includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *     produces HTML output
  21  * preSaveTransform().
  22  *     produces altered wiki markup.
  23  * preprocess()
  24  *     removes HTML comments and expands templates
  25  * cleanSig()
  26  *     Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *     Extracts sections from an article for section editing
  29  * getPreloadText()
  30  *     Removes <noinclude> sections, and <includeonly> tags.
  31  *
  32  * Globals used:
  33  *    objects:   $wgLang, $wgContLang
  34  *
  35  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  36  *
  37  * settings:
  38  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  39  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  40  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  41  *  $wgMaxArticleSize*
  42  *
  43  *  * only within ParserOptions
  44  * </pre>
  45  *
  46  * @ingroup Parser
  47  */
  48 class Parser {
  49         /**
  50          * Update this version number when the ParserOutput format
  51          * changes in an incompatible way, so the parser cache
  52          * can automatically discard old data.
  53          */
  54         const VERSION = '1.6.4';
  55
  56         # Flags for Parser::setFunctionHook
  57         # Also available as global constants from Defines.php
  58         const SFH_NO_HASH = 1;
  59         const SFH_OBJECT_ARGS = 2;
  60
  61         # Constants needed for external link processing
  62         # Everything except bracket, space, or control characters
  63         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  64         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  65                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  66
  67         # State constants for the definition list colon extraction
  68         const COLON_STATE_TEXT = 0;
  69         const COLON_STATE_TAG = 1;
  70         const COLON_STATE_TAGSTART = 2;
  71         const COLON_STATE_CLOSETAG = 3;
  72         const COLON_STATE_TAGSLASH = 4;
  73         const COLON_STATE_COMMENT = 5;
  74         const COLON_STATE_COMMENTDASH = 6;
  75         const COLON_STATE_COMMENTDASHDASH = 7;
  76
  77         # Flags for preprocessToDom
  78         const PTD_FOR_INCLUSION = 1;
  79
  80         # Allowed values for $this->mOutputType
  81         # Parameter to startExternalParse().
  82         const OT_HTML = 1; # like parse()
  83         const OT_WIKI = 2; # like preSaveTransform()
  84         const OT_PREPROCESS = 3; # like preprocess()
  85         const OT_MSG = 3;
  86         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  87
  88         # Marker Suffix needs to be accessible staticly.
  89         const MARKER_SUFFIX = "-QINU\x7f";
  90
  91         # Persistent:
  92         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  93         var $mSubstWords, $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex;
  94         var $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList;
  95         var $mVarCache, $mConf, $mFunctionTagHooks;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mLinkHolders, $mLinkID;
 102         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 103         var $mTplExpandCache; # empty-frame expansion cache
 104         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 105         var $mExpensiveFunctionCount; # number of expensive parser function calls
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions;      # ParserOptions object
 110         var $mTitle;        # Title context, used for self-link rendering and similar things
 111         var $mOutputType;   # Output type, one of the OT_xxx constants
 112         var $ot;            # Shortcut alias, see setOutputType()
 113         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 114         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 115         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 116
 117         /**
 118          * Constructor
 119          *
 120          * @public
 121          */
 122         function __construct( $conf = array() ) {
 123                 $this->mConf = $conf;
 124                 $this->mTagHooks = array();
 125                 $this->mTransparentTagHooks = array();
 126                 $this->mFunctionHooks = array();
 127                 $this->mFunctionTagHooks = array();
 128                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 129                 $this->mDefaultStripList = $this->mStripList = array();
 130                 $this->mUrlProtocols = wfUrlProtocols();
 131                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 132                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 133                 $this->mVarCache = array();
 134                 if ( isset( $conf['preprocessorClass'] ) ) {
 135                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 136                 } elseif ( extension_loaded( 'domxml' ) ) {
 137                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 138                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 139                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 140                 } elseif ( extension_loaded( 'dom' ) ) {
 141                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 142                 } else {
 143                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 144                 }
 145                 $this->mMarkerIndex = 0;
 146                 $this->mFirstCall = true;
 147         }
 148
 149         /**
 150          * Reduce memory usage to reduce the impact of circular references
 151          */
 152         function __destruct() {
 153                 if ( isset( $this->mLinkHolders ) ) {
 154                         $this->mLinkHolders->__destruct();
 155                 }
 156                 foreach ( $this as $name => $value ) {
 157                         unset( $this->$name );
 158                 }
 159         }
 160
 161         /**
 162          * Do various kinds of initialisation on the first call of the parser
 163          */
 164         function firstCallInit() {
 165                 if ( !$this->mFirstCall ) {
 166                         return;
 167                 }
 168                 $this->mFirstCall = false;
 169
 170                 wfProfileIn( __METHOD__ );
 171
 172                 CoreParserFunctions::register( $this );
 173                 CoreTagHooks::register( $this );
 174                 $this->initialiseVariables();
 175
 176                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 177                 wfProfileOut( __METHOD__ );
 178         }
 179
 180         /**
 181          * Clear Parser state
 182          *
 183          * @private
 184          */
 185         function clearState() {
 186                 wfProfileIn( __METHOD__ );
 187                 if ( $this->mFirstCall ) {
 188                         $this->firstCallInit();
 189                 }
 190                 $this->mOutput = new ParserOutput;
 191                 $this->mAutonumber = 0;
 192                 $this->mLastSection = '';
 193                 $this->mDTopen = false;
 194                 $this->mIncludeCount = array();
 195                 $this->mStripState = new StripState;
 196                 $this->mArgStack = false;
 197                 $this->mInPre = false;
 198                 $this->mLinkHolders = new LinkHolderArray( $this );
 199                 $this->mLinkID = 0;
 200                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 201                 $this->mVarCache = array();
 202
 203                 /**
 204                  * Prefix for temporary replacement strings for the multipass parser.
 205                  * \x07 should never appear in input as it's disallowed in XML.
 206                  * Using it at the front also gives us a little extra robustness
 207                  * since it shouldn't match when butted up against identifier-like
 208                  * string constructs.
 209                  *
 210                  * Must not consist of all title characters, or else it will change
 211                  * the behaviour of <nowiki> in a link.
 212                  */
 213                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 214                 # Changed to \x7f to allow XML double-parsing -- TS
 215                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 216
 217
 218                 # Clear these on every parse, bug 4549
 219                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 220
 221                 $this->mShowToc = true;
 222                 $this->mForceTocPosition = false;
 223                 $this->mIncludeSizes = array(
 224                         'post-expand' => 0,
 225                         'arg' => 0,
 226                 );
 227                 $this->mPPNodeCount = 0;
 228                 $this->mDefaultSort = false;
 229                 $this->mHeadings = array();
 230                 $this->mDoubleUnderscores = array();
 231                 $this->mExpensiveFunctionCount = 0;
 232
 233                 # Fix cloning
 234                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 235                         $this->mPreprocessor = null;
 236                 }
 237
 238                 wfRunHooks( 'ParserClearState', array( &$this ) );
 239                 wfProfileOut( __METHOD__ );
 240         }
 241
 242         /**
 243          * Convert wikitext to HTML
 244          * Do not call this function recursively.
 245          *
 246          * @param $text String: text we want to parse
 247          * @param $title A title object
 248          * @param $options ParserOptions
 249          * @param $linestart boolean
 250          * @param $clearState boolean
 251          * @param $revid Int: number to pass in {{REVISIONID}}
 252          * @return ParserOutput a ParserOutput
 253          */
 254         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 255                 /**
 256                  * First pass--just handle <nowiki> sections, pass the rest off
 257                  * to internalParse() which does all the real work.
 258                  */
 259
 260                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 261                 $fname = __METHOD__.'-' . wfGetCaller();
 262                 wfProfileIn( __METHOD__ );
 263                 wfProfileIn( $fname );
 264
 265                 if ( $clearState ) {
 266                         $this->clearState();
 267                 }
 268
 269                 $options->resetUsage();
 270                 $this->mOptions = $options;
 271                 $this->setTitle( $title ); # Page title has to be set for the pre-processor
 272
 273                 $oldRevisionId = $this->mRevisionId;
 274                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 275                 if ( $revid !== null ) {
 276                         $this->mRevisionId = $revid;
 277                         $this->mRevisionTimestamp = null;
 278                 }
 279                 $this->setOutputType( self::OT_HTML );
 280                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 281                 # No more strip!
 282                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 283                 $text = $this->internalParse( $text );
 284
 285                 $text = $this->mStripState->unstripGeneral( $text );
 286
 287                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 288                 $fixtags = array(
 289                         # french spaces, last one Guillemet-left
 290                         # only if there is something before the space
 291                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 292                         # french spaces, Guillemet-right
 293                         '/(\\302\\253) /' => '\\1&#160;',
 294                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 295                 );
 296                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 297
 298                 $text = $this->doBlockLevels( $text, $linestart );
 299
 300                 $this->replaceLinkHolders( $text );
 301
 302                 /**
 303                  * The page doesn't get language converted if
 304                  * a) It's disabled
 305                  * b) Content isn't converted
 306                  * c) It's a conversion table
 307                  */
 308                 if ( !( $wgDisableLangConversion
 309                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 310                                 || $this->mTitle->isConversionTable() ) ) {
 311
 312                         # The position of the convert() call should not be changed. it
 313                         # assumes that the links are all replaced and the only thing left
 314                         # is the <nowiki> mark.
 315
 316                         $text = $wgContLang->convert( $text );
 317                 }
 318
 319                 /**
 320                  * A page get its title converted except:
 321                  * a) Language conversion is globally disabled
 322                  * b) Title convert is globally disabled
 323                  * c) The page is a redirect page
 324                  * d) User request with a "linkconvert" set to "no"
 325                  * e) A "nocontentconvert" magic word has been set
 326                  * f) A "notitleconvert" magic word has been set
 327                  * g) User sets "noconvertlink" in his/her preference
 328                  *
 329                  * Note that if a user tries to set a title in a conversion
 330                  * rule but content conversion was not done, then the parser
 331                  * won't pick it up.  This is probably expected behavior.
 332                  */
 333                 if ( !( $wgDisableLangConversion
 334                                 || $wgDisableTitleConversion
 335                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 336                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 337                                 || $this->mOutput->getDisplayTitle() !== false ) )
 338                 {
 339                         $convruletitle = $wgContLang->getConvRuleTitle();
 340                         if ( $convruletitle ) {
 341                                 $this->mOutput->setTitleText( $convruletitle );
 342                         } else {
 343                                 $titleText = $wgContLang->convertTitle( $title );
 344                                 $this->mOutput->setTitleText( $titleText );
 345                         }
 346                 }
 347
 348                 $text = $this->mStripState->unstripNoWiki( $text );
 349
 350                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 351
 352 //!JF Move to its own function
 353
 354                 $uniq_prefix = $this->mUniqPrefix;
 355                 $matches = array();
 356                 $elements = array_keys( $this->mTransparentTagHooks );
 357                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 358
 359                 foreach ( $matches as $marker => $data ) {
 360                         list( $element, $content, $params, $tag ) = $data;
 361                         $tagName = strtolower( $element );
 362                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 363                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 364                         } else {
 365                                 $output = $tag;
 366                         }
 367                         $this->mStripState->general->setPair( $marker, $output );
 368                 }
 369                 $text = $this->mStripState->unstripGeneral( $text );
 370
 371                 $text = Sanitizer::normalizeCharReferences( $text );
 372
 373                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 374                         $text = MWTidy::tidy( $text );
 375                 } else {
 376                         # attempt to sanitize at least some nesting problems
 377                         # (bug #2702 and quite a few others)
 378                         $tidyregs = array(
 379                                 # ''Something [http://www.cool.com cool''] -->
 380                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 381                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 382                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 383                                 # fix up an anchor inside another anchor, only
 384                                 # at least for a single single nested link (bug 3695)
 385                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 386                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 387                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 388                                 # contains a div, so fix it up here; replace
 389                                 # div with escaped text
 390                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 391                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 392                                 # remove empty italic or bold tag pairs, some
 393                                 # introduced by rules above
 394                                 '/<([bi])><\/\\1>/' => '',
 395                         );
 396
 397                         $text = preg_replace(
 398                                 array_keys( $tidyregs ),
 399                                 array_values( $tidyregs ),
 400                                 $text );
 401                 }
 402                 global $wgExpensiveParserFunctionLimit;
 403                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 404                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 405                 }
 406
 407                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 408
 409                 # Information on include size limits, for the benefit of users who try to skirt them
 410                 if ( $this->mOptions->getEnableLimitReport() ) {
 411                         $max = $this->mOptions->getMaxIncludeSize();
 412                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 413                         $limitReport =
 414                                 "NewPP limit report\n" .
 415                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 416                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 417                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 418                                 $PFreport;
 419                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 420                         $text .= "\n<!-- \n$limitReport-->\n";
 421                 }
 422                 $this->mOutput->setText( $text );
 423
 424                 $this->mRevisionId = $oldRevisionId;
 425                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 426                 wfProfileOut( $fname );
 427                 wfProfileOut( __METHOD__ );
 428
 429                 return $this->mOutput;
 430         }
 431
 432         /**
 433          * Recursive parser entry point that can be called from an extension tag
 434          * hook.
 435          *
 436          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 437          *
 438          * @param $text String: text extension wants to have parsed
 439          * @param $frame PPFrame: The frame to use for expanding any template variables
 440          */
 441         function recursiveTagParse( $text, $frame=false ) {
 442                 wfProfileIn( __METHOD__ );
 443                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 444                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 445                 $text = $this->internalParse( $text, false, $frame );
 446                 wfProfileOut( __METHOD__ );
 447                 return $text;
 448         }
 449
 450         /**
 451          * Expand templates and variables in the text, producing valid, static wikitext.
 452          * Also removes comments.
 453          */
 454         function preprocess( $text, $title, $options, $revid = null ) {
 455                 wfProfileIn( __METHOD__ );
 456                 $this->clearState();
 457                 $this->setOutputType( self::OT_PREPROCESS );
 458                 $options->resetUsage();
 459                 $this->mOptions = $options;
 460                 $this->setTitle( $title );
 461                 if ( $revid !== null ) {
 462                         $this->mRevisionId = $revid;
 463                 }
 464                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 465                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 466                 $text = $this->replaceVariables( $text );
 467                 $text = $this->mStripState->unstripBoth( $text );
 468                 wfProfileOut( __METHOD__ );
 469                 return $text;
 470         }
 471
 472         /**
 473          * Process the wikitext for the ?preload= feature. (bug 5210)
 474          *
 475          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 476          * comments, templates, arguments, tags hooks and parser functions are untouched.
 477          */
 478         public function getPreloadText( $text, $title, $options ) {
 479                 # Parser (re)initialisation
 480                 $this->clearState();
 481                 $this->setOutputType( self::OT_PLAIN );
 482                 $options->resetUsage();
 483                 $this->mOptions = $options;
 484                 $this->setTitle( $title );
 485
 486                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 487                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 488                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 489         }
 490
 491         /**
 492          * Get a random string
 493          *
 494          * @private
 495          * @static
 496          */
 497         static function getRandomString() {
 498                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 499         }
 500
 501         /**
 502          * Accessor for mUniqPrefix.
 503          *
 504          * @return String
 505          */
 506         public function uniqPrefix() {
 507                 if ( !isset( $this->mUniqPrefix ) ) {
 508                         # @todo Fixme: this is probably *horribly wrong*
 509                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 510                         # if this is called for a parser cache hit, the parser may not
 511                         # have ever been initialized in the first place.
 512                         # Not really sure what the heck is supposed to be going on here.
 513                         return '';
 514                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 515                 }
 516                 return $this->mUniqPrefix;
 517         }
 518
 519         /**
 520          * Set the context title
 521          */
 522         function setTitle( $t ) {
 523                 if ( !$t || $t instanceof FakeTitle ) {
 524                         $t = Title::newFromText( 'NO TITLE' );
 525                 }
 526
 527                 if ( strval( $t->getFragment() ) !== '' ) {
 528                         # Strip the fragment to avoid various odd effects
 529                         $this->mTitle = clone $t;
 530                         $this->mTitle->setFragment( '' );
 531                 } else {
 532                         $this->mTitle = $t;
 533                 }
 534         }
 535
 536         /**
 537          * Accessor for the Title object
 538          *
 539          * @return Title object
 540          */
 541         function &getTitle() {
 542                 return $this->mTitle;
 543         }
 544
 545         /**
 546          * Accessor/mutator for the Title object
 547          *
 548          * @param $x New Title object or null to just get the current one
 549          * @return Title object
 550          */
 551         function Title( $x = null ) {
 552                 return wfSetVar( $this->mTitle, $x );
 553         }
 554
 555         /**
 556          * Set the output type
 557          *
 558          * @param $ot Integer: new value
 559          */
 560         function setOutputType( $ot ) {
 561                 $this->mOutputType = $ot;
 562                 # Shortcut alias
 563                 $this->ot = array(
 564                         'html' => $ot == self::OT_HTML,
 565                         'wiki' => $ot == self::OT_WIKI,
 566                         'pre' => $ot == self::OT_PREPROCESS,
 567                         'plain' => $ot == self::OT_PLAIN,
 568                 );
 569         }
 570
 571         /**
 572          * Accessor/mutator for the output type
 573          *
 574          * @param $x New value or null to just get the current one
 575          * @return Integer
 576          */
 577         function OutputType( $x = null ) {
 578                 return wfSetVar( $this->mOutputType, $x );
 579         }
 580
 581         /**
 582          * Get the ParserOutput object
 583          *
 584          * @return ParserOutput object
 585          */
 586         function getOutput() {
 587                 return $this->mOutput;
 588         }
 589
 590         /**
 591          * Get the ParserOptions object
 592          *
 593          * @return ParserOptions object
 594          */
 595         function getOptions() {
 596                 return $this->mOptions;
 597         }
 598
 599         /**
 600          * Accessor/mutator for the ParserOptions object
 601          *
 602          * @param $x New value or null to just get the current one
 603          * @return Current ParserOptions object
 604          */
 605         function Options( $x = null ) {
 606                 return wfSetVar( $this->mOptions, $x );
 607         }
 608
 609         function nextLinkID() {
 610                 return $this->mLinkID++;
 611         }
 612
 613         function getFunctionLang() {
 614                 global $wgLang, $wgContLang;
 615
 616                 $target = $this->mOptions->getTargetLanguage();
 617                 if ( $target !== null ) {
 618                         return $target;
 619                 } else {
 620                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 621                 }
 622         }
 623
 624         /**
 625          * Get a preprocessor object
 626          *
 627          * @return Preprocessor instance
 628          */
 629         function getPreprocessor() {
 630                 if ( !isset( $this->mPreprocessor ) ) {
 631                         $class = $this->mPreprocessorClass;
 632                         $this->mPreprocessor = new $class( $this );
 633                 }
 634                 return $this->mPreprocessor;
 635         }
 636
 637         /**
 638          * Replaces all occurrences of HTML-style comments and the given tags
 639          * in the text with a random marker and returns the next text. The output
 640          * parameter $matches will be an associative array filled with data in
 641          * the form:
 642          *   'UNIQ-xxxxx' => array(
 643          *     'element',
 644          *     'tag content',
 645          *     array( 'param' => 'x' ),
 646          *     '<element param="x">tag content</element>' ) )
 647          *
 648          * @param $elements list of element names. Comments are always extracted.
 649          * @param $text Source text string.
 650          * @param $matches Out parameter, Array: extarcted tags
 651          * @param $uniq_prefix
 652          * @return String: stripped text
 653          *
 654          * @static
 655          */
 656         public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 657                 static $n = 1;
 658                 $stripped = '';
 659                 $matches = array();
 660
 661                 $taglist = implode( '|', $elements );
 662                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 663
 664                 while ( $text != '' ) {
 665                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 666                         $stripped .= $p[0];
 667                         if ( count( $p ) < 5 ) {
 668                                 break;
 669                         }
 670                         if ( count( $p ) > 5 ) {
 671                                 # comment
 672                                 $element    = $p[4];
 673                                 $attributes = '';
 674                                 $close      = '';
 675                                 $inside     = $p[5];
 676                         } else {
 677                                 # tag
 678                                 $element    = $p[1];
 679                                 $attributes = $p[2];
 680                                 $close      = $p[3];
 681                                 $inside     = $p[4];
 682                         }
 683
 684                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 685                         $stripped .= $marker;
 686
 687                         if ( $close === '/>' ) {
 688                                 # Empty element tag, <tag />
 689                                 $content = null;
 690                                 $text = $inside;
 691                                 $tail = null;
 692                         } else {
 693                                 if ( $element === '!--' ) {
 694                                         $end = '/(-->)/';
 695                                 } else {
 696                                         $end = "/(<\\/$element\\s*>)/i";
 697                                 }
 698                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 699                                 $content = $q[0];
 700                                 if ( count( $q ) < 3 ) {
 701                                         # No end tag -- let it run out to the end of the text.
