607                                         $text = $sk->makeExternalImage( $url );
1608                                         break;
1609                                 }
1610                         }
1611                 }
1612                 return $text;
1613         }
1614
1615         /**
1616          * Process [[ ]] wikilinks
1617          * @return processed text
1618          *
1619          * @private
1620          */
1621         function replaceInternalLinks( $s ) {
1622                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1623                 return $s;
1624         }
1625
1626         /**
1627          * Process [[ ]] wikilinks (RIL)
1628          * @return LinkHolderArray
1629          *
1630          * @private
1631          */
1632         function replaceInternalLinks2( &$s ) {
1633                 global $wgContLang;
1634
1635                 wfProfileIn( __METHOD__ );
1636
1637                 wfProfileIn( __METHOD__.'-setup' );
1638                 static $tc = FALSE, $e1, $e1_img;
1639                 # the % is needed to support urlencoded titles as well
1640                 if ( !$tc ) {
1641                         $tc = Title::legalChars() . '#%';
1642                         # Match a link having the form [[namespace:link|alternate]]trail
1643                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1644                         # Match cases where there is no "]]", which might still be images
1645                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1646                 }
1647
1648                 $sk = $this->mOptions->getSkin();
1649                 $holders = new LinkHolderArray( $this );
1650
1651                 # split the entire text string on occurences of [[
1652                 $a = StringUtils::explode( '[[', ' ' . $s );
1653                 # get the first element (all text up to first [[), and remove the space we added
1654                 $s = $a->current();
1655                 $a->next();
1656                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1657                 $s = substr( $s, 1 );
1658
1659                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1660                 $e2 = null;
1661                 if ( $useLinkPrefixExtension ) {
1662                         # Match the end of a line for a word that's not followed by whitespace,
1663                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1664                         $e2 = wfMsgForContent( 'linkprefix' );
1665                 }
1666
1667                 if ( is_null( $this->mTitle ) ) {
1668                         wfProfileOut( __METHOD__.'-setup' );
1669                         wfProfileOut( __METHOD__ );
1670                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1671                 }
1672                 $nottalk = !$this->mTitle->isTalkPage();
1673
1674                 if ( $useLinkPrefixExtension ) {
1675                         $m = array();
1676                         if ( preg_match( $e2, $s, $m ) ) {
1677                                 $first_prefix = $m[2];
1678                         } else {
1679                                 $first_prefix = false;
1680                         }
1681                 } else {
1682                         $prefix = '';
1683                 }
1684
1685                 if ( $wgContLang->hasVariants() ) {
1686                         $selflink = $wgContLang->convertLinkToAllVariants( $this->mTitle->getPrefixedText() );
1687                 } else {
1688                         $selflink = array( $this->mTitle->getPrefixedText() );
1689                 }
1690                 $useSubpages = $this->areSubpagesAllowed();
1691                 wfProfileOut( __METHOD__.'-setup' );
1692
1693                 # Loop for each link
1694                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1695                         # Check for excessive memory usage
1696                         if ( $holders->isBig() ) {
1697                                 # Too big
1698                                 # Do the existence check, replace the link holders and clear the array
1699                                 $holders->replace( $s );
1700                                 $holders->clear();
1701                         }
1702
1703                         if ( $useLinkPrefixExtension ) {
1704                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1705                                 if ( preg_match( $e2, $s, $m ) ) {
1706                                         $prefix = $m[2];
1707                                         $s = $m[1];
1708                                 } else {
1709                                         $prefix='';
1710                                 }
1711                                 # first link
1712                                 if ( $first_prefix ) {
1713                                         $prefix = $first_prefix;
1714                                         $first_prefix = false;
1715                                 }
1716                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1717                         }
1718
1719                         $might_be_img = false;
1720
1721                         wfProfileIn( __METHOD__."-e1" );
1722                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1723                                 $text = $m[2];
1724                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1725                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1726                                 # the real problem is with the $e1 regex
1727                                 # See bug 1300.
1728                                 #
1729                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1730                                 # and no image is in sight. See bug 2095.
1731                                 #
1732                                 if ( $text !== '' &&
1733                                         substr( $m[3], 0, 1 ) === ']' &&
1734                                         strpos( $text, '[' ) !== false
1735                                 )
1736                                 {
1737                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1738                                         $m[3] = substr( $m[3], 1 );
1739                                 }
1740                                 # fix up urlencoded title texts
1741                                 if ( strpos( $m[1], '%' ) !== false ) {
1742                                         # Should anchors '#' also be rejected?
1743                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode( $m[1] ) );
1744                                 }
1745                                 $trail = $m[3];
1746                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1747                                 $might_be_img = true;
1748                                 $text = $m[2];
1749                                 if ( strpos( $m[1], '%' ) !== false ) {
1750                                         $m[1] = urldecode( $m[1] );
1751                                 }
1752                                 $trail = "";
1753                         } else { # Invalid form; output directly
1754                                 $s .= $prefix . '[[' . $line ;
1755                                 wfProfileOut( __METHOD__."-e1" );
1756                                 continue;
1757                         }
1758                         wfProfileOut( __METHOD__."-e1" );
1759                         wfProfileIn( __METHOD__."-misc" );
1760
1761                         # Don't allow internal links to pages containing
1762                         # PROTO: where PROTO is a valid URL protocol; these
1763                         # should be external links.
1764                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1765                                 $s .= $prefix . '[[' . $line ;
1766                                 wfProfileOut( __METHOD__."-misc" );
1767                                 continue;
1768                         }
1769
1770                         # Make subpage if necessary
1771                         if ( $useSubpages ) {
1772                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1773                         } else {
1774                                 $link = $m[1];
1775                         }
1776
1777                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1778                         if ( !$noforce ) {
1779                                 # Strip off leading ':'
1780                                 $link = substr( $link, 1 );
1781                         }
1782
1783                         wfProfileOut( __METHOD__."-misc" );
1784                         wfProfileIn( __METHOD__."-title" );
1785                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1786                         if ( $nt === null ) {
1787                                 $s .= $prefix . '[[' . $line;
1788                                 wfProfileOut( __METHOD__."-title" );
1789                                 continue;
1790                         }
1791
1792                         $ns = $nt->getNamespace();
1793                         $iw = $nt->getInterWiki();
1794                         wfProfileOut( __METHOD__."-title" );
1795
1796                         if ( $might_be_img ) { # if this is actually an invalid link
1797                                 wfProfileIn( __METHOD__."-might_be_img" );
1798                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1799                                         $found = false;
1800                                         while ( true ) {
1801                                                 # look at the next 'line' to see if we can close it there
1802                                                 $a->next();
1803                                                 $next_line = $a->current();
1804                                                 if ( $next_line === false || $next_line === null ) {
1805                                                         break;
1806                                                 }
1807                                                 $m = explode( ']]', $next_line, 3 );
1808                                                 if ( count( $m ) == 3 ) {
1809                                                         # the first ]] closes the inner link, the second the image
1810                                                         $found = true;
1811                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1812                                                         $trail = $m[2];
1813                                                         break;
1814                                                 } elseif ( count( $m ) == 2 ) {
1815                                                         # if there's exactly one ]] that's fine, we'll keep looking
1816                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1817                                                 } else {
1818                                                         # if $next_line is invalid too, we need look no further
1819                                                         $text .= '[[' . $next_line;
1820                                                         break;
1821                                                 }
1822                                         }
1823                                         if ( !$found ) {
1824                                                 # we couldn't find the end of this imageLink, so output it raw
1825                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1826                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1827                                                 $s .= "{$prefix}[[$link|$text";
1828                                                 # note: no $trail, because without an end, there *is* no trail
1829                                                 wfProfileOut( __METHOD__."-might_be_img" );
1830                                                 continue;
1831                                         }
1832                                 } else { # it's not an image, so output it raw
1833                                         $s .= "{$prefix}[[$link|$text";
1834                                         # note: no $trail, because without an end, there *is* no trail
1835                                         wfProfileOut( __METHOD__."-might_be_img" );
1836                                         continue;
1837                                 }
1838                                 wfProfileOut( __METHOD__."-might_be_img" );
1839                         }
1840
1841                         $wasblank = ( $text  == '' );
1842                         if ( $wasblank ) {
1843                                 $text = $link;
1844                         } else {
1845                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1846                                 # [[Lista d''e paise d''o munno]] -> <a href="">Lista d''e paise d''o munno</a>
1847                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']] -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1848                                 $text = $this->doQuotes($text);
1849                         }
1850
1851                         # Link not escaped by : , create the various objects
1852                         if ( $noforce ) {
1853
1854                                 # Interwikis
1855                                 wfProfileIn( __METHOD__."-interwiki" );
1856                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1857                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1858                                         $s = rtrim( $s . $prefix );
1859                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1860                                         wfProfileOut( __METHOD__."-interwiki" );
1861                                         continue;
1862                                 }
1863                                 wfProfileOut( __METHOD__."-interwiki" );
1864
1865                                 if ( $ns == NS_FILE ) {
1866                                         wfProfileIn( __METHOD__."-image" );
1867                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1868                                                 if ( $wasblank ) {
1869                                                         # if no parameters were passed, $text
1870                                                         # becomes something like "File:Foo.png",
1871                                                         # which we don't want to pass on to the
1872                                                         # image generator
1873                                                         $text = '';
1874                                                 } else {
1875                                                         # recursively parse links inside the image caption
1876                                                         # actually, this will parse them in any other parameters, too,
1877                                                         # but it might be hard to fix that, and it doesn't matter ATM
1878                                                         $text = $this->replaceExternalLinks( $text );
1879                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1880                                                 }
1881                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1882                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1883                                         } else {
1884                                                 $s .= $prefix . $trail;
1885                                         }
1886                                         $this->mOutput->addImage( $nt->getDBkey() );
1887                                         wfProfileOut( __METHOD__."-image" );
1888                                         continue;
1889
1890                                 }
1891
1892                                 if ( $ns == NS_CATEGORY ) {
1893                                         wfProfileIn( __METHOD__."-category" );
1894                                         $s = rtrim( $s . "\n" ); # bug 87
1895
1896                                         if ( $wasblank ) {
1897                                                 $sortkey = $this->getDefaultSort();
1898                                         } else {
1899                                                 $sortkey = $text;
1900                                         }
1901                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1902                                         $sortkey = str_replace( "\n", '', $sortkey );
1903                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1904                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1905
1906                                         /**
1907                                          * Strip the whitespace Category links produce, see bug 87
1908                                          * @todo We might want to use trim($tmp, "\n") here.
1909                                          */
1910                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '': $prefix . $trail;
1911
1912                                         wfProfileOut( __METHOD__."-category" );
1913                                         continue;
1914                                 }
1915                         }
1916
1917                         # Self-link checking
1918                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1919                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1920                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1921                                         continue;
1922                                 }
1923                         }
1924
1925                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1926                         # FIXME: Should do batch file existence checks, see comment below
1927                         if ( $ns == NS_MEDIA ) {
1928                                 wfProfileIn( __METHOD__."-media" );
1929                                 # Give extensions a chance to select the file revision for us
1930                                 $skip = $time = false;
1931                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1932                                 if ( $skip ) {
1933                                         $link = $sk->link( $nt );
1934                                 } else {
1935                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1936                                 }
1937                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1938                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1939                                 $this->mOutput->addImage( $nt->getDBkey() );
1940                                 wfProfileOut( __METHOD__."-media" );
1941                                 continue;
1942                         }
1943
1944                         wfProfileIn( __METHOD__."-always_known" );
1945                         # Some titles, such as valid special pages or files in foreign repos, should
1946                         # be shown as bluelinks even though they're not included in the page table
1947                         #
1948                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1949                         # batch file existence checks for NS_FILE and NS_MEDIA
1950                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1951                                 $this->mOutput->addLink( $nt );
1952                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1953                         } else {
1954                                 # Links will be added to the output link list after checking
1955                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1956                         }
1957                         wfProfileOut( __METHOD__."-always_known" );
1958                 }
1959                 wfProfileOut( __METHOD__ );
1960                 return $holders;
1961         }
1962
1963         /**
1964          * Make a link placeholder. The text returned can be later resolved to a real link with
1965          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1966          * parsing of interwiki links, and secondly to allow all existence checks and
1967          * article length checks (for stub links) to be bundled into a single query.
1968          *
1969          * @deprecated
1970          */
1971         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1972                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1973         }
1974
1975         /**
1976          * Render a forced-blue link inline; protect against double expansion of
1977          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1978          * Since this little disaster has to split off the trail text to avoid
1979          * breaking URLs in the following text without breaking trails on the
1980          * wiki links, it's been made into a horrible function.
1981          *
1982          * @param $nt Title
1983          * @param $text String
1984          * @param $query String
1985          * @param $trail String
1986          * @param $prefix String
1987          * @return String: HTML-wikitext mix oh yuck
1988          */
1989         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1990                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1991                 $sk = $this->mOptions->getSkin();
1992                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1993                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1994                 return $this->armorLinks( $link ) . $trail;
1995         }
1996
1997         /**
1998          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1999          * going to go through further parsing steps before inline URL expansion.
2000          *
2001          * Not needed quite as much as it used to be since free links are a bit
2002          * more sensible these days. But bracketed links are still an issue.
2003          *
2004          * @param $text String: more-or-less HTML
2005          * @return String: less-or-more HTML with NOPARSE bits
2006          */
2007         function armorLinks( $text ) {
2008                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2009                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2010         }
2011
2012         /**
2013          * Return true if subpage links should be expanded on this page.
2014          * @return Boolean
2015          */
2016         function areSubpagesAllowed() {
2017                 # Some namespaces don't allow subpages
2018                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2019         }
2020
2021         /**
2022          * Handle link to subpage if necessary
2023          *
2024          * @param $target String: the source of the link
2025          * @param &$text String: the link text, modified as necessary
2026          * @return string the full name of the link
2027          * @private
2028          */
2029         function maybeDoSubpageLink( $target, &$text ) {
2030                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2031         }
2032
2033         /**#@+
2034          * Used by doBlockLevels()
2035          * @private
2036          */
2037         function closeParagraph() {
2038                 $result = '';
2039                 if ( $this->mLastSection != '' ) {
2040                         $result = '</' . $this->mLastSection  . ">\n";
2041                 }
2042                 $this->mInPre = false;
2043                 $this->mLastSection = '';
2044                 return $result;
2045         }
2046
2047         /**
2048          * getCommon() returns the length of the longest common substring
2049          * of both arguments, starting at the beginning of both.
2050          * @private
2051          */
2052         function getCommon( $st1, $st2 ) {
2053                 $fl = strlen( $st1 );
2054                 $shorter = strlen( $st2 );
2055                 if ( $fl < $shorter ) {
2056                         $shorter = $fl;
2057                 }
2058
2059                 for ( $i = 0; $i < $shorter; ++$i ) {
2060                         if ( $st1{$i} != $st2{$i} ) {
2061                                 break;
2062                         }
2063                 }
2064                 return $i;
2065         }
2066
2067         /**
2068          * These next three functions open, continue, and close the list
2069          * element appropriate to the prefix character passed into them.
