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() / cleanSigInSig()
  26  *     Cleans a signature before saving it to preferences
  27  * getSection()
  28  *     Return the content of a section from an article for section editing
  29  * replaceSection()
  30  *     Replaces a section by number inside an article
  31  * getPreloadText()
  32  *     Removes <noinclude> sections, and <includeonly> tags.
  33  *
  34  * Globals used:
  35  *    objects:   $wgLang, $wgContLang
  36  *
  37  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  38  *
  39  * settings:
  40  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  41  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  42  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  43  *  $wgMaxArticleSize*
  44  *
  45  *  * only within ParserOptions
  46  * </pre>
  47  *
  48  * @ingroup Parser
  49  */
  50 class Parser {
  51         /**
  52          * Update this version number when the ParserOutput format
  53          * changes in an incompatible way, so the parser cache
  54          * can automatically discard old data.
  55          */
  56         const VERSION = '1.6.4';
  57
  58         # Flags for Parser::setFunctionHook
  59         # Also available as global constants from Defines.php
  60         const SFH_NO_HASH = 1;
  61         const SFH_OBJECT_ARGS = 2;
  62
  63         # Constants needed for external link processing
  64         # Everything except bracket, space, or control characters
  65         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  66         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  67                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  68
  69         # State constants for the definition list colon extraction
  70         const COLON_STATE_TEXT = 0;
  71         const COLON_STATE_TAG = 1;
  72         const COLON_STATE_TAGSTART = 2;
  73         const COLON_STATE_CLOSETAG = 3;
  74         const COLON_STATE_TAGSLASH = 4;
  75         const COLON_STATE_COMMENT = 5;
  76         const COLON_STATE_COMMENTDASH = 6;
  77         const COLON_STATE_COMMENTDASHDASH = 7;
  78
  79         # Flags for preprocessToDom
  80         const PTD_FOR_INCLUSION = 1;
  81
  82         # Allowed values for $this->mOutputType
  83         # Parameter to startExternalParse().
  84         const OT_HTML = 1; # like parse()
  85         const OT_WIKI = 2; # like preSaveTransform()
  86         const OT_PREPROCESS = 3; # like preprocess()
  87         const OT_MSG = 3;
  88         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  89
  90         # Marker Suffix needs to be accessible staticly.
  91         const MARKER_SUFFIX = "-QINU\x7f";
  92
  93         # Persistent:
  94         var $mTagHooks = array();
  95         var $mTransparentTagHooks = array();
  96         var $mFunctionHooks = array();
  97         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
  98         var $mFunctionTagHooks = array();
  99         var $mStripList  = array();
 100         var $mDefaultStripList  = array();
 101         var $mVarCache = array();
 102         var $mImageParams = array();
 103         var $mImageParamsMagicArray = array();
 104         var $mMarkerIndex = 0;
 105         var $mFirstCall = true;
 106         var $mVariables, $mSubstWords; # Initialised by initialiseVariables()
 107         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 108
 109
 110         # Cleared with clearState():
 111         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 112         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 113         var $mLinkHolders, $mLinkID;
 114         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 115         var $mTplExpandCache; # empty-frame expansion cache
 116         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 117         var $mExpensiveFunctionCount; # number of expensive parser function calls
 118         var $mUser; # User object; only used when doing pre-save transform
 119
 120         # Temporary
 121         # These are variables reset at least once per parse regardless of $clearState
 122         var $mOptions;      # ParserOptions object
 123         var $mTitle;        # Title context, used for self-link rendering and similar things
 124         var $mOutputType;   # Output type, one of the OT_xxx constants
 125         var $ot;            # Shortcut alias, see setOutputType()
 126         var $mRevisionObject; # The revision object of the specified revision ID
 127         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 128         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 129         var $mRevisionUser; # Userto display in {{REVISIONUSER}} tag
 130         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 131
 132         /**
 133          * Constructor
 134          *
 135          * @public
 136          */
 137         function __construct( $conf = array() ) {
 138                 $this->mConf = $conf;
 139                 $this->mUrlProtocols = wfUrlProtocols();
 140                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 141                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 142                 if ( isset( $conf['preprocessorClass'] ) ) {
 143                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 144                 } elseif ( extension_loaded( 'domxml' ) ) {
 145                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 146                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 147                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 148                 } elseif ( extension_loaded( 'dom' ) ) {
 149                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 150                 } else {
 151                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 152                 }
 153         }
 154
 155         /**
 156          * Reduce memory usage to reduce the impact of circular references
 157          */
 158         function __destruct() {
 159                 if ( isset( $this->mLinkHolders ) ) {
 160                         $this->mLinkHolders->__destruct();
 161                 }
 162                 foreach ( $this as $name => $value ) {
 163                         unset( $this->$name );
 164                 }
 165         }
 166
 167         /**
 168          * Do various kinds of initialisation on the first call of the parser
 169          */
 170         function firstCallInit() {
 171                 if ( !$this->mFirstCall ) {
 172                         return;
 173                 }
 174                 $this->mFirstCall = false;
 175
 176                 wfProfileIn( __METHOD__ );
 177
 178                 CoreParserFunctions::register( $this );
 179                 CoreTagHooks::register( $this );
 180                 $this->initialiseVariables();
 181
 182                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 183                 wfProfileOut( __METHOD__ );
 184         }
 185
 186         /**
 187          * Clear Parser state
 188          *
 189          * @private
 190          */
 191         function clearState() {
 192                 wfProfileIn( __METHOD__ );
 193                 if ( $this->mFirstCall ) {
 194                         $this->firstCallInit();
 195                 }
 196                 $this->mOutput = new ParserOutput;
 197                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 198                 $this->mAutonumber = 0;
 199                 $this->mLastSection = '';
 200                 $this->mDTopen = false;
 201                 $this->mIncludeCount = array();
 202                 $this->mStripState = new StripState;
 203                 $this->mArgStack = false;
 204                 $this->mInPre = false;
 205                 $this->mLinkHolders = new LinkHolderArray( $this );
 206                 $this->mLinkID = 0;
 207                 $this->mRevisionObject = $this->mRevisionTimestamp =
 208                         $this->mRevisionId = $this->mRevisionUser = null;
 209                 $this->mVarCache = array();
 210                 $this->mUser = null;
 211
 212                 /**
 213                  * Prefix for temporary replacement strings for the multipass parser.
 214                  * \x07 should never appear in input as it's disallowed in XML.
 215                  * Using it at the front also gives us a little extra robustness
 216                  * since it shouldn't match when butted up against identifier-like
 217                  * string constructs.
 218                  *
 219                  * Must not consist of all title characters, or else it will change
 220                  * the behaviour of <nowiki> in a link.
 221                  */
 222                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 223                 # Changed to \x7f to allow XML double-parsing -- TS
 224                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 225
 226
 227                 # Clear these on every parse, bug 4549
 228                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 229
 230                 $this->mShowToc = true;
 231                 $this->mForceTocPosition = false;
 232                 $this->mIncludeSizes = array(
 233                         'post-expand' => 0,
 234                         'arg' => 0,
 235                 );
 236                 $this->mPPNodeCount = 0;
 237                 $this->mDefaultSort = false;
 238                 $this->mHeadings = array();
 239                 $this->mDoubleUnderscores = array();
 240                 $this->mExpensiveFunctionCount = 0;
 241
 242                 # Fix cloning
 243                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 244                         $this->mPreprocessor = null;
 245                 }
 246
 247                 wfRunHooks( 'ParserClearState', array( &$this ) );
 248                 wfProfileOut( __METHOD__ );
 249         }
 250
 251         /**
 252          * Convert wikitext to HTML
 253          * Do not call this function recursively.