 702                                         $tail = '';
 703                                         $text = '';
 704                                 } else {
 705                                         $tail = $q[1];
 706                                         $text = $q[2];
 707                                 }
 708                         }
 709
 710                         $matches[$marker] = array( $element,
 711                                 $content,
 712                                 Sanitizer::decodeTagAttributes( $attributes ),
 713                                 "<$element$attributes$close$content$tail" );
 714                 }
 715                 return $stripped;
 716         }
 717
 718         /**
 719          * Get a list of strippable XML-like elements
 720          */
 721         function getStripList() {
 722                 return $this->mStripList;
 723         }
 724
 725         /**
 726          * @deprecated use replaceVariables
 727          */
 728         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 729                 return $text;
 730         }
 731
 732         /**
 733          * Restores pre, math, and other extensions removed by strip()
 734          *
 735          * always call unstripNoWiki() after this one
 736          * @private
 737          * @deprecated use $this->mStripState->unstrip()
 738          */
 739         function unstrip( $text, $state ) {
 740                 return $state->unstripGeneral( $text );
 741         }
 742
 743         /**
 744          * Always call this after unstrip() to preserve the order
 745          *
 746          * @private
 747          * @deprecated use $this->mStripState->unstrip()
 748          */
 749         function unstripNoWiki( $text, $state ) {
 750                 return $state->unstripNoWiki( $text );
 751         }
 752
 753         /**
 754          * @deprecated use $this->mStripState->unstripBoth()
 755          */
 756         function unstripForHTML( $text ) {
 757                 return $this->mStripState->unstripBoth( $text );
 758         }
 759
 760         /**
 761          * Add an item to the strip state
 762          * Returns the unique tag which must be inserted into the stripped text
 763          * The tag will be replaced with the original text in unstrip()
 764          *
 765          * @private
 766          */
 767         function insertStripItem( $text ) {
 768                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 769                 $this->mMarkerIndex++;
 770                 $this->mStripState->general->setPair( $rnd, $text );
 771                 return $rnd;
 772         }
 773
 774         /**
 775          * Interface with html tidy
 776          * @deprecated Use MWTidy::tidy()
 777          */
 778         public static function tidy( $text ) {
 779                 wfDeprecated( __METHOD__ );
 780                 return MWTidy::tidy( $text );
 781         }
 782
 783         /**
 784          * parse the wiki syntax used to render tables
 785          *
 786          * @private
 787          */
 788         function doTableStuff( $text ) {
 789                 wfProfileIn( __METHOD__ );
 790
 791                 $lines = StringUtils::explode( "\n", $text );
 792                 $text = null;
 793                 $out = '';
 794                 $td_history = array(); # Is currently a td tag open?
 795                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 796                 $tr_history = array(); # Is currently a tr tag open?
 797                 $tr_attributes = array(); # history of tr attributes
 798                 $has_opened_tr = array(); # Did this table open a <tr> element?
 799                 $indent_level = 0; # indent level of the table
 800
 801                 $table_tag = 'table';
 802                 $tr_tag = 'tr';
 803                 $th_tag = 'th';
 804                 $td_tag = 'td';
 805                 $caption_tag = 'caption';
 806
 807                 $extra_table_attribs = null;
 808                 $extra_tr_attribs = null;
 809                 $extra_td_attribs = null;
 810
 811                 $convert_style = false;
 812
 813                 foreach ( $lines as $outLine ) {
 814                         $line = trim( $outLine );
 815
 816                         if ( $line === '' ) { # empty line, go to next line
 817                                 $out .= $outLine."\n";
 818                                 continue;
 819                         }
 820
 821                         $first_character = $line[0];
 822                         $matches = array();
 823
 824                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 825                                 # First check if we are starting a new table
 826                                 $indent_level = strlen( $matches[1] );
 827
 828                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 829
 830                                 $attr = Sanitizer::decodeTagAttributes( $attributes );
 831
 832                                 $mode = @$attr['mode'];
 833                                 if ( !$mode ) $mode = 'data';
 834
 835                                 if ( $mode == 'grid' || $mode == 'layout' ) {
 836                                         $table_tag = 'div';
 837                                         $tr_tag = 'div';
 838                                         $th_tag = 'div';
 839                                         $td_tag = 'div';
 840                                         $caption_tag = 'div';
 841
 842                                         $extra_table_attribs = array( 'class' => 'grid-table' );
 843                                         $extra_tr_attribs = array( 'class' => 'grid-row' );
 844                                         $extra_td_attribs = array( 'class' => 'grid-cell' );
 845
 846                                         $convert_style = true;
 847                                 }
 848
 849                                 if ($convert_style) $attr['style'] = Sanitizer::styleFromAttributes( $attr );
 850                                 $attr = Sanitizer::validateTagAttributes( $attr, $table_tag );
 851                                 $attributes = Sanitizer::collapseTagAttributes( $attr, $extra_table_attribs );
 852
 853                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<$table_tag{$attributes}>";
 854                                 array_push( $td_history , false );
 855                                 array_push( $last_tag_history , '' );
 856                                 array_push( $tr_history , false );
 857                                 array_push( $tr_attributes , '' );
 858                                 array_push( $has_opened_tr , false );
 859                         } elseif ( count( $td_history ) == 0 ) {
 860                                 # Don't do any of the following
 861                                 $out .= $outLine."\n";
 862                                 continue;
 863                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 864                                 # We are ending a table
 865                                 $line = "</$table_tag>" . substr( $line , 2 );
 866                                 $last_tag = array_pop( $last_tag_history );
 867
 868                                 if ( !array_pop( $has_opened_tr ) ) {
 869                                         $line = "<$tr_tag><$td_tag></$td_tag></$tr_tag>{$line}";
 870                                 }
 871
 872                                 if ( array_pop( $tr_history ) ) {
 873                                         $line = "</$tr_tag>{$line}";
 874                                 }
 875
 876                                 if ( array_pop( $td_history ) ) {
 877                                         $line = "</{$last_tag}>{$line}";
 878                                 }
 879                                 array_pop( $tr_attributes );
 880                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 881                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 882                                 # Now we have a table row
 883                                 $line = preg_replace( '#^\|-+#', '', $line );
 884
 885                                 # Whats after the tag is now only attributes
 886                                 $attributes = $this->mStripState->unstripBoth( $line );
 887
 888                                 $attr = Sanitizer::decodeTagAttributes( $attributes );
 889                                 if ($convert_style) $attr['style'] = Sanitizer::styleFromAttributes( $attr );
 890                                 $attr = Sanitizer::validateTagAttributes( $attr, $tr_tag );
 891                                 $attributes = Sanitizer::collapseTagAttributes( $attr, $extra_tr_attribs );
 892
 893                                 array_pop( $tr_attributes );
 894                                 array_push( $tr_attributes, $attributes );
 895
 896                                 $line = '';
 897                                 $last_tag = array_pop( $last_tag_history );
 898                                 array_pop( $has_opened_tr );
 899                                 array_push( $has_opened_tr , true );
 900
 901                                 if ( array_pop( $tr_history ) ) {
 902                                         $line = "</$tr_tag>";
 903                                 }
 904
 905                                 if ( array_pop( $td_history ) ) {
 906                                         $line = "</{$last_tag}>{$line}";
 907                                 }
 908
 909                                 $outLine = $line;
 910                                 array_push( $tr_history , false );
 911                                 array_push( $td_history , false );
 912                                 array_push( $last_tag_history , '' );
 913                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 914                                 # This might be cell elements, td, th or captions
 915                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 916                                         $first_character = '+';
 917                                         $line = substr( $line , 1 );
 918                                 }
 919
 920                                 $line = substr( $line , 1 );
 921
 922                                 if ( $first_character === '!' ) {
 923                                         $line = str_replace( '!!' , '||' , $line );
 924                                 }
 925
 926                                 # Split up multiple cells on the same line.
 927                                 # FIXME : This can result in improper nesting of tags processed
 928                                 # by earlier parser steps, but should avoid splitting up eg
 929                                 # attribute values containing literal "||".
 930                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 931
 932                                 $outLine = '';
 933
 934                                 # Loop through each table cell
 935                                 foreach ( $cells as $cell ) {
 936                                         $previous = '';
 937                                         if ( $first_character !== '+' ) {
 938                                                 $tr_after = array_pop( $tr_attributes );
 939                                                 if ( !array_pop( $tr_history ) ) {
 940                                                         $previous = "<$tr_tag{$tr_after}>\n";
 941                                                 }
 942                                                 array_push( $tr_history , true );
 943                                                 array_push( $tr_attributes , '' );
 944                                                 array_pop( $has_opened_tr );
 945                                                 array_push( $has_opened_tr , true );
 946                                         }
 947
 948                                         $last_tag = array_pop( $last_tag_history );
 949
 950                                         if ( array_pop( $td_history ) ) {
 951                                                 $previous = "</{$last_tag}>\n{$previous}";
 952                                         }
 953
 954                                         if ( $first_character === '|' ) {
 955                                                 $last_tag = $td_tag;
 956                                         } elseif ( $first_character === '!' ) {
 957                                                 $last_tag = $th_tag;
 958                                         } elseif ( $first_character === '+' ) {
 959                                                 $last_tag = $caption_tag;
 960                                         } else {
 961                                                 $last_tag = '';
 962                                         }
 963
 964                                         array_push( $last_tag_history , $last_tag );
 965
 966                                         # A cell could contain both parameters and data
 967                                         $cell_data = explode( '|' , $cell , 2 );
 968
 969                                         $attributes = '';
 970
 971                                         # Bug 553: Note that a '|' inside an invalid link should not
 972                                         # be mistaken as delimiting cell parameters
 973                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 974                                                 if ($extra_td_attribs) $attributes = Sanitizer::collapseTagAttributes( $extra_td_attribs );
 975                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell}";
 976                                         } elseif ( count( $cell_data ) == 1 ) {
 977                                                 if ($extra_td_attribs) $attributes = Sanitizer::collapseTagAttributes( $extra_td_attribs );
 978                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[0]}";
 979                                         } else {
 980                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 981
 982                                                 $attr = Sanitizer::decodeTagAttributes( $attributes );
 983                                                 if ($convert_style) $attr['style'] = Sanitizer::styleFromAttributes( $attr );
 984                                                 $attr = Sanitizer::validateTagAttributes( $attr, $last_tag );
 985                                                 $attributes = Sanitizer::collapseTagAttributes( $attr, $extra_td_attribs );
 986
 987                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 988                                         }
 989
 990                                         $outLine .= $cell;
 991                                         array_push( $td_history , true );
 992                                 }
 993                         }
 994                         $out .= $outLine . "\n";
 995                 }
 996
 997                 # Closing open td, tr && table
 998                 while ( count( $td_history ) > 0 ) {
 999                         if ( array_pop( $td_history ) ) {
1000                                 $out .= "</$td_tag>\n";
1001                         }
1002                         if ( array_pop( $tr_history ) ) {
1003                                 $out .= "</$tr_tag>\n";
1004                         }
1005                         if ( !array_pop( $has_opened_tr ) ) {
1006                                 $out .= "<$tr_tag><$td_tag></$td_tag></$tr_tag>\n" ;
1007                         }
1008
1009                         $out .= "</$table_tag>\n";
1010                 }
1011
1012                 # Remove trailing line-ending (b/c)
1013                 if ( substr( $out, -1 ) === "\n" ) {
1014                         $out = substr( $out, 0, -1 );
1015                 }
1016
1017                 # special case: don't return empty table
1018                 if ( $out === "<$table_tag>\n<$tr_tag><$td_tag></$td_tag></$tr_tag>\n</$table_tag>" ) {
1019                         $out = '';
1020                 }
1021
1022                 wfProfileOut( __METHOD__ );
1023
1024                 return $out;
1025         }
1026
1027         /**
1028          * Helper function for parse() that transforms wiki markup into
1029          * HTML. Only called for $mOutputType == self::OT_HTML.
1030          *
1031          * @private
1032          */
1033         function internalParse( $text, $isMain = true, $frame=false ) {
1034                 wfProfileIn( __METHOD__ );
1035
1036                 $origText = $text;
1037
1038                 # Hook to suspend the parser in this state
1039                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1040                         wfProfileOut( __METHOD__ );
1041                         return $text ;
1042                 }
1043
1044                 # if $frame is provided, then use $frame for replacing any variables
1045                 if ( $frame ) {
1046                         # use frame depth to infer how include/noinclude tags should be handled
1047                         # depth=0 means this is the top-level document; otherwise it's an included document
1048                         if ( !$frame->depth ) {
1049                                 $flag = 0;
1050                         } else {
1051                                 $flag = Parser::PTD_FOR_INCLUSION;
1052                         }
1053                         $dom = $this->preprocessToDom( $text, $flag );
1054                         $text = $frame->expand( $dom );
1055                 } else {
1056                         # if $frame is not provided, then use old-style replaceVariables
1057                         $text = $this->replaceVariables( $text );
1058                 }
1059
1060                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1061                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1062
1063                 # Tables need to come after variable replacement for things to work
1064                 # properly; putting them before other transformations should keep
1065                 # exciting things like link expansions from showing up in surprising
1066                 # places.
1067                 $text = $this->doTableStuff( $text );
1068
1069                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1070
1071                 $text = $this->doDoubleUnderscore( $text );
1072
1073                 $text = $this->doHeadings( $text );
1074                 if ( $this->mOptions->getUseDynamicDates() ) {
1075                         $df = DateFormatter::getInstance();
1076                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1077                 }
1078                 $text = $this->replaceInternalLinks( $text );
1079                 $text = $this->doAllQuotes( $text );
1080                 $text = $this->replaceExternalLinks( $text );
1081
1082                 # replaceInternalLinks may sometimes leave behind
1083                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1084                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1085
1086                 $text = $this->doMagicLinks( $text );
1087                 $text = $this->formatHeadings( $text, $origText, $isMain );
1088
1089                 wfProfileOut( __METHOD__ );
1090                 return $text;
1091         }
1092
1093         /**
1094          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1095          * magic external links.
1096          *
1097          * DML
1098          * @private
1099          */
1100         function doMagicLinks( $text ) {
1101                 wfProfileIn( __METHOD__ );
1102                 $prots = $this->mUrlProtocols;
1103                 $urlChar = self::EXT_LINK_URL_CLASS;
1104                 $text = preg_replace_callback(
1105                         '!(?:                           # Start cases
1106                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1107                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1108                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1109                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1110                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1111                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1112                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1113                                     [0-9Xx]                 # check digit
1114                                     \b)
1115                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1116                 wfProfileOut( __METHOD__ );
1117                 return $text;
1118         }
1119
1120         function magicLinkCallback( $m ) {
1121                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1122                         # Skip anchor
1123                         return $m[0];
1124                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1125                         # Skip HTML element
1126                         return $m[0];
1127                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1128                         # Free external link
1129                         return $this->makeFreeExternalLink( $m[0] );
1130                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1131                         # RFC or PMID
1132                         $CssClass = '';
1133                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1134                                 $keyword = 'RFC';
1135                                 $urlmsg = 'rfcurl';
1136                                 $CssClass = 'mw-magiclink-rfc';
1137                                 $id = $m[4];
1138                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1139                                 $keyword = 'PMID';
1140                                 $urlmsg = 'pubmedurl';
1141                                 $CssClass = 'mw-magiclink-pmid';
1142                                 $id = $m[4];
1143                         } else {
1144                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1145                                         substr( $m[0], 0, 20 ) . '"' );
1146                         }
1147                         $url = wfMsgForContent( $urlmsg, $id);
1148                         $sk = $this->mOptions->getSkin( $this->mTitle );
1149                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1150                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1151                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1152                         # ISBN
1153                         $isbn = $m[5];
1154                         $num = strtr( $isbn, array(
1155                                 '-' => '',
1156                                 ' ' => '',
1157                                 'x' => 'X',
1158                         ));
1159                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1160                         return'<a href="' .
1161                                 $titleObj->escapeLocalUrl() .
1162                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1163                 } else {
1164                         return $m[0];
1165                 }
1166         }
1167
1168         /**
1169          * Make a free external link, given a user-supplied URL
1170          * @return HTML
1171          * @private
1172          */
1173         function makeFreeExternalLink( $url ) {
1174                 global $wgContLang;
1175                 wfProfileIn( __METHOD__ );
1176
1177                 $sk = $this->mOptions->getSkin( $this->mTitle );
1178                 $trail = '';
1179
1180                 # The characters '<' and '>' (which were escaped by
1181                 # removeHTMLtags()) should not be included in
1182                 # URLs, per RFC 2396.
1183                 $m2 = array();
1184                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1185                         $trail = substr( $url, $m2[0][1] ) . $trail;
1186                         $url = substr( $url, 0, $m2[0][1] );
1187                 }
1188
1189                 # Move trailing punctuation to $trail
1190                 $sep = ',;\.:!?';
1191                 # If there is no left bracket, then consider right brackets fair game too
1192                 if ( strpos( $url, '(' ) === false ) {
1193                         $sep .= ')';
1194                 }
1195
1196                 $numSepChars = strspn( strrev( $url ), $sep );
1197                 if ( $numSepChars ) {
1198                         $trail = substr( $url, -$numSepChars ) . $trail;
1199                         $url = substr( $url, 0, -$numSepChars );
1200                 }
1201
1202                 $url = Sanitizer::cleanUrl( $url );
1203
1204                 # Is this an external image?
1205                 $text = $this->maybeMakeExternalImage( $url );
1206                 if ( $text === false ) {
1207                         # Not an image, make a link
1208                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1209                                 $this->getExternalLinkAttribs( $url ) );
1210                         # Register it in the output object...
1211                         # Replace unnecessary URL escape codes with their equivalent characters
1212                         $pasteurized = self::replaceUnusualEscapes( $url );
1213                         $this->mOutput->addExternalLink( $pasteurized );
1214                 }
1215                 wfProfileOut( __METHOD__ );
1216                 return $text . $trail;
1217         }
1218
1219
1220         /**
1221          * Parse headers and return html
1222          *
1223          * @private
1224          */
1225         function doHeadings( $text ) {
1226                 wfProfileIn( __METHOD__ );
1227                 for ( $i = 6; $i >= 1; --$i ) {
1228                         $h = str_repeat( '=', $i );
1229                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1230                           "<h$i>\\1</h$i>", $text );
1231                 }
1232                 wfProfileOut( __METHOD__ );
1233                 return $text;
1234         }
1235
1236         /**
1237          * Replace single quotes with HTML markup
1238          * @private
1239          * @return string the altered text
1240          */
1241         function doAllQuotes( $text ) {
1242                 wfProfileIn( __METHOD__ );
1243                 $outtext = '';
1244                 $lines = StringUtils::explode( "\n", $text );
1245                 foreach ( $lines as $line ) {
1246                         $outtext .= $this->doQuotes( $line ) . "\n";
1247                 }
1248                 $outtext = substr( $outtext, 0,-1 );
1249                 wfProfileOut( __METHOD__ );
1250                 return $outtext;
1251         }
1252
1253         /**
1254          * Helper function for doAllQuotes()
1255          */
1256         public function doQuotes( $text ) {
1257                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1258                 if ( count( $arr ) == 1 ) {
1259                         return $text;
1260                 } else {
1261                         # First, do some preliminary work. This may shift some apostrophes from
1262                         # being mark-up to being text. It also counts the number of occurrences
1263                         # of bold and italics mark-ups.
1264                         $i = 0;
1265                         $numbold = 0;
1266                         $numitalics = 0;
1267                         foreach ( $arr as $r ) {
1268                                 if ( ( $i % 2 ) == 1 ) {
1269                                         # If there are ever four apostrophes, assume the first is supposed to
1270                                         # be text, and the remaining three constitute mark-up for bold text.
1271                                         if ( strlen( $arr[$i] ) == 4 ) {
1272                                                 $arr[$i-1] .= "'";
1273                                                 $arr[$i] = "'''";
1274                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1275                                                 # If there are more than 5 apostrophes in a row, assume they're all
1276                                                 # text except for the last 5.
1277                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1278                                                 $arr[$i] = "'''''";
1279                                         }
1280                                         # Count the number of occurrences of bold and italics mark-ups.
1281                                         # We are not counting sequences of five apostrophes.
1282                                         if ( strlen( $arr[$i] ) == 2 ) {
1283                                                 $numitalics++;
1284                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1285                                                 $numbold++;
1286                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1287                                                 $numitalics++;
1288                                                 $numbold++;
1289                                         }
1290                                 }
1291                                 $i++;
1292                         }
1293
1294                         # If there is an odd number of both bold and italics, it is likely
1295                         # that one of the bold ones was meant to be an apostrophe followed
1296                         # by italics. Which one we cannot know for certain, but it is more
1297                         # likely to be one that has a single-letter word before it.
1298                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1299                                 $i = 0;
1300                                 $firstsingleletterword = -1;
1301                                 $firstmultiletterword = -1;
1302                                 $firstspace = -1;
1303                                 foreach ( $arr as $r ) {
1304                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1305                                                 $x1 = substr( $arr[$i-1], -1 );
1306                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1307                                                 if ( $x1 === ' ' ) {
1308                                                         if ( $firstspace == -1 ) {
1309                                                                 $firstspace = $i;
1310                                                         }
1311                                                 } elseif ( $x2 === ' ') {
1312                                                         if ( $firstsingleletterword == -1 ) {
1313                                                                 $firstsingleletterword = $i;
1314                                                         }
1315                                                 } else {
1316                                                         if ( $firstmultiletterword == -1 ) {
1317                                                                 $firstmultiletterword = $i;
1318                                                         }
1319                                                 }
1320                                         }
1321                                         $i++;
1322                                 }
1323
1324                                 # If there is a single-letter word, use it!
1325                                 if ( $firstsingleletterword > -1 ) {
1326                                         $arr[$firstsingleletterword] = "''";
1327                                         $arr[$firstsingleletterword-1] .= "'";
1328                                 } elseif ( $firstmultiletterword > -1 ) {
1329                                         # If not, but there's a multi-letter word, use that one.
1330                                         $arr[$firstmultiletterword] = "''";
1331                                         $arr[$firstmultiletterword-1] .= "'";
1332                                 } elseif ( $firstspace > -1 ) {
1333                                         # ... otherwise use the first one that has neither.
1334                                         # (notice that it is possible for all three to be -1 if, for example,
1335                                         # there is only one pentuple-apostrophe in the line)
1336                                         $arr[$firstspace] = "''";
1337                                         $arr[$firstspace-1] .= "'";
1338                                 }
1339                         }
1340
1341                         # Now let's actually convert our apostrophic mush to HTML!