2070          * @private
2071          */
2072         function openList( $char ) {
2073                 $result = $this->closeParagraph();
2074
2075                 if ( '*' === $char ) {
2076                         $result .= '<ul><li>';
2077                 } elseif ( '#' === $char ) {
2078                         $result .= '<ol><li>';
2079                 } elseif ( ':' === $char ) {
2080                         $result .= '<dl><dd>';
2081                 } elseif ( ';' === $char ) {
2082                         $result .= '<dl><dt>';
2083                         $this->mDTopen = true;
2084                 } else {
2085                         $result = '<!-- ERR 1 -->';
2086                 }
2087
2088                 return $result;
2089         }
2090
2091         /**
2092          * TODO: document
2093          * @param $char String
2094          * @private
2095          */
2096         function nextItem( $char ) {
2097                 if ( '*' === $char || '#' === $char ) {
2098                         return '</li><li>';
2099                 } elseif ( ':' === $char || ';' === $char ) {
2100                         $close = '</dd>';
2101                         if ( $this->mDTopen ) {
2102                                 $close = '</dt>';
2103                         }
2104                         if ( ';' === $char ) {
2105                                 $this->mDTopen = true;
2106                                 return $close . '<dt>';
2107                         } else {
2108                                 $this->mDTopen = false;
2109                                 return $close . '<dd>';
2110                         }
2111                 }
2112                 return '<!-- ERR 2 -->';
2113         }
2114
2115         /**
2116          * TODO: document
2117          * @param $char String
2118          * @private
2119          */
2120         function closeList( $char ) {
2121                 if ( '*' === $char ) {
2122                         $text = '</li></ul>';
2123                 } elseif ( '#' === $char ) {
2124                         $text = '</li></ol>';
2125                 } elseif ( ':' === $char ) {
2126                         if ( $this->mDTopen ) {
2127                                 $this->mDTopen = false;
2128                                 $text = '</dt></dl>';
2129                         } else {
2130                                 $text = '</dd></dl>';
2131                         }
2132                 } else {
2133                         return '<!-- ERR 3 -->';
2134                 }
2135                 return $text."\n";
2136         }
2137         /**#@-*/
2138
2139         /**
2140          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2141          *
2142          * @param $text String
2143          * @param $linestart Boolean: whether or not this is at the start of a line.
2144          * @private
2145          * @return string the lists rendered as HTML
2146          */
2147         function doBlockLevels( $text, $linestart ) {
2148                 wfProfileIn( __METHOD__ );
2149
2150                 # Parsing through the text line by line.  The main thing
2151                 # happening here is handling of block-level elements p, pre,
2152                 # and making lists from lines starting with * # : etc.
2153                 #
2154                 $textLines = StringUtils::explode( "\n", $text );
2155
2156                 $lastPrefix = $output = '';
2157                 $this->mDTopen = $inBlockElem = false;
2158                 $prefixLength = 0;
2159                 $paragraphStack = false;
2160
2161                 foreach ( $textLines as $oLine ) {
2162                         # Fix up $linestart
2163                         if ( !$linestart ) {
2164                                 $output .= $oLine;
2165                                 $linestart = true;
2166                                 continue;
2167                         }
2168                         # * = ul
2169                         # # = ol
2170                         # ; = dt
2171                         # : = dd
2172
2173                         $lastPrefixLength = strlen( $lastPrefix );
2174                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2175                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2176                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2177                         if ( !$this->mInPre ) {
2178                                 # Multiple prefixes may abut each other for nested lists.
2179                                 $prefixLength = strspn( $oLine, '*#:;' );
2180                                 $prefix = substr( $oLine, 0, $prefixLength );
2181
2182                                 # eh?
2183                                 # ; and : are both from definition-lists, so they're equivalent
2184                                 #  for the purposes of determining whether or not we need to open/close
2185                                 #  elements.
2186                                 $prefix2 = str_replace( ';', ':', $prefix );
2187                                 $t = substr( $oLine, $prefixLength );
2188                                 $this->mInPre = (bool)$preOpenMatch;
2189                         } else {
2190                                 # Don't interpret any other prefixes in preformatted text
2191                                 $prefixLength = 0;
2192                                 $prefix = $prefix2 = '';
2193                                 $t = $oLine;
2194                         }
2195
2196                         # List generation
2197                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2198                                 # Same as the last item, so no need to deal with nesting or opening stuff
2199                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2200                                 $paragraphStack = false;
2201
2202                                 if ( substr( $prefix, -1 ) === ';') {
2203                                         # The one nasty exception: definition lists work like this:
2204                                         # ; title : definition text
2205                                         # So we check for : in the remainder text to split up the
2206                                         # title and definition, without b0rking links.
2207                                         $term = $t2 = '';
2208                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2209                                                 $t = $t2;
2210                                                 $output .= $term . $this->nextItem( ':' );
2211                                         }
2212                                 }
2213                         } elseif ( $prefixLength || $lastPrefixLength ) {
2214                                 # We need to open or close prefixes, or both.
2215
2216                                 # Either open or close a level...
2217                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2218                                 $paragraphStack = false;
2219
2220                                 # Close all the prefixes which aren't shared.
2221                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2222                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2223                                         --$lastPrefixLength;
2224                                 }
2225
2226                                 # Continue the current prefix if appropriate.
2227                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2228                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2229                                 }
2230
2231                                 # Open prefixes where appropriate.
2232                                 while ( $prefixLength > $commonPrefixLength ) {
2233                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2234                                         $output .= $this->openList( $char );
2235
2236                                         if ( ';' === $char ) {
2237                                                 # FIXME: This is dupe of code above
2238                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2239                                                         $t = $t2;
2240                                                         $output .= $term . $this->nextItem( ':' );
2241                                                 }
2242                                         }
2243                                         ++$commonPrefixLength;
2244                                 }
2245                                 $lastPrefix = $prefix2;
2246                         }
2247
2248                         # If we have no prefixes, go to paragraph mode.
2249                         if ( 0 == $prefixLength ) {
2250                                 wfProfileIn( __METHOD__."-paragraph" );
2251                                 # No prefix (not in list)--go to paragraph mode
2252                                 # XXX: use a stack for nestable elements like span, table and div
2253                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2254                                 $closematch = preg_match(
2255                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2256                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2257                                 if ( $openmatch or $closematch ) {
2258                                         $paragraphStack = false;
2259                                         # TODO bug 5718: paragraph closed
2260                                         $output .= $this->closeParagraph();
2261                                         if ( $preOpenMatch and !$preCloseMatch ) {
2262                                                 $this->mInPre = true;
2263                                         }
2264                                         if ( $closematch ) {
2265                                                 $inBlockElem = false;
2266                                         } else {
2267                                                 $inBlockElem = true;
2268                                         }
2269                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2270                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2271                                                 # pre
2272                                                 if ( $this->mLastSection !== 'pre' ) {
2273                                                         $paragraphStack = false;
2274                                                         $output .= $this->closeParagraph().'<pre>';
2275                                                         $this->mLastSection = 'pre';
2276                                                 }
2277                                                 $t = substr( $t, 1 );
2278                                         } else {
2279                                                 # paragraph
2280                                                 if ( trim( $t ) === '' ) {
2281                                                         if ( $paragraphStack ) {
2282                                                                 $output .= $paragraphStack.'<br />';
2283                                                                 $paragraphStack = false;
2284                                                                 $this->mLastSection = 'p';
2285                                                         } else {
2286                                                                 if ( $this->mLastSection !== 'p' ) {
2287                                                                         $output .= $this->closeParagraph();
2288                                                                         $this->mLastSection = '';
2289                                                                         $paragraphStack = '<p>';
2290                                                                 } else {
2291                                                                         $paragraphStack = '</p><p>';
2292                                                                 }
2293                                                         }
2294                                                 } else {
2295                                                         if ( $paragraphStack ) {
2296                                                                 $output .= $paragraphStack;
2297                                                                 $paragraphStack = false;
2298                                                                 $this->mLastSection = 'p';
2299                                                         } elseif ( $this->mLastSection !== 'p' ) {
2300                                                                 $output .= $this->closeParagraph().'<p>';
2301                                                                 $this->mLastSection = 'p';
2302                                                         }
2303                                                 }
2304                                         }
2305                                 }
2306                                 wfProfileOut( __METHOD__."-paragraph" );
2307                         }
2308                         # somewhere above we forget to get out of pre block (bug 785)
2309                         if ( $preCloseMatch && $this->mInPre ) {
2310                                 $this->mInPre = false;
2311                         }
2312                         if ( $paragraphStack === false ) {
2313                                 $output .= $t."\n";
2314                         }
2315                 }
2316                 while ( $prefixLength ) {
2317                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2318                         --$prefixLength;
2319                 }
2320                 if ( $this->mLastSection != '' ) {
2321                         $output .= '</' . $this->mLastSection . '>';
2322                         $this->mLastSection = '';
2323                 }
2324
2325                 wfProfileOut( __METHOD__ );
2326                 return $output;
2327         }
2328
2329         /**
2330          * Split up a string on ':', ignoring any occurences inside tags
2331          * to prevent illegal overlapping.
2332          *
2333          * @param $str String: the string to split
2334          * @param &$before String: set to everything before the ':'
2335          * @param &$after String: set to everything after the ':'
2336          * return String: the position of the ':', or false if none found
2337          */
2338         function findColonNoLinks( $str, &$before, &$after ) {
2339                 wfProfileIn( __METHOD__ );
2340
2341                 $pos = strpos( $str, ':' );
2342                 if ( $pos === false ) {
2343                         # Nothing to find!
2344                         wfProfileOut( __METHOD__ );
2345                         return false;
2346                 }
2347
2348                 $lt = strpos( $str, '<' );
2349                 if ( $lt === false || $lt > $pos ) {
2350                         # Easy; no tag nesting to worry about
2351                         $before = substr( $str, 0, $pos );
2352                         $after = substr( $str, $pos+1 );
2353                         wfProfileOut( __METHOD__ );
2354                         return $pos;
2355                 }
2356
2357                 # Ugly state machine to walk through avoiding tags.
2358                 $state = self::COLON_STATE_TEXT;
2359                 $stack = 0;
2360                 $len = strlen( $str );
2361                 for( $i = 0; $i < $len; $i++ ) {
2362                         $c = $str{$i};
2363
2364                         switch( $state ) {
2365                         # (Using the number is a performance hack for common cases)
2366                         case 0: # self::COLON_STATE_TEXT:
2367                                 switch( $c ) {
2368                                 case "<":
2369                                         # Could be either a <start> tag or an </end> tag
2370                                         $state = self::COLON_STATE_TAGSTART;
2371                                         break;
2372                                 case ":":
2373                                         if ( $stack == 0 ) {
2374                                                 # We found it!
2375                                                 $before = substr( $str, 0, $i );
2376                                                 $after = substr( $str, $i + 1 );
2377                                                 wfProfileOut( __METHOD__ );
2378                                                 return $i;
2379                                         }
2380                                         # Embedded in a tag; don't break it.
2381                                         break;
2382                                 default:
2383                                         # Skip ahead looking for something interesting
2384                                         $colon = strpos( $str, ':', $i );
2385                                         if ( $colon === false ) {
2386                                                 # Nothing else interesting
2387                                                 wfProfileOut( __METHOD__ );
2388                                                 return false;
2389                                         }
2390                                         $lt = strpos( $str, '<', $i );
2391                                         if ( $stack === 0 ) {
2392                                                 if ( $lt === false || $colon < $lt ) {
2393                                                         # We found it!
2394                                                         $before = substr( $str, 0, $colon );
2395                                                         $after = substr( $str, $colon + 1 );
2396                                                         wfProfileOut( __METHOD__ );
2397                                                         return $i;
2398                                                 }
2399                                         }
2400                                         if ( $lt === false ) {
2401                                                 # Nothing else interesting to find; abort!
2402                                                 # We're nested, but there's no close tags left. Abort!
2403                                                 break 2;
2404                                         }
2405                                         # Skip ahead to next tag start
2406                                         $i = $lt;
2407                                         $state = self::COLON_STATE_TAGSTART;
2408                                 }
2409                                 break;
2410                         case 1: # self::COLON_STATE_TAG:
2411                                 # In a <tag>
2412                                 switch( $c ) {
2413                                 case ">":
2414                                         $stack++;
2415                                         $state = self::COLON_STATE_TEXT;
2416                                         break;
2417                                 case "/":
2418                                         # Slash may be followed by >?
2419                                         $state = self::COLON_STATE_TAGSLASH;
2420                                         break;
2421                                 default:
2422                                         # ignore
2423                                 }
2424                                 break;
2425                         case 2: # self::COLON_STATE_TAGSTART:
2426                                 switch( $c ) {
2427                                 case "/":
2428                                         $state = self::COLON_STATE_CLOSETAG;
2429                                         break;
2430                                 case "!":
2431                                         $state = self::COLON_STATE_COMMENT;
2432                                         break;
2433                                 case ">":
2434                                         # Illegal early close? This shouldn't happen D:
2435                                         $state = self::COLON_STATE_TEXT;
2436                                         break;
2437                                 default:
2438                                         $state = self::COLON_STATE_TAG;
2439                                 }
2440                                 break;
2441                         case 3: # self::COLON_STATE_CLOSETAG:
2442                                 # In a </tag>
2443                                 if ( $c === ">" ) {
2444                                         $stack--;
2445                                         if ( $stack < 0 ) {
2446                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2447                                                 wfProfileOut( __METHOD__ );
2448                                                 return false;
2449                                         }
2450                                         $state = self::COLON_STATE_TEXT;
2451                                 }
2452                                 break;
2453                         case self::COLON_STATE_TAGSLASH:
2454                                 if ( $c === ">" ) {
2455                                         # Yes, a self-closed tag <blah/>
2456                                         $state = self::COLON_STATE_TEXT;
2457                                 } else {
2458                                         # Probably we're jumping the gun, and this is an attribute
2459                                         $state = self::COLON_STATE_TAG;
2460                                 }
2461                                 break;
2462                         case 5: # self::COLON_STATE_COMMENT:
2463                                 if ( $c === "-" ) {
2464                                         $state = self::COLON_STATE_COMMENTDASH;
2465                                 }
2466                                 break;
2467                         case self::COLON_STATE_COMMENTDASH:
2468                                 if ( $c === "-" ) {
2469                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2470                                 } else {
2471                                         $state = self::COLON_STATE_COMMENT;
2472                                 }
2473                                 break;
2474                         case self::COLON_STATE_COMMENTDASHDASH:
2475                                 if ( $c === ">" ) {
2476                                         $state = self::COLON_STATE_TEXT;
2477                                 } else {
2478                                         $state = self::COLON_STATE_COMMENT;
2479                                 }
2480                                 break;
2481                         default:
2482                                 throw new MWException( "State machine error in " . __METHOD__ );
2483                         }
2484                 }
2485                 if ( $stack > 0 ) {
2486                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2487                         return false;
2488                 }
2489                 wfProfileOut( __METHOD__ );
2490                 return false;
2491         }
2492
2493         /**
2494          * Return value of a magic variable (like PAGENAME)
2495          *
2496          * @private
2497          */
2498         function getVariableValue( $index, $frame=false ) {
2499                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2500                 global $wgScriptPath, $wgStylePath;
2501
2502                 /**
2503                  * Some of these require message or data lookups and can be
2504                  * expensive to check many times.