 254          *
 255          * @param $text String: text we want to parse
 256          * @param $title A title object
 257          * @param $options ParserOptions
 258          * @param $linestart boolean
 259          * @param $clearState boolean
 260          * @param $revid Int: number to pass in {{REVISIONID}}
 261          * @return ParserOutput a ParserOutput
 262          */
 263         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 264                 /**
 265                  * First pass--just handle <nowiki> sections, pass the rest off
 266                  * to internalParse() which does all the real work.
 267                  */
 268
 269                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 270                 $fname = __METHOD__.'-' . wfGetCaller();
 271                 wfProfileIn( __METHOD__ );
 272                 wfProfileIn( $fname );
 273
 274                 $this->startExternalParse( $title, $options, self::OT_HTML, $clearState );
 275
 276                 $oldRevisionId = $this->mRevisionId;
 277                 $oldRevisionObject = $this->mRevisionObject;
 278                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 279                 $oldRevisionUser = $this->mRevisionUser;
 280                 if ( $revid !== null ) {
 281                         $this->mRevisionId = $revid;
 282                         $this->mRevisionObject = null;
 283                         $this->mRevisionTimestamp = null;
 284                         $this->mRevisionUser = null;
 285                 }
 286
 287                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 288                 # No more strip!
 289                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 290                 $text = $this->internalParse( $text );
 291
 292                 $text = $this->mStripState->unstripGeneral( $text );
 293
 294                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 295                 $fixtags = array(
 296                         # french spaces, last one Guillemet-left
 297                         # only if there is something before the space
 298                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 299                         # french spaces, Guillemet-right
 300                         '/(\\302\\253) /' => '\\1&#160;',
 301                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 302                 );
 303                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 304
 305                 $text = $this->doBlockLevels( $text, $linestart );
 306
 307                 $this->replaceLinkHolders( $text );
 308
 309                 /**
 310                  * The page doesn't get language converted if
 311                  * a) It's disabled
 312                  * b) Content isn't converted
 313                  * c) It's a conversion table
 314                  */
 315                 if ( !( $wgDisableLangConversion
 316                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 317                                 || $this->mTitle->isConversionTable() ) ) {
 318
 319                         # The position of the convert() call should not be changed. it
 320                         # assumes that the links are all replaced and the only thing left
 321                         # is the <nowiki> mark.
 322
 323                         $text = $wgContLang->convert( $text );
 324                 }
 325
 326                 /**
 327                  * A converted title will be provided in the output object if title and
 328                  * content conversion are enabled, the article text does not contain
 329                  * a conversion-suppressing double-underscore tag, and no
 330                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 331                  * automatic link conversion.
 332                  */
 333                 if ( !( $wgDisableLangConversion
 334                                 || $wgDisableTitleConversion
 335                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 336                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 337                                 || $this->mOutput->getDisplayTitle() !== false ) )
 338                 {
 339                         $convruletitle = $wgContLang->getConvRuleTitle();
 340                         if ( $convruletitle ) {
 341                                 $this->mOutput->setTitleText( $convruletitle );
 342                         } else {
 343                                 $titleText = $wgContLang->convertTitle( $title );
 344                                 $this->mOutput->setTitleText( $titleText );
 345                         }
 346                 }
 347
 348                 $text = $this->mStripState->unstripNoWiki( $text );
 349
 350                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 351
 352 //!JF Move to its own function
 353
 354                 $uniq_prefix = $this->mUniqPrefix;
 355                 $matches = array();
 356                 $elements = array_keys( $this->mTransparentTagHooks );
 357                 $text = $this->extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 358
 359                 foreach ( $matches as $marker => $data ) {
 360                         list( $element, $content, $params, $tag ) = $data;
 361                         $tagName = strtolower( $element );
 362                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 363                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 364                         } else {
 365                                 $output = $tag;
 366                         }
 367                         $this->mStripState->general->setPair( $marker, $output );
 368                 }
 369                 $text = $this->mStripState->unstripGeneral( $text );
 370
 371                 $text = Sanitizer::normalizeCharReferences( $text );
 372
 373                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 374                         $text = MWTidy::tidy( $text );
 375                 } else {
 376                         # attempt to sanitize at least some nesting problems
 377                         # (bug #2702 and quite a few others)
 378                         $tidyregs = array(
 379                                 # ''Something [http://www.cool.com cool''] -->
 380                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 381                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 382                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 383                                 # fix up an anchor inside another anchor, only
 384                                 # at least for a single single nested link (bug 3695)
 385                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 386                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 387                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 388                                 # contains a div, so fix it up here; replace
 389                                 # div with escaped text
 390                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 391                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 392                                 # remove empty italic or bold tag pairs, some
 393                                 # introduced by rules above
 394                                 '/<([bi])><\/\\1>/' => '',
 395                         );
 396
 397                         $text = preg_replace(
 398                                 array_keys( $tidyregs ),
 399                                 array_values( $tidyregs ),
 400                                 $text );
 401                 }
 402                 global $wgExpensiveParserFunctionLimit;
 403                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 404                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 405                 }
 406
 407                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 408
 409                 # Information on include size limits, for the benefit of users who try to skirt them
 410                 if ( $this->mOptions->getEnableLimitReport() ) {
 411                         $max = $this->mOptions->getMaxIncludeSize();
 412                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 413                         $limitReport =
 414                                 "NewPP limit report\n" .
 415                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 416                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 417                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 418                                 $PFreport;
 419                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 420                         $text .= "\n<!-- \n$limitReport-->\n";
 421                 }
 422                 $this->mOutput->setText( $text );
 423
 424                 $this->mRevisionId = $oldRevisionId;
 425                 $this->mRevisionObject = $oldRevisionObject;
 426                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 427                 $this->mRevisionUser = $oldRevisionUser;
 428                 wfProfileOut( $fname );
 429                 wfProfileOut( __METHOD__ );
 430
 431                 return $this->mOutput;
 432         }
 433
 434         /**
 435          * Recursive parser entry point that can be called from an extension tag
 436          * hook.
 437          *
 438          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 439          *
 440          * @param $text String: text extension wants to have parsed
 441          * @param $frame PPFrame: The frame to use for expanding any template variables
 442          */
 443         function recursiveTagParse( $text, $frame=false ) {
 444                 wfProfileIn( __METHOD__ );
 445                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 446                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 447                 $text = $this->internalParse( $text, false, $frame );
 448                 wfProfileOut( __METHOD__ );
 449                 return $text;
 450         }
 451
 452         /**
 453          * Expand templates and variables in the text, producing valid, static wikitext.
 454          * Also removes comments.
 455          */
 456         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 457                 wfProfileIn( __METHOD__ );
 458                 $this->startExternalParse( $title, $options, self::OT_PREPROCESS, true );
 459                 if ( $revid !== null ) {
 460                         $this->mRevisionId = $revid;
 461                 }
 462                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 463                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 464                 $text = $this->replaceVariables( $text );
 465                 $text = $this->mStripState->unstripBoth( $text );
 466                 wfProfileOut( __METHOD__ );
 467                 return $text;
 468         }
 469
 470         /**
 471          * Process the wikitext for the ?preload= feature. (bug 5210)
 472          *
 473          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 474          * comments, templates, arguments, tags hooks and parser functions are untouched.