1342                         $output = '';
1343                         $buffer = '';
1344                         $state = '';
1345                         $i = 0;
1346                         foreach ( $arr as $r ) {
1347                                 if ( ( $i % 2 ) == 0 ) {
1348                                         if ( $state === 'both' ) {
1349                                                 $buffer .= $r;
1350                                         } else {
1351                                                 $output .= $r;
1352                                         }
1353                                 } else {
1354                                         if ( strlen( $r ) == 2 ) {
1355                                                 if ( $state === 'i' ) {
1356                                                         $output .= '</i>'; $state = '';
1357                                                 } elseif ( $state === 'bi' ) {
1358                                                         $output .= '</i>'; $state = 'b';
1359                                                 } elseif ( $state === 'ib' ) {
1360                                                         $output .= '</b></i><b>'; $state = 'b';
1361                                                 } elseif ( $state === 'both' ) {
1362                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1363                                                 } else { # $state can be 'b' or ''
1364                                                         $output .= '<i>'; $state .= 'i';
1365                                                 }
1366                                         } elseif ( strlen( $r ) == 3 ) {
1367                                                 if ( $state === 'b' ) {
1368                                                         $output .= '</b>'; $state = '';
1369                                                 } elseif ( $state === 'bi' ) {
1370                                                         $output .= '</i></b><i>'; $state = 'i';
1371                                                 } elseif ( $state === 'ib' ) {
1372                                                         $output .= '</b>'; $state = 'i';
1373                                                 } elseif ( $state === 'both' ) {
1374                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1375                                                 } else { # $state can be 'i' or ''
1376                                                         $output .= '<b>'; $state .= 'b';
1377                                                 }
1378                                         } elseif ( strlen( $r ) == 5 ) {
1379                                                 if ( $state === 'b' ) {
1380                                                         $output .= '</b><i>'; $state = 'i';
1381                                                 } elseif ( $state === 'i' ) {
1382                                                         $output .= '</i><b>'; $state = 'b';
1383                                                 } elseif ( $state === 'bi' ) {
1384                                                         $output .= '</i></b>'; $state = '';
1385                                                 } elseif ( $state === 'ib' ) {
1386                                                         $output .= '</b></i>'; $state = '';
1387                                                 } elseif ( $state === 'both' ) {
1388                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1389                                                 } else { # ($state == '')
1390                                                         $buffer = ''; $state = 'both';
1391                                                 }
1392                                         }
1393                                 }
1394                                 $i++;
1395                         }
1396                         # Now close all remaining tags.  Notice that the order is important.
1397                         if ( $state === 'b' || $state === 'ib' ) {
1398                                 $output .= '</b>';
1399                         }
1400                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1401                                 $output .= '</i>';
1402                         }
1403                         if ( $state === 'bi' ) {
1404                                 $output .= '</b>';
1405                         }
1406                         # There might be lonely ''''', so make sure we have a buffer
1407                         if ( $state === 'both' && $buffer ) {
1408                                 $output .= '<b><i>'.$buffer.'</i></b>';
1409                         }
1410                         return $output;
1411                 }
1412         }
1413
1414         /**
1415          * Replace external links (REL)
1416          *
1417          * Note: this is all very hackish and the order of execution matters a lot.
1418          * Make sure to run maintenance/parserTests.php if you change this code.
1419          *
1420          * @private
1421          */
1422         function replaceExternalLinks( $text ) {
1423                 global $wgContLang;
1424                 wfProfileIn( __METHOD__ );
1425
1426                 $sk = $this->mOptions->getSkin( $this->mTitle );
1427
1428                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1429                 $s = array_shift( $bits );
1430
1431                 $i = 0;
1432                 while ( $i<count( $bits ) ) {
1433                         $url = $bits[$i++];
1434                         $protocol = $bits[$i++];
1435                         $text = $bits[$i++];
1436                         $trail = $bits[$i++];
1437
1438                         # The characters '<' and '>' (which were escaped by
1439                         # removeHTMLtags()) should not be included in
1440                         # URLs, per RFC 2396.
1441                         $m2 = array();
1442                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1443                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1444                                 $url = substr( $url, 0, $m2[0][1] );
1445                         }
1446
1447                         # If the link text is an image URL, replace it with an <img> tag
1448                         # This happened by accident in the original parser, but some people used it extensively
1449                         $img = $this->maybeMakeExternalImage( $text );
1450                         if ( $img !== false ) {
1451                                 $text = $img;
1452                         }
1453
1454                         $dtrail = '';
1455
1456                         # Set linktype for CSS - if URL==text, link is essentially free
1457                         $linktype = ( $text === $url ) ? 'free' : 'text';
1458
1459                         # No link text, e.g. [http://domain.tld/some.link]
1460                         if ( $text == '' ) {
1461                                 # Autonumber if allowed. See bug #5918
1462                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1463                                         $langObj = $this->getFunctionLang();
1464                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1465                                         $linktype = 'autonumber';
1466                                 } else {
1467                                         # Otherwise just use the URL
1468                                         $text = htmlspecialchars( $url );
1469                                         $linktype = 'free';
1470                                 }
1471                         } else {
1472                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1473                                 # Check for trail
1474                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1475                         }
1476
1477                         $text = $wgContLang->markNoConversion( $text );
1478
1479                         $url = Sanitizer::cleanUrl( $url );
1480
1481                         # Use the encoded URL
1482                         # This means that users can paste URLs directly into the text
1483                         # Funny characters like ö aren't valid in URLs anyway
1484                         # This was changed in August 2004
1485                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1486                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1487
1488                         # Register link in the output object.
1489                         # Replace unnecessary URL escape codes with the referenced character
1490                         # This prevents spammers from hiding links from the filters
1491                         $pasteurized = self::replaceUnusualEscapes( $url );
1492                         $this->mOutput->addExternalLink( $pasteurized );
1493                 }
1494
1495                 wfProfileOut( __METHOD__ );
1496                 return $s;
1497         }
1498
1499         /**
1500          * Get an associative array of additional HTML attributes appropriate for a
1501          * particular external link.  This currently may include rel => nofollow
1502          * (depending on configuration, namespace, and the URL's domain) and/or a
1503          * target attribute (depending on configuration).
1504          *
1505          * @param $url String: optional URL, to extract the domain from for rel =>
1506          *   nofollow if appropriate
1507          * @return Array: associative array of HTML attributes
1508          */
1509         function getExternalLinkAttribs( $url = false ) {
1510                 $attribs = array();
1511                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1512                 $ns = $this->mTitle->getNamespace();
1513                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1514                         $attribs['rel'] = 'nofollow';
1515
1516                         global $wgNoFollowDomainExceptions;
1517                         if ( $wgNoFollowDomainExceptions ) {
1518                                 $bits = wfParseUrl( $url );
1519                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1520                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1521                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1522                                                         unset( $attribs['rel'] );
1523                                                         break;
1524                                                 }
1525                                         }
1526                                 }
1527                         }
1528                 }
1529                 if ( $this->mOptions->getExternalLinkTarget() ) {
1530                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1531                 }
1532                 return $attribs;
1533         }
1534
1535
1536         /**
1537          * Replace unusual URL escape codes with their equivalent characters
1538          *
1539          * @param $url String
1540          * @return String
1541          *
1542          * @todo  This can merge genuinely required bits in the path or query string,
1543          *        breaking legit URLs. A proper fix would treat the various parts of
1544          *        the URL differently; as a workaround, just use the output for
1545          *        statistical records, not for actual linking/output.
1546          */
1547         static function replaceUnusualEscapes( $url ) {
1548                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1549                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1550         }
1551
1552         /**
1553          * Callback function used in replaceUnusualEscapes().
1554          * Replaces unusual URL escape codes with their equivalent character
1555          */
1556         private static function replaceUnusualEscapesCallback( $matches ) {
1557                 $char = urldecode( $matches[0] );
1558                 $ord = ord( $char );
1559                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1560                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1561                         # No, shouldn't be escaped
1562                         return $char;
1563                 } else {
1564                         # Yes, leave it escaped
1565                         return $matches[0];
1566                 }
1567         }
1568
1569         /**
1570          * make an image if it's allowed, either through the global
1571          * option, through the exception, or through the on-wiki whitelist
1572          * @private
1573          */
1574         function maybeMakeExternalImage( $url ) {
1575                 $sk = $this->mOptions->getSkin( $this->mTitle );
1576                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1577                 $imagesexception = !empty( $imagesfrom );
1578                 $text = false;
1579                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1580                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1581                         $imagematch = false;
1582                         foreach ( $imagesfrom as $match ) {
1583                                 if ( strpos( $url, $match ) === 0 ) {
1584                                         $imagematch = true;
1585                                         break;
1586                                 }
1587                         }
1588                 } elseif ( $imagesexception ) {
1589                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1590                 } else {
1591                         $imagematch = false;
1592                 }
1593                 if ( $this->mOptions->getAllowExternalImages()
1594                      || ( $imagesexception && $imagematch ) ) {
1595                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1596                                 # Image found
1597                                 $text = $sk->makeExternalImage( $url );
1598                         }
1599                 }
1600                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1601                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1602                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1603                         foreach ( $whitelist as $entry ) {
1604                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1605                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1606                                         continue;
1607                                 }
1608                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1609                                         # Image matches a whitelist entry
1610                                         $text = $sk->makeExternalImage( $url );
1611                                         break;
1612                                 }
1613                         }
1614                 }
1615                 return $text;
1616         }
1617
1618         /**
1619          * Process [[ ]] wikilinks
1620          * @return processed text
1621          *
1622          * @private
1623          */
1624         function replaceInternalLinks( $s ) {
1625                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1626                 return $s;
1627         }
1628
1629         /**
1630          * Process [[ ]] wikilinks (RIL)
1631          * @return LinkHolderArray
1632          *
1633          * @private
1634          */
1635         function replaceInternalLinks2( &$s ) {
1636                 global $wgContLang;
1637
1638                 wfProfileIn( __METHOD__ );
1639
1640                 wfProfileIn( __METHOD__.'-setup' );
1641                 static $tc = FALSE, $e1, $e1_img;
1642                 # the % is needed to support urlencoded titles as well
1643                 if ( !$tc ) {
1644                         $tc = Title::legalChars() . '#%';
1645                         # Match a link having the form [[namespace:link|alternate]]trail
1646                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1647                         # Match cases where there is no "]]", which might still be images
1648                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1649                 }
1650
1651                 $sk = $this->mOptions->getSkin( $this->mTitle );
1652                 $holders = new LinkHolderArray( $this );
1653
1654                 # split the entire text string on occurences of [[
1655                 $a = StringUtils::explode( '[[', ' ' . $s );
1656                 # get the first element (all text up to first [[), and remove the space we added
1657                 $s = $a->current();
1658                 $a->next();
1659                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1660                 $s = substr( $s, 1 );
1661
1662                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1663                 $e2 = null;
1664                 if ( $useLinkPrefixExtension ) {
1665                         # Match the end of a line for a word that's not followed by whitespace,
1666                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1667                         $e2 = wfMsgForContent( 'linkprefix' );
1668                 }
1669
1670                 if ( is_null( $this->mTitle ) ) {
1671                         wfProfileOut( __METHOD__.'-setup' );
1672                         wfProfileOut( __METHOD__ );
1673                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1674                 }
1675                 $nottalk = !$this->mTitle->isTalkPage();
1676
1677                 if ( $useLinkPrefixExtension ) {
1678                         $m = array();
1679                         if ( preg_match( $e2, $s, $m ) ) {
1680                                 $first_prefix = $m[2];
1681                         } else {
1682                                 $first_prefix = false;
1683                         }
1684                 } else {
1685                         $prefix = '';
1686                 }
1687
1688                 if ( $wgContLang->hasVariants() ) {
1689                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1690                 } else {
1691                         $selflink = array( $this->mTitle->getPrefixedText() );
1692                 }
1693                 $useSubpages = $this->areSubpagesAllowed();
1694                 wfProfileOut( __METHOD__.'-setup' );
1695
1696                 # Loop for each link
1697                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1698                         # Check for excessive memory usage
1699                         if ( $holders->isBig() ) {
1700                                 # Too big
1701                                 # Do the existence check, replace the link holders and clear the array
1702                                 $holders->replace( $s );
1703                                 $holders->clear();
1704                         }
1705
1706                         if ( $useLinkPrefixExtension ) {
1707                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1708                                 if ( preg_match( $e2, $s, $m ) ) {
1709                                         $prefix = $m[2];
1710                                         $s = $m[1];
1711                                 } else {
1712                                         $prefix='';
1713                                 }
1714                                 # first link
1715                                 if ( $first_prefix ) {
1716                                         $prefix = $first_prefix;
1717                                         $first_prefix = false;
1718                                 }
1719                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1720                         }
1721
1722                         $might_be_img = false;
1723
1724                         wfProfileIn( __METHOD__."-e1" );
1725                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1726                                 $text = $m[2];
1727                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1728                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1729                                 # the real problem is with the $e1 regex
1730                                 # See bug 1300.
1731                                 #
1732                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1733                                 # and no image is in sight. See bug 2095.
1734                                 #
1735                                 if ( $text !== '' &&
1736                                         substr( $m[3], 0, 1 ) === ']' &&
1737                                         strpos( $text, '[' ) !== false
1738                                 )
1739                                 {
1740                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1741                                         $m[3] = substr( $m[3], 1 );
1742                                 }
1743                                 # fix up urlencoded title texts
1744                                 if ( strpos( $m[1], '%' ) !== false ) {
1745                                         # Should anchors '#' also be rejected?
1746                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode( $m[1] ) );
1747                                 }
1748                                 $trail = $m[3];
1749                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1750                                 $might_be_img = true;
1751                                 $text = $m[2];
1752                                 if ( strpos( $m[1], '%' ) !== false ) {
1753                                         $m[1] = urldecode( $m[1] );
1754                                 }
1755                                 $trail = "";
1756                         } else { # Invalid form; output directly
1757                                 $s .= $prefix . '[[' . $line ;
1758                                 wfProfileOut( __METHOD__."-e1" );
1759                                 continue;
1760                         }
1761                         wfProfileOut( __METHOD__."-e1" );
1762                         wfProfileIn( __METHOD__."-misc" );
1763
1764                         # Don't allow internal links to pages containing
1765                         # PROTO: where PROTO is a valid URL protocol; these
1766                         # should be external links.
1767                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1768                                 $s .= $prefix . '[[' . $line ;
1769                                 wfProfileOut( __METHOD__."-misc" );
1770                                 continue;
1771                         }
1772
1773                         # Make subpage if necessary
1774                         if ( $useSubpages ) {
1775                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1776                         } else {
1777                                 $link = $m[1];
1778                         }
1779
1780                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1781                         if ( !$noforce ) {
1782                                 # Strip off leading ':'
1783                                 $link = substr( $link, 1 );
1784                         }
1785
1786                         wfProfileOut( __METHOD__."-misc" );
1787                         wfProfileIn( __METHOD__."-title" );
1788                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1789                         if ( $nt === null ) {
1790                                 $s .= $prefix . '[[' . $line;
1791                                 wfProfileOut( __METHOD__."-title" );
1792                                 continue;
1793                         }
1794
1795                         $ns = $nt->getNamespace();
1796                         $iw = $nt->getInterWiki();
1797                         wfProfileOut( __METHOD__."-title" );
1798
1799                         if ( $might_be_img ) { # if this is actually an invalid link
1800                                 wfProfileIn( __METHOD__."-might_be_img" );
1801                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1802                                         $found = false;
1803                                         while ( true ) {
1804                                                 # look at the next 'line' to see if we can close it there
1805                                                 $a->next();
1806                                                 $next_line = $a->current();
1807                                                 if ( $next_line === false || $next_line === null ) {
1808                                                         break;
1809                                                 }
1810                                                 $m = explode( ']]', $next_line, 3 );
1811                                                 if ( count( $m ) == 3 ) {
1812                                                         # the first ]] closes the inner link, the second the image
1813                                                         $found = true;
1814                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1815                                                         $trail = $m[2];
1816                                                         break;
1817                                                 } elseif ( count( $m ) == 2 ) {
1818                                                         # if there's exactly one ]] that's fine, we'll keep looking
1819                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1820                                                 } else {
1821                                                         # if $next_line is invalid too, we need look no further
1822                                                         $text .= '[[' . $next_line;
1823                                                         break;
1824                                                 }
1825                                         }
1826                                         if ( !$found ) {
1827                                                 # we couldn't find the end of this imageLink, so output it raw
1828                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1829                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1830                                                 $s .= "{$prefix}[[$link|$text";
1831                                                 # note: no $trail, because without an end, there *is* no trail
1832                                                 wfProfileOut( __METHOD__."-might_be_img" );
1833                                                 continue;
1834                                         }
1835                                 } else { # it's not an image, so output it raw
1836                                         $s .= "{$prefix}[[$link|$text";
1837                                         # note: no $trail, because without an end, there *is* no trail
1838                                         wfProfileOut( __METHOD__."-might_be_img" );
1839                                         continue;
1840                                 }
1841                                 wfProfileOut( __METHOD__."-might_be_img" );
1842                         }
1843
1844                         $wasblank = ( $text  == '' );
1845                         if ( $wasblank ) {
1846                                 $text = $link;
1847                         } else {
1848                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1849                                 # [[Lista d''e paise d''o munno]] -> <a href="">Lista d''e paise d''o munno</a>
1850                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']] -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1851                                 $text = $this->doQuotes($text);
1852                         }
1853
1854                         # Link not escaped by : , create the various objects
1855                         if ( $noforce ) {
1856
1857                                 # Interwikis
1858                                 wfProfileIn( __METHOD__."-interwiki" );
1859                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1860                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1861                                         $s = rtrim( $s . $prefix );
1862                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1863                                         wfProfileOut( __METHOD__."-interwiki" );
1864                                         continue;
1865                                 }
1866                                 wfProfileOut( __METHOD__."-interwiki" );
1867
1868                                 if ( $ns == NS_FILE ) {
1869                                         wfProfileIn( __METHOD__."-image" );
1870                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1871                                                 if ( $wasblank ) {
1872                                                         # if no parameters were passed, $text
1873                                                         # becomes something like "File:Foo.png",
1874                                                         # which we don't want to pass on to the
1875                                                         # image generator
1876                                                         $text = '';
1877                                                 } else {
1878                                                         # recursively parse links inside the image caption
1879                                                         # actually, this will parse them in any other parameters, too,
1880                                                         # but it might be hard to fix that, and it doesn't matter ATM
1881                                                         $text = $this->replaceExternalLinks( $text );
1882                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1883                                                 }
1884                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1885                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1886                                         } else {
1887                                                 $s .= $prefix . $trail;
1888                                         }
1889                                         $this->mOutput->addImage( $nt->getDBkey() );
1890                                         wfProfileOut( __METHOD__."-image" );
1891                                         continue;
1892
1893                                 }
1894
1895                                 if ( $ns == NS_CATEGORY ) {
1896                                         wfProfileIn( __METHOD__."-category" );
1897                                         $s = rtrim( $s . "\n" ); # bug 87
1898
1899                                         if ( $wasblank ) {
1900                                                 $sortkey = $this->getDefaultSort();
1901                                         } else {
1902                                                 $sortkey = $text;
1903                                         }
1904                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1905                                         $sortkey = str_replace( "\n", '', $sortkey );
1906                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1907                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1908
1909                                         /**
1910                                          * Strip the whitespace Category links produce, see bug 87
1911                                          * @todo We might want to use trim($tmp, "\n") here.
1912                                          */
1913                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '': $prefix . $trail;
1914
1915                                         wfProfileOut( __METHOD__."-category" );
1916                                         continue;
1917                                 }
1918                         }
1919
1920                         # Self-link checking
1921                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1922                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1923                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1924                                         continue;
1925                                 }
1926                         }
1927
1928                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1929                         # FIXME: Should do batch file existence checks, see comment below
1930                         if ( $ns == NS_MEDIA ) {
1931                                 wfProfileIn( __METHOD__."-media" );
1932                                 # Give extensions a chance to select the file revision for us
1933                                 $skip = $time = false;
1934                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1935                                 if ( $skip ) {
1936                                         $link = $sk->link( $nt );
1937                                 } else {
1938                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1939                                 }
1940                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1941                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1942                                 $this->mOutput->addImage( $nt->getDBkey() );
1943                                 wfProfileOut( __METHOD__."-media" );
1944                                 continue;
1945                         }
1946
1947                         wfProfileIn( __METHOD__."-always_known" );
1948                         # Some titles, such as valid special pages or files in foreign repos, should
1949                         # be shown as bluelinks even though they're not included in the page table
1950                         #
1951                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1952                         # batch file existence checks for NS_FILE and NS_MEDIA
1953                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1954                                 $this->mOutput->addLink( $nt );
1955                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1956                         } else {
1957                                 # Links will be added to the output link list after checking
1958                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1959                         }
1960                         wfProfileOut( __METHOD__."-always_known" );
1961                 }
1962                 wfProfileOut( __METHOD__ );
1963                 return $holders;
1964         }
1965
1966         /**
1967          * Make a link placeholder. The text returned can be later resolved to a real link with
1968          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1969          * parsing of interwiki links, and secondly to allow all existence checks and
1970          * article length checks (for stub links) to be bundled into a single query.
1971          *
1972          * @deprecated
1973          */
1974         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1975                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1976         }
1977
1978         /**
1979          * Render a forced-blue link inline; protect against double expansion of
1980          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1981          * Since this little disaster has to split off the trail text to avoid
1982          * breaking URLs in the following text without breaking trails on the
1983          * wiki links, it's been made into a horrible function.
1984          *
1985          * @param $nt Title
1986          * @param $text String
1987          * @param $query String
1988          * @param $trail String
1989          * @param $prefix String
1990          * @return String: HTML-wikitext mix oh yuck
1991          */
1992         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1993                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1994                 $sk = $this->mOptions->getSkin( $this->mTitle );
1995                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1996                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1997                 return $this->armorLinks( $link ) . $trail;
1998         }
1999
2000         /**
2001          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2002          * going to go through further parsing steps before inline URL expansion.