2505                  */
2506                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2507                         if ( isset( $this->mVarCache[$index] ) ) {
2508                                 return $this->mVarCache[$index];
2509                         }
2510                 }
2511
2512                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2513                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2514
2515                 # Use the time zone
2516                 global $wgLocaltimezone;
2517                 if ( isset( $wgLocaltimezone ) ) {
2518                         $oldtz = date_default_timezone_get();
2519                         date_default_timezone_set( $wgLocaltimezone );
2520                 }
2521
2522                 $localTimestamp = date( 'YmdHis', $ts );
2523                 $localMonth = date( 'm', $ts );
2524                 $localMonth1 = date( 'n', $ts );
2525                 $localMonthName = date( 'n', $ts );
2526                 $localDay = date( 'j', $ts );
2527                 $localDay2 = date( 'd', $ts );
2528                 $localDayOfWeek = date( 'w', $ts );
2529                 $localWeek = date( 'W', $ts );
2530                 $localYear = date( 'Y', $ts );
2531                 $localHour = date( 'H', $ts );
2532                 if ( isset( $wgLocaltimezone ) ) {
2533                         date_default_timezone_set( $oldtz );
2534                 }
2535
2536                 switch ( $index ) {
2537                         case 'currentmonth':
2538                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2539                                 break;
2540                         case 'currentmonth1':
2541                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2542                                 break;
2543                         case 'currentmonthname':
2544                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2545                                 break;
2546                         case 'currentmonthnamegen':
2547                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2548                                 break;
2549                         case 'currentmonthabbrev':
2550                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2551                                 break;
2552                         case 'currentday':
2553                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2554                                 break;
2555                         case 'currentday2':
2556                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2557                                 break;
2558                         case 'localmonth':
2559                                 $value = $wgContLang->formatNum( $localMonth );
2560                                 break;
2561                         case 'localmonth1':
2562                                 $value = $wgContLang->formatNum( $localMonth1 );
2563                                 break;
2564                         case 'localmonthname':
2565                                 $value = $wgContLang->getMonthName( $localMonthName );
2566                                 break;
2567                         case 'localmonthnamegen':
2568                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2569                                 break;
2570                         case 'localmonthabbrev':
2571                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2572                                 break;
2573                         case 'localday':
2574                                 $value = $wgContLang->formatNum( $localDay );
2575                                 break;
2576                         case 'localday2':
2577                                 $value = $wgContLang->formatNum( $localDay2 );
2578                                 break;
2579                         case 'pagename':
2580                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2581                                 break;
2582                         case 'pagenamee':
2583                                 $value = $this->mTitle->getPartialURL();
2584                                 break;
2585                         case 'fullpagename':
2586                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2587                                 break;
2588                         case 'fullpagenamee':
2589                                 $value = $this->mTitle->getPrefixedURL();
2590                                 break;
2591                         case 'subpagename':
2592                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2593                                 break;
2594                         case 'subpagenamee':
2595                                 $value = $this->mTitle->getSubpageUrlForm();
2596                                 break;
2597                         case 'basepagename':
2598                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2599                                 break;
2600                         case 'basepagenamee':
2601                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2602                                 break;
2603                         case 'talkpagename':
2604                                 if ( $this->mTitle->canTalk() ) {
2605                                         $talkPage = $this->mTitle->getTalkPage();
2606                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2607                                 } else {
2608                                         $value = '';
2609                                 }
2610                                 break;
2611                         case 'talkpagenamee':
2612                                 if ( $this->mTitle->canTalk() ) {
2613                                         $talkPage = $this->mTitle->getTalkPage();
2614                                         $value = $talkPage->getPrefixedUrl();
2615                                 } else {
2616                                         $value = '';
2617                                 }
2618                                 break;
2619                         case 'subjectpagename':
2620                                 $subjPage = $this->mTitle->getSubjectPage();
2621                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2622                                 break;
2623                         case 'subjectpagenamee':
2624                                 $subjPage = $this->mTitle->getSubjectPage();
2625                                 $value = $subjPage->getPrefixedUrl();
2626                                 break;
2627                         case 'revisionid':
2628                                 # Let the edit saving system know we should parse the page
2629                                 # *after* a revision ID has been assigned.
2630                                 $this->mOutput->setFlag( 'vary-revision' );
2631                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2632                                 $value = $this->mRevisionId;
2633                                 break;
2634                         case 'revisionday':
2635                                 # Let the edit saving system know we should parse the page
2636                                 # *after* a revision ID has been assigned. This is for null edits.
2637                                 $this->mOutput->setFlag( 'vary-revision' );
2638                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2639                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2640                                 break;
2641                         case 'revisionday2':
2642                                 # Let the edit saving system know we should parse the page
2643                                 # *after* a revision ID has been assigned. This is for null edits.
2644                                 $this->mOutput->setFlag( 'vary-revision' );
2645                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2646                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2647                                 break;
2648                         case 'revisionmonth':
2649                                 # Let the edit saving system know we should parse the page
2650                                 # *after* a revision ID has been assigned. This is for null edits.
2651                                 $this->mOutput->setFlag( 'vary-revision' );
2652                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2653                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2654                                 break;
2655                         case 'revisionmonth1':
2656                                 # Let the edit saving system know we should parse the page
2657                                 # *after* a revision ID has been assigned. This is for null edits.
2658                                 $this->mOutput->setFlag( 'vary-revision' );
2659                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2660                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2661                                 break;
2662                         case 'revisionyear':
2663                                 # Let the edit saving system know we should parse the page
2664                                 # *after* a revision ID has been assigned. This is for null edits.
2665                                 $this->mOutput->setFlag( 'vary-revision' );
2666                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2667                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2668                                 break;
2669                         case 'revisiontimestamp':
2670                                 # Let the edit saving system know we should parse the page
2671                                 # *after* a revision ID has been assigned. This is for null edits.
2672                                 $this->mOutput->setFlag( 'vary-revision' );
2673                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2674                                 $value = $this->getRevisionTimestamp();
2675                                 break;
2676                         case 'revisionuser':
2677                                 # Let the edit saving system know we should parse the page
2678                                 # *after* a revision ID has been assigned. This is for null edits.
2679                                 $this->mOutput->setFlag( 'vary-revision' );
2680                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2681                                 $value = $this->getRevisionUser();
2682                                 break;
2683                         case 'namespace':
2684                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2685                                 break;
2686                         case 'namespacee':
2687                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2688                                 break;
2689                         case 'talkspace':
2690                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2691                                 break;
2692                         case 'talkspacee':
2693                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2694                                 break;
2695                         case 'subjectspace':
2696                                 $value = $this->mTitle->getSubjectNsText();
2697                                 break;
2698                         case 'subjectspacee':
2699                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2700                                 break;
2701                         case 'currentdayname':
2702                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2703                                 break;
2704                         case 'currentyear':
2705                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2706                                 break;
2707                         case 'currenttime':
2708                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2709                                 break;
2710                         case 'currenthour':
2711                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2712                                 break;
2713                         case 'currentweek':
2714                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2715                                 # int to remove the padding
2716                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2717                                 break;
2718                         case 'currentdow':
2719                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2720                                 break;
2721                         case 'localdayname':
2722                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2723                                 break;
2724                         case 'localyear':
2725                                 $value = $wgContLang->formatNum( $localYear, true );
2726                                 break;
2727                         case 'localtime':
2728                                 $value = $wgContLang->time( $localTimestamp, false, false );
2729                                 break;
2730                         case 'localhour':
2731                                 $value = $wgContLang->formatNum( $localHour, true );
2732                                 break;
2733                         case 'localweek':
2734                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2735                                 # int to remove the padding
2736                                 $value = $wgContLang->formatNum( (int)$localWeek );
2737                                 break;
2738                         case 'localdow':
2739                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2740                                 break;
2741                         case 'numberofarticles':
2742                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2743                                 break;
2744                         case 'numberoffiles':
2745                                 $value = $wgContLang->formatNum( SiteStats::images() );
2746                                 break;
2747                         case 'numberofusers':
2748                                 $value = $wgContLang->formatNum( SiteStats::users() );
2749                                 break;
2750                         case 'numberofactiveusers':
2751                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2752                                 break;
2753                         case 'numberofpages':
2754                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2755                                 break;
2756                         case 'numberofadmins':
2757                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2758                                 break;
2759                         case 'numberofedits':
2760                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2761                                 break;
2762                         case 'numberofviews':
2763                                 $value = $wgContLang->formatNum( SiteStats::views() );
2764                                 break;
2765                         case 'currenttimestamp':
2766                                 $value = wfTimestamp( TS_MW, $ts );
2767                                 break;
2768                         case 'localtimestamp':
2769                                 $value = $localTimestamp;
2770                                 break;
2771                         case 'currentversion':
2772                                 $value = SpecialVersion::getVersion();
2773                                 break;
2774                         case 'sitename':
2775                                 return $wgSitename;
2776                         case 'server':
2777                                 return $wgServer;
2778                         case 'servername':
2779                                 return $wgServerName;
2780                         case 'scriptpath':
2781                                 return $wgScriptPath;
2782                         case 'stylepath':
2783                                 return $wgStylePath;
2784                         case 'directionmark':
2785                                 return $wgContLang->getDirMark();
2786                         case 'contentlanguage':
2787                                 global $wgContLanguageCode;
2788                                 return $wgContLanguageCode;
2789                         default:
2790                                 $ret = null;
2791                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2792                                         return $ret;
2793                                 } else {
2794                                         return null;
2795                                 }
2796                 }
2797
2798                 if ( $index )
2799                         $this->mVarCache[$index] = $value;
2800
2801                 return $value;
2802         }
2803
2804         /**
2805          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2806          *
2807          * @private
2808          */
2809         function initialiseVariables() {
2810                 wfProfileIn( __METHOD__ );
2811                 $variableIDs = MagicWord::getVariableIDs();
2812                 $substIDs = MagicWord::getSubstIDs();
2813
2814                 $this->mVariables = new MagicWordArray( $variableIDs );
2815                 $this->mSubstWords = new MagicWordArray( $substIDs );
2816                 wfProfileOut( __METHOD__ );
2817         }
2818
2819         /**
2820          * Preprocess some wikitext and return the document tree.
2821          * This is the ghost of replace_variables().
2822          *
2823          * @param $text String: The text to parse
2824          * @param $flags Integer: bitwise combination of:
2825          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2826          *                                     included. Default is to assume a direct page view.
2827          *
2828          * The generated DOM tree must depend only on the input text and the flags.
2829          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2830          *
2831          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2832          * change in the DOM tree for a given text, must be passed through the section identifier
2833          * in the section edit link and thus back to extractSections().
2834          *
2835          * The output of this function is currently only cached in process memory, but a persistent
2836          * cache may be implemented at a later date which takes further advantage of these strict
2837          * dependency requirements.
2838          *
2839          * @private
2840          */
2841         function preprocessToDom( $text, $flags = 0 ) {
2842                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2843                 return $dom;
2844         }
2845
2846         /**
2847          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2848          */
2849         public static function splitWhitespace( $s ) {
2850                 $ltrimmed = ltrim( $s );
2851                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2852                 $trimmed = rtrim( $ltrimmed );
2853                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2854                 if ( $diff > 0 ) {
2855                         $w2 = substr( $ltrimmed, -$diff );
2856                 } else {
2857                         $w2 = '';
2858                 }
2859                 return array( $w1, $trimmed, $w2 );
2860         }
2861
2862         /**
2863          * Replace magic variables, templates, and template arguments
2864          * with the appropriate text. Templates are substituted recursively,
2865          * taking care to avoid infinite loops.
2866          *
2867          * Note that the substitution depends on value of $mOutputType:
2868          *  self::OT_WIKI: only {{subst:}} templates
2869          *  self::OT_PREPROCESS: templates but not extension tags
2870          *  self::OT_HTML: all templates and extension tags
2871          *
2872          * @param $text String: the text to transform
2873          * @param $frame PPFrame Object describing the arguments passed to the template.
2874          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2875          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2876          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2877          * @private
2878          */
2879         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2880                 # Is there any text? Also, Prevent too big inclusions!
2881                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2882                         return $text;
2883                 }
2884                 wfProfileIn( __METHOD__ );
2885
2886                 if ( $frame === false ) {
2887                         $frame = $this->getPreprocessor()->newFrame();
2888                 } elseif ( !( $frame instanceof PPFrame ) ) {
2889                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2890                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2891                 }
2892
2893                 $dom = $this->preprocessToDom( $text );
2894                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2895                 $text = $frame->expand( $dom, $flags );
2896
2897                 wfProfileOut( __METHOD__ );
2898                 return $text;
2899         }
2900
2901         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2902         static function createAssocArgs( $args ) {
2903                 $assocArgs = array();
2904                 $index = 1;
2905                 foreach ( $args as $arg ) {
2906                         $eqpos = strpos( $arg, '=' );
2907                         if ( $eqpos === false ) {
2908                                 $assocArgs[$index++] = $arg;
2909                         } else {
2910                                 $name = trim( substr( $arg, 0, $eqpos ) );
2911                                 $value = trim( substr( $arg, $eqpos+1 ) );
2912                                 if ( $value === false ) {
2913                                         $value = '';
2914                                 }
2915                                 if ( $name !== false ) {
2916                                         $assocArgs[$name] = $value;
2917                                 }
2918                         }
2919                 }
2920
2921                 return $assocArgs;
2922         }
2923
2924         /**
2925          * Warn the user when a parser limitation is reached
2926          * Will warn at most once the user per limitation type
2927          *
2928          * @param $limitationType String: should be one of:
2929          *   'expensive-parserfunction' (corresponding messages:
2930          *       'expensive-parserfunction-warning',
2931          *       'expensive-parserfunction-category')
2932          *   'post-expand-template-argument' (corresponding messages:
2933          *       'post-expand-template-argument-warning',
2934          *       'post-expand-template-argument-category')
2935          *   'post-expand-template-inclusion' (corresponding messages:
2936          *       'post-expand-template-inclusion-warning',
2937          *       'post-expand-template-inclusion-category')
2938          * @param $current Current value
2939          * @param $max Maximum allowed, when an explicit limit has been
2940          *       exceeded, provide the values (optional)
2941          */
2942         function limitationWarn( $limitationType, $current=null, $max=null) {
2943                 # does no harm if $current and $max are present but are unnecessary for the message
2944                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2945                 $this->mOutput->addWarning( $warning );
2946                 $this->addTrackingCategory( "$limitationType-category" );
2947         }
2948
2949         /**
2950          * Return the text of a template, after recursively
2951          * replacing any variables or templates within the template.
2952          *
2953          * @param $piece Array: the parts of the template
2954          *  $piece['title']: the title, i.e. the part before the |
2955          *  $piece['parts']: the parameter array
2956          *  $piece['lineStart']: whether the brace was at the start of a line
2957          * @param $frame PPFrame The current frame, contains template arguments
2958          * @return String: the text of the template
2959          * @private
2960          */
2961         function braceSubstitution( $piece, $frame ) {
2962                 global $wgContLang, $wgNonincludableNamespaces;
2963                 wfProfileIn( __METHOD__ );
2964                 wfProfileIn( __METHOD__.'-setup' );
2965
2966                 # Flags
2967                 $found = false;             # $text has been filled
2968                 $nowiki = false;            # wiki markup in $text should be escaped
2969                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2970                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2971                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2972                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2973
2974                 # Title object, where $text came from
2975                 $title = null;
2976
2977                 # $part1 is the bit before the first |, and must contain only title characters.