 475          */
 476         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 477                 # Parser (re)initialisation
 478                 $this->startExternalParse( $title, $options, self::OT_PLAIN, true );
 479
 480                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 481                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 482                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 483         }
 484
 485         /**
 486          * Get a random string
 487          *
 488          * @private
 489          * @static
 490          */
 491         static private function getRandomString() {
 492                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 493         }
 494
 495         /**
 496          * Set the current user.
 497          * Should only be used when doing pre-save transform.
 498          *
 499          * @param $user Mixed: User object or null (to reset)
 500          */
 501         function setUser( $user ) {
 502                 $this->mUser = $user;
 503         }
 504
 505         /**
 506          * Accessor for mUniqPrefix.
 507          *
 508          * @return String
 509          */
 510         public function uniqPrefix() {
 511                 if ( !isset( $this->mUniqPrefix ) ) {
 512                         # @todo Fixme: this is probably *horribly wrong*
 513                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 514                         # if this is called for a parser cache hit, the parser may not
 515                         # have ever been initialized in the first place.
 516                         # Not really sure what the heck is supposed to be going on here.
 517                         return '';
 518                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 519                 }
 520                 return $this->mUniqPrefix;
 521         }
 522
 523         /**
 524          * Set the context title
 525          */
 526         function setTitle( $t ) {
 527                 if ( !$t || $t instanceof FakeTitle ) {
 528                         $t = Title::newFromText( 'NO TITLE' );
 529                 }
 530
 531                 if ( strval( $t->getFragment() ) !== '' ) {
 532                         # Strip the fragment to avoid various odd effects
 533                         $this->mTitle = clone $t;
 534                         $this->mTitle->setFragment( '' );
 535                 } else {
 536                         $this->mTitle = $t;
 537                 }
 538         }
 539
 540         /**
 541          * Accessor for the Title object
 542          *
 543          * @return Title object
 544          */
 545         function getTitle() {
 546                 return $this->mTitle;
 547         }
 548
 549         /**
 550          * Accessor/mutator for the Title object
 551          *
 552          * @param $x New Title object or null to just get the current one
 553          * @return Title object
 554          */
 555         function Title( $x = null ) {
 556                 return wfSetVar( $this->mTitle, $x );
 557         }
 558
 559         /**
 560          * Set the output type
 561          *
 562          * @param $ot Integer: new value
 563          */
 564         function setOutputType( $ot ) {
 565                 $this->mOutputType = $ot;
 566                 # Shortcut alias
 567                 $this->ot = array(
 568                         'html' => $ot == self::OT_HTML,
 569                         'wiki' => $ot == self::OT_WIKI,
 570                         'pre' => $ot == self::OT_PREPROCESS,
 571                         'plain' => $ot == self::OT_PLAIN,
 572                 );
 573         }
 574
 575         /**
 576          * Accessor/mutator for the output type
 577          *
 578          * @param $x New value or null to just get the current one
 579          * @return Integer
 580          */
 581         function OutputType( $x = null ) {
 582                 return wfSetVar( $this->mOutputType, $x );
 583         }
 584
 585         /**
 586          * Get the ParserOutput object
 587          *
 588          * @return ParserOutput object
 589          */
 590         function getOutput() {
 591                 return $this->mOutput;
 592         }
 593
 594         /**
 595          * Get the ParserOptions object
 596          *
 597          * @return ParserOptions object
 598          */
 599         function getOptions() {
 600                 return $this->mOptions;
 601         }
 602
 603         /**
 604          * Accessor/mutator for the ParserOptions object
 605          *
 606          * @param $x New value or null to just get the current one
 607          * @return Current ParserOptions object
 608          */
 609         function Options( $x = null ) {
 610                 return wfSetVar( $this->mOptions, $x );
 611         }
 612
 613         function nextLinkID() {
 614                 return $this->mLinkID++;
 615         }
 616
 617         function getFunctionLang() {
 618                 global $wgLang, $wgContLang;
 619
 620                 $target = $this->mOptions->getTargetLanguage();
 621                 if ( $target !== null ) {
 622                         return $target;
 623                 } else {
 624                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 625                 }
 626         }
 627
 628         /**
 629          * Get a User object either from $this->mUser, if set, or from the
 630          * ParserOptions object otherwise
 631          *
 632          * @return User object
 633          */
 634         function getUser() {
 635                 if ( !is_null( $this->mUser ) ) {
 636                         return $this->mUser;
 637                 }
 638                 return $this->mOptions->getUser();
 639         }
 640
 641         /**
 642          * Get a preprocessor object
 643          *
 644          * @return Preprocessor instance
 645          */
 646         function getPreprocessor() {
 647                 if ( !isset( $this->mPreprocessor ) ) {
 648                         $class = $this->mPreprocessorClass;
 649                         $this->mPreprocessor = new $class( $this );
 650                 }
 651                 return $this->mPreprocessor;
 652         }
 653
 654         /**
 655          * Replaces all occurrences of HTML-style comments and the given tags
 656          * in the text with a random marker and returns the next text. The output
 657          * parameter $matches will be an associative array filled with data in
 658          * the form:
 659          *   'UNIQ-xxxxx' => array(
 660          *     'element',
 661          *     'tag content',
 662          *     array( 'param' => 'x' ),
 663          *     '<element param="x">tag content</element>' ) )
 664          *
 665          * @param $elements list of element names. Comments are always extracted.
 666          * @param $text Source text string.
 667          * @param $matches Out parameter, Array: extracted tags
 668          * @param $uniq_prefix
 669          * @return String: stripped text
 670          *
 671          * @static
 672          */
 673         public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 674                 static $n = 1;
 675                 $stripped = '';
 676                 $matches = array();
 677
 678                 $taglist = implode( '|', $elements );
 679                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 680
 681                 while ( $text != '' ) {
 682                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 683                         $stripped .= $p[0];
 684                         if ( count( $p ) < 5 ) {
 685                                 break;
 686                         }
 687                         if ( count( $p ) > 5 ) {
 688                                 # comment
 689                                 $element    = $p[4];
 690                                 $attributes = '';
 691                                 $close      = '';
 692                                 $inside     = $p[5];
 693                         } else {
 694                                 # tag
 695                                 $element    = $p[1];
 696                                 $attributes = $p[2];
 697                                 $close      = $p[3];
 698                                 $inside     = $p[4];
 699                         }
 700
 701                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 702                         $stripped .= $marker;
 703
 704                         if ( $close === '/>' ) {
 705                                 # Empty element tag, <tag />
 706                                 $content = null;
 707                                 $text = $inside;
 708                                 $tail = null;
 709                         } else {
 710                                 if ( $element === '!--' ) {
 711                                         $end = '/(-->)/';
 712                                 } else {
 713                                         $end = "/(<\\/$element\\s*>)/i";
 714                                 }
 715                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 716                                 $content = $q[0];
 717                                 if ( count( $q ) < 3 ) {
 718                                         # No end tag -- let it run out to the end of the text.