2003          *
2004          * Not needed quite as much as it used to be since free links are a bit
2005          * more sensible these days. But bracketed links are still an issue.
2006          *
2007          * @param $text String: more-or-less HTML
2008          * @return String: less-or-more HTML with NOPARSE bits
2009          */
2010         function armorLinks( $text ) {
2011                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2012                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2013         }
2014
2015         /**
2016          * Return true if subpage links should be expanded on this page.
2017          * @return Boolean
2018          */
2019         function areSubpagesAllowed() {
2020                 # Some namespaces don't allow subpages
2021                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2022         }
2023
2024         /**
2025          * Handle link to subpage if necessary
2026          *
2027          * @param $target String: the source of the link
2028          * @param &$text String: the link text, modified as necessary
2029          * @return string the full name of the link
2030          * @private
2031          */
2032         function maybeDoSubpageLink( $target, &$text ) {
2033                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2034         }
2035
2036         /**#@+
2037          * Used by doBlockLevels()
2038          * @private
2039          */
2040         function closeParagraph() {
2041                 $result = '';
2042                 if ( $this->mLastSection != '' ) {
2043                         $result = '</' . $this->mLastSection  . ">\n";
2044                 }
2045                 $this->mInPre = false;
2046                 $this->mLastSection = '';
2047                 return $result;
2048         }
2049
2050         /**
2051          * getCommon() returns the length of the longest common substring
2052          * of both arguments, starting at the beginning of both.
2053          * @private
2054          */
2055         function getCommon( $st1, $st2 ) {
2056                 $fl = strlen( $st1 );
2057                 $shorter = strlen( $st2 );
2058                 if ( $fl < $shorter ) {
2059                         $shorter = $fl;
2060                 }
2061
2062                 for ( $i = 0; $i < $shorter; ++$i ) {
2063                         if ( $st1{$i} != $st2{$i} ) {
2064                                 break;
2065                         }
2066                 }
2067                 return $i;
2068         }
2069
2070         /**
2071          * These next three functions open, continue, and close the list
2072          * element appropriate to the prefix character passed into them.
2073          * @private
2074          */
2075         function openList( $char ) {
2076                 $result = $this->closeParagraph();
2077
2078                 if ( '*' === $char ) {
2079                         $result .= '<ul><li>';
2080                 } elseif ( '#' === $char ) {
2081                         $result .= '<ol><li>';
2082                 } elseif ( ':' === $char ) {
2083                         $result .= '<dl><dd>';
2084                 } elseif ( ';' === $char ) {
2085                         $result .= '<dl><dt>';
2086                         $this->mDTopen = true;
2087                 } else {
2088                         $result = '<!-- ERR 1 -->';
2089                 }
2090
2091                 return $result;
2092         }
2093
2094         /**
2095          * TODO: document
2096          * @param $char String
2097          * @private
2098          */
2099         function nextItem( $char ) {
2100                 if ( '*' === $char || '#' === $char ) {
2101                         return '</li><li>';
2102                 } elseif ( ':' === $char || ';' === $char ) {
2103                         $close = '</dd>';
2104                         if ( $this->mDTopen ) {
2105                                 $close = '</dt>';
2106                         }
2107                         if ( ';' === $char ) {
2108                                 $this->mDTopen = true;
2109                                 return $close . '<dt>';
2110                         } else {
2111                                 $this->mDTopen = false;
2112                                 return $close . '<dd>';
2113                         }
2114                 }
2115                 return '<!-- ERR 2 -->';
2116         }
2117
2118         /**
2119          * TODO: document
2120          * @param $char String
2121          * @private
2122          */
2123         function closeList( $char ) {
2124                 if ( '*' === $char ) {
2125                         $text = '</li></ul>';
2126                 } elseif ( '#' === $char ) {
2127                         $text = '</li></ol>';
2128                 } elseif ( ':' === $char ) {
2129                         if ( $this->mDTopen ) {
2130                                 $this->mDTopen = false;
2131                                 $text = '</dt></dl>';
2132                         } else {
2133                                 $text = '</dd></dl>';
2134                         }
2135                 } else {
2136                         return '<!-- ERR 3 -->';
2137                 }
2138                 return $text."\n";
2139         }
2140         /**#@-*/
2141
2142         /**
2143          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2144          *
2145          * @param $text String
2146          * @param $linestart Boolean: whether or not this is at the start of a line.
2147          * @private
2148          * @return string the lists rendered as HTML
2149          */
2150         function doBlockLevels( $text, $linestart ) {
2151                 wfProfileIn( __METHOD__ );
2152
2153                 # Parsing through the text line by line.  The main thing
2154                 # happening here is handling of block-level elements p, pre,
2155                 # and making lists from lines starting with * # : etc.
2156                 #
2157                 $textLines = StringUtils::explode( "\n", $text );
2158
2159                 $lastPrefix = $output = '';
2160                 $this->mDTopen = $inBlockElem = false;
2161                 $prefixLength = 0;
2162                 $paragraphStack = false;
2163
2164                 foreach ( $textLines as $oLine ) {
2165                         # Fix up $linestart
2166                         if ( !$linestart ) {
2167                                 $output .= $oLine;
2168                                 $linestart = true;
2169                                 continue;
2170                         }
2171                         # * = ul
2172                         # # = ol
2173                         # ; = dt
2174                         # : = dd
2175
2176                         $lastPrefixLength = strlen( $lastPrefix );
2177                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2178                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2179                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2180                         if ( !$this->mInPre ) {
2181                                 # Multiple prefixes may abut each other for nested lists.
2182                                 $prefixLength = strspn( $oLine, '*#:;' );
2183                                 $prefix = substr( $oLine, 0, $prefixLength );
2184
2185                                 # eh?
2186                                 # ; and : are both from definition-lists, so they're equivalent
2187                                 #  for the purposes of determining whether or not we need to open/close
2188                                 #  elements.
2189                                 $prefix2 = str_replace( ';', ':', $prefix );
2190                                 $t = substr( $oLine, $prefixLength );
2191                                 $this->mInPre = (bool)$preOpenMatch;
2192                         } else {
2193                                 # Don't interpret any other prefixes in preformatted text
2194                                 $prefixLength = 0;
2195                                 $prefix = $prefix2 = '';
2196                                 $t = $oLine;
2197                         }
2198
2199                         # List generation
2200                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2201                                 # Same as the last item, so no need to deal with nesting or opening stuff
2202                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2203                                 $paragraphStack = false;
2204
2205                                 if ( substr( $prefix, -1 ) === ';') {
2206                                         # The one nasty exception: definition lists work like this:
2207                                         # ; title : definition text
2208                                         # So we check for : in the remainder text to split up the
2209                                         # title and definition, without b0rking links.
2210                                         $term = $t2 = '';
2211                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2212                                                 $t = $t2;
2213                                                 $output .= $term . $this->nextItem( ':' );
2214                                         }
2215                                 }
2216                         } elseif ( $prefixLength || $lastPrefixLength ) {
2217                                 # We need to open or close prefixes, or both.
2218
2219                                 # Either open or close a level...
2220                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2221                                 $paragraphStack = false;
2222
2223                                 # Close all the prefixes which aren't shared.
2224                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2225                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2226                                         --$lastPrefixLength;
2227                                 }
2228
2229                                 # Continue the current prefix if appropriate.
2230                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2231                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2232                                 }
2233
2234                                 # Open prefixes where appropriate.
2235                                 while ( $prefixLength > $commonPrefixLength ) {
2236                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2237                                         $output .= $this->openList( $char );
2238
2239                                         if ( ';' === $char ) {
2240                                                 # FIXME: This is dupe of code above
2241                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2242                                                         $t = $t2;
2243                                                         $output .= $term . $this->nextItem( ':' );
2244                                                 }
2245                                         }
2246                                         ++$commonPrefixLength;
2247                                 }
2248                                 $lastPrefix = $prefix2;
2249                         }
2250
2251                         # If we have no prefixes, go to paragraph mode.
2252                         if ( 0 == $prefixLength ) {
2253                                 wfProfileIn( __METHOD__."-paragraph" );
2254                                 # No prefix (not in list)--go to paragraph mode
2255                                 # XXX: use a stack for nestable elements like span, table and div
2256                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2257                                 $closematch = preg_match(
2258                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2259                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2260                                 if ( $openmatch or $closematch ) {
2261                                         $paragraphStack = false;
2262                                         # TODO bug 5718: paragraph closed
2263                                         $output .= $this->closeParagraph();
2264                                         if ( $preOpenMatch and !$preCloseMatch ) {
2265                                                 $this->mInPre = true;
2266                                         }
2267                                         $inBlockElem = !$closematch;
2268                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2269                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2270                                                 # pre
2271                                                 if ( $this->mLastSection !== 'pre' ) {
2272                                                         $paragraphStack = false;
2273                                                         $output .= $this->closeParagraph().'<pre>';
2274                                                         $this->mLastSection = 'pre';
2275                                                 }
2276                                                 $t = substr( $t, 1 );
2277                                         } else {
2278                                                 # paragraph
2279                                                 if ( trim( $t ) === '' ) {
2280                                                         if ( $paragraphStack ) {
2281                                                                 $output .= $paragraphStack.'<br />';
2282                                                                 $paragraphStack = false;
2283                                                                 $this->mLastSection = 'p';
2284                                                         } else {
2285                                                                 if ( $this->mLastSection !== 'p' ) {
2286                                                                         $output .= $this->closeParagraph();
2287                                                                         $this->mLastSection = '';
2288                                                                         $paragraphStack = '<p>';
2289                                                                 } else {
2290                                                                         $paragraphStack = '</p><p>';
2291                                                                 }
2292                                                         }
2293                                                 } else {
2294                                                         if ( $paragraphStack ) {
2295                                                                 $output .= $paragraphStack;
2296                                                                 $paragraphStack = false;
2297                                                                 $this->mLastSection = 'p';
2298                                                         } elseif ( $this->mLastSection !== 'p' ) {
2299                                                                 $output .= $this->closeParagraph().'<p>';
2300                                                                 $this->mLastSection = 'p';
2301                                                         }
2302                                                 }
2303                                         }
2304                                 }
2305                                 wfProfileOut( __METHOD__."-paragraph" );
2306                         }
2307                         # somewhere above we forget to get out of pre block (bug 785)
2308                         if ( $preCloseMatch && $this->mInPre ) {
2309                                 $this->mInPre = false;
2310                         }
2311                         if ( $paragraphStack === false ) {
2312                                 $output .= $t."\n";
2313                         }
2314                 }
2315                 while ( $prefixLength ) {
2316                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2317                         --$prefixLength;
2318                 }
2319                 if ( $this->mLastSection != '' ) {
2320                         $output .= '</' . $this->mLastSection . '>';
2321                         $this->mLastSection = '';
2322                 }
2323
2324                 wfProfileOut( __METHOD__ );
2325                 return $output;
2326         }
2327
2328         /**
2329          * Split up a string on ':', ignoring any occurences inside tags
2330          * to prevent illegal overlapping.
2331          *
2332          * @param $str String: the string to split
2333          * @param &$before String: set to everything before the ':'
2334          * @param &$after String: set to everything after the ':'
2335          * return String: the position of the ':', or false if none found
2336          */
2337         function findColonNoLinks( $str, &$before, &$after ) {
2338                 wfProfileIn( __METHOD__ );
2339
2340                 $pos = strpos( $str, ':' );
2341                 if ( $pos === false ) {
2342                         # Nothing to find!
2343                         wfProfileOut( __METHOD__ );
2344                         return false;
2345                 }
2346
2347                 $lt = strpos( $str, '<' );
2348                 if ( $lt === false || $lt > $pos ) {
2349                         # Easy; no tag nesting to worry about
2350                         $before = substr( $str, 0, $pos );
2351                         $after = substr( $str, $pos+1 );
2352                         wfProfileOut( __METHOD__ );
2353                         return $pos;
2354                 }
2355
2356                 # Ugly state machine to walk through avoiding tags.
2357                 $state = self::COLON_STATE_TEXT;
2358                 $stack = 0;
2359                 $len = strlen( $str );
2360                 for( $i = 0; $i < $len; $i++ ) {
2361                         $c = $str{$i};
2362
2363                         switch( $state ) {
2364                         # (Using the number is a performance hack for common cases)
2365                         case 0: # self::COLON_STATE_TEXT:
2366                                 switch( $c ) {
2367                                 case "<":
2368                                         # Could be either a <start> tag or an </end> tag
2369                                         $state = self::COLON_STATE_TAGSTART;
2370                                         break;
2371                                 case ":":
2372                                         if ( $stack == 0 ) {
2373                                                 # We found it!
2374                                                 $before = substr( $str, 0, $i );
2375                                                 $after = substr( $str, $i + 1 );
2376                                                 wfProfileOut( __METHOD__ );
2377                                                 return $i;
2378                                         }
2379                                         # Embedded in a tag; don't break it.
2380                                         break;
2381                                 default:
2382                                         # Skip ahead looking for something interesting
2383                                         $colon = strpos( $str, ':', $i );
2384                                         if ( $colon === false ) {
2385                                                 # Nothing else interesting
2386                                                 wfProfileOut( __METHOD__ );
2387                                                 return false;
2388                                         }
2389                                         $lt = strpos( $str, '<', $i );
2390                                         if ( $stack === 0 ) {
2391                                                 if ( $lt === false || $colon < $lt ) {
2392                                                         # We found it!
2393                                                         $before = substr( $str, 0, $colon );
2394                                                         $after = substr( $str, $colon + 1 );
2395                                                         wfProfileOut( __METHOD__ );
2396                                                         return $i;
2397                                                 }
2398                                         }
2399                                         if ( $lt === false ) {
2400                                                 # Nothing else interesting to find; abort!
2401                                                 # We're nested, but there's no close tags left. Abort!
2402                                                 break 2;
2403                                         }
2404                                         # Skip ahead to next tag start
2405                                         $i = $lt;
2406                                         $state = self::COLON_STATE_TAGSTART;
2407                                 }
2408                                 break;
2409                         case 1: # self::COLON_STATE_TAG:
2410                                 # In a <tag>
2411                                 switch( $c ) {
2412                                 case ">":
2413                                         $stack++;
2414                                         $state = self::COLON_STATE_TEXT;
2415                                         break;
2416                                 case "/":
2417                                         # Slash may be followed by >?
2418                                         $state = self::COLON_STATE_TAGSLASH;
2419                                         break;
2420                                 default:
2421                                         # ignore
2422                                 }
2423                                 break;
2424                         case 2: # self::COLON_STATE_TAGSTART:
2425                                 switch( $c ) {
2426                                 case "/":
2427                                         $state = self::COLON_STATE_CLOSETAG;
2428                                         break;
2429                                 case "!":
2430                                         $state = self::COLON_STATE_COMMENT;
2431                                         break;
2432                                 case ">":
2433                                         # Illegal early close? This shouldn't happen D:
2434                                         $state = self::COLON_STATE_TEXT;
2435                                         break;
2436                                 default:
2437                                         $state = self::COLON_STATE_TAG;
2438                                 }
2439                                 break;
2440                         case 3: # self::COLON_STATE_CLOSETAG:
2441                                 # In a </tag>
2442                                 if ( $c === ">" ) {
2443                                         $stack--;
2444                                         if ( $stack < 0 ) {
2445                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2446                                                 wfProfileOut( __METHOD__ );
2447                                                 return false;
2448                                         }
2449                                         $state = self::COLON_STATE_TEXT;
2450                                 }
2451                                 break;
2452                         case self::COLON_STATE_TAGSLASH:
2453                                 if ( $c === ">" ) {
2454                                         # Yes, a self-closed tag <blah/>
2455                                         $state = self::COLON_STATE_TEXT;
2456                                 } else {
2457                                         # Probably we're jumping the gun, and this is an attribute
2458                                         $state = self::COLON_STATE_TAG;
2459                                 }
2460                                 break;
2461                         case 5: # self::COLON_STATE_COMMENT:
2462                                 if ( $c === "-" ) {
2463                                         $state = self::COLON_STATE_COMMENTDASH;
2464                                 }
2465                                 break;
2466                         case self::COLON_STATE_COMMENTDASH:
2467                                 if ( $c === "-" ) {
2468                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2469                                 } else {
2470                                         $state = self::COLON_STATE_COMMENT;
2471                                 }
2472                                 break;
2473                         case self::COLON_STATE_COMMENTDASHDASH:
2474                                 if ( $c === ">" ) {
2475                                         $state = self::COLON_STATE_TEXT;
2476                                 } else {
2477                                         $state = self::COLON_STATE_COMMENT;
2478                                 }
2479                                 break;
2480                         default:
2481                                 throw new MWException( "State machine error in " . __METHOD__ );
2482                         }
2483                 }
2484                 if ( $stack > 0 ) {
2485                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2486                         return false;
2487                 }
2488                 wfProfileOut( __METHOD__ );
2489                 return false;
2490         }
2491
2492         /**
2493          * Return value of a magic variable (like PAGENAME)
2494          *
2495          * @private
2496          */
2497         function getVariableValue( $index, $frame=false ) {
2498                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2499                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2500
2501                 /**
2502                  * Some of these require message or data lookups and can be
2503                  * expensive to check many times.
2504                  */
2505                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2506                         if ( isset( $this->mVarCache[$index] ) ) {
2507                                 return $this->mVarCache[$index];
2508                         }
2509                 }
2510
2511                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2512                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2513
2514                 # Use the time zone
2515                 global $wgLocaltimezone;
2516                 if ( isset( $wgLocaltimezone ) ) {
2517                         $oldtz = date_default_timezone_get();
2518                         date_default_timezone_set( $wgLocaltimezone );
2519                 }
2520
2521                 $localTimestamp = date( 'YmdHis', $ts );
2522                 $localMonth = date( 'm', $ts );
2523                 $localMonth1 = date( 'n', $ts );
2524                 $localMonthName = date( 'n', $ts );
2525                 $localDay = date( 'j', $ts );
2526                 $localDay2 = date( 'd', $ts );
2527                 $localDayOfWeek = date( 'w', $ts );
2528                 $localWeek = date( 'W', $ts );
2529                 $localYear = date( 'Y', $ts );
2530                 $localHour = date( 'H', $ts );
2531                 if ( isset( $wgLocaltimezone ) ) {
2532                         date_default_timezone_set( $oldtz );
2533                 }
2534
2535                 switch ( $index ) {
2536                         case 'currentmonth':
2537                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2538                                 break;
2539                         case 'currentmonth1':
2540                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2541                                 break;
2542                         case 'currentmonthname':
2543                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2544                                 break;
2545                         case 'currentmonthnamegen':
2546                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2547                                 break;
2548                         case 'currentmonthabbrev':
2549                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2550                                 break;
2551                         case 'currentday':
2552                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2553                                 break;
2554                         case 'currentday2':
2555                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2556                                 break;
2557                         case 'localmonth':
2558                                 $value = $wgContLang->formatNum( $localMonth );
2559                                 break;
2560                         case 'localmonth1':
2561                                 $value = $wgContLang->formatNum( $localMonth1 );
2562                                 break;
2563                         case 'localmonthname':
2564                                 $value = $wgContLang->getMonthName( $localMonthName );
2565                                 break;
2566                         case 'localmonthnamegen':
2567                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2568                                 break;
2569                         case 'localmonthabbrev':
2570                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2571                                 break;
2572                         case 'localday':
2573                                 $value = $wgContLang->formatNum( $localDay );
2574                                 break;
2575                         case 'localday2':
2576                                 $value = $wgContLang->formatNum( $localDay2 );
2577                                 break;
2578                         case 'pagename':
2579                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2580                                 break;
2581                         case 'pagenamee':
2582                                 $value = $this->mTitle->getPartialURL();
2583                                 break;
2584                         case 'fullpagename':
2585                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2586                                 break;
2587                         case 'fullpagenamee':
2588                                 $value = $this->mTitle->getPrefixedURL();
2589                                 break;
2590                         case 'subpagename':
2591                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2592                                 break;
2593                         case 'subpagenamee':
2594                                 $value = $this->mTitle->getSubpageUrlForm();
2595                                 break;
2596                         case 'basepagename':
2597                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2598                                 break;
2599                         case 'basepagenamee':
2600                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2601                                 break;
2602                         case 'talkpagename':
2603                                 if ( $this->mTitle->canTalk() ) {
2604                                         $talkPage = $this->mTitle->getTalkPage();
2605                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2606                                 } else {
2607                                         $value = '';
2608                                 }
2609                                 break;
2610                         case 'talkpagenamee':
2611                                 if ( $this->mTitle->canTalk() ) {
2612                                         $talkPage = $this->mTitle->getTalkPage();
2613                                         $value = $talkPage->getPrefixedUrl();
2614                                 } else {
2615                                         $value = '';
2616                                 }
2617                                 break;
2618                         case 'subjectpagename':
2619                                 $subjPage = $this->mTitle->getSubjectPage();
2620                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2621                                 break;
2622                         case 'subjectpagenamee':
2623                                 $subjPage = $this->mTitle->getSubjectPage();
2624                                 $value = $subjPage->getPrefixedUrl();
2625                                 break;
2626                         case 'revisionid':
2627                                 # Let the edit saving system know we should parse the page
2628                                 # *after* a revision ID has been assigned.
2629                                 $this->mOutput->setFlag( 'vary-revision' );
2630                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2631                                 $value = $this->mRevisionId;
2632                                 break;
2633                         case 'revisionday':
2634                                 # Let the edit saving system know we should parse the page
2635                                 # *after* a revision ID has been assigned. This is for null edits.