2978                 # Various prefixes will be stripped from it later.
2979                 $titleWithSpaces = $frame->expand( $piece['title'] );
2980                 $part1 = trim( $titleWithSpaces );
2981                 $titleText = false;
2982
2983                 # Original title text preserved for various purposes
2984                 $originalTitle = $part1;
2985
2986                 # $args is a list of argument nodes, starting from index 0, not including $part1
2987                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2988                 wfProfileOut( __METHOD__.'-setup' );
2989
2990                 # SUBST
2991                 wfProfileIn( __METHOD__.'-modifiers' );
2992                 if ( !$found ) {
2993
2994                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2995
2996                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2997                         # Decide whether to expand template or keep wikitext as-is.
2998                         if ( $this->ot['wiki'] ) {
2999                                 if ( $substMatch === false ) {
3000                                         $literal = true;  # literal when in PST with no prefix
3001                                 } else {
3002                                         $literal = false; # expand when in PST with subst: or safesubst:
3003                                 }
3004                         } else {
3005                                 if ( $substMatch == 'subst' ) {
3006                                         $literal = true;  # literal when not in PST with plain subst:
3007                                 } else {
3008                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3009                                 }
3010                         }
3011                         if ( $literal ) {
3012                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3013                                 $isLocalObj = true;
3014                                 $found = true;
3015                         }
3016                 }
3017
3018                 # Variables
3019                 if ( !$found && $args->getLength() == 0 ) {
3020                         $id = $this->mVariables->matchStartToEnd( $part1 );
3021                         if ( $id !== false ) {
3022                                 $text = $this->getVariableValue( $id, $frame );
3023                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3024                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3025                                 }
3026                                 $found = true;
3027                         }
3028                 }
3029
3030                 # MSG, MSGNW and RAW
3031                 if ( !$found ) {
3032                         # Check for MSGNW:
3033                         $mwMsgnw = MagicWord::get( 'msgnw' );
3034                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3035                                 $nowiki = true;
3036                         } else {
3037                                 # Remove obsolete MSG:
3038                                 $mwMsg = MagicWord::get( 'msg' );
3039                                 $mwMsg->matchStartAndRemove( $part1 );
3040                         }
3041
3042                         # Check for RAW:
3043                         $mwRaw = MagicWord::get( 'raw' );
3044                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3045                                 $forceRawInterwiki = true;
3046                         }
3047                 }
3048                 wfProfileOut( __METHOD__.'-modifiers' );
3049
3050                 # Parser functions
3051                 if ( !$found ) {
3052                         wfProfileIn( __METHOD__ . '-pfunc' );
3053
3054                         $colonPos = strpos( $part1, ':' );
3055                         if ( $colonPos !== false ) {
3056                                 # Case sensitive functions
3057                                 $function = substr( $part1, 0, $colonPos );
3058                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3059                                         $function = $this->mFunctionSynonyms[1][$function];
3060                                 } else {
3061                                         # Case insensitive functions
3062                                         $function = $wgContLang->lc( $function );
3063                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3064                                                 $function = $this->mFunctionSynonyms[0][$function];
3065                                         } else {
3066                                                 $function = false;
3067                                         }
3068                                 }
3069                                 if ( $function ) {
3070                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3071                                         $initialArgs = array( &$this );
3072                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3073                                         if ( $flags & SFH_OBJECT_ARGS ) {
3074                                                 # Add a frame parameter, and pass the arguments as an array
3075                                                 $allArgs = $initialArgs;
3076                                                 $allArgs[] = $frame;
3077                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3078                                                         $funcArgs[] = $args->item( $i );
3079                                                 }
3080                                                 $allArgs[] = $funcArgs;
3081                                         } else {
3082                                                 # Convert arguments to plain text
3083                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3084                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3085                                                 }
3086                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3087                                         }
3088
3089                                         # Workaround for PHP bug 35229 and similar
3090                                         if ( !is_callable( $callback ) ) {
3091                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3092                                                 wfProfileOut( __METHOD__ );
3093                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3094                                         }
3095                                         $result = call_user_func_array( $callback, $allArgs );
3096                                         $found = true;
3097                                         $noparse = true;
3098                                         $preprocessFlags = 0;
3099
3100                                         if ( is_array( $result ) ) {
3101                                                 if ( isset( $result[0] ) ) {
3102                                                         $text = $result[0];
3103                                                         unset( $result[0] );
3104                                                 }
3105
3106                                                 # Extract flags into the local scope
3107                                                 # This allows callers to set flags such as nowiki, found, etc.
3108                                                 extract( $result );
3109                                         } else {
3110                                                 $text = $result;
3111                                         }
3112                                         if ( !$noparse ) {
3113                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3114                                                 $isChildObj = true;
3115                                         }
3116                                 }
3117                         }
3118                         wfProfileOut( __METHOD__ . '-pfunc' );
3119                 }
3120
3121                 # Finish mangling title and then check for loops.
3122                 # Set $title to a Title object and $titleText to the PDBK
3123                 if ( !$found ) {
3124                         $ns = NS_TEMPLATE;
3125                         # Split the title into page and subpage
3126                         $subpage = '';
3127                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3128                         if ( $subpage !== '' ) {
3129                                 $ns = $this->mTitle->getNamespace();
3130                         }
3131                         $title = Title::newFromText( $part1, $ns );
3132                         if ( $title ) {
3133                                 $titleText = $title->getPrefixedText();
3134                                 # Check for language variants if the template is not found
3135                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3136                                         $wgContLang->findVariantLink( $part1, $title, true );
3137                                 }
3138                                 # Do recursion depth check
3139                                 $limit = $this->mOptions->getMaxTemplateDepth();
3140                                 if ( $frame->depth >= $limit ) {
3141                                         $found = true;
3142                                         $text = '<span class="error">'
3143                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3144                                                 . '</span>';
3145                                 }
3146                         }
3147                 }
3148
3149                 # Load from database
3150                 if ( !$found && $title ) {
3151                         wfProfileIn( __METHOD__ . '-loadtpl' );
3152                         if ( !$title->isExternal() ) {
3153                                 if ( $title->getNamespace() == NS_SPECIAL
3154                                         && $this->mOptions->getAllowSpecialInclusion()
3155                                         && $this->ot['html'] )
3156                                 {
3157                                         $text = SpecialPage::capturePath( $title );
3158                                         if ( is_string( $text ) ) {
3159                                                 $found = true;
3160                                                 $isHTML = true;
3161                                                 $this->disableCache();
3162                                         }
3163                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3164                                         $found = false; # access denied
3165                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3166                                 } else {
3167                                         list( $text, $title ) = $this->getTemplateDom( $title );
3168                                         if ( $text !== false ) {
3169                                                 $found = true;
3170                                                 $isChildObj = true;
3171                                         }
3172                                 }
3173
3174                                 # If the title is valid but undisplayable, make a link to it
3175                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3176                                         $text = "[[:$titleText]]";
3177                                         $found = true;
3178                                 }
3179                         } elseif ( $title->isTrans() ) {
3180                                 # Interwiki transclusion
3181                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3182                                         $text = $this->interwikiTransclude( $title, 'render' );
3183                                         $isHTML = true;
3184                                 } else {
3185                                         $text = $this->interwikiTransclude( $title, 'raw' );
3186                                         # Preprocess it like a template
3187                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3188                                         $isChildObj = true;
3189                                 }
3190                                 $found = true;
3191                         }
3192
3193                         # Do infinite loop check
3194                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3195                         if ( !$frame->loopCheck( $title ) ) {
3196                                 $found = true;
3197                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3198                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3199                         }
3200                         wfProfileOut( __METHOD__ . '-loadtpl' );
3201                 }
3202
3203                 # If we haven't found text to substitute by now, we're done
3204                 # Recover the source wikitext and return it
3205                 if ( !$found ) {
3206                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3207                         wfProfileOut( __METHOD__ );
3208                         return array( 'object' => $text );
3209                 }
3210
3211                 # Expand DOM-style return values in a child frame
3212                 if ( $isChildObj ) {
3213                         # Clean up argument array
3214                         $newFrame = $frame->newChild( $args, $title );
3215
3216                         if ( $nowiki ) {
3217                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3218                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3219                                 # Expansion is eligible for the empty-frame cache
3220                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3221                                         $text = $this->mTplExpandCache[$titleText];
3222                                 } else {
3223                                         $text = $newFrame->expand( $text );
3224                                         $this->mTplExpandCache[$titleText] = $text;
3225                                 }
3226                         } else {
3227                                 # Uncached expansion
3228                                 $text = $newFrame->expand( $text );
3229                         }
3230                 }
3231                 if ( $isLocalObj && $nowiki ) {
3232                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3233                         $isLocalObj = false;
3234                 }
3235
3236                 # Replace raw HTML by a placeholder
3237                 # Add a blank line preceding, to prevent it from mucking up
3238                 # immediately preceding headings
3239                 if ( $isHTML ) {
3240                         $text = "\n\n" . $this->insertStripItem( $text );
3241                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3242                         # Escape nowiki-style return values
3243                         $text = wfEscapeWikiText( $text );
3244                 } elseif ( is_string( $text )
3245                         && !$piece['lineStart']
3246                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3247                 {
3248                         # Bug 529: if the template begins with a table or block-level
3249                         # element, it should be treated as beginning a new line.
3250                         # This behaviour is somewhat controversial.
3251                         $text = "\n" . $text;
3252                 }
3253
3254                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3255                         # Error, oversize inclusion
3256                         if ( $titleText !== false ) {
3257                                 # Make a working, properly escaped link if possible (bug 23588)
3258                                 $text = "[[:$titleText]]";
3259                         } else {
3260                                 # This will probably not be a working link, but at least it may
3261                                 # provide some hint of where the problem is
3262                                 preg_replace( '/^:/', '', $originalTitle );
3263                                 $text = "[[:$originalTitle]]";
3264                         }
3265                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3266                         $this->limitationWarn( 'post-expand-template-inclusion' );
3267                 }
3268
3269                 if ( $isLocalObj ) {
3270                         $ret = array( 'object' => $text );
3271                 } else {
3272                         $ret = array( 'text' => $text );
3273                 }
3274
3275                 wfProfileOut( __METHOD__ );
3276                 return $ret;
3277         }
3278
3279         /**
3280          * Get the semi-parsed DOM representation of a template with a given title,
3281          * and its redirect destination title. Cached.
3282          */
3283         function getTemplateDom( $title ) {
3284                 $cacheTitle = $title;
3285                 $titleText = $title->getPrefixedDBkey();
3286
3287                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3288                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3289                         $title = Title::makeTitle( $ns, $dbk );
3290                         $titleText = $title->getPrefixedDBkey();
3291                 }
3292                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3293                         return array( $this->mTplDomCache[$titleText], $title );
3294                 }
3295
3296                 # Cache miss, go to the database
3297                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3298
3299                 if ( $text === false ) {
3300                         $this->mTplDomCache[$titleText] = false;
3301                         return array( false, $title );
3302                 }
3303
3304                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3305                 $this->mTplDomCache[ $titleText ] = $dom;
3306
3307                 if ( !$title->equals( $cacheTitle ) ) {
3308                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3309                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3310                 }
3311
3312                 return array( $dom, $title );
3313         }
3314
3315         /**
3316          * Fetch the unparsed text of a template and register a reference to it.
3317          */
3318         function fetchTemplateAndTitle( $title ) {
3319                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3320                 $stuff = call_user_func( $templateCb, $title, $this );
3321                 $text = $stuff['text'];
3322                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3323                 if ( isset( $stuff['deps'] ) ) {
3324                         foreach ( $stuff['deps'] as $dep ) {
3325                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3326                         }
3327                 }
3328                 return array( $text, $finalTitle );
3329         }
3330
3331         function fetchTemplate( $title ) {
3332                 $rv = $this->fetchTemplateAndTitle( $title );
3333                 return $rv[0];
3334         }
3335
3336         /**
3337          * Static function to get a template
3338          * Can be overridden via ParserOptions::setTemplateCallback().
3339          */
3340         static function statelessFetchTemplate( $title, $parser=false ) {
3341                 $text = $skip = false;
3342                 $finalTitle = $title;
3343                 $deps = array();
3344
3345                 # Loop to fetch the article, with up to 1 redirect
3346                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3347                         # Give extensions a chance to select the revision instead
3348                         $id = false; # Assume current
3349                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3350
3351                         if ( $skip ) {
3352                                 $text = false;
3353                                 $deps[] = array(
3354                                         'title' => $title,
3355                                         'page_id' => $title->getArticleID(),
3356                                         'rev_id' => null );
3357                                 break;
3358                         }
3359                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3360                         $rev_id = $rev ? $rev->getId() : 0;
3361                         # If there is no current revision, there is no page
3362                         if ( $id === false && !$rev ) {
3363                                 $linkCache = LinkCache::singleton();
3364                                 $linkCache->addBadLinkObj( $title );
3365                         }
3366
3367                         $deps[] = array(
3368                                 'title' => $title,
3369                                 'page_id' => $title->getArticleID(),
3370                                 'rev_id' => $rev_id );
3371
3372                         if ( $rev ) {
3373                                 $text = $rev->getText();
3374                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3375                                 global $wgContLang;
3376                                 $message = $wgContLang->lcfirst( $title->getText() );
3377                                 $text = wfMsgForContentNoTrans( $message );
3378                                 if ( wfEmptyMsg( $message, $text ) ) {
3379                                         $text = false;
3380                                         break;
3381                                 }
3382                         } else {
3383                                 break;
3384                         }
3385                         if ( $text === false ) {
3386                                 break;
3387                         }
3388                         # Redirect?
3389                         $finalTitle = $title;
3390                         $title = Title::newFromRedirect( $text );
3391                 }
3392                 return array(
3393                         'text' => $text,
3394                         'finalTitle' => $finalTitle,
3395                         'deps' => $deps );
3396         }
3397
3398         /**
3399          * Transclude an interwiki link.