 719                                         $tail = '';
 720                                         $text = '';
 721                                 } else {
 722                                         $tail = $q[1];
 723                                         $text = $q[2];
 724                                 }
 725                         }
 726
 727                         $matches[$marker] = array( $element,
 728                                 $content,
 729                                 Sanitizer::decodeTagAttributes( $attributes ),
 730                                 "<$element$attributes$close$content$tail" );
 731                 }
 732                 return $stripped;
 733         }
 734
 735         /**
 736          * Get a list of strippable XML-like elements
 737          */
 738         function getStripList() {
 739                 return $this->mStripList;
 740         }
 741
 742         /**
 743          * @deprecated use replaceVariables
 744          */
 745         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 746                 return $text;
 747         }
 748
 749         /**
 750          * Restores pre, math, and other extensions removed by strip()
 751          *
 752          * always call unstripNoWiki() after this one
 753          * @private
 754          * @deprecated use $this->mStripState->unstrip()
 755          */
 756         function unstrip( $text, $state ) {
 757                 return $state->unstripGeneral( $text );
 758         }
 759
 760         /**
 761          * Always call this after unstrip() to preserve the order
 762          *
 763          * @private
 764          * @deprecated use $this->mStripState->unstrip()
 765          */
 766         function unstripNoWiki( $text, $state ) {
 767                 return $state->unstripNoWiki( $text );
 768         }
 769
 770         /**
 771          * @deprecated use $this->mStripState->unstripBoth()
 772          */
 773         function unstripForHTML( $text ) {
 774                 return $this->mStripState->unstripBoth( $text );
 775         }
 776
 777         /**
 778          * Add an item to the strip state
 779          * Returns the unique tag which must be inserted into the stripped text
 780          * The tag will be replaced with the original text in unstrip()
 781          *
 782          * @private
 783          */
 784         function insertStripItem( $text ) {
 785                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 786                 $this->mMarkerIndex++;
 787                 $this->mStripState->general->setPair( $rnd, $text );
 788                 return $rnd;
 789         }
 790
 791         /**
 792          * Interface with html tidy
 793          * @deprecated Use MWTidy::tidy()
 794          */
 795         public static function tidy( $text ) {
 796                 wfDeprecated( __METHOD__ );
 797                 return MWTidy::tidy( $text );
 798         }
 799
 800         /**
 801          * parse the wiki syntax used to render tables
 802          *
 803          * @private
 804          */
 805         function doTableStuff( $text ) {
 806                 wfProfileIn( __METHOD__ );
 807
 808                 $lines = StringUtils::explode( "\n", $text );
 809                 $out = '';
 810                 $td_history = array(); # Is currently a td tag open?
 811                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 812                 $tr_history = array(); # Is currently a tr tag open?
 813                 $tr_attributes = array(); # history of tr attributes
 814                 $has_opened_tr = array(); # Did this table open a <tr> element?
 815                 $indent_level = 0; # indent level of the table
 816
 817                 # Keep pulling lines off the front of the array until they're all gone.
 818                 # we want to be able to push lines back on to the front of the stream,
 819                 # but StringUtils::explode() returns funky optimised Iterators which don't
 820                 # support insertion.  So maintain a separate buffer and draw on that first if
 821                 # there's anything in it
 822                 $extraLines = array();
 823                 $lines->rewind();
 824                 do {
 825                         if( $extraLines ){
 826                                 $outLine = array_shift( $extraLines );
 827                         } elseif( $lines->valid() ) {
 828                                 $outLine = $lines->current();
 829                                 $lines->next();
 830                         } else {
 831                                 break;
 832                         }
 833                         $line = trim( $outLine );
 834
 835                         if ( $line === '' ) { # empty line, go to next line
 836                                 $out .= $outLine."\n";
 837                                 continue;
 838                         }
 839
 840                         $first_character = $line[0];
 841                         $matches = array();
 842
 843                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 844                                 # First check if we are starting a new table
 845                                 $indent_level = strlen( $matches[1] );
 846
 847                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 848                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 849
 850                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 851                                 array_push( $td_history , false );
 852                                 array_push( $last_tag_history , '' );
 853                                 array_push( $tr_history , false );
 854                                 array_push( $tr_attributes , '' );
 855                                 array_push( $has_opened_tr , false );
 856                         } elseif ( count( $td_history ) == 0 ) {
 857                                 # Don't do any of the following
 858                                 $out .= $outLine."\n";
 859                                 continue;
 860                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 861                                 # We are ending a table
 862                                 $line = '</table>' . substr( $line , 2 );
 863                                 $last_tag = array_pop( $last_tag_history );
 864
 865                                 if ( !array_pop( $has_opened_tr ) ) {
 866                                         $line = "<tr><td></td></tr>{$line}";
 867                                 }
 868
 869                                 if ( array_pop( $tr_history ) ) {
 870                                         $line = "</tr>{$line}";
 871                                 }
 872
 873                                 if ( array_pop( $td_history ) ) {
 874                                         $line = "</{$last_tag}>{$line}";
 875                                 }
 876                                 array_pop( $tr_attributes );
 877                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 878                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 879                                 # Now we have a table row
 880                                 $line = preg_replace( '#^\|-+#', '', $line );
 881
 882                                 # Whats after the tag is now only attributes
 883                                 $attributes = $this->mStripState->unstripBoth( $line );
 884                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 885                                 array_pop( $tr_attributes );
 886                                 array_push( $tr_attributes, $attributes );
 887
 888                                 $line = '';
 889                                 $last_tag = array_pop( $last_tag_history );
 890                                 array_pop( $has_opened_tr );
 891                                 array_push( $has_opened_tr , true );
 892
 893                                 if ( array_pop( $tr_history ) ) {
 894                                         $line = '</tr>';
 895                                 }
 896
 897                                 if ( array_pop( $td_history ) ) {
 898                                         $line = "</{$last_tag}>{$line}";
 899                                 }
 900
 901                                 $outLine = $line;
 902                                 array_push( $tr_history , false );
 903                                 array_push( $td_history , false );
 904                                 array_push( $last_tag_history , '' );
 905                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 906                                 # This might be cell elements, td, th or captions
 907                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 908                                         $first_character = '|+';
 909                                 }
 910
 911                                 $line = substr( $line , strlen( $first_character ) );
 912
 913                                 if ( $first_character === '!' ) {
 914                                         $line = str_replace( '!!' , '||' , $line );
 915                                 }
 916
 917                                 # Split up multiple cells on the same line.
 918                                 # FIXME : This can result in improper nesting of tags processed
 919                                 # by earlier parser steps, but should avoid splitting up eg
 920                                 # attribute values containing literal "||".
 921                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 922                                 $cell = array_shift( $cells );
 923
 924                                 # Inject cells back into the stream to be dealt with later
 925                                 # TODO: really we should do the whole thing as a stream...