2636                                 $this->mOutput->setFlag( 'vary-revision' );
2637                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2638                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2639                                 break;
2640                         case 'revisionday2':
2641                                 # Let the edit saving system know we should parse the page
2642                                 # *after* a revision ID has been assigned. This is for null edits.
2643                                 $this->mOutput->setFlag( 'vary-revision' );
2644                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2645                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2646                                 break;
2647                         case 'revisionmonth':
2648                                 # Let the edit saving system know we should parse the page
2649                                 # *after* a revision ID has been assigned. This is for null edits.
2650                                 $this->mOutput->setFlag( 'vary-revision' );
2651                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2652                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2653                                 break;
2654                         case 'revisionmonth1':
2655                                 # Let the edit saving system know we should parse the page
2656                                 # *after* a revision ID has been assigned. This is for null edits.
2657                                 $this->mOutput->setFlag( 'vary-revision' );
2658                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2659                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2660                                 break;
2661                         case 'revisionyear':
2662                                 # Let the edit saving system know we should parse the page
2663                                 # *after* a revision ID has been assigned. This is for null edits.
2664                                 $this->mOutput->setFlag( 'vary-revision' );
2665                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2666                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2667                                 break;
2668                         case 'revisiontimestamp':
2669                                 # Let the edit saving system know we should parse the page
2670                                 # *after* a revision ID has been assigned. This is for null edits.
2671                                 $this->mOutput->setFlag( 'vary-revision' );
2672                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2673                                 $value = $this->getRevisionTimestamp();
2674                                 break;
2675                         case 'revisionuser':
2676                                 # Let the edit saving system know we should parse the page
2677                                 # *after* a revision ID has been assigned. This is for null edits.
2678                                 $this->mOutput->setFlag( 'vary-revision' );
2679                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2680                                 $value = $this->getRevisionUser();
2681                                 break;
2682                         case 'namespace':
2683                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2684                                 break;
2685                         case 'namespacee':
2686                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2687                                 break;
2688                         case 'talkspace':
2689                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2690                                 break;
2691                         case 'talkspacee':
2692                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2693                                 break;
2694                         case 'subjectspace':
2695                                 $value = $this->mTitle->getSubjectNsText();
2696                                 break;
2697                         case 'subjectspacee':
2698                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2699                                 break;
2700                         case 'currentdayname':
2701                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2702                                 break;
2703                         case 'currentyear':
2704                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2705                                 break;
2706                         case 'currenttime':
2707                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2708                                 break;
2709                         case 'currenthour':
2710                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2711                                 break;
2712                         case 'currentweek':
2713                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2714                                 # int to remove the padding
2715                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2716                                 break;
2717                         case 'currentdow':
2718                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2719                                 break;
2720                         case 'localdayname':
2721                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2722                                 break;
2723                         case 'localyear':
2724                                 $value = $wgContLang->formatNum( $localYear, true );
2725                                 break;
2726                         case 'localtime':
2727                                 $value = $wgContLang->time( $localTimestamp, false, false );
2728                                 break;
2729                         case 'localhour':
2730                                 $value = $wgContLang->formatNum( $localHour, true );
2731                                 break;
2732                         case 'localweek':
2733                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2734                                 # int to remove the padding
2735                                 $value = $wgContLang->formatNum( (int)$localWeek );
2736                                 break;
2737                         case 'localdow':
2738                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2739                                 break;
2740                         case 'numberofarticles':
2741                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2742                                 break;
2743                         case 'numberoffiles':
2744                                 $value = $wgContLang->formatNum( SiteStats::images() );
2745                                 break;
2746                         case 'numberofusers':
2747                                 $value = $wgContLang->formatNum( SiteStats::users() );
2748                                 break;
2749                         case 'numberofactiveusers':
2750                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2751                                 break;
2752                         case 'numberofpages':
2753                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2754                                 break;
2755                         case 'numberofadmins':
2756                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2757                                 break;
2758                         case 'numberofedits':
2759                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2760                                 break;
2761                         case 'numberofviews':
2762                                 $value = $wgContLang->formatNum( SiteStats::views() );
2763                                 break;
2764                         case 'currenttimestamp':
2765                                 $value = wfTimestamp( TS_MW, $ts );
2766                                 break;
2767                         case 'localtimestamp':
2768                                 $value = $localTimestamp;
2769                                 break;
2770                         case 'currentversion':
2771                                 $value = SpecialVersion::getVersion();
2772                                 break;
2773                         case 'articlepath':
2774                                 return $wgArticlePath;
2775                         case 'sitename':
2776                                 return $wgSitename;
2777                         case 'server':
2778                                 return $wgServer;
2779                         case 'servername':
2780                                 return $wgServerName;
2781                         case 'scriptpath':
2782                                 return $wgScriptPath;
2783                         case 'stylepath':
2784                                 return $wgStylePath;
2785                         case 'directionmark':
2786                                 return $wgContLang->getDirMark();
2787                         case 'contentlanguage':
2788                                 global $wgLanguageCode;
2789                                 return $wgLanguageCode;
2790                         default:
2791                                 $ret = null;
2792                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2793                                         return $ret;
2794                                 } else {
2795                                         return null;
2796                                 }
2797                 }
2798
2799                 if ( $index )
2800                         $this->mVarCache[$index] = $value;
2801
2802                 return $value;
2803         }
2804
2805         /**
2806          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2807          *
2808          * @private
2809          */
2810         function initialiseVariables() {
2811                 wfProfileIn( __METHOD__ );
2812                 $variableIDs = MagicWord::getVariableIDs();
2813                 $substIDs = MagicWord::getSubstIDs();
2814
2815                 $this->mVariables = new MagicWordArray( $variableIDs );
2816                 $this->mSubstWords = new MagicWordArray( $substIDs );
2817                 wfProfileOut( __METHOD__ );
2818         }
2819
2820         /**
2821          * Preprocess some wikitext and return the document tree.
2822          * This is the ghost of replace_variables().
2823          *
2824          * @param $text String: The text to parse
2825          * @param $flags Integer: bitwise combination of:
2826          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2827          *                                     included. Default is to assume a direct page view.
2828          *
2829          * The generated DOM tree must depend only on the input text and the flags.
2830          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2831          *
2832          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2833          * change in the DOM tree for a given text, must be passed through the section identifier
2834          * in the section edit link and thus back to extractSections().
2835          *
2836          * The output of this function is currently only cached in process memory, but a persistent
2837          * cache may be implemented at a later date which takes further advantage of these strict
2838          * dependency requirements.
2839          *
2840          * @private
2841          */
2842         function preprocessToDom( $text, $flags = 0 ) {
2843                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2844                 return $dom;
2845         }
2846
2847         /**
2848          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2849          */
2850         public static function splitWhitespace( $s ) {
2851                 $ltrimmed = ltrim( $s );
2852                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2853                 $trimmed = rtrim( $ltrimmed );
2854                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2855                 if ( $diff > 0 ) {
2856                         $w2 = substr( $ltrimmed, -$diff );
2857                 } else {
2858                         $w2 = '';
2859                 }
2860                 return array( $w1, $trimmed, $w2 );
2861         }
2862
2863         /**
2864          * Replace magic variables, templates, and template arguments
2865          * with the appropriate text. Templates are substituted recursively,
2866          * taking care to avoid infinite loops.
2867          *
2868          * Note that the substitution depends on value of $mOutputType:
2869          *  self::OT_WIKI: only {{subst:}} templates
2870          *  self::OT_PREPROCESS: templates but not extension tags
2871          *  self::OT_HTML: all templates and extension tags
2872          *
2873          * @param $text String: the text to transform
2874          * @param $frame PPFrame Object describing the arguments passed to the template.
2875          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2876          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2877          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2878          * @private
2879          */
2880         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2881                 # Is there any text? Also, Prevent too big inclusions!
2882                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2883                         return $text;
2884                 }
2885                 wfProfileIn( __METHOD__ );
2886
2887                 if ( $frame === false ) {
2888                         $frame = $this->getPreprocessor()->newFrame();
2889                 } elseif ( !( $frame instanceof PPFrame ) ) {
2890                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2891                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2892                 }
2893
2894                 $dom = $this->preprocessToDom( $text );
2895                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2896                 $text = $frame->expand( $dom, $flags );
2897
2898                 wfProfileOut( __METHOD__ );
2899                 return $text;
2900         }
2901
2902         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2903         static function createAssocArgs( $args ) {
2904                 $assocArgs = array();
2905                 $index = 1;
2906                 foreach ( $args as $arg ) {
2907                         $eqpos = strpos( $arg, '=' );
2908                         if ( $eqpos === false ) {
2909                                 $assocArgs[$index++] = $arg;
2910                         } else {
2911                                 $name = trim( substr( $arg, 0, $eqpos ) );
2912                                 $value = trim( substr( $arg, $eqpos+1 ) );
2913                                 if ( $value === false ) {
2914                                         $value = '';
2915                                 }
2916                                 if ( $name !== false ) {
2917                                         $assocArgs[$name] = $value;
2918                                 }
2919                         }
2920                 }
2921
2922                 return $assocArgs;
2923         }
2924
2925         /**
2926          * Warn the user when a parser limitation is reached
2927          * Will warn at most once the user per limitation type
2928          *
2929          * @param $limitationType String: should be one of:
2930          *   'expensive-parserfunction' (corresponding messages:
2931          *       'expensive-parserfunction-warning',
2932          *       'expensive-parserfunction-category')
2933          *   'post-expand-template-argument' (corresponding messages:
2934          *       'post-expand-template-argument-warning',
2935          *       'post-expand-template-argument-category')
2936          *   'post-expand-template-inclusion' (corresponding messages:
2937          *       'post-expand-template-inclusion-warning',
2938          *       'post-expand-template-inclusion-category')
2939          * @param $current Current value
2940          * @param $max Maximum allowed, when an explicit limit has been
2941          *       exceeded, provide the values (optional)
2942          */
2943         function limitationWarn( $limitationType, $current=null, $max=null) {
2944                 # does no harm if $current and $max are present but are unnecessary for the message
2945                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2946                 $this->mOutput->addWarning( $warning );
2947                 $this->addTrackingCategory( "$limitationType-category" );
2948         }
2949
2950         /**
2951          * Return the text of a template, after recursively
2952          * replacing any variables or templates within the template.
2953          *
2954          * @param $piece Array: the parts of the template
2955          *  $piece['title']: the title, i.e. the part before the |
2956          *  $piece['parts']: the parameter array
2957          *  $piece['lineStart']: whether the brace was at the start of a line
2958          * @param $frame PPFrame The current frame, contains template arguments
2959          * @return String: the text of the template
2960          * @private
2961          */
2962         function braceSubstitution( $piece, $frame ) {
2963                 global $wgContLang, $wgNonincludableNamespaces;
2964                 wfProfileIn( __METHOD__ );
2965                 wfProfileIn( __METHOD__.'-setup' );
2966
2967                 # Flags
2968                 $found = false;             # $text has been filled
2969                 $nowiki = false;            # wiki markup in $text should be escaped
2970                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2971                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2972                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2973                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2974
2975                 # Title object, where $text came from
2976                 $title = null;
2977
2978                 # $part1 is the bit before the first |, and must contain only title characters.
2979                 # Various prefixes will be stripped from it later.
2980                 $titleWithSpaces = $frame->expand( $piece['title'] );
2981                 $part1 = trim( $titleWithSpaces );
2982                 $titleText = false;
2983
2984                 # Original title text preserved for various purposes
2985                 $originalTitle = $part1;
2986
2987                 # $args is a list of argument nodes, starting from index 0, not including $part1
2988                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2989                 wfProfileOut( __METHOD__.'-setup' );
2990
2991                 # SUBST
2992                 wfProfileIn( __METHOD__.'-modifiers' );
2993                 if ( !$found ) {
2994
2995                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2996
2997                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2998                         # Decide whether to expand template or keep wikitext as-is.
2999                         if ( $this->ot['wiki'] ) {
3000                                 if ( $substMatch === false ) {
3001                                         $literal = true;  # literal when in PST with no prefix
3002                                 } else {
3003                                         $literal = false; # expand when in PST with subst: or safesubst:
3004                                 }
3005                         } else {
3006                                 if ( $substMatch == 'subst' ) {
3007                                         $literal = true;  # literal when not in PST with plain subst:
3008                                 } else {
3009                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3010                                 }
3011                         }
3012                         if ( $literal ) {
3013                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3014                                 $isLocalObj = true;
3015                                 $found = true;
3016                         }
3017                 }
3018
3019                 # Variables
3020                 if ( !$found && $args->getLength() == 0 ) {
3021                         $id = $this->mVariables->matchStartToEnd( $part1 );
3022                         if ( $id !== false ) {
3023                                 $text = $this->getVariableValue( $id, $frame );
3024                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3025                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3026                                 }
3027                                 $found = true;
3028                         }
3029                 }
3030
3031                 # MSG, MSGNW and RAW
3032                 if ( !$found ) {
3033                         # Check for MSGNW:
3034                         $mwMsgnw = MagicWord::get( 'msgnw' );
3035                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3036                                 $nowiki = true;
3037                         } else {
3038                                 # Remove obsolete MSG:
3039                                 $mwMsg = MagicWord::get( 'msg' );
3040                                 $mwMsg->matchStartAndRemove( $part1 );
3041                         }
3042
3043                         # Check for RAW:
3044                         $mwRaw = MagicWord::get( 'raw' );
3045                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3046                                 $forceRawInterwiki = true;
3047                         }
3048                 }
3049                 wfProfileOut( __METHOD__.'-modifiers' );
3050
3051                 # Parser functions
3052                 if ( !$found ) {
3053                         wfProfileIn( __METHOD__ . '-pfunc' );
3054
3055                         $colonPos = strpos( $part1, ':' );
3056                         if ( $colonPos !== false ) {
3057                                 # Case sensitive functions
3058                                 $function = substr( $part1, 0, $colonPos );
3059                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3060                                         $function = $this->mFunctionSynonyms[1][$function];
3061                                 } else {
3062                                         # Case insensitive functions
3063                                         $function = $wgContLang->lc( $function );
3064                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3065                                                 $function = $this->mFunctionSynonyms[0][$function];
3066                                         } else {
3067                                                 $function = false;
3068                                         }
3069                                 }
3070                                 if ( $function ) {
3071                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3072                                         $initialArgs = array( &$this );
3073                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3074                                         if ( $flags & SFH_OBJECT_ARGS ) {
3075                                                 # Add a frame parameter, and pass the arguments as an array
3076                                                 $allArgs = $initialArgs;
3077                                                 $allArgs[] = $frame;
3078                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3079                                                         $funcArgs[] = $args->item( $i );
3080                                                 }
3081                                                 $allArgs[] = $funcArgs;
3082                                         } else {
3083                                                 # Convert arguments to plain text
3084                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3085                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3086                                                 }
3087                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3088                                         }
3089
3090                                         # Workaround for PHP bug 35229 and similar
3091                                         if ( !is_callable( $callback ) ) {
3092                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3093                                                 wfProfileOut( __METHOD__ );
3094                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3095                                         }
3096                                         $result = call_user_func_array( $callback, $allArgs );
3097                                         $found = true;
3098                                         $noparse = true;
3099                                         $preprocessFlags = 0;
3100
3101                                         if ( is_array( $result ) ) {
3102                                                 if ( isset( $result[0] ) ) {
3103                                                         $text = $result[0];
3104                                                         unset( $result[0] );
3105                                                 }
3106
3107                                                 # Extract flags into the local scope
3108                                                 # This allows callers to set flags such as nowiki, found, etc.
3109                                                 extract( $result );
3110                                         } else {
3111                                                 $text = $result;
3112                                         }
3113                                         if ( !$noparse ) {
3114                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3115                                                 $isChildObj = true;
3116                                         }
3117                                 }
3118                         }
3119                         wfProfileOut( __METHOD__ . '-pfunc' );
3120                 }
3121
3122                 # Finish mangling title and then check for loops.
3123                 # Set $title to a Title object and $titleText to the PDBK
3124                 if ( !$found ) {
3125                         $ns = NS_TEMPLATE;
3126                         # Split the title into page and subpage
3127                         $subpage = '';
3128                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3129                         if ( $subpage !== '' ) {
3130                                 $ns = $this->mTitle->getNamespace();
3131                         }
3132                         $title = Title::newFromText( $part1, $ns );
3133                         if ( $title ) {
3134                                 $titleText = $title->getPrefixedText();
3135                                 # Check for language variants if the template is not found
3136                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3137                                         $wgContLang->findVariantLink( $part1, $title, true );
3138                                 }
3139                                 # Do recursion depth check
3140                                 $limit = $this->mOptions->getMaxTemplateDepth();
3141                                 if ( $frame->depth >= $limit ) {
3142                                         $found = true;
3143                                         $text = '<span class="error">'
3144                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3145                                                 . '</span>';
3146                                 }
3147                         }
3148                 }
3149
3150                 # Load from database
3151                 if ( !$found && $title ) {
3152                         wfProfileIn( __METHOD__ . '-loadtpl' );
3153                         if ( !$title->isExternal() ) {
3154                                 if ( $title->getNamespace() == NS_SPECIAL
3155                                         && $this->mOptions->getAllowSpecialInclusion()
3156                                         && $this->ot['html'] )
3157                                 {
3158                                         $text = SpecialPage::capturePath( $title );
3159                                         if ( is_string( $text ) ) {
3160                                                 $found = true;
3161                                                 $isHTML = true;
3162                                                 $this->disableCache();
3163                                         }
3164                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3165                                         $found = false; # access denied
3166                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3167                                 } else {
3168                                         list( $text, $title ) = $this->getTemplateDom( $title );
3169                                         if ( $text !== false ) {
3170                                                 $found = true;
3171                                                 $isChildObj = true;
3172                                         }
3173                                 }
3174
3175                                 # If the title is valid but undisplayable, make a link to it
3176                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3177                                         $text = "[[:$titleText]]";
3178                                         $found = true;
3179                                 }
3180                         } elseif ( $title->isTrans() ) {
3181                                 # Interwiki transclusion
3182                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3183                                         $text = $this->interwikiTransclude( $title, 'render' );
3184                                         $isHTML = true;
3185                                 } else {
3186                                         $text = $this->interwikiTransclude( $title, 'raw' );
3187                                         # Preprocess it like a template
3188                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3189                                         $isChildObj = true;
3190                                 }
3191                                 $found = true;
3192                         }
3193
3194                         # Do infinite loop check
3195                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3196                         if ( !$frame->loopCheck( $title ) ) {
3197                                 $found = true;
3198                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3199                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3200                         }
3201                         wfProfileOut( __METHOD__ . '-loadtpl' );
3202                 }
3203
3204                 # If we haven't found text to substitute by now, we're done
3205                 # Recover the source wikitext and return it
3206                 if ( !$found ) {
3207                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3208                         wfProfileOut( __METHOD__ );
3209                         return array( 'object' => $text );
3210                 }
3211
3212                 # Expand DOM-style return values in a child frame
3213                 if ( $isChildObj ) {
3214                         # Clean up argument array
3215                         $newFrame = $frame->newChild( $args, $title );
3216
3217                         if ( $nowiki ) {
3218                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3219                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3220                                 # Expansion is eligible for the empty-frame cache
3221                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3222                                         $text = $this->mTplExpandCache[$titleText];
3223                                 } else {
3224                                         $text = $newFrame->expand( $text );
3225                                         $this->mTplExpandCache[$titleText] = $text;
3226                                 }
3227                         } else {
3228                                 # Uncached expansion
3229                                 $text = $newFrame->expand( $text );
3230                         }
3231                 }
3232                 if ( $isLocalObj && $nowiki ) {
3233                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3234                         $isLocalObj = false;
3235                 }
3236
3237                 # Replace raw HTML by a placeholder
3238                 # Add a blank line preceding, to prevent it from mucking up
3239                 # immediately preceding headings
3240                 if ( $isHTML ) {
3241                         $text = "\n\n" . $this->insertStripItem( $text );
3242                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3243                         # Escape nowiki-style return values
3244                         $text = wfEscapeWikiText( $text );
3245                 } elseif ( is_string( $text )
3246                         && !$piece['lineStart']
3247                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3248                 {
3249                         # Bug 529: if the template begins with a table or block-level
3250                         # element, it should be treated as beginning a new line.
3251                         # This behaviour is somewhat controversial.
3252                         $text = "\n" . $text;
3253                 }
3254
3255                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3256                         # Error, oversize inclusion
3257                         if ( $titleText !== false ) {
3258                                 # Make a working, properly escaped link if possible (bug 23588)
3259                                 $text = "[[:$titleText]]";
3260                         } else {
3261                                 # This will probably not be a working link, but at least it may
3262                                 # provide some hint of where the problem is
3263                                 preg_replace( '/^:/', '', $originalTitle );
3264                                 $text = "[[:$originalTitle]]";
3265                         }
3266                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3267                         $this->limitationWarn( 'post-expand-template-inclusion' );
3268                 }
3269
3270                 if ( $isLocalObj ) {
3271                         $ret = array( 'object' => $text );
3272                 } else {
3273                         $ret = array( 'text' => $text );
3274                 }
3275
3276                 wfProfileOut( __METHOD__ );
3277                 return $ret;
3278         }
3279
3280         /**
3281          * Get the semi-parsed DOM representation of a template with a given title,
3282          * and its redirect destination title. Cached.