3400          */
3401         function interwikiTransclude( $title, $action ) {
3402                 global $wgEnableScaryTranscluding;
3403
3404                 if ( !$wgEnableScaryTranscluding ) {
3405                         return wfMsg('scarytranscludedisabled');
3406                 }
3407
3408                 $url = $title->getFullUrl( "action=$action" );
3409
3410                 if ( strlen( $url ) > 255 ) {
3411                         return wfMsg( 'scarytranscludetoolong' );
3412                 }
3413                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3414         }
3415
3416         function fetchScaryTemplateMaybeFromCache( $url ) {
3417                 global $wgTranscludeCacheExpiry;
3418                 $dbr = wfGetDB( DB_SLAVE );
3419                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3420                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3421                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3422                 if ( $obj ) {
3423                         return $obj->tc_contents;
3424                 }
3425
3426                 $text = Http::get( $url );
3427                 if ( !$text ) {
3428                         return wfMsg( 'scarytranscludefailed', $url );
3429                 }
3430
3431                 $dbw = wfGetDB( DB_MASTER );
3432                 $dbw->replace( 'transcache', array('tc_url'), array(
3433                         'tc_url' => $url,
3434                         'tc_time' => $dbw->timestamp( time() ),
3435                         'tc_contents' => $text)
3436                 );
3437                 return $text;
3438         }
3439
3440
3441         /**
3442          * Triple brace replacement -- used for template arguments
3443          * @private
3444          */
3445         function argSubstitution( $piece, $frame ) {
3446                 wfProfileIn( __METHOD__ );
3447
3448                 $error = false;
3449                 $parts = $piece['parts'];
3450                 $nameWithSpaces = $frame->expand( $piece['title'] );
3451                 $argName = trim( $nameWithSpaces );
3452                 $object = false;
3453                 $text = $frame->getArgument( $argName );
3454                 if (  $text === false && $parts->getLength() > 0
3455                   && (
3456                     $this->ot['html']
3457                     || $this->ot['pre']
3458                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3459                   )
3460                 ) {
3461                         # No match in frame, use the supplied default
3462                         $object = $parts->item( 0 )->getChildren();
3463                 }
3464                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3465                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3466                         $this->limitationWarn( 'post-expand-template-argument' );
3467                 }
3468
3469                 if ( $text === false && $object === false ) {
3470                         # No match anywhere
3471                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3472                 }
3473                 if ( $error !== false ) {
3474                         $text .= $error;
3475                 }
3476                 if ( $object !== false ) {
3477                         $ret = array( 'object' => $object );
3478                 } else {
3479                         $ret = array( 'text' => $text );
3480                 }
3481
3482                 wfProfileOut( __METHOD__ );
3483                 return $ret;
3484         }
3485
3486         /**
3487          * Return the text to be used for a given extension tag.
3488          * This is the ghost of strip().
3489          *
3490          * @param $params Associative array of parameters:
3491          *     name       PPNode for the tag name
3492          *     attr       PPNode for unparsed text where tag attributes are thought to be
3493          *     attributes Optional associative array of parsed attributes
3494          *     inner      Contents of extension element
3495          *     noClose    Original text did not have a close tag
3496          * @param $frame PPFrame
3497          */
3498         function extensionSubstitution( $params, $frame ) {
3499                 $name = $frame->expand( $params['name'] );
3500                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3501                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3502                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3503
3504                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3505                         ( $this->ot['html'] || $this->ot['pre'] );
3506                 if ( $isFunctionTag ) {
3507                         $markerType = 'none';
3508                 } else {
3509                         $markerType = 'general';
3510                 }
3511                 if ( $this->ot['html'] || $isFunctionTag ) {
3512                         $name = strtolower( $name );
3513                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3514                         if ( isset( $params['attributes'] ) ) {
3515                                 $attributes = $attributes + $params['attributes'];
3516                         }
3517
3518                         if ( isset( $this->mTagHooks[$name] ) ) {
3519                                 # Workaround for PHP bug 35229 and similar
3520                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3521                                         throw new MWException( "Tag hook for $name is not callable\n" );
3522                                 }
3523                                 $output = call_user_func_array( $this->mTagHooks[$name],
3524                                         array( $content, $attributes, $this, $frame ) );
3525                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3526                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3527                                 if ( !is_callable( $callback ) ) {
3528                                         throw new MWException( "Tag hook for $name is not callable\n" );
3529                                 }
3530
3531                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3532                         } else {
3533                                 $output = '<span class="error">Invalid tag extension name: ' .
3534                                         htmlspecialchars( $name ) . '</span>';
3535                         }
3536
3537                         if ( is_array( $output ) ) {
3538                                 # Extract flags to local scope (to override $markerType)
3539                                 $flags = $output;
3540                                 $output = $flags[0];
3541                                 unset( $flags[0] );
3542                                 extract( $flags );
3543                         }
3544                 } else {
3545                         if ( is_null( $attrText ) ) {
3546                                 $attrText = '';
3547                         }
3548                         if ( isset( $params['attributes'] ) ) {
3549                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3550                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3551                                                 htmlspecialchars( $attrValue ) . '"';
3552                                 }
3553                         }
3554                         if ( $content === null ) {
3555                                 $output = "<$name$attrText/>";
3556                         } else {
3557                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3558                                 $output = "<$name$attrText>$content$close";
3559                         }
3560                 }
3561
3562                 if ( $markerType === 'none' ) {
3563                         return $output;
3564                 } elseif ( $markerType === 'nowiki' ) {
3565                         $this->mStripState->nowiki->setPair( $marker, $output );
3566                 } elseif ( $markerType === 'general' ) {
3567                         $this->mStripState->general->setPair( $marker, $output );
3568                 } else {
3569                         throw new MWException( __METHOD__.': invalid marker type' );
3570                 }
3571                 return $marker;
3572         }
3573
3574         /**
3575          * Increment an include size counter
3576          *
3577          * @param $type String: the type of expansion
3578          * @param $size Integer: the size of the text
3579          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3580          */
3581         function incrementIncludeSize( $type, $size ) {
3582                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3583                         return false;
3584                 } else {
3585                         $this->mIncludeSizes[$type] += $size;
3586                         return true;
3587                 }
3588         }
3589
3590         /**
3591          * Increment the expensive function count
3592          *
3593          * @return Boolean: false if the limit has been exceeded
3594          */
3595         function incrementExpensiveFunctionCount() {
3596                 global $wgExpensiveParserFunctionLimit;
3597                 $this->mExpensiveFunctionCount++;
3598                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3599                         return true;
3600                 }
3601                 return false;
3602         }
3603
3604         /**
3605          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3606          * Fills $this->mDoubleUnderscores, returns the modified text
3607          */
3608         function doDoubleUnderscore( $text ) {
3609                 wfProfileIn( __METHOD__ );
3610
3611                 # The position of __TOC__ needs to be recorded
3612                 $mw = MagicWord::get( 'toc' );
3613                 if ( $mw->match( $text ) ) {
3614                         $this->mShowToc = true;
3615                         $this->mForceTocPosition = true;
3616
3617                         # Set a placeholder. At the end we'll fill it in with the TOC.
3618                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3619
3620                         # Only keep the first one.
3621                         $text = $mw->replace( '', $text );
3622                 }
3623
3624                 # Now match and remove the rest of them
3625                 $mwa = MagicWord::getDoubleUnderscoreArray();
3626                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3627
3628                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3629                         $this->mOutput->mNoGallery = true;
3630                 }
3631                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3632                         $this->mShowToc = false;
3633                 }
3634                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3635                         $this->addTrackingCategory( 'hidden-category-category' );
3636                 }
3637                 # (bug 8068) Allow control over whether robots index a page.
3638                 #
3639                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3640                 # is not desirable, the last one on the page should win.
3641                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3642                         $this->mOutput->setIndexPolicy( 'noindex' );
3643                         $this->addTrackingCategory( 'noindex-category' );
3644                 }
3645                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3646                         $this->mOutput->setIndexPolicy( 'index' );
3647                         $this->addTrackingCategory( 'index-category' );
3648                 }
3649
3650                 # Cache all double underscores in the database
3651                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3652                         $this->mOutput->setProperty( $key, '' );
3653                 }
3654
3655                 wfProfileOut( __METHOD__ );
3656                 return $text;
3657         }
3658
3659         /**
3660          * Add a tracking category, getting the title from a system message,
3661          * or print a debug message if the title is invalid.
3662          *
3663          * @param $msg String: message key
3664          * @return Boolean: whether the addition was successful
3665          */
3666         protected function addTrackingCategory( $msg ) {
3667                 $cat = wfMsgForContent( $msg );
3668
3669                 # Allow tracking categories to be disabled by setting them to "-"
3670                 if ( $cat === '-' ) {
3671                         return false;
3672                 }
3673
3674                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3675                 if ( $containerCategory ) {
3676                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3677                         return true;
3678                 } else {
3679                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3680                         return false;
3681                 }
3682         }
3683
3684         /**
3685          * This function accomplishes several tasks:
3686          * 1) Auto-number headings if that option is enabled
3687          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3688          * 3) Add a Table of contents on the top for users who have enabled the option
3689          * 4) Auto-anchor headings
3690          *
3691          * It loops through all headlines, collects the necessary data, then splits up the
3692          * string and re-inserts the newly formatted headlines.
3693          *
3694          * @param $text String
3695          * @param $origText String: original, untouched wikitext
3696          * @param $isMain Boolean
3697          * @private
3698          */
3699         function formatHeadings( $text, $origText, $isMain=true ) {
3700                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3701
3702                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3703                 $showEditLink = $this->mOptions->getEditSection();
3704
3705                 # Do not call quickUserCan unless necessary
3706                 if ( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3707                         $showEditLink = 0;
3708                 }
3709
3710                 # Inhibit editsection links if requested in the page
3711                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3712                         $showEditLink = 0;
3713                 }
3714
3715                 # Get all headlines for numbering them and adding funky stuff like [edit]
3716                 # links - this is for later, but we need the number of headlines right now
3717                 $matches = array();
3718                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3719
3720                 # if there are fewer than 4 headlines in the article, do not show TOC
3721                 # unless it's been explicitly enabled.
3722                 $enoughToc = $this->mShowToc &&
3723                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3724
3725                 # Allow user to stipulate that a page should have a "new section"
3726                 # link added via __NEWSECTIONLINK__
3727                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3728                         $this->mOutput->setNewSection( true );
3729                 }
3730
3731                 # Allow user to remove the "new section"
3732                 # link via __NONEWSECTIONLINK__
3733                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3734                         $this->mOutput->hideNewSection( true );
3735                 }
3736
3737                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3738                 # override above conditions and always show TOC above first header
3739                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3740                         $this->mShowToc = true;
3741                         $enoughToc = true;
3742                 }
3743
3744                 # We need this to perform operations on the HTML
3745                 $sk = $this->mOptions->getSkin();
3746
3747                 # headline counter
3748                 $headlineCount = 0;
3749                 $numVisible = 0;
3750
3751                 # Ugh .. the TOC should have neat indentation levels which can be
3752                 # passed to the skin functions. These are determined here
3753                 $toc = '';
3754                 $full = '';
3755                 $head = array();
3756                 $sublevelCount = array();
3757                 $levelCount = array();
3758                 $toclevel = 0;
3759                 $level = 0;
3760                 $prevlevel = 0;
3761                 $toclevel = 0;
3762                 $prevtoclevel = 0;
3763                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3764                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3765                 $oldType = $this->mOutputType;
3766                 $this->setOutputType( self::OT_WIKI );
3767                 $frame = $this->getPreprocessor()->newFrame();
3768                 $root = $this->preprocessToDom( $origText );
3769                 $node = $root->getFirstChild();
3770                 $byteOffset = 0;
3771                 $tocraw = array();
3772
3773                 foreach ( $matches[3] as $headline ) {
3774                         $isTemplate = false;
3775                         $titleText = false;
3776                         $sectionIndex = false;
3777                         $numbering = '';
3778                         $markerMatches = array();
3779                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3780                                 $serial = $markerMatches[1];
3781                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3782                                 $isTemplate = ( $titleText != $baseTitleText );
3783                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3784                         }
3785
3786                         if ( $toclevel ) {
3787                                 $prevlevel = $level;
3788                                 $prevtoclevel = $toclevel;
3789                         }
3790                         $level = $matches[1][$headlineCount];
3791
3792                         if ( $level > $prevlevel ) {
3793                                 # Increase TOC level
3794                                 $toclevel++;
3795                                 $sublevelCount[$toclevel] = 0;
3796                                 if ( $toclevel<$wgMaxTocLevel ) {
3797                                         $prevtoclevel = $toclevel;
3798                                         $toc .= $sk->tocIndent();
3799                                         $numVisible++;
3800                                 }
3801                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3802                                 # Decrease TOC level, find level to jump to
3803
3804                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3805                                         if ( $levelCount[$i] == $level ) {
3806                                                 # Found last matching level
3807                                                 $toclevel = $i;
3808                                                 break;
3809                                         } elseif ( $levelCount[$i] < $level ) {
3810                                                 # Found first matching level below current level
3811                                                 $toclevel = $i + 1;
3812                                                 break;
3813                                         }
3814                                 }
3815                                 if ( $i == 0 ) {
3816                                         $toclevel = 1;
3817                                 }
3818                                 if ( $toclevel<$wgMaxTocLevel ) {
3819                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3820                                                 # Unindent only if the previous toc level was shown :p
3821                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3822                                                 $prevtoclevel = $toclevel;
3823                                         } else {
3824                                                 $toc .= $sk->tocLineEnd();
3825                                         }
3826                                 }
3827                         } else {
3828                                 # No change in level, end TOC line
3829                                 if ( $toclevel<$wgMaxTocLevel ) {
3830                                         $toc .= $sk->tocLineEnd();
3831                                 }
3832                         }
3833
3834                         $levelCount[$toclevel] = $level;
3835
3836                         # count number of headlines for each level
3837                         @$sublevelCount[$toclevel]++;
3838                         $dot = 0;
3839                         for( $i = 1; $i <= $toclevel; $i++ ) {
3840                                 if ( !empty( $sublevelCount[$i] ) ) {
3841                                         if ( $dot ) {
3842                                                 $numbering .= '.';
3843                                         }
3844                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3845                                         $dot = 1;
3846                                 }
3847                         }
3848
3849                         # The safe header is a version of the header text safe to use for links
3850                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3851                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3852
3853                         # Remove link placeholders by the link text.
3854                         #     <!--LINK number-->
3855                         # turns into
3856                         #     link text with suffix
3857                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3858
3859                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3860                         $tocline = preg_replace(
3861                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3862                                 array( '',                          '<$1>' ),
3863                                 $safeHeadline
3864                         );
3865                         $tocline = trim( $tocline );
3866
3867                         # For the anchor, strip out HTML-y stuff period
3868                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3869                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
3870
3871                         # Save headline for section edit hint before it's escaped
3872                         $headlineHint = $safeHeadline;
3873
3874                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3875                                 # For reverse compatibility, provide an id that's
3876                                 # HTML4-compatible, like we used to.
3877                                 #
3878                                 # It may be worth noting, academically, that it's possible for
3879                                 # the legacy anchor to conflict with a non-legacy headline
3880                                 # anchor on the page.  In this case likely the "correct" thing
3881                                 # would be to either drop the legacy anchors or make sure
3882                                 # they're numbered first.  However, this would require people
3883                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3884                                 # manually, so let's not bother worrying about it.
3885                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3886                                         array( 'noninitial', 'legacy' ) );
3887                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3888
3889                                 if ( $legacyHeadline == $safeHeadline ) {
3890                                         # No reason to have both (in fact, we can't)
3891                                         $legacyHeadline = false;
3892                                 }
3893                         } else {
3894                                 $legacyHeadline = false;
3895                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3896                                         'noninitial' );
3897                         }
3898
3899                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3900                         # Does this apply to Unicode characters?  Because we aren't
3901                         # handling those here.