 926                                 # but that would be too much like a sensible implementation :P
 927                                 if( count( $cells ) ){
 928                                         foreach( array_reverse( $cells ) as $extraCell ){
 929                                                 array_unshift( $extraLines, $first_character . $extraCell );
 930                                         }
 931                                 }
 932
 933                                 $outLine = '';
 934
 935                                 $previous = '';
 936                                 if ( $first_character !== '|+' ) {
 937                                         $tr_after = array_pop( $tr_attributes );
 938                                         if ( !array_pop( $tr_history ) ) {
 939                                                 $previous = "<tr{$tr_after}>\n";
 940                                         }
 941                                         array_push( $tr_history , true );
 942                                         array_push( $tr_attributes , '' );
 943                                         array_pop( $has_opened_tr );
 944                                         array_push( $has_opened_tr , true );
 945                                 }
 946
 947                                 $last_tag = array_pop( $last_tag_history );
 948
 949                                 if ( array_pop( $td_history ) ) {
 950                                         $previous = "</{$last_tag}>\n{$previous}";
 951                                 }
 952
 953                                 if ( $first_character === '|' ) {
 954                                         $last_tag = 'td';
 955                                 } elseif ( $first_character === '!' ) {
 956                                         $last_tag = 'th';
 957                                 } elseif ( $first_character === '|+' ) {
 958                                         $last_tag = 'caption';
 959                                 } else {
 960                                         $last_tag = '';
 961                                 }
 962
 963                                 array_push( $last_tag_history , $last_tag );
 964
 965                                 # A cell could contain both parameters and data... but the pipe could
 966                                 # also be the start of a nested table, or a raw pipe inside an invalid
 967                                 # link (bug 553).
 968                                 $cell_data = preg_split( '/(?<!\{)\|/', $cell, 2 );
 969
 970                                 # Bug 553: a '|' inside an invalid link should not
 971                                 # be mistaken as delimiting cell parameters
 972                                 if ( strpos( $cell_data[0], '[[' ) !== false ) {
 973                                         $data = $cell;
 974                                         $cell = "{$previous}<{$last_tag}>";
 975                                 } elseif ( count( $cell_data ) == 1 ) {
 976                                         $cell = "{$previous}<{$last_tag}>";
 977                                         $data = $cell_data[0];
 978                                 } else {
 979                                         $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 980                                         $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 981                                         $cell = "{$previous}<{$last_tag}{$attributes}>";
 982                                         $data = $cell_data[1];
 983                                 }
 984
 985                                 # Bug 529: the start of a table cell should be a linestart context for
 986                                 # processing other block markup, including nested tables.  The original
 987                                 # implementation of this was to add a newline before every brace construct,
 988                                 # which broke all manner of other things.  Instead, push the contents
 989                                 # of the cell back into the stream and come back to it later.  But don't
 990                                 # do that if the first line is empty, or you may get extra whitespace
 991                                 if( $data ){
 992                                         array_unshift( $extraLines, trim( $data ) );
 993                                 }
 994
 995                                 $outLine .= $cell;
 996                                 array_push( $td_history , true );
 997                         }
 998                         $out .= $outLine . "\n";
 999                 } while( $lines->valid() || count( $extraLines ) );
1000
1001                 # Closing open td, tr && table
1002                 while ( count( $td_history ) > 0 ) {
1003                         if ( array_pop( $td_history ) ) {
1004                                 $out .= "</td>\n";
1005                         }
1006                         if ( array_pop( $tr_history ) ) {
1007                                 $out .= "</tr>\n";
1008                         }
1009                         if ( !array_pop( $has_opened_tr ) ) {
1010                                 $out .= "<tr><td></td></tr>\n" ;
1011                         }
1012
1013                         $out .= "</table>\n";
1014                 }
1015
1016                 # Remove trailing line-ending (b/c)
1017                 if ( substr( $out, -1 ) === "\n" ) {
1018                         $out = substr( $out, 0, -1 );
1019                 }
1020
1021                 # special case: don't return empty table
1022                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1023                         $out = '';
1024                 }
1025
1026                 wfProfileOut( __METHOD__ );
1027
1028                 return $out;
1029         }
1030
1031         /**
1032          * Helper function for parse() that transforms wiki markup into
1033          * HTML. Only called for $mOutputType == self::OT_HTML.
1034          *
1035          * @private
1036          */
1037         function internalParse( $text, $isMain = true, $frame=false ) {
1038                 wfProfileIn( __METHOD__ );
1039
1040                 $origText = $text;
1041
1042                 # Hook to suspend the parser in this state
1043                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1044                         wfProfileOut( __METHOD__ );
1045                         return $text ;
1046                 }
1047
1048                 # if $frame is provided, then use $frame for replacing any variables
1049                 if ( $frame ) {
1050                         # use frame depth to infer how include/noinclude tags should be handled
1051                         # depth=0 means this is the top-level document; otherwise it's an included document
1052                         if ( !$frame->depth ) {
1053                                 $flag = 0;
1054                         } else {
1055                                 $flag = Parser::PTD_FOR_INCLUSION;
1056                         }
1057                         $dom = $this->preprocessToDom( $text, $flag );
1058                         $text = $frame->expand( $dom );
1059                 } else {
1060                         # if $frame is not provided, then use old-style replaceVariables
1061                         $text = $this->replaceVariables( $text );
1062                 }
1063
1064                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1065                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1066
1067                 # Tables need to come after variable replacement for things to work
1068                 # properly; putting them before other transformations should keep
1069                 # exciting things like link expansions from showing up in surprising
1070                 # places.
1071                 $text = $this->doTableStuff( $text );
1072
1073                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1074
1075                 $text = $this->doDoubleUnderscore( $text );
1076
1077                 $text = $this->doHeadings( $text );
1078                 if ( $this->mOptions->getUseDynamicDates() ) {
1079                         $df = DateFormatter::getInstance();
1080                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1081                 }
1082                 $text = $this->replaceInternalLinks( $text );
1083                 $text = $this->doAllQuotes( $text );
1084                 $text = $this->replaceExternalLinks( $text );
1085
1086                 # replaceInternalLinks may sometimes leave behind
1087                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1088                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1089
1090                 $text = $this->doMagicLinks( $text );
1091                 $text = $this->formatHeadings( $text, $origText, $isMain );
1092
1093                 wfProfileOut( __METHOD__ );
1094                 return $text;
1095         }
1096
1097         /**
1098          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1099          * magic external links.
1100          *
1101          * DML
1102          * @private
1103          */
1104         function doMagicLinks( $text ) {
1105                 wfProfileIn( __METHOD__ );
1106                 $prots = $this->mUrlProtocols;
1107                 $urlChar = self::EXT_LINK_URL_CLASS;
1108                 $text = preg_replace_callback(
1109                         '!(?:                           # Start cases
1110                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1111                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1112                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1113                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1114                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1115                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1116                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1117                                         [0-9Xx]                 # check digit
1118                                         \b)
1119                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1120                 wfProfileOut( __METHOD__ );
1121                 return $text;
1122         }
1123
1124         function magicLinkCallback( $m ) {
1125                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1126                         # Skip anchor
1127                         return $m[0];
1128                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1129                         # Skip HTML element
1130                         return $m[0];
1131                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1132                         # Free external link
1133                         return $this->makeFreeExternalLink( $m[0] );
1134                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1135                         # RFC or PMID
1136                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1137                                 $keyword = 'RFC';
1138                                 $urlmsg = 'rfcurl';
1139                                 $CssClass = 'mw-magiclink-rfc';
1140                                 $id = $m[4];
1141                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1142                                 $keyword = 'PMID';
1143                                 $urlmsg = 'pubmedurl';
1144                                 $CssClass = 'mw-magiclink-pmid';
1145                                 $id = $m[4];
1146                         } else {
1147                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1148                                         substr( $m[0], 0, 20 ) . '"' );
1149                         }
1150                         $url = wfMsgForContent( $urlmsg, $id);
1151                         $sk = $this->mOptions->getSkin( $this->mTitle );
1152                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1153                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1154                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1155                         # ISBN
1156                         $isbn = $m[5];
1157                         $num = strtr( $isbn, array(
1158                                 '-' => '',
1159                                 ' ' => '',
1160                                 'x' => 'X',
1161                         ));
1162                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1163                         return'<a href="' .