3283          */
3284         function getTemplateDom( $title ) {
3285                 $cacheTitle = $title;
3286                 $titleText = $title->getPrefixedDBkey();
3287
3288                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3289                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3290                         $title = Title::makeTitle( $ns, $dbk );
3291                         $titleText = $title->getPrefixedDBkey();
3292                 }
3293                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3294                         return array( $this->mTplDomCache[$titleText], $title );
3295                 }
3296
3297                 # Cache miss, go to the database
3298                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3299
3300                 if ( $text === false ) {
3301                         $this->mTplDomCache[$titleText] = false;
3302                         return array( false, $title );
3303                 }
3304
3305                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3306                 $this->mTplDomCache[ $titleText ] = $dom;
3307
3308                 if ( !$title->equals( $cacheTitle ) ) {
3309                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3310                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3311                 }
3312
3313                 return array( $dom, $title );
3314         }
3315
3316         /**
3317          * Fetch the unparsed text of a template and register a reference to it.
3318          */
3319         function fetchTemplateAndTitle( $title ) {
3320                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3321                 $stuff = call_user_func( $templateCb, $title, $this );
3322                 $text = $stuff['text'];
3323                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3324                 if ( isset( $stuff['deps'] ) ) {
3325                         foreach ( $stuff['deps'] as $dep ) {
3326                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3327                         }
3328                 }
3329                 return array( $text, $finalTitle );
3330         }
3331
3332         function fetchTemplate( $title ) {
3333                 $rv = $this->fetchTemplateAndTitle( $title );
3334                 return $rv[0];
3335         }
3336
3337         /**
3338          * Static function to get a template
3339          * Can be overridden via ParserOptions::setTemplateCallback().
3340          */
3341         static function statelessFetchTemplate( $title, $parser=false ) {
3342                 $text = $skip = false;
3343                 $finalTitle = $title;
3344                 $deps = array();
3345
3346                 # Loop to fetch the article, with up to 1 redirect
3347                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3348                         # Give extensions a chance to select the revision instead
3349                         $id = false; # Assume current
3350                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3351
3352                         if ( $skip ) {
3353                                 $text = false;
3354                                 $deps[] = array(
3355                                         'title' => $title,
3356                                         'page_id' => $title->getArticleID(),
3357                                         'rev_id' => null );
3358                                 break;
3359                         }
3360                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3361                         $rev_id = $rev ? $rev->getId() : 0;
3362                         # If there is no current revision, there is no page
3363                         if ( $id === false && !$rev ) {
3364                                 $linkCache = LinkCache::singleton();
3365                                 $linkCache->addBadLinkObj( $title );
3366                         }
3367
3368                         $deps[] = array(
3369                                 'title' => $title,
3370                                 'page_id' => $title->getArticleID(),
3371                                 'rev_id' => $rev_id );
3372
3373                         if ( $rev ) {
3374                                 $text = $rev->getText();
3375                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3376                                 global $wgContLang;
3377                                 $message = $wgContLang->lcfirst( $title->getText() );
3378                                 $text = wfMsgForContentNoTrans( $message );
3379                                 if ( wfEmptyMsg( $message, $text ) ) {
3380                                         $text = false;
3381                                         break;
3382                                 }
3383                         } else {
3384                                 break;
3385                         }
3386                         if ( $text === false ) {
3387                                 break;
3388                         }
3389                         # Redirect?
3390                         $finalTitle = $title;
3391                         $title = Title::newFromRedirect( $text );
3392                 }
3393                 return array(
3394                         'text' => $text,
3395                         'finalTitle' => $finalTitle,
3396                         'deps' => $deps );
3397         }
3398
3399         /**
3400          * Transclude an interwiki link.
3401          */
3402         function interwikiTransclude( $title, $action ) {
3403                 global $wgEnableScaryTranscluding;
3404
3405                 if ( !$wgEnableScaryTranscluding ) {
3406                         return wfMsgForContent('scarytranscludedisabled');
3407                 }
3408
3409                 $url = $title->getFullUrl( "action=$action" );
3410
3411                 if ( strlen( $url ) > 255 ) {
3412                         return wfMsgForContent( 'scarytranscludetoolong' );
3413                 }
3414                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3415         }
3416
3417         function fetchScaryTemplateMaybeFromCache( $url ) {
3418                 global $wgTranscludeCacheExpiry;
3419                 $dbr = wfGetDB( DB_SLAVE );
3420                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3421                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3422                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3423                 if ( $obj ) {
3424                         return $obj->tc_contents;
3425                 }
3426
3427                 $text = Http::get( $url );
3428                 if ( !$text ) {
3429                         return wfMsgForContent( 'scarytranscludefailed', $url );
3430                 }
3431
3432                 $dbw = wfGetDB( DB_MASTER );
3433                 $dbw->replace( 'transcache', array('tc_url'), array(
3434                         'tc_url' => $url,
3435                         'tc_time' => $dbw->timestamp( time() ),
3436                         'tc_contents' => $text)
3437                 );
3438                 return $text;
3439         }
3440
3441
3442         /**
3443          * Triple brace replacement -- used for template arguments
3444          * @private
3445          */
3446         function argSubstitution( $piece, $frame ) {
3447                 wfProfileIn( __METHOD__ );
3448
3449                 $error = false;
3450                 $parts = $piece['parts'];
3451                 $nameWithSpaces = $frame->expand( $piece['title'] );
3452                 $argName = trim( $nameWithSpaces );
3453                 $object = false;
3454                 $text = $frame->getArgument( $argName );
3455                 if (  $text === false && $parts->getLength() > 0
3456                   && (
3457                     $this->ot['html']
3458                     || $this->ot['pre']
3459                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3460                   )
3461                 ) {
3462                         # No match in frame, use the supplied default
3463                         $object = $parts->item( 0 )->getChildren();
3464                 }
3465                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3466                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3467                         $this->limitationWarn( 'post-expand-template-argument' );
3468                 }
3469
3470                 if ( $text === false && $object === false ) {
3471                         # No match anywhere
3472                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3473                 }
3474                 if ( $error !== false ) {
3475                         $text .= $error;
3476                 }
3477                 if ( $object !== false ) {
3478                         $ret = array( 'object' => $object );
3479                 } else {
3480                         $ret = array( 'text' => $text );
3481                 }
3482
3483                 wfProfileOut( __METHOD__ );
3484                 return $ret;
3485         }
3486
3487         /**
3488          * Return the text to be used for a given extension tag.
3489          * This is the ghost of strip().
3490          *
3491          * @param $params Associative array of parameters:
3492          *     name       PPNode for the tag name
3493          *     attr       PPNode for unparsed text where tag attributes are thought to be
3494          *     attributes Optional associative array of parsed attributes
3495          *     inner      Contents of extension element
3496          *     noClose    Original text did not have a close tag
3497          * @param $frame PPFrame
3498          */
3499         function extensionSubstitution( $params, $frame ) {
3500                 $name = $frame->expand( $params['name'] );
3501                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3502                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3503                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3504
3505                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3506                         ( $this->ot['html'] || $this->ot['pre'] );
3507                 if ( $isFunctionTag ) {
3508                         $markerType = 'none';
3509                 } else {
3510                         $markerType = 'general';
3511                 }
3512                 if ( $this->ot['html'] || $isFunctionTag ) {
3513                         $name = strtolower( $name );
3514                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3515                         if ( isset( $params['attributes'] ) ) {
3516                                 $attributes = $attributes + $params['attributes'];
3517                         }
3518
3519                         if ( isset( $this->mTagHooks[$name] ) ) {
3520                                 # Workaround for PHP bug 35229 and similar
3521                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3522                                         throw new MWException( "Tag hook for $name is not callable\n" );
3523                                 }
3524                                 $output = call_user_func_array( $this->mTagHooks[$name],
3525                                         array( $content, $attributes, $this, $frame ) );
3526                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3527                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3528                                 if ( !is_callable( $callback ) ) {
3529                                         throw new MWException( "Tag hook for $name is not callable\n" );
3530                                 }
3531
3532                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3533                         } else {
3534                                 $output = '<span class="error">Invalid tag extension name: ' .
3535                                         htmlspecialchars( $name ) . '</span>';
3536                         }
3537
3538                         if ( is_array( $output ) ) {
3539                                 # Extract flags to local scope (to override $markerType)
3540                                 $flags = $output;
3541                                 $output = $flags[0];
3542                                 unset( $flags[0] );
3543                                 extract( $flags );
3544                         }
3545                 } else {
3546                         if ( is_null( $attrText ) ) {
3547                                 $attrText = '';
3548                         }
3549                         if ( isset( $params['attributes'] ) ) {
3550                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3551                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3552                                                 htmlspecialchars( $attrValue ) . '"';
3553                                 }
3554                         }
3555                         if ( $content === null ) {
3556                                 $output = "<$name$attrText/>";
3557                         } else {
3558                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3559                                 $output = "<$name$attrText>$content$close";
3560                         }
3561                 }
3562
3563                 if ( $markerType === 'none' ) {
3564                         return $output;
3565                 } elseif ( $markerType === 'nowiki' ) {
3566                         $this->mStripState->nowiki->setPair( $marker, $output );
3567                 } elseif ( $markerType === 'general' ) {
3568                         $this->mStripState->general->setPair( $marker, $output );
3569                 } else {
3570                         throw new MWException( __METHOD__.': invalid marker type' );
3571                 }
3572                 return $marker;
3573         }
3574
3575         /**
3576          * Increment an include size counter
3577          *
3578          * @param $type String: the type of expansion
3579          * @param $size Integer: the size of the text
3580          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3581          */
3582         function incrementIncludeSize( $type, $size ) {
3583                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3584                         return false;
3585                 } else {
3586                         $this->mIncludeSizes[$type] += $size;
3587                         return true;
3588                 }
3589         }
3590
3591         /**
3592          * Increment the expensive function count
3593          *
3594          * @return Boolean: false if the limit has been exceeded
3595          */
3596         function incrementExpensiveFunctionCount() {
3597                 global $wgExpensiveParserFunctionLimit;
3598                 $this->mExpensiveFunctionCount++;
3599                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3600                         return true;
3601                 }
3602                 return false;
3603         }
3604
3605         /**
3606          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3607          * Fills $this->mDoubleUnderscores, returns the modified text
3608          */
3609         function doDoubleUnderscore( $text ) {
3610                 wfProfileIn( __METHOD__ );
3611
3612                 # The position of __TOC__ needs to be recorded
3613                 $mw = MagicWord::get( 'toc' );
3614                 if ( $mw->match( $text ) ) {
3615                         $this->mShowToc = true;
3616                         $this->mForceTocPosition = true;
3617
3618                         # Set a placeholder. At the end we'll fill it in with the TOC.
3619                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3620
3621                         # Only keep the first one.
3622                         $text = $mw->replace( '', $text );
3623                 }
3624
3625                 # Now match and remove the rest of them
3626                 $mwa = MagicWord::getDoubleUnderscoreArray();
3627                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3628
3629                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3630                         $this->mOutput->mNoGallery = true;
3631                 }
3632                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3633                         $this->mShowToc = false;
3634                 }
3635                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3636                         $this->addTrackingCategory( 'hidden-category-category' );
3637                 }
3638                 # (bug 8068) Allow control over whether robots index a page.
3639                 #
3640                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3641                 # is not desirable, the last one on the page should win.
3642                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3643                         $this->mOutput->setIndexPolicy( 'noindex' );
3644                         $this->addTrackingCategory( 'noindex-category' );
3645                 }
3646                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3647                         $this->mOutput->setIndexPolicy( 'index' );
3648                         $this->addTrackingCategory( 'index-category' );
3649                 }
3650
3651                 # Cache all double underscores in the database
3652                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3653                         $this->mOutput->setProperty( $key, '' );
3654                 }
3655
3656                 wfProfileOut( __METHOD__ );
3657                 return $text;
3658         }
3659
3660         /**
3661          * Add a tracking category, getting the title from a system message,
3662          * or print a debug message if the title is invalid.
3663          *
3664          * @param $msg String: message key
3665          * @return Boolean: whether the addition was successful
3666          */
3667         protected function addTrackingCategory( $msg ) {
3668                 $cat = wfMsgForContent( $msg );
3669
3670                 # Allow tracking categories to be disabled by setting them to "-"
3671                 if ( $cat === '-' ) {
3672                         return false;
3673                 }
3674
3675                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3676                 if ( $containerCategory ) {
3677                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3678                         return true;
3679                 } else {
3680                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3681                         return false;
3682                 }
3683         }
3684
3685         /**
3686          * This function accomplishes several tasks:
3687          * 1) Auto-number headings if that option is enabled
3688          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3689          * 3) Add a Table of contents on the top for users who have enabled the option
3690          * 4) Auto-anchor headings
3691          *
3692          * It loops through all headlines, collects the necessary data, then splits up the
3693          * string and re-inserts the newly formatted headlines.
3694          *
3695          * @param $text String
3696          * @param $origText String: original, untouched wikitext
3697          * @param $isMain Boolean
3698          * @private
3699          */
3700         function formatHeadings( $text, $origText, $isMain=true ) {
3701                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3702
3703                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3704
3705                 # Inhibit editsection links if requested in the page
3706                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3707                         $showEditLink = 0;
3708                 } else {
3709                         $showEditLink = $this->mOptions->getEditSection();
3710                 }
3711
3712                 # Get all headlines for numbering them and adding funky stuff like [edit]
3713                 # links - this is for later, but we need the number of headlines right now
3714                 $matches = array();
3715                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3716
3717                 # if there are fewer than 4 headlines in the article, do not show TOC
3718                 # unless it's been explicitly enabled.
3719                 $enoughToc = $this->mShowToc &&
3720                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3721
3722                 # Allow user to stipulate that a page should have a "new section"
3723                 # link added via __NEWSECTIONLINK__
3724                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3725                         $this->mOutput->setNewSection( true );
3726                 }
3727
3728                 # Allow user to remove the "new section"
3729                 # link via __NONEWSECTIONLINK__
3730                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3731                         $this->mOutput->hideNewSection( true );
3732                 }
3733
3734                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3735                 # override above conditions and always show TOC above first header
3736                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3737                         $this->mShowToc = true;
3738                         $enoughToc = true;
3739                 }
3740
3741                 # We need this to perform operations on the HTML
3742                 $sk = $this->mOptions->getSkin( $this->mTitle );
3743
3744                 # headline counter
3745                 $headlineCount = 0;
3746                 $numVisible = 0;
3747
3748                 # Ugh .. the TOC should have neat indentation levels which can be
3749                 # passed to the skin functions. These are determined here
3750                 $toc = '';
3751                 $full = '';
3752                 $head = array();
3753                 $sublevelCount = array();
3754                 $levelCount = array();
3755                 $toclevel = 0;
3756                 $level = 0;
3757                 $prevlevel = 0;
3758                 $toclevel = 0;
3759                 $prevtoclevel = 0;
3760                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3761                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3762                 $oldType = $this->mOutputType;
3763                 $this->setOutputType( self::OT_WIKI );
3764                 $frame = $this->getPreprocessor()->newFrame();
3765                 $root = $this->preprocessToDom( $origText );
3766                 $node = $root->getFirstChild();
3767                 $byteOffset = 0;
3768                 $tocraw = array();
3769
3770                 foreach ( $matches[3] as $headline ) {
3771                         $isTemplate = false;
3772                         $titleText = false;
3773                         $sectionIndex = false;
3774                         $numbering = '';
3775                         $markerMatches = array();
3776                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3777                                 $serial = $markerMatches[1];
3778                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3779                                 $isTemplate = ( $titleText != $baseTitleText );
3780                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3781                         }
3782
3783                         if ( $toclevel ) {
3784                                 $prevlevel = $level;
3785                         }
3786                         $level = $matches[1][$headlineCount];
3787
3788                         if ( $level > $prevlevel ) {
3789                                 # Increase TOC level
3790                                 $toclevel++;
3791                                 $sublevelCount[$toclevel] = 0;
3792                                 if ( $toclevel<$wgMaxTocLevel ) {
3793                                         $prevtoclevel = $toclevel;
3794                                         $toc .= $sk->tocIndent();
3795                                         $numVisible++;
3796                                 }
3797                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3798                                 # Decrease TOC level, find level to jump to
3799
3800                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3801                                         if ( $levelCount[$i] == $level ) {
3802                                                 # Found last matching level
3803                                                 $toclevel = $i;
3804                                                 break;
3805                                         } elseif ( $levelCount[$i] < $level ) {
3806                                                 # Found first matching level below current level
3807                                                 $toclevel = $i + 1;
3808                                                 break;
3809                                         }
3810                                 }
3811                                 if ( $i == 0 ) {
3812                                         $toclevel = 1;
3813                                 }
3814                                 if ( $toclevel<$wgMaxTocLevel ) {
3815                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3816                                                 # Unindent only if the previous toc level was shown :p
3817                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3818                                                 $prevtoclevel = $toclevel;
3819                                         } else {
3820                                                 $toc .= $sk->tocLineEnd();
3821                                         }
3822                                 }
3823                         } else {
3824                                 # No change in level, end TOC line
3825                                 if ( $toclevel<$wgMaxTocLevel ) {
3826                                         $toc .= $sk->tocLineEnd();
3827                                 }
3828                         }
3829
3830                         $levelCount[$toclevel] = $level;
3831
3832                         # count number of headlines for each level
3833                         @$sublevelCount[$toclevel]++;
3834                         $dot = 0;
3835                         for( $i = 1; $i <= $toclevel; $i++ ) {
3836                                 if ( !empty( $sublevelCount[$i] ) ) {
3837                                         if ( $dot ) {
3838                                                 $numbering .= '.';
3839                                         }
3840                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3841                                         $dot = 1;
3842                                 }
3843                         }
3844
3845                         # The safe header is a version of the header text safe to use for links
3846                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3847                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3848
3849                         # Remove link placeholders by the link text.
3850                         #     <!--LINK number-->
3851                         # turns into
3852                         #     link text with suffix
3853                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3854
3855                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3856                         $tocline = preg_replace(
3857                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3858                                 array( '',                          '<$1>' ),
3859                                 $safeHeadline
3860                         );
3861                         $tocline = trim( $tocline );
3862
3863                         # For the anchor, strip out HTML-y stuff period
3864                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3865                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
3866
3867                         # Save headline for section edit hint before it's escaped
3868                         $headlineHint = $safeHeadline;
3869
3870                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3871                                 # For reverse compatibility, provide an id that's
3872                                 # HTML4-compatible, like we used to.
3873                                 #
3874                                 # It may be worth noting, academically, that it's possible for
3875                                 # the legacy anchor to conflict with a non-legacy headline
3876                                 # anchor on the page.  In this case likely the "correct" thing
3877                                 # would be to either drop the legacy anchors or make sure
3878                                 # they're numbered first.  However, this would require people
3879                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3880                                 # manually, so let's not bother worrying about it.
3881                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3882                                         array( 'noninitial', 'legacy' ) );
3883                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3884
3885                                 if ( $legacyHeadline == $safeHeadline ) {
3886                                         # No reason to have both (in fact, we can't)
3887                                         $legacyHeadline = false;
3888                                 }
3889                         } else {
3890                                 $legacyHeadline = false;
3891                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3892                                         'noninitial' );
3893                         }
3894
3895                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3896                         # Does this apply to Unicode characters?  Because we aren't
3897                         # handling those here.
3898                         $arrayKey = strtolower( $safeHeadline );
3899                         if ( $legacyHeadline === false ) {
3900                                 $legacyArrayKey = false;
3901                         } else {
3902                                 $legacyArrayKey = strtolower( $legacyHeadline );
3903                         }
3904
3905                         # count how many in assoc. array so we can track dupes in anchors
3906                         if ( isset( $refers[$arrayKey] ) ) {
3907                                 $refers[$arrayKey]++;
3908                         } else {
3909                                 $refers[$arrayKey] = 1;
3910                         }
3911                         if ( isset( $refers[$legacyArrayKey] ) ) {
3912                                 $refers[$legacyArrayKey]++;
3913                         } else {
3914                                 $refers[$legacyArrayKey] = 1;
3915                         }
3916
3917                         # Don't number the heading if it is the only one (looks silly)
3918                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3919                                 # the two are different if the line contains a link
3920                                 $headline = $numbering . ' ' . $headline;
3921                         }
3922
3923                         # Create the anchor for linking from the TOC to the section
3924                         $anchor = $safeHeadline;
3925                         $legacyAnchor = $legacyHeadline;
3926                         if ( $refers[$arrayKey] > 1 ) {
3927                                 $anchor .= '_' . $refers[$arrayKey];
3928                         }
3929                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3930                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3931                         }
3932                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3933                                 $toc .= $sk->tocLine( $anchor, $tocline,
3934                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3935                         }
3936
3937                         # Add the section to the section tree
3938                         # Find the DOM node for this header
3939                         while ( $node && !$isTemplate ) {
3940                                 if ( $node->getName() === 'h' ) {
3941                                         $bits = $node->splitHeading();
3942                                         if ( $bits['i'] == $sectionIndex )
3943                                                 break;
3944                                 }
3945                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3946                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3947                                 $node = $node->getNextSibling();
3948                         }
3949                         $tocraw[] = array(
3950                                 'toclevel' => $toclevel,
3951                                 'level' => $level,
3952                                 'line' => $tocline,
3953                                 'number' => $numbering,
3954                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3955                                 'fromtitle' => $titleText,
3956                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3957                                 'anchor' => $anchor,
3958                         );
3959
3960                         # give headline the correct <h#> tag
3961                         if ( $showEditLink && $sectionIndex !== false ) {
3962                                 if ( $isTemplate ) {
3963                                         # Put a T flag in the section identifier, to indicate to extractSections()
3964                                         # that sections inside <includeonly> should be counted.