3902                         $arrayKey = strtolower( $safeHeadline );
3903                         if ( $legacyHeadline === false ) {
3904                                 $legacyArrayKey = false;
3905                         } else {
3906                                 $legacyArrayKey = strtolower( $legacyHeadline );
3907                         }
3908
3909                         # count how many in assoc. array so we can track dupes in anchors
3910                         if ( isset( $refers[$arrayKey] ) ) {
3911                                 $refers[$arrayKey]++;
3912                         } else {
3913                                 $refers[$arrayKey] = 1;
3914                         }
3915                         if ( isset( $refers[$legacyArrayKey] ) ) {
3916                                 $refers[$legacyArrayKey]++;
3917                         } else {
3918                                 $refers[$legacyArrayKey] = 1;
3919                         }
3920
3921                         # Don't number the heading if it is the only one (looks silly)
3922                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3923                                 # the two are different if the line contains a link
3924                                 $headline = $numbering . ' ' . $headline;
3925                         }
3926
3927                         # Create the anchor for linking from the TOC to the section
3928                         $anchor = $safeHeadline;
3929                         $legacyAnchor = $legacyHeadline;
3930                         if ( $refers[$arrayKey] > 1 ) {
3931                                 $anchor .= '_' . $refers[$arrayKey];
3932                         }
3933                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3934                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3935                         }
3936                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3937                                 $toc .= $sk->tocLine( $anchor, $tocline,
3938                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3939                         }
3940
3941                         # Add the section to the section tree
3942                         # Find the DOM node for this header
3943                         while ( $node && !$isTemplate ) {
3944                                 if ( $node->getName() === 'h' ) {
3945                                         $bits = $node->splitHeading();
3946                                         if ( $bits['i'] == $sectionIndex )
3947                                                 break;
3948                                 }
3949                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3950                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3951                                 $node = $node->getNextSibling();
3952                         }
3953                         $tocraw[] = array(
3954                                 'toclevel' => $toclevel,
3955                                 'level' => $level,
3956                                 'line' => $tocline,
3957                                 'number' => $numbering,
3958                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3959                                 'fromtitle' => $titleText,
3960                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3961                                 'anchor' => $anchor,
3962                         );
3963
3964                         # give headline the correct <h#> tag
3965                         if ( $showEditLink && $sectionIndex !== false ) {
3966                                 if ( $isTemplate ) {
3967                                         # Put a T flag in the section identifier, to indicate to extractSections()
3968                                         # that sections inside <includeonly> should be counted.
3969                                         $editlink = $sk->doEditSectionLink( Title::newFromText( $titleText ), "T-$sectionIndex" );
3970                                 } else {
3971                                         $editlink = $sk->doEditSectionLink( $this->mTitle, $sectionIndex, $headlineHint );
3972                                 }
3973                         } else {
3974                                 $editlink = '';
3975                         }
3976                         $head[$headlineCount] = $sk->makeHeadline( $level,
3977                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3978                                 $editlink, $legacyAnchor );
3979
3980                         $headlineCount++;
3981                 }
3982
3983                 $this->setOutputType( $oldType );
3984
3985                 # Never ever show TOC if no headers
3986                 if ( $numVisible < 1 ) {
3987                         $enoughToc = false;
3988                 }
3989
3990                 if ( $enoughToc ) {
3991                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3992                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3993                         }
3994                         $toc = $sk->tocList( $toc );
3995                         $this->mOutput->setTOCHTML( $toc );
3996                 }
3997
3998                 if ( $isMain ) {
3999                         $this->mOutput->setSections( $tocraw );
4000                 }
4001
4002                 # split up and insert constructed headlines
4003
4004                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4005                 $i = 0;
4006
4007                 foreach ( $blocks as $block ) {
4008                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4009                                 # This is the [edit] link that appears for the top block of text when
4010                                 # section editing is enabled
4011
4012                                 # Disabled because it broke block formatting
4013                                 # For example, a bullet point in the top line
4014                                 # $full .= $sk->editSectionLink(0);
4015                         }
4016                         $full .= $block;
4017                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4018                                 # Top anchor now in skin
4019                                 $full = $full.$toc;
4020                         }
4021
4022                         if ( !empty( $head[$i] ) ) {
4023                                 $full .= $head[$i];
4024                         }
4025                         $i++;
4026                 }
4027                 if ( $this->mForceTocPosition ) {
4028                         return str_replace( '<!--MWTOC-->', $toc, $full );
4029                 } else {
4030                         return $full;
4031                 }
4032         }
4033
4034         /**
4035          * Transform wiki markup when saving a page by doing \r\n -> \n
4036          * conversion, substitting signatures, {{subst:}} templates, etc.
4037          *
4038          * @param $text String: the text to transform
4039          * @param &$title Title: the Title object for the current article
4040          * @param $user User: the User object describing the current user
4041          * @param $options ParserOptions: parsing options
4042          * @param $clearState Boolean: whether to clear the parser state first
4043          * @return String: the altered wiki markup
4044          */
4045         public function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
4046                 $this->mOptions = $options;
4047                 $this->setTitle( $title );
4048                 $this->setOutputType( self::OT_WIKI );
4049
4050                 if ( $clearState ) {
4051                         $this->clearState();
4052                 }
4053
4054                 $pairs = array(
4055                         "\r\n" => "\n",
4056                 );
4057                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4058                 $text = $this->pstPass2( $text, $user );
4059                 $text = $this->mStripState->unstripBoth( $text );
4060                 return $text;
4061         }
4062
4063         /**
4064          * Pre-save transform helper function
4065          * @private
4066          */
4067         function pstPass2( $text, $user ) {
4068                 global $wgContLang, $wgLocaltimezone;
4069
4070                 # Note: This is the timestamp saved as hardcoded wikitext to
4071                 # the database, we use $wgContLang here in order to give
4072                 # everyone the same signature and use the default one rather
4073                 # than the one selected in each user's preferences.
4074                 # (see also bug 12815)
4075                 $ts = $this->mOptions->getTimestamp();
4076                 if ( isset( $wgLocaltimezone ) ) {
4077                         $tz = $wgLocaltimezone;
4078                 } else {
4079                         $tz = date_default_timezone_get();
4080                 }
4081
4082                 $unixts = wfTimestamp( TS_UNIX, $ts );
4083                 $oldtz = date_default_timezone_get();
4084                 date_default_timezone_set( $tz );
4085                 $ts = date( 'YmdHis', $unixts );
4086                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4087
4088                 # Allow translation of timezones through wiki. date() can return
4089                 # whatever crap the system uses, localised or not, so we cannot
4090                 # ship premade translations.
4091                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4092                 $value = wfMsgForContent( $key );
4093                 if ( !wfEmptyMsg( $key, $value ) ) {
4094                         $tzMsg = $value;
4095                 }
4096
4097                 date_default_timezone_set( $oldtz );
4098
4099                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4100
4101                 # Variable replacement
4102                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4103                 $text = $this->replaceVariables( $text );
4104
4105                 # Signatures
4106                 $sigText = $this->getUserSig( $user );
4107                 $text = strtr( $text, array(
4108                         '~~~~~' => $d,
4109                         '~~~~' => "$sigText $d",
4110                         '~~~' => $sigText
4111                 ) );
4112
4113                 # Context links: [[|name]] and [[name (context)|]]
4114                 global $wgLegalTitleChars;
4115                 $tc = "[$wgLegalTitleChars]";
4116                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4117
4118                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4119                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4120                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4121                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4122
4123                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4124                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4125                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4126                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4127
4128                 $t = $this->mTitle->getText();
4129                 $m = array();
4130                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4131                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4132                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4133                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4134                 } else {
4135                         # if there's no context, don't bother duplicating the title
4136                         $text = preg_replace( $p2, '[[\\1]]', $text );
4137                 }
4138
4139                 # Trim trailing whitespace
4140                 $text = rtrim( $text );
4141
4142                 return $text;
4143         }
4144
4145         /**
4146          * Fetch the user's signature text, if any, and normalize to
4147          * validated, ready-to-insert wikitext.
4148          * If you have pre-fetched the nickname or the fancySig option, you can
4149          * specify them here to save a database query.
4150          *
4151          * @param $user User
4152          * @param $nickname String: nickname to use or false to use user's default nickname
4153          * @param $fancySig Boolean: whether the nicknname is the complete signature
4154          *                  or null to use default value
4155          * @return string
4156          */
4157         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4158                 global $wgMaxSigChars;
4159
4160                 $username = $user->getName();
4161
4162                 # If not given, retrieve from the user object.
4163                 if ( $nickname === false )
4164                         $nickname = $user->getOption( 'nickname' );
4165
4166                 if ( is_null( $fancySig ) ) {
4167                         $fancySig = $user->getBoolOption( 'fancysig' );
4168                 }
4169
4170                 $nickname = $nickname == null ? $username : $nickname;
4171
4172                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4173                         $nickname = $username;
4174                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4175                 } elseif ( $fancySig !== false ) {
4176                         # Sig. might contain markup; validate this
4177                         if ( $this->validateSig( $nickname ) !== false ) {
4178                                 # Validated; clean up (if needed) and return it
4179                                 return $this->cleanSig( $nickname, true );
4180                         } else {
4181                                 # Failed to validate; fall back to the default
4182                                 $nickname = $username;
4183                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4184                         }
4185                 }
4186
4187                 # Make sure nickname doesnt get a sig in a sig
4188                 $nickname = $this->cleanSigInSig( $nickname );
4189
4190                 # If we're still here, make it a link to the user page
4191                 $userText = wfEscapeWikiText( $username );
4192                 $nickText = wfEscapeWikiText( $nickname );
4193                 if ( $user->isAnon() )  {
4194                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4195                 } else {
4196                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4197                 }
4198         }
4199
4200         /**
4201          * Check that the user's signature contains no bad XML
4202          *
4203          * @param $text String
4204          * @return mixed An expanded string, or false if invalid.
4205          */
4206         function validateSig( $text ) {
4207                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4208         }
4209
4210         /**
4211          * Clean up signature text
4212          *
4213          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4214          * 2) Substitute all transclusions
4215          *
4216          * @param $text String
4217          * @param $parsing Whether we're cleaning (preferences save) or parsing
4218          * @return String: signature text
4219          */
4220         function cleanSig( $text, $parsing = false ) {
4221                 if ( !$parsing ) {
4222                         global $wgTitle;
4223                         $this->clearState();
4224                         $this->setTitle( $wgTitle );
4225                         $this->mOptions = new ParserOptions;
4226                         $this->setOutputType = self::OT_PREPROCESS;
4227                 }
4228
4229                 # Option to disable this feature
4230                 if ( !$this->mOptions->getCleanSignatures() ) {
4231                         return $text;
4232                 }
4233
4234                 # FIXME: regex doesn't respect extension tags or nowiki
4235                 #  => Move this logic to braceSubstitution()
4236                 $substWord = MagicWord::get( 'subst' );
4237                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4238                 $substText = '{{' . $substWord->getSynonym( 0 );
4239
4240                 $text = preg_replace( $substRegex, $substText, $text );
4241                 $text = $this->cleanSigInSig( $text );
4242                 $dom = $this->preprocessToDom( $text );
4243                 $frame = $this->getPreprocessor()->newFrame();
4244                 $text = $frame->expand( $dom );
4245
4246                 if ( !$parsing ) {
4247                         $text = $this->mStripState->unstripBoth( $text );
4248                 }
4249
4250                 return $text;
4251         }
4252
4253         /**
4254          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4255          *
4256          * @param $text String
4257          * @return String: signature text with /~{3,5}/ removed
4258          */
4259         function cleanSigInSig( $text ) {
4260                 $text = preg_replace( '/~{3,5}/', '', $text );
4261                 return $text;
4262         }
4263
4264         /**
4265          * Set up some variables which are usually set up in parse()
4266          * so that an external function can call some class members with confidence
4267          */
4268         public function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4269                 $this->setTitle( $title );
4270                 $this->mOptions = $options;
4271                 $this->setOutputType( $outputType );
4272                 if ( $clearState ) {
4273                         $this->clearState();
4274                 }
4275         }
4276
4277         /**
4278          * Wrapper for preprocess()
4279          *
4280          * @param $text String: the text to preprocess
4281          * @param $options ParserOptions: options
4282          * @return String
4283          */
4284         public function transformMsg( $text, $options ) {
4285                 global $wgTitle;
4286                 static $executing = false;
4287
4288                 # Guard against infinite recursion
4289                 if ( $executing ) {
4290                         return $text;
4291                 }
4292                 $executing = true;
4293
4294                 wfProfileIn( __METHOD__ );
4295                 $text = $this->preprocess( $text, $wgTitle, $options );
4296
4297                 $executing = false;
4298                 wfProfileOut( __METHOD__ );
4299                 return $text;
4300         }
4301
4302         /**
4303          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4304          * The callback should have the following form:
4305          *    function myParserHook( $text, $params, $parser ) { ... }
4306          *
4307          * Transform and return $text. Use $parser for any required context, e.g. use
4308          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4309          *
4310          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4311          * @param $callback Mixed: the callback function (and object) to use for the tag
4312          * @return The old value of the mTagHooks array associated with the hook
4313          */
4314         public function setHook( $tag, $callback ) {
4315                 $tag = strtolower( $tag );
4316                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4317                 $this->mTagHooks[$tag] = $callback;
4318                 if ( !in_array( $tag, $this->mStripList ) ) {
4319                         $this->mStripList[] = $tag;
4320                 }
4321
4322                 return $oldVal;
4323         }
4324
4325         function setTransparentTagHook( $tag, $callback ) {
4326                 $tag = strtolower( $tag );
4327                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4328                 $this->mTransparentTagHooks[$tag] = $callback;
4329
4330                 return $oldVal;
4331         }
4332
4333         /**
4334          * Remove all tag hooks
4335          */
4336         function clearTagHooks() {
4337                 $this->mTagHooks = array();
4338                 $this->mStripList = $this->mDefaultStripList;
4339         }
4340
4341         /**
4342          * Create a function, e.g. {{sum:1|2|3}}
4343          * The callback function should have the form:
4344          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4345          *
4346          * Or with SFH_OBJECT_ARGS:
4347          *    function myParserFunction( $parser, $frame, $args ) { ... }
4348          *
4349          * The callback may either return the text result of the function, or an array with the text
4350          * in element 0, and a number of flags in the other elements. The names of the flags are
4351          * specified in the keys. Valid flags are:
4352          *   found                     The text returned is valid, stop processing the template. This
4353          *                             is on by default.
4354          *   nowiki                    Wiki markup in the return value should be escaped
4355          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4356          *
4357          * @param $id String: The magic word ID
4358          * @param $callback Mixed: the callback function (and object) to use
4359          * @param $flags Integer: a combination of the following flags:
4360          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4361          *
4362          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4363          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4364          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4365          *     the arguments, and to control the way they are expanded.
4366          *
4367          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4368          *     arguments, for instance:
4369          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4370          *
4371          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4372          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4373          *     working if/when this is changed.
4374          *
4375          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4376          *     expansion.
4377          *
4378          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4379          *     about the methods available in PPFrame and PPNode.