1164                                 $titleObj->escapeLocalUrl() .
1165                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1166                 } else {
1167                         return $m[0];
1168                 }
1169         }
1170
1171         /**
1172          * Make a free external link, given a user-supplied URL
1173          * @return HTML
1174          * @private
1175          */
1176         function makeFreeExternalLink( $url ) {
1177                 global $wgContLang;
1178                 wfProfileIn( __METHOD__ );
1179
1180                 $sk = $this->mOptions->getSkin( $this->mTitle );
1181                 $trail = '';
1182
1183                 # The characters '<' and '>' (which were escaped by
1184                 # removeHTMLtags()) should not be included in
1185                 # URLs, per RFC 2396.
1186                 $m2 = array();
1187                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1188                         $trail = substr( $url, $m2[0][1] ) . $trail;
1189                         $url = substr( $url, 0, $m2[0][1] );
1190                 }
1191
1192                 # Move trailing punctuation to $trail
1193                 $sep = ',;\.:!?';
1194                 # If there is no left bracket, then consider right brackets fair game too
1195                 if ( strpos( $url, '(' ) === false ) {
1196                         $sep .= ')';
1197                 }
1198
1199                 $numSepChars = strspn( strrev( $url ), $sep );
1200                 if ( $numSepChars ) {
1201                         $trail = substr( $url, -$numSepChars ) . $trail;
1202                         $url = substr( $url, 0, -$numSepChars );
1203                 }
1204
1205                 $url = Sanitizer::cleanUrl( $url );
1206
1207                 # Is this an external image?
1208                 $text = $this->maybeMakeExternalImage( $url );
1209                 if ( $text === false ) {
1210                         # Not an image, make a link
1211                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1212                                 $this->getExternalLinkAttribs( $url ) );
1213                         # Register it in the output object...
1214                         # Replace unnecessary URL escape codes with their equivalent characters
1215                         $pasteurized = self::replaceUnusualEscapes( $url );
1216                         $this->mOutput->addExternalLink( $pasteurized );
1217                 }
1218                 wfProfileOut( __METHOD__ );
1219                 return $text . $trail;
1220         }
1221
1222
1223         /**
1224          * Parse headers and return html
1225          *
1226          * @private
1227          */
1228         function doHeadings( $text ) {
1229                 wfProfileIn( __METHOD__ );
1230                 for ( $i = 6; $i >= 1; --$i ) {
1231                         $h = str_repeat( '=', $i );
1232                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1233                           "<h$i>\\1</h$i>", $text );
1234                 }
1235                 wfProfileOut( __METHOD__ );
1236                 return $text;
1237         }
1238
1239         /**
1240          * Replace single quotes with HTML markup
1241          * @private
1242          * @return string the altered text
1243          */
1244         function doAllQuotes( $text ) {
1245                 wfProfileIn( __METHOD__ );
1246                 $outtext = '';
1247                 $lines = StringUtils::explode( "\n", $text );
1248                 foreach ( $lines as $line ) {
1249                         $outtext .= $this->doQuotes( $line ) . "\n";
1250                 }
1251                 $outtext = substr( $outtext, 0,-1 );
1252                 wfProfileOut( __METHOD__ );
1253                 return $outtext;
1254         }
1255
1256         /**
1257          * Helper function for doAllQuotes()
1258          */
1259         public function doQuotes( $text ) {
1260                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1261                 if ( count( $arr ) == 1 ) {
1262                         return $text;
1263                 } else {
1264                         # First, do some preliminary work. This may shift some apostrophes from
1265                         # being mark-up to being text. It also counts the number of occurrences
1266                         # of bold and italics mark-ups.
1267                         $numbold = 0;
1268                         $numitalics = 0;
1269                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1270                                 if ( ( $i % 2 ) == 1 ) {
1271                                         # If there are ever four apostrophes, assume the first is supposed to
1272                                         # be text, and the remaining three constitute mark-up for bold text.
1273                                         if ( strlen( $arr[$i] ) == 4 ) {
1274                                                 $arr[$i-1] .= "'";
1275                                                 $arr[$i] = "'''";
1276                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1277                                                 # If there are more than 5 apostrophes in a row, assume they're all
1278                                                 # text except for the last 5.
1279                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1280                                                 $arr[$i] = "'''''";
1281                                         }
1282                                         # Count the number of occurrences of bold and italics mark-ups.
1283                                         # We are not counting sequences of five apostrophes.
1284                                         if ( strlen( $arr[$i] ) == 2 ) {
1285                                                 $numitalics++;
1286                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1287                                                 $numbold++;
1288                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1289                                                 $numitalics++;
1290                                                 $numbold++;
1291                                         }
1292                                 }
1293                         }
1294
1295                         # If there is an odd number of both bold and italics, it is likely
1296                         # that one of the bold ones was meant to be an apostrophe followed
1297                         # by italics. Which one we cannot know for certain, but it is more
1298                         # likely to be one that has a single-letter word before it.
1299                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1300                                 $i = 0;
1301                                 $firstsingleletterword = -1;
1302                                 $firstmultiletterword = -1;
1303                                 $firstspace = -1;
1304                                 foreach ( $arr as $r ) {
1305                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1306                                                 $x1 = substr( $arr[$i-1], -1 );
1307                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1308                                                 if ( $x1 === ' ' ) {
1309                                                         if ( $firstspace == -1 ) {
1310                                                                 $firstspace = $i;
1311                                                         }
1312                                                 } elseif ( $x2 === ' ') {
1313                                                         if ( $firstsingleletterword == -1 ) {
1314                                                                 $firstsingleletterword = $i;
1315                                                         }
1316                                                 } else {
1317                                                         if ( $firstmultiletterword == -1 ) {
1318                                                                 $firstmultiletterword = $i;
1319                                                         }
1320                                                 }
1321                                         }
1322                                         $i++;
1323                                 }
1324
1325                                 # If there is a single-letter word, use it!
1326                                 if ( $firstsingleletterword > -1 ) {
1327                                         $arr[$firstsingleletterword] = "''";
1328                                         $arr[$firstsingleletterword-1] .= "'";
1329                                 } elseif ( $firstmultiletterword > -1 ) {
1330                                         # If not, but there's a multi-letter word, use that one.
1331                                         $arr[$firstmultiletterword] = "''";
1332                                         $arr[$firstmultiletterword-1] .= "'";
1333                                 } elseif ( $firstspace > -1 ) {
1334                                         # ... otherwise use the first one that has neither.
1335                                         # (notice that it is possible for all three to be -1 if, for example,
1336                                         # there is only one pentuple-apostrophe in the line)
1337                                         $arr[$firstspace] = "''";
1338                                         $arr[$firstspace-1] .= "'";
1339                                 }
1340                         }
1341
1342                         # Now let's actually convert our apostrophic mush to HTML!