3965                                         $editlink = $sk->doEditSectionLink( Title::newFromText( $titleText ), "T-$sectionIndex", null, $this->mOptions->getUserLang() );
3966                                 } else {
3967                                         $editlink = $sk->doEditSectionLink( $this->mTitle, $sectionIndex, $headlineHint, $this->mOptions->getUserLang() );
3968                                 }
3969                         } else {
3970                                 $editlink = '';
3971                         }
3972                         $head[$headlineCount] = $sk->makeHeadline( $level,
3973                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3974                                 $editlink, $legacyAnchor );
3975
3976                         $headlineCount++;
3977                 }
3978
3979                 $this->setOutputType( $oldType );
3980
3981                 # Never ever show TOC if no headers
3982                 if ( $numVisible < 1 ) {
3983                         $enoughToc = false;
3984                 }
3985
3986                 if ( $enoughToc ) {
3987                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3988                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3989                         }
3990                         $toc = $sk->tocList( $toc );
3991                         $this->mOutput->setTOCHTML( $toc );
3992                 }
3993
3994                 if ( $isMain ) {
3995                         $this->mOutput->setSections( $tocraw );
3996                 }
3997
3998                 # split up and insert constructed headlines
3999
4000                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4001                 $i = 0;
4002
4003                 foreach ( $blocks as $block ) {
4004                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4005                                 # This is the [edit] link that appears for the top block of text when
4006                                 # section editing is enabled
4007
4008                                 # Disabled because it broke block formatting
4009                                 # For example, a bullet point in the top line
4010                                 # $full .= $sk->editSectionLink(0);
4011                         }
4012                         $full .= $block;
4013                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4014                                 # Top anchor now in skin
4015                                 $full = $full.$toc;
4016                         }
4017
4018                         if ( !empty( $head[$i] ) ) {
4019                                 $full .= $head[$i];
4020                         }
4021                         $i++;
4022                 }
4023                 if ( $this->mForceTocPosition ) {
4024                         return str_replace( '<!--MWTOC-->', $toc, $full );
4025                 } else {
4026                         return $full;
4027                 }
4028         }
4029
4030         /**
4031          * Transform wiki markup when saving a page by doing \r\n -> \n
4032          * conversion, substitting signatures, {{subst:}} templates, etc.
4033          *
4034          * @param $text String: the text to transform
4035          * @param &$title Title: the Title object for the current article
4036          * @param $user User: the User object describing the current user
4037          * @param $options ParserOptions: parsing options
4038          * @param $clearState Boolean: whether to clear the parser state first
4039          * @return String: the altered wiki markup
4040          */
4041         public function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
4042                 $options->resetUsage();
4043                 $this->mOptions = $options;
4044                 $this->setTitle( $title );
4045                 $this->setOutputType( self::OT_WIKI );
4046
4047                 if ( $clearState ) {
4048                         $this->clearState();
4049                 }
4050
4051                 $pairs = array(
4052                         "\r\n" => "\n",
4053                 );
4054                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4055                 $text = $this->pstPass2( $text, $user );
4056                 $text = $this->mStripState->unstripBoth( $text );
4057                 return $text;
4058         }
4059
4060         /**
4061          * Pre-save transform helper function
4062          * @private
4063          */
4064         function pstPass2( $text, $user ) {
4065                 global $wgContLang, $wgLocaltimezone;
4066
4067                 # Note: This is the timestamp saved as hardcoded wikitext to
4068                 # the database, we use $wgContLang here in order to give
4069                 # everyone the same signature and use the default one rather
4070                 # than the one selected in each user's preferences.
4071                 # (see also bug 12815)
4072                 $ts = $this->mOptions->getTimestamp();
4073                 if ( isset( $wgLocaltimezone ) ) {
4074                         $tz = $wgLocaltimezone;
4075                 } else {
4076                         $tz = date_default_timezone_get();
4077                 }
4078
4079                 $unixts = wfTimestamp( TS_UNIX, $ts );
4080                 $oldtz = date_default_timezone_get();
4081                 date_default_timezone_set( $tz );
4082                 $ts = date( 'YmdHis', $unixts );
4083                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4084
4085                 # Allow translation of timezones through wiki. date() can return
4086                 # whatever crap the system uses, localised or not, so we cannot
4087                 # ship premade translations.
4088                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4089                 $value = wfMsgForContent( $key );
4090                 if ( !wfEmptyMsg( $key, $value ) ) {
4091                         $tzMsg = $value;
4092                 }
4093
4094                 date_default_timezone_set( $oldtz );
4095
4096                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4097
4098                 # Variable replacement
4099                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4100                 $text = $this->replaceVariables( $text );
4101
4102                 # Signatures
4103                 $sigText = $this->getUserSig( $user );
4104                 $text = strtr( $text, array(
4105                         '~~~~~' => $d,
4106                         '~~~~' => "$sigText $d",
4107                         '~~~' => $sigText
4108                 ) );
4109
4110                 # Context links: [[|name]] and [[name (context)|]]
4111                 global $wgLegalTitleChars;
4112                 $tc = "[$wgLegalTitleChars]";
4113                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4114
4115                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4116                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4117                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4118                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4119
4120                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4121                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4122                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4123                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4124
4125                 $t = $this->mTitle->getText();
4126                 $m = array();
4127                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4128                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4129                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4130                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4131                 } else {
4132                         # if there's no context, don't bother duplicating the title
4133                         $text = preg_replace( $p2, '[[\\1]]', $text );
4134                 }
4135
4136                 # Trim trailing whitespace
4137                 $text = rtrim( $text );
4138
4139                 return $text;
4140         }
4141
4142         /**
4143          * Fetch the user's signature text, if any, and normalize to
4144          * validated, ready-to-insert wikitext.
4145          * If you have pre-fetched the nickname or the fancySig option, you can
4146          * specify them here to save a database query.
4147          *
4148          * @param $user User
4149          * @param $nickname String: nickname to use or false to use user's default nickname
4150          * @param $fancySig Boolean: whether the nicknname is the complete signature
4151          *                  or null to use default value
4152          * @return string
4153          */
4154         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4155                 global $wgMaxSigChars;
4156
4157                 $username = $user->getName();
4158
4159                 # If not given, retrieve from the user object.
4160                 if ( $nickname === false )
4161                         $nickname = $user->getOption( 'nickname' );
4162
4163                 if ( is_null( $fancySig ) ) {
4164                         $fancySig = $user->getBoolOption( 'fancysig' );
4165                 }
4166
4167                 $nickname = $nickname == null ? $username : $nickname;
4168
4169                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4170                         $nickname = $username;
4171                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4172                 } elseif ( $fancySig !== false ) {
4173                         # Sig. might contain markup; validate this
4174                         if ( $this->validateSig( $nickname ) !== false ) {
4175                                 # Validated; clean up (if needed) and return it
4176                                 return $this->cleanSig( $nickname, true );
4177                         } else {
4178                                 # Failed to validate; fall back to the default
4179                                 $nickname = $username;
4180                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4181                         }
4182                 }
4183
4184                 # Make sure nickname doesnt get a sig in a sig
4185                 $nickname = $this->cleanSigInSig( $nickname );
4186
4187                 # If we're still here, make it a link to the user page
4188                 $userText = wfEscapeWikiText( $username );
4189                 $nickText = wfEscapeWikiText( $nickname );
4190                 if ( $user->isAnon() )  {
4191                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4192                 } else {
4193                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4194                 }
4195         }
4196
4197         /**
4198          * Check that the user's signature contains no bad XML
4199          *
4200          * @param $text String
4201          * @return mixed An expanded string, or false if invalid.
4202          */
4203         function validateSig( $text ) {
4204                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4205         }
4206
4207         /**
4208          * Clean up signature text
4209          *
4210          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4211          * 2) Substitute all transclusions
4212          *
4213          * @param $text String
4214          * @param $parsing Whether we're cleaning (preferences save) or parsing
4215          * @return String: signature text
4216          */
4217         function cleanSig( $text, $parsing = false ) {
4218                 if ( !$parsing ) {
4219                         global $wgTitle;
4220                         $this->clearState();
4221                         $this->setTitle( $wgTitle );
4222                         $this->mOptions = new ParserOptions;
4223                         $this->setOutputType = self::OT_PREPROCESS;
4224                 }
4225
4226                 # Option to disable this feature
4227                 if ( !$this->mOptions->getCleanSignatures() ) {
4228                         return $text;
4229                 }
4230
4231                 # FIXME: regex doesn't respect extension tags or nowiki
4232                 #  => Move this logic to braceSubstitution()
4233                 $substWord = MagicWord::get( 'subst' );
4234                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4235                 $substText = '{{' . $substWord->getSynonym( 0 );
4236
4237                 $text = preg_replace( $substRegex, $substText, $text );
4238                 $text = $this->cleanSigInSig( $text );
4239                 $dom = $this->preprocessToDom( $text );
4240                 $frame = $this->getPreprocessor()->newFrame();
4241                 $text = $frame->expand( $dom );
4242
4243                 if ( !$parsing ) {
4244                         $text = $this->mStripState->unstripBoth( $text );
4245                 }
4246
4247                 return $text;
4248         }
4249
4250         /**
4251          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4252          *
4253          * @param $text String
4254          * @return String: signature text with /~{3,5}/ removed
4255          */
4256         function cleanSigInSig( $text ) {
4257                 $text = preg_replace( '/~{3,5}/', '', $text );
4258                 return $text;
4259         }
4260
4261         /**
4262          * Set up some variables which are usually set up in parse()
4263          * so that an external function can call some class members with confidence
4264          */
4265         public function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4266                 $this->setTitle( $title );
4267                 $options->resetUsage();
4268                 $this->mOptions = $options;
4269                 $this->setOutputType( $outputType );
4270                 if ( $clearState ) {
4271                         $this->clearState();
4272                 }
4273         }
4274
4275         /**
4276          * Wrapper for preprocess()
4277          *
4278          * @param $text String: the text to preprocess
4279          * @param $options ParserOptions: options
4280          * @return String
4281          */
4282         public function transformMsg( $text, $options ) {
4283                 global $wgTitle;
4284                 static $executing = false;
4285
4286                 # Guard against infinite recursion
4287                 if ( $executing ) {
4288                         return $text;
4289                 }
4290                 $executing = true;
4291
4292                 wfProfileIn( __METHOD__ );
4293                 $text = $this->preprocess( $text, $wgTitle, $options );
4294
4295                 $executing = false;
4296                 wfProfileOut( __METHOD__ );
4297                 return $text;
4298         }
4299
4300         /**
4301          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4302          * The callback should have the following form:
4303          *    function myParserHook( $text, $params, $parser ) { ... }
4304          *
4305          * Transform and return $text. Use $parser for any required context, e.g. use
4306          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4307          *
4308          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4309          * @param $callback Mixed: the callback function (and object) to use for the tag
4310          * @return The old value of the mTagHooks array associated with the hook
4311          */
4312         public function setHook( $tag, $callback ) {
4313                 $tag = strtolower( $tag );
4314                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4315                 $this->mTagHooks[$tag] = $callback;
4316                 if ( !in_array( $tag, $this->mStripList ) ) {
4317                         $this->mStripList[] = $tag;
4318                 }
4319
4320                 return $oldVal;
4321         }
4322
4323         function setTransparentTagHook( $tag, $callback ) {
4324                 $tag = strtolower( $tag );
4325                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4326                 $this->mTransparentTagHooks[$tag] = $callback;
4327
4328                 return $oldVal;
4329         }
4330
4331         /**
4332          * Remove all tag hooks
4333          */
4334         function clearTagHooks() {
4335                 $this->mTagHooks = array();
4336                 $this->mStripList = $this->mDefaultStripList;
4337         }
4338
4339         /**
4340          * Create a function, e.g. {{sum:1|2|3}}
4341          * The callback function should have the form:
4342          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4343          *
4344          * Or with SFH_OBJECT_ARGS:
4345          *    function myParserFunction( $parser, $frame, $args ) { ... }
4346          *
4347          * The callback may either return the text result of the function, or an array with the text
4348          * in element 0, and a number of flags in the other elements. The names of the flags are
4349          * specified in the keys. Valid flags are:
4350          *   found                     The text returned is valid, stop processing the template. This
4351          *                             is on by default.
4352          *   nowiki                    Wiki markup in the return value should be escaped
4353          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4354          *
4355          * @param $id String: The magic word ID
4356          * @param $callback Mixed: the callback function (and object) to use
4357          * @param $flags Integer: a combination of the following flags:
4358          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4359          *
4360          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4361          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4362          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4363          *     the arguments, and to control the way they are expanded.
4364          *
4365          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4366          *     arguments, for instance:
4367          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4368          *
4369          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4370          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4371          *     working if/when this is changed.
4372          *
4373          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4374          *     expansion.
4375          *
4376          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4377          *     about the methods available in PPFrame and PPNode.
4378          *
4379          * @return The old callback function for this name, if any
4380          */
4381         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4382                 global $wgContLang;
4383
4384                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4385                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4386
4387                 # Add to function cache
4388                 $mw = MagicWord::get( $id );
4389                 if ( !$mw )
4390                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4391
4392                 $synonyms = $mw->getSynonyms();
4393                 $sensitive = intval( $mw->isCaseSensitive() );
4394
4395                 foreach ( $synonyms as $syn ) {
4396                         # Case
4397                         if ( !$sensitive ) {
4398                                 $syn = $wgContLang->lc( $syn );
4399                         }
4400                         # Add leading hash
4401                         if ( !( $flags & SFH_NO_HASH ) ) {
4402                                 $syn = '#' . $syn;
4403                         }
4404                         # Remove trailing colon
4405                         if ( substr( $syn, -1, 1 ) === ':' ) {
4406                                 $syn = substr( $syn, 0, -1 );
4407                         }
4408                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4409                 }
4410                 return $oldVal;
4411         }
4412
4413         /**
4414          * Get all registered function hook identifiers
4415          *
4416          * @return Array
4417          */
4418         function getFunctionHooks() {
4419                 return array_keys( $this->mFunctionHooks );
4420         }
4421
4422         /**
4423          * Create a tag function, e.g. <test>some stuff</test>.
4424          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4425          * Unlike parser functions, their content is not preprocessed.
4426          */
4427         function setFunctionTagHook( $tag, $callback, $flags ) {
4428                 $tag = strtolower( $tag );
4429                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4430                         $this->mFunctionTagHooks[$tag] : null;
4431                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4432
4433                 if ( !in_array( $tag, $this->mStripList ) ) {
4434                         $this->mStripList[] = $tag;
4435                 }
4436
4437                 return $old;
4438         }
4439
4440         /**
4441          * FIXME: update documentation. makeLinkObj() is deprecated.
4442          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4443          * Placeholders created in Skin::makeLinkObj()
4444          * Returns an array of link CSS classes, indexed by PDBK.
4445          */
4446         function replaceLinkHolders( &$text, $options = 0 ) {
4447                 return $this->mLinkHolders->replace( $text );
4448         }
4449
4450         /**
4451          * Replace <!--LINK--> link placeholders with plain text of links
4452          * (not HTML-formatted).
4453          *
4454          * @param $text String
4455          * @return String
4456          */
4457         function replaceLinkHoldersText( $text ) {
4458                 return $this->mLinkHolders->replaceText( $text );
4459         }
4460
4461         /**
4462          * Renders an image gallery from a text with one line per image.
4463          * text labels may be given by using |-style alternative text. E.g.
4464          *   Image:one.jpg|The number "1"
4465          *   Image:tree.jpg|A tree
4466          * given as text will return the HTML of a gallery with two images,
4467          * labeled 'The number "1"' and
4468          * 'A tree'.
4469          */
4470         function renderImageGallery( $text, $params ) {
4471                 $ig = new ImageGallery();
4472                 $ig->setContextTitle( $this->mTitle );
4473                 $ig->setShowBytes( false );
4474                 $ig->setShowFilename( false );
4475                 $ig->setParser( $this );
4476                 $ig->setHideBadImages();
4477                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4478                 $ig->useSkin( $this->mOptions->getSkin( $this->mTitle ) );
4479                 $ig->mRevisionId = $this->mRevisionId;
4480
4481                 if ( isset( $params['showfilename'] ) ) {
4482                         $ig->setShowFilename( true );
4483                 } else {
4484                         $ig->setShowFilename( false );
4485                 }
4486                 if ( isset( $params['caption'] ) ) {
4487                         $caption = $params['caption'];
4488                         $caption = htmlspecialchars( $caption );
4489                         $caption = $this->replaceInternalLinks( $caption );
4490                         $ig->setCaptionHtml( $caption );
4491                 }
4492                 if ( isset( $params['perrow'] ) ) {
4493                         $ig->setPerRow( $params['perrow'] );
4494                 }
4495                 if ( isset( $params['widths'] ) ) {
4496                         $ig->setWidths( $params['widths'] );
4497                 }
4498                 if ( isset( $params['heights'] ) ) {
4499                         $ig->setHeights( $params['heights'] );
4500                 }
4501
4502                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4503
4504                 $lines = StringUtils::explode( "\n", $text );
4505                 foreach ( $lines as $line ) {
4506                         # match lines like these:
4507                         # Image:someimage.jpg|This is some image
4508                         $matches = array();
4509                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4510                         # Skip empty lines
4511                         if ( count( $matches ) == 0 ) {
4512                                 continue;
4513                         }
4514
4515                         if ( strpos( $matches[0], '%' ) !== false ) {
4516                                 $matches[1] = urldecode( $matches[1] );
4517                         }
4518                         $tp = Title::newFromText( $matches[1] );
4519                         $nt =& $tp;
4520                         if ( is_null( $nt ) ) {
4521                                 # Bogus title. Ignore these so we don't bomb out later.
4522                                 continue;
4523                         }
4524                         if ( isset( $matches[3] ) ) {
4525                                 $label = $matches[3];
4526                         } else {
4527                                 $label = '';
4528                         }
4529
4530                         $html = $this->recursiveTagParse( trim( $label ) );
4531
4532                         $ig->add( $nt, $html );
4533
4534                         # Only add real images (bug #5586)
4535                         if ( $nt->getNamespace() == NS_FILE ) {
4536                                 $this->mOutput->addImage( $nt->getDBkey() );
4537                         }
4538                 }
4539                 return $ig->toHTML();
4540         }
4541
4542         function getImageParams( $handler ) {
4543                 if ( $handler ) {
4544                         $handlerClass = get_class( $handler );
4545                 } else {
4546                         $handlerClass = '';
4547                 }
4548                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4549                         # Initialise static lists
4550                         static $internalParamNames = array(
4551                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4552                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4553                                         'bottom', 'text-bottom' ),
4554                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4555                                         'upright', 'border', 'link', 'alt' ),
4556                         );
4557                         static $internalParamMap;
4558                         if ( !$internalParamMap ) {
4559                                 $internalParamMap = array();
4560                                 foreach ( $internalParamNames as $type => $names ) {
4561                                         foreach ( $names as $name ) {
4562                                                 $magicName = str_replace( '-', '_', "img_$name" );
4563                                                 $internalParamMap[$magicName] = array( $type, $name );
4564                                         }
4565                                 }
4566                         }
4567
4568                         # Add handler params
4569                         $paramMap = $internalParamMap;
4570                         if ( $handler ) {
4571                                 $handlerParamMap = $handler->getParamMap();
4572                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4573                                         $paramMap[$magic] = array( 'handler', $paramName );
4574                                 }
4575                         }
4576                         $this->mImageParams[$handlerClass] = $paramMap;
4577                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4578                 }
4579                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4580         }
4581
4582         /**
4583          * Parse image options text and use it to make an image
4584          *
4585          * @param $title Title
4586          * @param $options String
4587          * @param $holders LinkHolderArray
4588          */
4589         function makeImage( $title, $options, $holders = false ) {
4590                 # Check if the options text is of the form "options|alt text"
4591                 # Options are:
4592                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4593                 #  * left       no resizing, just left align. label is used for alt= only
4594                 #  * right      same, but right aligned
4595                 #  * none       same, but not aligned
4596                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4597                 #  * center     center the image
4598                 #  * frame      Keep original image size, no magnify-button.
4599                 #  * framed     Same as "frame"
4600                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4601                 #  * upright    reduce width for upright images, rounded to full __0 px
4602                 #  * border     draw a 1px border around the image
4603                 #  * alt        Text for HTML alt attribute (defaults to empty)
4604                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4605                 # vertical-align values (no % or length right now):
4606                 #  * baseline
4607                 #  * sub
4608                 #  * super
4609                 #  * top
4610                 #  * text-top
4611                 #  * middle
4612                 #  * bottom
4613                 #  * text-bottom
4614
4615                 $parts = StringUtils::explode( "|", $options );
4616                 $sk = $this->mOptions->getSkin( $this->mTitle );
4617
4618                 # Give extensions a chance to select the file revision for us
4619                 $skip = $time = $descQuery = false;
4620                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4621
4622                 if ( $skip ) {
4623                         return $sk->link( $title );
4624                 }
4625
4626                 # Get the file
4627                 $file = wfFindFile( $title, array( 'time' => $time ) );
4628                 # Get parameter map
4629                 $handler = $file ? $file->getHandler() : false;
4630
4631                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4632
4633                 # Process the input parameters
4634                 $caption = '';
4635                 $params = array( 'frame' => array(), 'handler' => array(),
4636                         'horizAlign' => array(), 'vertAlign' => array() );
4637                 foreach ( $parts as $part ) {
4638                         $part = trim( $part );
4639                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4640                         $validated = false;
4641                         if ( isset( $paramMap[$magicName] ) ) {
4642                                 list( $type, $paramName ) = $paramMap[$magicName];
4643
4644                                 # Special case; width and height come in one variable together
4645                                 if ( $type === 'handler' && $paramName === 'width' ) {
4646                                         $m = array();
4647                                         # (bug 13500) In both cases (width/height and width only),
4648                                         # permit trailing "px" for backward compatibility.