4380          *
4381          * @return The old callback function for this name, if any
4382          */
4383         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4384                 global $wgContLang;
4385
4386                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4387                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4388
4389                 # Add to function cache
4390                 $mw = MagicWord::get( $id );
4391                 if ( !$mw )
4392                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4393
4394                 $synonyms = $mw->getSynonyms();
4395                 $sensitive = intval( $mw->isCaseSensitive() );
4396
4397                 foreach ( $synonyms as $syn ) {
4398                         # Case
4399                         if ( !$sensitive ) {
4400                                 $syn = $wgContLang->lc( $syn );
4401                         }
4402                         # Add leading hash
4403                         if ( !( $flags & SFH_NO_HASH ) ) {
4404                                 $syn = '#' . $syn;
4405                         }
4406                         # Remove trailing colon
4407                         if ( substr( $syn, -1, 1 ) === ':' ) {
4408                                 $syn = substr( $syn, 0, -1 );
4409                         }
4410                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4411                 }
4412                 return $oldVal;
4413         }
4414
4415         /**
4416          * Get all registered function hook identifiers
4417          *
4418          * @return Array
4419          */
4420         function getFunctionHooks() {
4421                 return array_keys( $this->mFunctionHooks );
4422         }
4423
4424         /**
4425          * Create a tag function, e.g. <test>some stuff</test>.
4426          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4427          * Unlike parser functions, their content is not preprocessed.
4428          */
4429         function setFunctionTagHook( $tag, $callback, $flags ) {
4430                 $tag = strtolower( $tag );
4431                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4432                         $this->mFunctionTagHooks[$tag] : null;
4433                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4434
4435                 if ( !in_array( $tag, $this->mStripList ) ) {
4436                         $this->mStripList[] = $tag;
4437                 }
4438
4439                 return $old;
4440         }
4441
4442         /**
4443          * FIXME: update documentation. makeLinkObj() is deprecated.
4444          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4445          * Placeholders created in Skin::makeLinkObj()
4446          * Returns an array of link CSS classes, indexed by PDBK.
4447          */
4448         function replaceLinkHolders( &$text, $options = 0 ) {
4449                 return $this->mLinkHolders->replace( $text );
4450         }
4451
4452         /**
4453          * Replace <!--LINK--> link placeholders with plain text of links
4454          * (not HTML-formatted).
4455          *
4456          * @param $text String
4457          * @return String
4458          */
4459         function replaceLinkHoldersText( $text ) {
4460                 return $this->mLinkHolders->replaceText( $text );
4461         }
4462
4463         /**
4464          * Renders an image gallery from a text with one line per image.
4465          * text labels may be given by using |-style alternative text. E.g.
4466          *   Image:one.jpg|The number "1"
4467          *   Image:tree.jpg|A tree
4468          * given as text will return the HTML of a gallery with two images,
4469          * labeled 'The number "1"' and
4470          * 'A tree'.
4471          */
4472         function renderImageGallery( $text, $params ) {
4473                 $ig = new ImageGallery();
4474                 $ig->setContextTitle( $this->mTitle );
4475                 $ig->setShowBytes( false );
4476                 $ig->setShowFilename( false );
4477                 $ig->setParser( $this );
4478                 $ig->setHideBadImages();
4479                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4480                 $ig->useSkin( $this->mOptions->getSkin() );
4481                 $ig->mRevisionId = $this->mRevisionId;
4482
4483                 if ( isset( $params['showfilename'] ) ) {
4484                         $ig->setShowFilename( true );
4485                 } else {
4486                         $ig->setShowFilename( false );
4487                 }
4488                 if ( isset( $params['caption'] ) ) {
4489                         $caption = $params['caption'];
4490                         $caption = htmlspecialchars( $caption );
4491                         $caption = $this->replaceInternalLinks( $caption );
4492                         $ig->setCaptionHtml( $caption );
4493                 }
4494                 if ( isset( $params['perrow'] ) ) {
4495                         $ig->setPerRow( $params['perrow'] );
4496                 }
4497                 if ( isset( $params['widths'] ) ) {
4498                         $ig->setWidths( $params['widths'] );
4499                 }
4500                 if ( isset( $params['heights'] ) ) {
4501                         $ig->setHeights( $params['heights'] );
4502                 }
4503
4504                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4505
4506                 $lines = StringUtils::explode( "\n", $text );
4507                 foreach ( $lines as $line ) {
4508                         # match lines like these:
4509                         # Image:someimage.jpg|This is some image
4510                         $matches = array();
4511                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4512                         # Skip empty lines
4513                         if ( count( $matches ) == 0 ) {
4514                                 continue;
4515                         }
4516
4517                         if ( strpos( $matches[0], '%' ) !== false ) {
4518                                 $matches[1] = urldecode( $matches[1] );
4519                         }
4520                         $tp = Title::newFromText( $matches[1] );
4521                         $nt =& $tp;
4522                         if ( is_null( $nt ) ) {
4523                                 # Bogus title. Ignore these so we don't bomb out later.
4524                                 continue;
4525                         }
4526                         if ( isset( $matches[3] ) ) {
4527                                 $label = $matches[3];
4528                         } else {
4529                                 $label = '';
4530                         }
4531
4532                         $html = $this->recursiveTagParse( trim( $label ) );
4533
4534                         $ig->add( $nt, $html );
4535
4536                         # Only add real images (bug #5586)
4537                         if ( $nt->getNamespace() == NS_FILE ) {
4538                                 $this->mOutput->addImage( $nt->getDBkey() );
4539                         }
4540                 }
4541                 return $ig->toHTML();
4542         }
4543
4544         function getImageParams( $handler ) {
4545                 if ( $handler ) {
4546                         $handlerClass = get_class( $handler );
4547                 } else {
4548                         $handlerClass = '';
4549                 }
4550                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4551                         # Initialise static lists
4552                         static $internalParamNames = array(
4553                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4554                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4555                                         'bottom', 'text-bottom' ),
4556                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4557                                         'upright', 'border', 'link', 'alt' ),
4558                         );
4559                         static $internalParamMap;
4560                         if ( !$internalParamMap ) {
4561                                 $internalParamMap = array();
4562                                 foreach ( $internalParamNames as $type => $names ) {
4563                                         foreach ( $names as $name ) {
4564                                                 $magicName = str_replace( '-', '_', "img_$name" );
4565                                                 $internalParamMap[$magicName] = array( $type, $name );
4566                                         }
4567                                 }
4568                         }
4569
4570                         # Add handler params
4571                         $paramMap = $internalParamMap;
4572                         if ( $handler ) {
4573                                 $handlerParamMap = $handler->getParamMap();
4574                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4575                                         $paramMap[$magic] = array( 'handler', $paramName );
4576                                 }
4577                         }
4578                         $this->mImageParams[$handlerClass] = $paramMap;
4579                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4580                 }
4581                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4582         }
4583
4584         /**
4585          * Parse image options text and use it to make an image
4586          *
4587          * @param $title Title
4588          * @param $options String
4589          * @param $holders LinkHolderArray
4590          */
4591         function makeImage( $title, $options, $holders = false ) {
4592                 # Check if the options text is of the form "options|alt text"
4593                 # Options are:
4594                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4595                 #  * left       no resizing, just left align. label is used for alt= only
4596                 #  * right      same, but right aligned
4597                 #  * none       same, but not aligned
4598                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4599                 #  * center     center the image
4600                 #  * frame      Keep original image size, no magnify-button.
4601                 #  * framed     Same as "frame"
4602                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4603                 #  * upright    reduce width for upright images, rounded to full __0 px
4604                 #  * border     draw a 1px border around the image
4605                 #  * alt        Text for HTML alt attribute (defaults to empty)
4606                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4607                 # vertical-align values (no % or length right now):
4608                 #  * baseline
4609                 #  * sub
4610                 #  * super
4611                 #  * top
4612                 #  * text-top
4613                 #  * middle
4614                 #  * bottom
4615                 #  * text-bottom
4616
4617                 $parts = StringUtils::explode( "|", $options );
4618                 $sk = $this->mOptions->getSkin();
4619
4620                 # Give extensions a chance to select the file revision for us
4621                 $skip = $time = $descQuery = false;
4622                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4623
4624                 if ( $skip ) {
4625                         return $sk->link( $title );
4626                 }
4627
4628                 # Get the file
4629                 $imagename = $title->getDBkey();
4630                 $file = wfFindFile( $title, array( 'time' => $time ) );
4631                 # Get parameter map
4632                 $handler = $file ? $file->getHandler() : false;
4633
4634                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4635
4636                 # Process the input parameters
4637                 $caption = '';
4638                 $params = array( 'frame' => array(), 'handler' => array(),
4639                         'horizAlign' => array(), 'vertAlign' => array() );
4640                 foreach ( $parts as $part ) {
4641                         $part = trim( $part );
4642                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4643                         $validated = false;
4644                         if ( isset( $paramMap[$magicName] ) ) {
4645                                 list( $type, $paramName ) = $paramMap[$magicName];
4646
4647                                 # Special case; width and height come in one variable together
4648                                 if ( $type === 'handler' && $paramName === 'width' ) {
4649                                         $m = array();
4650                                         # (bug 13500) In both cases (width/height and width only),
4651                                         # permit trailing "px" for backward compatibility.
4652                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4653                                                 $width = intval( $m[1] );
4654                                                 $height = intval( $m[2] );
4655                                                 if ( $handler->validateParam( 'width', $width ) ) {
4656                                                         $params[$type]['width'] = $width;
4657                                                         $validated = true;
4658                                                 }
4659                                                 if ( $handler->validateParam( 'height', $height ) ) {
4660                                                         $params[$type]['height'] = $height;
4661                                                         $validated = true;
4662                                                 }
4663                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4664                                                 $width = intval( $value );
4665                                                 if ( $handler->validateParam( 'width', $width ) ) {
4666                                                         $params[$type]['width'] = $width;
4667                                                         $validated = true;
4668                                                 }
4669                                         } # else no validation -- bug 13436
4670                                 } else {
4671                                         if ( $type === 'handler' ) {
4672                                                 # Validate handler parameter
4673                                                 $validated = $handler->validateParam( $paramName, $value );
4674                                         } else {
4675                                                 # Validate internal parameters
4676                                                 switch( $paramName ) {
4677                                                 case 'manualthumb':
4678                                                 case 'alt':
4679                                                         # @todo Fixme: possibly check validity here for
4680                                                         # manualthumb? downstream behavior seems odd with
4681                                                         # missing manual thumbs.
4682                                                         $validated = true;
4683                                                         $value = $this->stripAltText( $value, $holders );
4684                                                         break;
4685                                                 case 'link':
4686                                                         $chars = self::EXT_LINK_URL_CLASS;
4687                                                         $prots = $this->mUrlProtocols;
4688                                                         if ( $value === '' ) {
4689                                                                 $paramName = 'no-link';
4690                                                                 $value = true;
4691                                                                 $validated = true;
4692                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4693                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4694                                                                         $paramName = 'link-url';
4695                                                                         $this->mOutput->addExternalLink( $value );
4696                                                                         $validated = true;
4697                                                                 }
4698                                                         } else {
4699                                                                 $linkTitle = Title::newFromText( $value );
4700                                                                 if ( $linkTitle ) {
4701                                                                         $paramName = 'link-title';
4702                                                                         $value = $linkTitle;
4703                                                                         $this->mOutput->addLink( $linkTitle );
4704                                                                         $validated = true;
4705                                                                 }
4706                                                         }
4707                                                         break;
4708                                                 default:
4709                                                         # Most other things appear to be empty or numeric...
4710                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4711                                                 }
4712                                         }
4713
4714                                         if ( $validated ) {
4715                                                 $params[$type][$paramName] = $value;
4716                                         }
4717                                 }
4718                         }
4719                         if ( !$validated ) {
4720                                 $caption = $part;
4721                         }
4722                 }
4723
4724                 # Process alignment parameters
4725                 if ( $params['horizAlign'] ) {
4726                         $params['frame']['align'] = key( $params['horizAlign'] );
4727                 }
4728                 if ( $params['vertAlign'] ) {
4729                         $params['frame']['valign'] = key( $params['vertAlign'] );
4730                 }
4731
4732                 $params['frame']['caption'] = $caption;
4733
4734                 # Will the image be presented in a frame, with the caption below?
4735                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4736                                  isset( $params['frame']['framed'] ) ||
4737                                  isset( $params['frame']['thumbnail'] ) ||
4738                                  isset( $params['frame']['manualthumb'] );
4739
4740                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4741                 # came to also set the caption, ordinary text after the image -- which
4742                 # makes no sense, because that just repeats the text multiple times in
4743                 # screen readers.  It *also* came to set the title attribute.
4744                 #
4745                 # Now that we have an alt attribute, we should not set the alt text to
4746                 # equal the caption: that's worse than useless, it just repeats the
4747                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4748                 # use the unnamed parameter for alt text as well, just for the time be-
4749                 # ing, if the unnamed param is set and the alt param is not.
4750                 #
4751                 # For the future, we need to figure out if we want to tweak this more,
4752                 # e.g., introducing a title= parameter for the title; ignoring the un-
4753                 # named parameter entirely for images without a caption; adding an ex-
4754                 # plicit caption= parameter and preserving the old magic unnamed para-
4755                 # meter for BC; ...
4756                 if ( $imageIsFramed ) { # Framed image
4757                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4758                                 # No caption or alt text, add the filename as the alt text so
4759                                 # that screen readers at least get some description of the image
4760                                 $params['frame']['alt'] = $title->getText();
4761                         }
4762                         # Do not set $params['frame']['title'] because tooltips don't make sense
4763                         # for framed images
4764                 } else { # Inline image
4765                         if ( !isset( $params['frame']['alt'] ) ) {
4766                                 # No alt text, use the "caption" for the alt text
4767                                 if ( $caption !== '') {
4768                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4769                                 } else {
4770                                         # No caption, fall back to using the filename for the
4771                                         # alt text
4772                                         $params['frame']['alt'] = $title->getText();
4773                                 }
4774                         }
4775                         # Use the "caption" for the tooltip text
4776                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4777                 }
4778
4779                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4780
4781                 # Linker does the rest
4782                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4783
4784                 # Give the handler a chance to modify the parser object
4785                 if ( $handler ) {
4786                         $handler->parserTransformHook( $this, $file );
4787                 }
4788
4789                 return $ret;
4790         }
4791
4792         protected function stripAltText( $caption, $holders ) {
4793                 # Strip bad stuff out of the title (tooltip).  We can't just use
4794                 # replaceLinkHoldersText() here, because if this function is called
4795                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4796                 if ( $holders ) {
4797                         $tooltip = $holders->replaceText( $caption );
4798                 } else {
4799                         $tooltip = $this->replaceLinkHoldersText( $caption );
4800                 }
4801
4802                 # make sure there are no placeholders in thumbnail attributes
4803                 # that are later expanded to html- so expand them now and
4804                 # remove the tags
4805                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4806                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4807
4808                 return $tooltip;
4809         }
4810
4811         /**
4812          * Set a flag in the output object indicating that the content is dynamic and
4813          * shouldn't be cached.
4814          */
4815         function disableCache() {
4816                 wfDebug( "Parser output marked as uncacheable.\n" );
4817                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
4818                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
4819         }
4820
4821         /**
4822          * Callback from the Sanitizer for expanding items found in HTML attribute
4823          * values, so they can be safely tested and escaped.
4824          *
4825          * @param $text String
4826          * @param $frame PPFrame
4827          * @return String
4828          * @private
4829          */
4830         function attributeStripCallback( &$text, $frame = false ) {
4831                 $text = $this->replaceVariables( $text, $frame );
4832                 $text = $this->mStripState->unstripBoth( $text );
4833                 return $text;
4834         }
4835
4836         /**
4837          * Accessor
4838          */
4839         function getTags() {
4840                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4841         }
4842
4843         /**
4844          * Break wikitext input into sections, and either pull or replace
4845          * some particular section's text.