1343                         $output = '';
1344                         $buffer = '';
1345                         $state = '';
1346                         $i = 0;
1347                         foreach ( $arr as $r ) {
1348                                 if ( ( $i % 2 ) == 0 ) {
1349                                         if ( $state === 'both' ) {
1350                                                 $buffer .= $r;
1351                                         } else {
1352                                                 $output .= $r;
1353                                         }
1354                                 } else {
1355                                         if ( strlen( $r ) == 2 ) {
1356                                                 if ( $state === 'i' ) {
1357                                                         $output .= '</i>'; $state = '';
1358                                                 } elseif ( $state === 'bi' ) {
1359                                                         $output .= '</i>'; $state = 'b';
1360                                                 } elseif ( $state === 'ib' ) {
1361                                                         $output .= '</b></i><b>'; $state = 'b';
1362                                                 } elseif ( $state === 'both' ) {
1363                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1364                                                 } else { # $state can be 'b' or ''
1365                                                         $output .= '<i>'; $state .= 'i';
1366                                                 }
1367                                         } elseif ( strlen( $r ) == 3 ) {
1368                                                 if ( $state === 'b' ) {
1369                                                         $output .= '</b>'; $state = '';
1370                                                 } elseif ( $state === 'bi' ) {
1371                                                         $output .= '</i></b><i>'; $state = 'i';
1372                                                 } elseif ( $state === 'ib' ) {
1373                                                         $output .= '</b>'; $state = 'i';
1374                                                 } elseif ( $state === 'both' ) {
1375                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1376                                                 } else { # $state can be 'i' or ''
1377                                                         $output .= '<b>'; $state .= 'b';
1378                                                 }
1379                                         } elseif ( strlen( $r ) == 5 ) {
1380                                                 if ( $state === 'b' ) {
1381                                                         $output .= '</b><i>'; $state = 'i';
1382                                                 } elseif ( $state === 'i' ) {
1383                                                         $output .= '</i><b>'; $state = 'b';
1384                                                 } elseif ( $state === 'bi' ) {
1385                                                         $output .= '</i></b>'; $state = '';
1386                                                 } elseif ( $state === 'ib' ) {
1387                                                         $output .= '</b></i>'; $state = '';
1388                                                 } elseif ( $state === 'both' ) {
1389                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1390                                                 } else { # ($state == '')
1391                                                         $buffer = ''; $state = 'both';
1392                                                 }
1393                                         }
1394                                 }
1395                                 $i++;
1396                         }
1397                         # Now close all remaining tags.  Notice that the order is important.
1398                         if ( $state === 'b' || $state === 'ib' ) {
1399                                 $output .= '</b>';
1400                         }
1401                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1402                                 $output .= '</i>';
1403                         }
1404                         if ( $state === 'bi' ) {
1405                                 $output .= '</b>';
1406                         }
1407                         # There might be lonely ''''', so make sure we have a buffer
1408                         if ( $state === 'both' && $buffer ) {
1409                                 $output .= '<b><i>'.$buffer.'</i></b>';
1410                         }
1411                         return $output;
1412                 }
1413         }
1414
1415         /**
1416          * Replace external links (REL)
1417          *
1418          * Note: this is all very hackish and the order of execution matters a lot.
1419          * Make sure to run maintenance/parserTests.php if you change this code.
1420          *
1421          * @private
1422          */
1423         function replaceExternalLinks( $text ) {
1424                 global $wgContLang;
1425                 wfProfileIn( __METHOD__ );
1426
1427                 $sk = $this->mOptions->getSkin( $this->mTitle );
1428
1429                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1430                 $s = array_shift( $bits );
1431
1432                 $i = 0;
1433                 while ( $i<count( $bits ) ) {
1434                         $url = $bits[$i++];
1435                         $protocol = $bits[$i++];
1436                         $text = $bits[$i++];
1437                         $trail = $bits[$i++];
1438
1439                         # The characters '<' and '>' (which were escaped by
1440                         # removeHTMLtags()) should not be included in
1441                         # URLs, per RFC 2396.
1442                         $m2 = array();
1443                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1444                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1445                                 $url = substr( $url, 0, $m2[0][1] );
1446                         }
1447
1448                         # If the link text is an image URL, replace it with an <img> tag
1449                         # This happened by accident in the original parser, but some people used it extensively
1450                         $img = $this->maybeMakeExternalImage( $text );
1451                         if ( $img !== false ) {
1452                                 $text = $img;
1453                         }
1454
1455                         $dtrail = '';
1456
1457                         # Set linktype for CSS - if URL==text, link is essentially free
1458                         $linktype = ( $text === $url ) ? 'free' : 'text';
1459
1460                         # No link text, e.g. [http://domain.tld/some.link]
1461                         if ( $text == '' ) {
1462                                 # Autonumber if allowed. See bug #5918
1463                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1464                                         $langObj = $this->getFunctionLang();
1465                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1466                                         $linktype = 'autonumber';
1467                                 } else {
1468                                         # Otherwise just use the URL
1469                                         $text = htmlspecialchars( $url );
1470                                         $linktype = 'free';
1471                                 }
1472                         } else {
1473                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1474                                 # Check for trail
1475                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1476                         }
1477
1478                         $text = $wgContLang->markNoConversion( $text );
1479
1480                         $url = Sanitizer::cleanUrl( $url );
1481
1482                         # Use the encoded URL
1483                         # This means that users can paste URLs directly into the text
1484                         # Funny characters like ö aren't valid in URLs anyway
1485                         # This was changed in August 2004
1486                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1487                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1488
1489                         # Register link in the output object.
1490                         # Replace unnecessary URL escape codes with the referenced character
1491                         # This prevents spammers from hiding links from the filters
1492                         $pasteurized = self::replaceUnusualEscapes( $url );
1493                         $this->mOutput->addExternalLink( $pasteurized );
1494                 }
1495
1496                 wfProfileOut( __METHOD__ );
1497                 return $s;
1498         }
1499
1500         /**
1501          * Get an associative array of additional HTML attributes appropriate for a
1502          * particular external link.  This currently may include rel => nofollow
1503          * (depending on configuration, namespace, and the URL's domain) and/or a
1504          * target attribute (depending on configuration).
1505          *
1506          * @param $url String: optional URL, to extract the domain from for rel =>
1507          *   nofollow if appropriate
1508          * @return Array: associative array of HTML attributes
1509          */
1510         function getExternalLinkAttribs( $url = false ) {
1511                 $attribs = array();
1512                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1513                 $ns = $this->mTitle->getNamespace();
1514                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1515                         $attribs['rel'] = 'nofollow';
1516
1517                         global $wgNoFollowDomainExceptions;
1518                         if ( $wgNoFollowDomainExceptions ) {
1519                                 $bits = wfParseUrl( $url );
1520                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1521                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1522                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1523                                                         unset( $attribs['rel'] );
1524                                                         break;
1525                                                 }
1526                                         }
1527                                 }
1528                         }
1529                 }
1530                 if ( $this->mOptions->getExternalLinkTarget() ) {
1531                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1532                 }
1533                 return $attribs;
1534         }
1535
1536
1537         /**
1538          * Replace unusual URL escape codes with their equivalent characters
1539          *
1540          * @param $url String
1541          * @return String
1542          *
1543          * @todo  This can merge genuinely required bits in the path or query string,
1544          *        breaking legit URLs. A proper fix would treat the various parts of
1545          *        the URL differently; as a workaround, just use the output for
1546          *        statistical records, not for actual linking/output.