4649                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4650                                                 $width = intval( $m[1] );
4651                                                 $height = intval( $m[2] );
4652                                                 if ( $handler->validateParam( 'width', $width ) ) {
4653                                                         $params[$type]['width'] = $width;
4654                                                         $validated = true;
4655                                                 }
4656                                                 if ( $handler->validateParam( 'height', $height ) ) {
4657                                                         $params[$type]['height'] = $height;
4658                                                         $validated = true;
4659                                                 }
4660                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4661                                                 $width = intval( $value );
4662                                                 if ( $handler->validateParam( 'width', $width ) ) {
4663                                                         $params[$type]['width'] = $width;
4664                                                         $validated = true;
4665                                                 }
4666                                         } # else no validation -- bug 13436
4667                                 } else {
4668                                         if ( $type === 'handler' ) {
4669                                                 # Validate handler parameter
4670                                                 $validated = $handler->validateParam( $paramName, $value );
4671                                         } else {
4672                                                 # Validate internal parameters
4673                                                 switch( $paramName ) {
4674                                                 case 'manualthumb':
4675                                                 case 'alt':
4676                                                         # @todo Fixme: possibly check validity here for
4677                                                         # manualthumb? downstream behavior seems odd with
4678                                                         # missing manual thumbs.
4679                                                         $validated = true;
4680                                                         $value = $this->stripAltText( $value, $holders );
4681                                                         break;
4682                                                 case 'link':
4683                                                         $chars = self::EXT_LINK_URL_CLASS;
4684                                                         $prots = $this->mUrlProtocols;
4685                                                         if ( $value === '' ) {
4686                                                                 $paramName = 'no-link';
4687                                                                 $value = true;
4688                                                                 $validated = true;
4689                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4690                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4691                                                                         $paramName = 'link-url';
4692                                                                         $this->mOutput->addExternalLink( $value );
4693                                                                         $validated = true;
4694                                                                 }
4695                                                         } else {
4696                                                                 $linkTitle = Title::newFromText( $value );
4697                                                                 if ( $linkTitle ) {
4698                                                                         $paramName = 'link-title';
4699                                                                         $value = $linkTitle;
4700                                                                         $this->mOutput->addLink( $linkTitle );
4701                                                                         $validated = true;
4702                                                                 }
4703                                                         }
4704                                                         break;
4705                                                 default:
4706                                                         # Most other things appear to be empty or numeric...
4707                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4708                                                 }
4709                                         }
4710
4711                                         if ( $validated ) {
4712                                                 $params[$type][$paramName] = $value;
4713                                         }
4714                                 }
4715                         }
4716                         if ( !$validated ) {
4717                                 $caption = $part;
4718                         }
4719                 }
4720
4721                 # Process alignment parameters
4722                 if ( $params['horizAlign'] ) {
4723                         $params['frame']['align'] = key( $params['horizAlign'] );
4724                 }
4725                 if ( $params['vertAlign'] ) {
4726                         $params['frame']['valign'] = key( $params['vertAlign'] );
4727                 }
4728
4729                 $params['frame']['caption'] = $caption;
4730
4731                 # Will the image be presented in a frame, with the caption below?
4732                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4733                                  isset( $params['frame']['framed'] ) ||
4734                                  isset( $params['frame']['thumbnail'] ) ||
4735                                  isset( $params['frame']['manualthumb'] );
4736
4737                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4738                 # came to also set the caption, ordinary text after the image -- which
4739                 # makes no sense, because that just repeats the text multiple times in
4740                 # screen readers.  It *also* came to set the title attribute.
4741                 #
4742                 # Now that we have an alt attribute, we should not set the alt text to
4743                 # equal the caption: that's worse than useless, it just repeats the
4744                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4745                 # use the unnamed parameter for alt text as well, just for the time be-
4746                 # ing, if the unnamed param is set and the alt param is not.
4747                 #
4748                 # For the future, we need to figure out if we want to tweak this more,
4749                 # e.g., introducing a title= parameter for the title; ignoring the un-
4750                 # named parameter entirely for images without a caption; adding an ex-
4751                 # plicit caption= parameter and preserving the old magic unnamed para-
4752                 # meter for BC; ...
4753                 if ( $imageIsFramed ) { # Framed image
4754                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4755                                 # No caption or alt text, add the filename as the alt text so
4756                                 # that screen readers at least get some description of the image
4757                                 $params['frame']['alt'] = $title->getText();
4758                         }
4759                         # Do not set $params['frame']['title'] because tooltips don't make sense
4760                         # for framed images
4761                 } else { # Inline image
4762                         if ( !isset( $params['frame']['alt'] ) ) {
4763                                 # No alt text, use the "caption" for the alt text
4764                                 if ( $caption !== '') {
4765                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4766                                 } else {
4767                                         # No caption, fall back to using the filename for the
4768                                         # alt text
4769                                         $params['frame']['alt'] = $title->getText();
4770                                 }
4771                         }
4772                         # Use the "caption" for the tooltip text
4773                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4774                 }
4775
4776                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4777
4778                 # Linker does the rest
4779                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery, $this->mOptions->getThumbSize() );
4780
4781                 # Give the handler a chance to modify the parser object
4782                 if ( $handler ) {
4783                         $handler->parserTransformHook( $this, $file );
4784                 }
4785
4786                 return $ret;
4787         }
4788
4789         protected function stripAltText( $caption, $holders ) {
4790                 # Strip bad stuff out of the title (tooltip).  We can't just use
4791                 # replaceLinkHoldersText() here, because if this function is called
4792                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4793                 if ( $holders ) {
4794                         $tooltip = $holders->replaceText( $caption );
4795                 } else {
4796                         $tooltip = $this->replaceLinkHoldersText( $caption );
4797                 }
4798
4799                 # make sure there are no placeholders in thumbnail attributes
4800                 # that are later expanded to html- so expand them now and
4801                 # remove the tags
4802                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4803                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4804
4805                 return $tooltip;
4806         }
4807
4808         /**
4809          * Set a flag in the output object indicating that the content is dynamic and
4810          * shouldn't be cached.
4811          */
4812         function disableCache() {
4813                 wfDebug( "Parser output marked as uncacheable.\n" );
4814                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
4815                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
4816         }
4817
4818         /**
4819          * Callback from the Sanitizer for expanding items found in HTML attribute
4820          * values, so they can be safely tested and escaped.
4821          *
4822          * @param $text String
4823          * @param $frame PPFrame
4824          * @return String
4825          * @private
4826          */
4827         function attributeStripCallback( &$text, $frame = false ) {
4828                 $text = $this->replaceVariables( $text, $frame );
4829                 $text = $this->mStripState->unstripBoth( $text );
4830                 return $text;
4831         }
4832
4833         /**
4834          * Accessor
4835          */
4836         function getTags() {
4837                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4838         }
4839
4840         /**
4841          * Break wikitext input into sections, and either pull or replace
4842          * some particular section's text.
4843          *
4844          * External callers should use the getSection and replaceSection methods.
4845          *
4846          * @param $text String: Page wikitext
4847          * @param $section String: a section identifier string of the form:
4848          *   <flag1> - <flag2> - ... - <section number>
4849          *
4850          * Currently the only recognised flag is "T", which means the target section number
4851          * was derived during a template inclusion parse, in other words this is a template
4852          * section edit link. If no flags are given, it was an ordinary section edit link.
4853          * This flag is required to avoid a section numbering mismatch when a section is
4854          * enclosed by <includeonly> (bug 6563).
4855          *
4856          * The section number 0 pulls the text before the first heading; other numbers will
4857          * pull the given section along with its lower-level subsections. If the section is
4858          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4859          *
4860          * @param $mode String: one of "get" or "replace"
4861          * @param $newText String: replacement text for section data.
4862          * @return String: for "get", the extracted section text.
4863          *                 for "replace", the whole page with the section replaced.
4864          */
4865         private function extractSections( $text, $section, $mode, $newText='' ) {
4866                 global $wgTitle;
4867                 $this->clearState();
4868                 $this->setTitle( $wgTitle ); # not generally used but removes an ugly failure mode
4869                 $this->mOptions = new ParserOptions;
4870                 $this->setOutputType( self::OT_PLAIN );
4871                 $outText = '';
4872                 $frame = $this->getPreprocessor()->newFrame();
4873
4874                 # Process section extraction flags
4875                 $flags = 0;
4876                 $sectionParts = explode( '-', $section );
4877                 $sectionIndex = array_pop( $sectionParts );
4878                 foreach ( $sectionParts as $part ) {
4879                         if ( $part === 'T' ) {
4880                                 $flags |= self::PTD_FOR_INCLUSION;
4881                         }
4882                 }
4883                 # Preprocess the text
4884                 $root = $this->preprocessToDom( $text, $flags );
4885
4886                 # <h> nodes indicate section breaks
4887                 # They can only occur at the top level, so we can find them by iterating the root's children
4888                 $node = $root->getFirstChild();
4889
4890                 # Find the target section
4891                 if ( $sectionIndex == 0 ) {
4892                         # Section zero doesn't nest, level=big
4893                         $targetLevel = 1000;
4894                 } else {
4895                         while ( $node ) {
4896                                 if ( $node->getName() === 'h' ) {
4897                                         $bits = $node->splitHeading();
4898                                         if ( $bits['i'] == $sectionIndex ) {
4899                                                 $targetLevel = $bits['level'];
4900                                                 break;
4901                                         }
4902                                 }
4903                                 if ( $mode === 'replace' ) {
4904                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4905                                 }
4906                                 $node = $node->getNextSibling();
4907                         }
4908                 }
4909
4910                 if ( !$node ) {
4911                         # Not found
4912                         if ( $mode === 'get' ) {
4913                                 return $newText;
4914                         } else {
4915                                 return $text;
4916                         }
4917                 }
4918
4919                 # Find the end of the section, including nested sections
4920                 do {
4921                         if ( $node->getName() === 'h' ) {
4922                                 $bits = $node->splitHeading();
4923                                 $curLevel = $bits['level'];
4924                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4925                                         break;
4926                                 }
4927                         }
4928                         if ( $mode === 'get' ) {
4929                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4930                         }
4931                         $node = $node->getNextSibling();
4932                 } while ( $node );
4933
4934                 # Write out the remainder (in replace mode only)
4935                 if ( $mode === 'replace' ) {
4936                         # Output the replacement text
4937                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4938                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4939                         # Only add trailing whitespace if there is newText
4940                         if ( $newText != "" ) {
4941                                 $outText .= $newText . "\n\n";
4942                         }
4943
4944                         while ( $node ) {
4945                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4946                                 $node = $node->getNextSibling();
4947                         }
4948                 }
4949
4950                 if ( is_string( $outText ) ) {
4951                         # Re-insert stripped tags
4952                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4953                 }
4954
4955                 return $outText;
4956         }
4957
4958         /**
4959          * This function returns the text of a section, specified by a number ($section).
4960          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4961          * the first section before any such heading (section 0).
4962          *
4963          * If a section contains subsections, these are also returned.
4964          *
4965          * @param $text String: text to look in
4966          * @param $section String: section identifier
4967          * @param $deftext String: default to return if section is not found
4968          * @return string text of the requested section
4969          */
4970         public function getSection( $text, $section, $deftext='' ) {
4971                 return $this->extractSections( $text, $section, "get", $deftext );
4972         }
4973
4974         public function replaceSection( $oldtext, $section, $text ) {
4975                 return $this->extractSections( $oldtext, $section, "replace", $text );
4976         }
4977
4978         /**
4979          * Get the ID of the revision we are parsing
4980          *
4981          * @return Mixed: integer or null
4982          */
4983         function getRevisionId() {
4984                 return $this->mRevisionId;
4985         }
4986
4987         /**
4988          * Get the timestamp associated with the current revision, adjusted for
4989          * the default server-local timestamp
4990          */
4991         function getRevisionTimestamp() {
4992                 if ( is_null( $this->mRevisionTimestamp ) ) {
4993                         wfProfileIn( __METHOD__ );
4994                         global $wgContLang;
4995                         $dbr = wfGetDB( DB_SLAVE );
4996                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4997                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4998
4999                         # Normalize timestamp to internal MW format for timezone processing.
5000                         # This has the added side-effect of replacing a null value with
5001                         # the current time, which gives us more sensible behavior for
5002                         # previews.
5003                         $timestamp = wfTimestamp( TS_MW, $timestamp );
5004
5005                         # The cryptic '' timezone parameter tells to use the site-default
5006                         # timezone offset instead of the user settings.
5007                         #
5008                         # Since this value will be saved into the parser cache, served
5009                         # to other users, and potentially even used inside links and such,
5010                         # it needs to be consistent for all visitors.
5011                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5012
5013                         wfProfileOut( __METHOD__ );
5014                 }
5015                 return $this->mRevisionTimestamp;
5016         }
5017
5018         /**
5019          * Get the name of the user that edited the last revision
5020          *
5021          * @return String: user name
5022          */
5023         function getRevisionUser() {
5024                 # if this template is subst: the revision id will be blank,
5025                 # so just use the current user's name
5026                 if ( $this->mRevisionId ) {
5027                         $revision = Revision::newFromId( $this->mRevisionId );
5028                         $revuser = $revision->getUserText();
5029                 } else {
5030                         global $wgUser;
5031                         $revuser = $wgUser->getName();
5032                 }
5033                 return $revuser;
5034         }
5035
5036         /**
5037          * Mutator for $mDefaultSort
5038          *
5039          * @param $sort New value
5040          */
5041         public function setDefaultSort( $sort ) {
5042                 $this->mDefaultSort = $sort;
5043                 $this->mOutput->setProperty( 'defaultsort', $sort );
5044         }
5045
5046         /**
5047          * Accessor for $mDefaultSort
5048          * Will use the title/prefixed title if none is set
5049          *
5050          * @return string
5051          */
5052         public function getDefaultSort() {
5053                 if ( $this->mDefaultSort !== false ) {
5054                         return $this->mDefaultSort;
5055                 } else {
5056                         return $this->mTitle->getCategorySortkey();
5057                 }
5058         }
5059
5060         /**
5061          * Accessor for $mDefaultSort
5062          * Unlike getDefaultSort(), will return false if none is set
5063          *
5064          * @return string or false
5065          */
5066         public function getCustomDefaultSort() {
5067                 return $this->mDefaultSort;
5068         }
5069
5070         /**
5071          * Try to guess the section anchor name based on a wikitext fragment
5072          * presumably extracted from a heading, for example "Header" from
5073          * "== Header ==".
5074          */
5075         public function guessSectionNameFromWikiText( $text ) {
5076                 # Strip out wikitext links(they break the anchor)
5077                 $text = $this->stripSectionName( $text );
5078                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5079                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5080         }
5081
5082         /**
5083          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5084          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5085          * as something other than UTF-8 (apparently?), resulting in breakage.
5086          *
5087          * @param $text String: The section name
5088          * @return string An anchor
5089          */
5090         public function guessLegacySectionNameFromWikiText( $text ) {
5091                 # Strip out wikitext links(they break the anchor)
5092                 $text = $this->stripSectionName( $text );
5093                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5094                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5095         }
5096
5097         /**
5098          * Strips a text string of wikitext for use in a section anchor
5099          *
5100          * Accepts a text string and then removes all wikitext from the
5101          * string and leaves only the resultant text (i.e. the result of
5102          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5103          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5104          * to create valid section anchors by mimicing the output of the
5105          * parser when headings are parsed.
5106          *
5107          * @param $text String: text string to be stripped of wikitext
5108          * for use in a Section anchor
5109          * @return Filtered text string
5110          */
5111         public function stripSectionName( $text ) {
5112                 # Strip internal link markup
5113                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5114                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5115
5116                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5117                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5118                 # on how many empty links there are on the page - need to figure that out.
5119                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5120
5121                 # Parse wikitext quotes (italics & bold)
5122                 $text = $this->doQuotes( $text );
5123
5124                 # Strip HTML tags
5125                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5126                 return $text;
5127         }
5128
5129         /**
5130          * strip/replaceVariables/unstrip for preprocessor regression testing
5131          */
5132         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5133                 $this->clearState();
5134                 if ( !$title instanceof Title ) {
5135                         $title = Title::newFromText( $title );
5136                 }
5137                 $this->mTitle = $title;
5138                 $options->resetUsage();
5139                 $this->mOptions = $options;
5140                 $this->setOutputType( $outputType );
5141                 $text = $this->replaceVariables( $text );
5142                 $text = $this->mStripState->unstripBoth( $text );
5143                 $text = Sanitizer::removeHTMLtags( $text );
5144                 return $text;
5145         }
5146
5147         function testPst( $text, $title, $options ) {
5148                 global $wgUser;
5149                 if ( !$title instanceof Title ) {
5150                         $title = Title::newFromText( $title );
5151                 }
5152                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5153         }
5154
5155         function testPreprocess( $text, $title, $options ) {
5156                 if ( !$title instanceof Title ) {
5157                         $title = Title::newFromText( $title );
5158                 }
5159                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5160         }
5161
5162         function markerSkipCallback( $s, $callback ) {
5163                 $i = 0;
5164                 $out = '';
5165                 while ( $i < strlen( $s ) ) {
5166                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5167                         if ( $markerStart === false ) {
5168                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5169                                 break;
5170                         } else {
5171                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5172                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5173                                 if ( $markerEnd === false ) {
5174                                         $out .= substr( $s, $markerStart );
5175                                         break;
5176                                 } else {
5177                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5178                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5179                                         $i = $markerEnd;
5180                                 }
5181                         }
5182                 }
5183                 return $out;
5184         }
5185
5186         function serialiseHalfParsedText( $text ) {
5187                 $data = array();
5188                 $data['text'] = $text;
5189
5190                 # First, find all strip markers, and store their
5191                 #  data in an array.
5192                 $stripState = new StripState;
5193                 $pos = 0;
5194                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5195                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5196                 {
5197                         $end_pos += strlen( self::MARKER_SUFFIX );
5198                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5199
5200                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5201                                 $replaceArray = $stripState->general;
5202                                 $stripText = $this->mStripState->general->data[$marker];
5203                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5204                                 $replaceArray = $stripState->nowiki;
5205                                 $stripText = $this->mStripState->nowiki->data[$marker];
5206                         } else {
5207                                 throw new MWException( "Hanging strip marker: '$marker'." );
5208                         }
5209
5210                         $replaceArray->setPair( $marker, $stripText );
5211                         $pos = $end_pos;
5212                 }
5213                 $data['stripstate'] = $stripState;
5214
5215                 # Now, find all of our links, and store THEIR
5216                 #  data in an array! :)
5217                 $links = array( 'internal' => array(), 'interwiki' => array() );
5218                 $pos = 0;
5219
5220                 # Internal links
5221                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5222                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5223
5224                         $ns = trim( $ns );
5225                         if ( empty( $links['internal'][$ns] ) ) {
5226                                 $links['internal'][$ns] = array();
5227                         }
5228
5229                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5230                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5231                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5232                 }
5233
5234                 $pos = 0;
5235
5236                 # Interwiki links
5237                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5238                         $data = substr( $text, $start_pos );
5239                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5240                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5241                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5242                 }
5243
5244                 $data['linkholder'] = $links;
5245
5246                 return $data;
5247         }
5248
5249         /**
5250          * TODO: document
5251          * @param $data Array
5252          * @param $intPrefix String unique identifying prefix
5253          * @return String
5254          */
5255         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5256                 if ( !$intPrefix ) {
5257                         $intPrefix = $this->getRandomString();
5258                 }
5259
5260                 # First, extract the strip state.
5261                 $stripState = $data['stripstate'];
5262                 $this->mStripState->general->merge( $stripState->general );
5263                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5264
5265                 # Now, extract the text, and renumber links
5266                 $text = $data['text'];
5267                 $links = $data['linkholder'];
5268
5269                 # Internal...
5270                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5271                         foreach ( $nsLinks as $key => $entry ) {
5272                                 $newKey = $intPrefix . '-' . $key;
5273                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5274
5275                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5276                         }
5277                 }
5278
5279                 # Interwiki...
5280                 foreach ( $links['interwiki'] as $key => $entry ) {
5281                         $newKey = "$intPrefix-$key";
5282                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5283
5284                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5285                 }
5286
5287                 # Should be good to go.
5288                 return $text;
5289         }
5290 }
5291
5292 /**
5293  * @todo document, briefly.
5294  * @ingroup Parser
5295  */
5296 class StripState {
5297         var $general, $nowiki;
5298
5299         function __construct() {
5300                 $this->general = new ReplacementArray;
5301                 $this->nowiki = new ReplacementArray;
5302         }
5303
5304         function unstripGeneral( $text ) {
5305                 wfProfileIn( __METHOD__ );
5306                 do {
5307                         $oldText = $text;
5308                         $text = $this->general->replace( $text );
5309                 } while ( $text !== $oldText );
5310                 wfProfileOut( __METHOD__ );
5311                 return $text;
5312         }
5313
5314         function unstripNoWiki( $text ) {
5315                 wfProfileIn( __METHOD__ );
5316                 do {
5317                         $oldText = $text;
5318                         $text = $this->nowiki->replace( $text );
5319                 } while ( $text !== $oldText );
5320                 wfProfileOut( __METHOD__ );
5321                 return $text;
5322         }
5323
5324         function unstripBoth( $text ) {
5325                 wfProfileIn( __METHOD__ );
5326                 do {
5327                         $oldText = $text;
5328                         $text = $this->general->replace( $text );
5329                         $text = $this->nowiki->replace( $text );
5330                 } while ( $text !== $oldText );
5331                 wfProfileOut( __METHOD__ );
5332                 return $text;
5333         }
5334 }
5335
5336 /**
5337  * @todo document, briefly.
5338  * @ingroup Parser
5339  */
5340 class OnlyIncludeReplacer {
5341         var $output = '';
5342
5343         function replace( $matches ) {
5344                 if ( substr( $matches[1], -1 ) === "\n" ) {
5345                         $this->output .= substr( $matches[1], 0, -1 );
5346                 } else {
5347                         $this->output .= $matches[1];
5348                 }
5349         }
5350 }