4846          *
4847          * External callers should use the getSection and replaceSection methods.
4848          *
4849          * @param $text String: Page wikitext
4850          * @param $section String: a section identifier string of the form:
4851          *   <flag1> - <flag2> - ... - <section number>
4852          *
4853          * Currently the only recognised flag is "T", which means the target section number
4854          * was derived during a template inclusion parse, in other words this is a template
4855          * section edit link. If no flags are given, it was an ordinary section edit link.
4856          * This flag is required to avoid a section numbering mismatch when a section is
4857          * enclosed by <includeonly> (bug 6563).
4858          *
4859          * The section number 0 pulls the text before the first heading; other numbers will
4860          * pull the given section along with its lower-level subsections. If the section is
4861          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4862          *
4863          * @param $mode String: one of "get" or "replace"
4864          * @param $newText String: replacement text for section data.
4865          * @return String: for "get", the extracted section text.
4866          *                 for "replace", the whole page with the section replaced.
4867          */
4868         private function extractSections( $text, $section, $mode, $newText='' ) {
4869                 global $wgTitle;
4870                 $this->clearState();
4871                 $this->setTitle( $wgTitle ); # not generally used but removes an ugly failure mode
4872                 $this->mOptions = new ParserOptions;
4873                 $this->setOutputType( self::OT_PLAIN );
4874                 $outText = '';
4875                 $frame = $this->getPreprocessor()->newFrame();
4876
4877                 # Process section extraction flags
4878                 $flags = 0;
4879                 $sectionParts = explode( '-', $section );
4880                 $sectionIndex = array_pop( $sectionParts );
4881                 foreach ( $sectionParts as $part ) {
4882                         if ( $part === 'T' ) {
4883                                 $flags |= self::PTD_FOR_INCLUSION;
4884                         }
4885                 }
4886                 # Preprocess the text
4887                 $root = $this->preprocessToDom( $text, $flags );
4888
4889                 # <h> nodes indicate section breaks
4890                 # They can only occur at the top level, so we can find them by iterating the root's children
4891                 $node = $root->getFirstChild();
4892
4893                 # Find the target section
4894                 if ( $sectionIndex == 0 ) {
4895                         # Section zero doesn't nest, level=big
4896                         $targetLevel = 1000;
4897                 } else {
4898                         while ( $node ) {
4899                                 if ( $node->getName() === 'h' ) {
4900                                         $bits = $node->splitHeading();
4901                                         if ( $bits['i'] == $sectionIndex ) {
4902                                                 $targetLevel = $bits['level'];
4903                                                 break;
4904                                         }
4905                                 }
4906                                 if ( $mode === 'replace' ) {
4907                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4908                                 }
4909                                 $node = $node->getNextSibling();
4910                         }
4911                 }
4912
4913                 if ( !$node ) {
4914                         # Not found
4915                         if ( $mode === 'get' ) {
4916                                 return $newText;
4917                         } else {
4918                                 return $text;
4919                         }
4920                 }
4921
4922                 # Find the end of the section, including nested sections
4923                 do {
4924                         if ( $node->getName() === 'h' ) {
4925                                 $bits = $node->splitHeading();
4926                                 $curLevel = $bits['level'];
4927                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4928                                         break;
4929                                 }
4930                         }
4931                         if ( $mode === 'get' ) {
4932                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4933                         }
4934                         $node = $node->getNextSibling();
4935                 } while ( $node );
4936
4937                 # Write out the remainder (in replace mode only)
4938                 if ( $mode === 'replace' ) {
4939                         # Output the replacement text
4940                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4941                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4942                         # Only add trailing whitespace if there is newText
4943                         if ( $newText != "" ) {
4944                                 $outText .= $newText . "\n\n";
4945                         }
4946
4947                         while ( $node ) {
4948                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4949                                 $node = $node->getNextSibling();
4950                         }
4951                 }
4952
4953                 if ( is_string( $outText ) ) {
4954                         # Re-insert stripped tags
4955                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4956                 }
4957
4958                 return $outText;
4959         }
4960
4961         /**
4962          * This function returns the text of a section, specified by a number ($section).
4963          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4964          * the first section before any such heading (section 0).
4965          *
4966          * If a section contains subsections, these are also returned.
4967          *
4968          * @param $text String: text to look in
4969          * @param $section String: section identifier
4970          * @param $deftext String: default to return if section is not found
4971          * @return string text of the requested section
4972          */
4973         public function getSection( $text, $section, $deftext='' ) {
4974                 return $this->extractSections( $text, $section, "get", $deftext );
4975         }
4976
4977         public function replaceSection( $oldtext, $section, $text ) {
4978                 return $this->extractSections( $oldtext, $section, "replace", $text );
4979         }
4980
4981         /**
4982          * Get the ID of the revision we are parsing
4983          *
4984          * @return Mixed: integer or null
4985          */
4986         function getRevisionId() {
4987                 return $this->mRevisionId;
4988         }
4989
4990         /**
4991          * Get the timestamp associated with the current revision, adjusted for
4992          * the default server-local timestamp
4993          */
4994         function getRevisionTimestamp() {
4995                 if ( is_null( $this->mRevisionTimestamp ) ) {
4996                         wfProfileIn( __METHOD__ );
4997                         global $wgContLang;
4998                         $dbr = wfGetDB( DB_SLAVE );
4999                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
5000                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
5001
5002                         # Normalize timestamp to internal MW format for timezone processing.
5003                         # This has the added side-effect of replacing a null value with
5004                         # the current time, which gives us more sensible behavior for
5005                         # previews.
5006                         $timestamp = wfTimestamp( TS_MW, $timestamp );
5007
5008                         # The cryptic '' timezone parameter tells to use the site-default
5009                         # timezone offset instead of the user settings.
5010                         #
5011                         # Since this value will be saved into the parser cache, served
5012                         # to other users, and potentially even used inside links and such,
5013                         # it needs to be consistent for all visitors.
5014                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5015
5016                         wfProfileOut( __METHOD__ );
5017                 }
5018                 return $this->mRevisionTimestamp;
5019         }
5020
5021         /**
5022          * Get the name of the user that edited the last revision
5023          *
5024          * @return String: user name
5025          */
5026         function getRevisionUser() {
5027                 # if this template is subst: the revision id will be blank,
5028                 # so just use the current user's name
5029                 if ( $this->mRevisionId ) {
5030                         $revision = Revision::newFromId( $this->mRevisionId );
5031                         $revuser = $revision->getUserText();
5032                 } else {
5033                         global $wgUser;
5034                         $revuser = $wgUser->getName();
5035                 }
5036                 return $revuser;
5037         }
5038
5039         /**
5040          * Mutator for $mDefaultSort
5041          *
5042          * @param $sort New value
5043          */
5044         public function setDefaultSort( $sort ) {
5045                 $this->mDefaultSort = $sort;
5046                 $this->mOutput->setProperty( 'defaultsort', $sort );
5047         }
5048
5049         /**
5050          * Accessor for $mDefaultSort
5051          * Will use the title/prefixed title if none is set
5052          *
5053          * @return string
5054          */
5055         public function getDefaultSort() {
5056                 global $wgCategoryPrefixedDefaultSortkey;
5057                 if ( $this->mDefaultSort !== false ) {
5058                         return $this->mDefaultSort;
5059                 } else {
5060                         return $this->mTitle->getCategorySortkey();
5061                 }
5062         }
5063
5064         /**
5065          * Accessor for $mDefaultSort
5066          * Unlike getDefaultSort(), will return false if none is set
5067          *
5068          * @return string or false
5069          */
5070         public function getCustomDefaultSort() {
5071                 return $this->mDefaultSort;
5072         }
5073
5074         /**
5075          * Try to guess the section anchor name based on a wikitext fragment
5076          * presumably extracted from a heading, for example "Header" from
5077          * "== Header ==".
5078          */
5079         public function guessSectionNameFromWikiText( $text ) {
5080                 # Strip out wikitext links(they break the anchor)
5081                 $text = $this->stripSectionName( $text );
5082                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5083                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5084         }
5085
5086         /**
5087          * Strips a text string of wikitext for use in a section anchor
5088          *
5089          * Accepts a text string and then removes all wikitext from the
5090          * string and leaves only the resultant text (i.e. the result of
5091          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5092          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5093          * to create valid section anchors by mimicing the output of the
5094          * parser when headings are parsed.
5095          *
5096          * @param $text String: text string to be stripped of wikitext
5097          * for use in a Section anchor
5098          * @return Filtered text string
5099          */
5100         public function stripSectionName( $text ) {
5101                 # Strip internal link markup
5102                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5103                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5104
5105                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5106                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5107                 # on how many empty links there are on the page - need to figure that out.
5108                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5109
5110                 # Parse wikitext quotes (italics & bold)
5111                 $text = $this->doQuotes( $text );
5112
5113                 # Strip HTML tags
5114                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5115                 return $text;
5116         }
5117
5118         function srvus( $text ) {
5119                 return $this->testSrvus( $text, $this->mOutputType );
5120         }
5121
5122         /**
5123          * strip/replaceVariables/unstrip for preprocessor regression testing
5124          */
5125         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5126                 $this->clearState();
5127                 if ( !$title instanceof Title ) {
5128                         $title = Title::newFromText( $title );
5129                 }
5130                 $this->mTitle = $title;
5131                 $this->mOptions = $options;
5132                 $this->setOutputType( $outputType );
5133                 $text = $this->replaceVariables( $text );
5134                 $text = $this->mStripState->unstripBoth( $text );
5135                 $text = Sanitizer::removeHTMLtags( $text );
5136                 return $text;
5137         }
5138
5139         function testPst( $text, $title, $options ) {
5140                 global $wgUser;
5141                 if ( !$title instanceof Title ) {
5142                         $title = Title::newFromText( $title );
5143                 }
5144                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5145         }
5146
5147         function testPreprocess( $text, $title, $options ) {
5148                 if ( !$title instanceof Title ) {
5149                         $title = Title::newFromText( $title );
5150                 }
5151                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5152         }
5153
5154         function markerSkipCallback( $s, $callback ) {
5155                 $i = 0;
5156                 $out = '';
5157                 while ( $i < strlen( $s ) ) {
5158                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5159                         if ( $markerStart === false ) {
5160                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5161                                 break;
5162                         } else {
5163                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5164                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5165                                 if ( $markerEnd === false ) {
5166                                         $out .= substr( $s, $markerStart );
5167                                         break;
5168                                 } else {
5169                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5170                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5171                                         $i = $markerEnd;
5172                                 }
5173                         }
5174                 }
5175                 return $out;
5176         }
5177
5178         function serialiseHalfParsedText( $text ) {
5179                 $data = array();
5180                 $data['text'] = $text;
5181
5182                 # First, find all strip markers, and store their
5183                 #  data in an array.
5184                 $stripState = new StripState;
5185                 $pos = 0;
5186                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5187                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5188                 {
5189                         $end_pos += strlen( self::MARKER_SUFFIX );
5190                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5191
5192                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5193                                 $replaceArray = $stripState->general;
5194                                 $stripText = $this->mStripState->general->data[$marker];
5195                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5196                                 $replaceArray = $stripState->nowiki;
5197                                 $stripText = $this->mStripState->nowiki->data[$marker];
5198                         } else {
5199                                 throw new MWException( "Hanging strip marker: '$marker'." );
5200                         }
5201
5202                         $replaceArray->setPair( $marker, $stripText );
5203                         $pos = $end_pos;
5204                 }
5205                 $data['stripstate'] = $stripState;
5206
5207                 # Now, find all of our links, and store THEIR
5208                 #  data in an array! :)
5209                 $links = array( 'internal' => array(), 'interwiki' => array() );
5210                 $pos = 0;
5211
5212                 # Internal links
5213                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5214                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5215
5216                         $ns = trim( $ns );
5217                         if ( empty( $links['internal'][$ns] ) ) {
5218                                 $links['internal'][$ns] = array();
5219                         }
5220
5221                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5222                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5223                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5224                 }
5225
5226                 $pos = 0;
5227
5228                 # Interwiki links
5229                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5230                         $data = substr( $text, $start_pos );
5231                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5232                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5233                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5234                 }
5235
5236                 $data['linkholder'] = $links;
5237
5238                 return $data;
5239         }
5240
5241         /**
5242          * TODO: document
5243          * @param $data Array
5244          * @param $intPrefix String unique identifying prefix
5245          * @return String
5246          */
5247         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5248                 if ( !$intPrefix ) {
5249                         $intPrefix = $this->getRandomString();
5250                 }
5251
5252                 # First, extract the strip state.
5253                 $stripState = $data['stripstate'];
5254                 $this->mStripState->general->merge( $stripState->general );
5255                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5256
5257                 # Now, extract the text, and renumber links
5258                 $text = $data['text'];
5259                 $links = $data['linkholder'];
5260
5261                 # Internal...
5262                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5263                         foreach ( $nsLinks as $key => $entry ) {
5264                                 $newKey = $intPrefix . '-' . $key;
5265                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5266
5267                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5268                         }
5269                 }
5270
5271                 # Interwiki...
5272                 foreach ( $links['interwiki'] as $key => $entry ) {
5273                         $newKey = "$intPrefix-$key";
5274                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5275
5276                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5277                 }
5278
5279                 # Should be good to go.
5280                 return $text;
5281         }
5282 }
5283
5284 /**
5285  * @todo document, briefly.
5286  * @ingroup Parser
5287  */
5288 class StripState {
5289         var $general, $nowiki;
5290
5291         function __construct() {
5292                 $this->general = new ReplacementArray;
5293                 $this->nowiki = new ReplacementArray;
5294         }
5295
5296         function unstripGeneral( $text ) {
5297                 wfProfileIn( __METHOD__ );
5298                 do {
5299                         $oldText = $text;
5300                         $text = $this->general->replace( $text );
5301                 } while ( $text !== $oldText );
5302                 wfProfileOut( __METHOD__ );
5303                 return $text;
5304         }
5305
5306         function unstripNoWiki( $text ) {
5307                 wfProfileIn( __METHOD__ );
5308                 do {
5309                         $oldText = $text;
5310                         $text = $this->nowiki->replace( $text );
5311                 } while ( $text !== $oldText );
5312                 wfProfileOut( __METHOD__ );
5313                 return $text;
5314         }
5315
5316         function unstripBoth( $text ) {
5317                 wfProfileIn( __METHOD__ );
5318                 do {
5319                         $oldText = $text;
5320                         $text = $this->general->replace( $text );
5321                         $text = $this->nowiki->replace( $text );
5322                 } while ( $text !== $oldText );
5323                 wfProfileOut( __METHOD__ );
5324                 return $text;
5325         }
5326 }
5327
5328 /**
5329  * @todo document, briefly.
5330  * @ingroup Parser
5331  */
5332 class OnlyIncludeReplacer {
5333         var $output = '';
5334
5335         function replace( $matches ) {
5336                 if ( substr( $matches[1], -1 ) === "\n" ) {
5337                         $this->output .= substr( $matches[1], 0, -1 );
5338                 } else {
5339                         $this->output .= $matches[1];
5340                 }
5341         }
5342 }