1547          */
1548         static function replaceUnusualEscapes( $url ) {
1549                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1550                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1551         }
1552
1553         /**
1554          * Callback function used in replaceUnusualEscapes().
1555          * Replaces unusual URL escape codes with their equivalent character
1556          */
1557         private static function replaceUnusualEscapesCallback( $matches ) {
1558                 $char = urldecode( $matches[0] );
1559                 $ord = ord( $char );
1560                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1561                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1562                         # No, shouldn't be escaped
1563                         return $char;
1564                 } else {
1565                         # Yes, leave it escaped
1566                         return $matches[0];
1567                 }
1568         }
1569
1570         /**
1571          * make an image if it's allowed, either through the global
1572          * option, through the exception, or through the on-wiki whitelist
1573          * @private
1574          */
1575         function maybeMakeExternalImage( $url ) {
1576                 $sk = $this->mOptions->getSkin( $this->mTitle );
1577                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1578                 $imagesexception = !empty( $imagesfrom );
1579                 $text = false;
1580                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1581                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1582                         $imagematch = false;
1583                         foreach ( $imagesfrom as $match ) {
1584                                 if ( strpos( $url, $match ) === 0 ) {
1585                                         $imagematch = true;
1586                                         break;
1587                                 }
1588                         }
1589                 } elseif ( $imagesexception ) {
1590                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1591                 } else {
1592                         $imagematch = false;
1593                 }
1594                 if ( $this->mOptions->getAllowExternalImages()
1595                          || ( $imagesexception && $imagematch ) ) {
1596                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1597                                 # Image found
1598                                 $text = $sk->makeExternalImage( $url );
1599                         }
1600                 }
1601                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1602                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1603                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1604                         foreach ( $whitelist as $entry ) {
1605                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1606                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1607                                         continue;
1608                                 }
1609                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1610                                         # Image matches a whitelist entry
1611                                         $text = $sk->makeExternalImage( $url );
1612                                         break;
1613                                 }
1614                         }
1615                 }
1616                 return $text;
1617         }
1618
1619         /**
1620          * Process [[ ]] wikilinks
1621          * @return String: processed text
1622          *
1623          * @private
1624          */
1625         function replaceInternalLinks( $s ) {
1626                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1627                 return $s;
1628         }
1629
1630         /**
1631          * Process [[ ]] wikilinks (RIL)
1632          * @return LinkHolderArray
1633          *
1634          * @private
1635          */
1636         function replaceInternalLinks2( &$s ) {
1637                 global $wgContLang;
1638
1639                 wfProfileIn( __METHOD__ );
1640
1641                 wfProfileIn( __METHOD__.'-setup' );
1642                 static $tc = FALSE, $e1, $e1_img;
1643                 # the % is needed to support urlencoded titles as well
1644                 if ( !$tc ) {
1645                         $tc = Title::legalChars() . '#%';
1646                         # Match a link having the form [[namespace:link|alternate]]trail
1647                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1648                         # Match cases where there is no "]]", which might still be images
1649                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1650                 }
1651
1652                 $sk = $this->mOptions->getSkin( $this->mTitle );
1653                 $holders = new LinkHolderArray( $this );
1654
1655                 # split the entire text string on occurences of [[
1656                 $a = StringUtils::explode( '[[', ' ' . $s );
1657                 # get the first element (all text up to first [[), and remove the space we added
1658                 $s = $a->current();
1659                 $a->next();
1660                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1661                 $s = substr( $s, 1 );
1662
1663                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1664                 $e2 = null;
1665                 if ( $useLinkPrefixExtension ) {
1666                         # Match the end of a line for a word that's not followed by whitespace,
1667                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1668                         $e2 = wfMsgForContent( 'linkprefix' );
1669                 }
1670
1671                 if ( is_null( $this->mTitle ) ) {
1672                         wfProfileOut( __METHOD__.'-setup' );
1673                         wfProfileOut( __METHOD__ );
1674                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1675                 }
1676                 $nottalk = !$this->mTitle->isTalkPage();
1677
1678                 if ( $useLinkPrefixExtension ) {
1679                         $m = array();
1680                         if ( preg_match( $e2, $s, $m ) ) {
1681                                 $first_prefix = $m[2];
1682                         } else {
1683                                 $first_prefix = false;
1684                         }
1685                 } else {
1686                         $prefix = '';
1687                 }
1688
1689                 if ( $wgContLang->hasVariants() ) {
1690                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1691                 } else {
1692                         $selflink = array( $this->mTitle->getPrefixedText() );
1693                 }
1694                 $useSubpages = $this->areSubpagesAllowed();
1695                 wfProfileOut( __METHOD__.'-setup' );
1696
1697                 # Loop for each link
1698                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1699                         # Check for excessive memory usage
1700                         if ( $holders->isBig() ) {
1701                                 # Too big
1702                                 # Do the existence check, replace the link holders and clear the array
1703                                 $holders->replace( $s );
1704                                 $holders->clear();
1705                         }
1706
1707                         if ( $useLinkPrefixExtension ) {
1708                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1709                                 if ( preg_match( $e2, $s, $m ) ) {
1710                                         $prefix = $m[2];
1711                                         $s = $m[1];
1712                                 } else {
1713                                         $prefix='';
1714                                 }
1715                                 # first link
1716                                 if ( $first_prefix ) {
1717                                         $prefix = $first_prefix;
1718                                         $first_prefix = false;
1719                                 }
1720                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1721                         }
1722
1723                         $might_be_img = false;
1724
1725                         wfProfileIn( __METHOD__."-e1" );
1726                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1727                                 $text = $m[2];
1728                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1729                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1730                                 # the real problem is with the $e1 regex
1731                                 # See bug 1300.
1732                                 #
1733                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1734                                 # and no image is in sight. See bug 2095.
1735                                 #
1736                                 if ( $text !== '' &&
1737                                         substr( $m[3], 0, 1 ) === ']' &&
1738                                         strpos( $text, '[' ) !== false
1739                                 )
1740                                 {
1741                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1742                                         $m[3] = substr( $m[3], 1 );
1743                                 }
1744                                 # fix up urlencoded title texts
1745                                 if ( strpos( $m[1], '%' ) !== false ) {
1746                                         # Should anchors '#' also be rejected?
1747                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1748                                 }
1749                                 $trail = $m[3];
1750                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1751                                 $might_be_img = true;
1752                                 $text = $m[2];
1753                                 if ( strpos( $m[1], '%' ) !== false ) {
1754                                         $m[1] = rawurldecode( $m[1] );
1755                                 }
1756                                 $trail = "";
1757                         } else { # Invalid form; output directly
1758                                 $s .= $prefix . '[[' . $line ;
1759                                 wfProfileOut( __METHOD__."-e1" );
1760                                 continue;
1761                         }
1762                         wfProfileOut( __METHOD__."-e1" );
1763                         wfProfileIn( __METHOD__."-misc" );
1764
1765                         # Don't allow internal links to pages containing
1766                         # PROTO: where PROTO is a valid URL protocol; these
1767                         # should be external links.
1768                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1769                                 $s .= $prefix . '[[' . $line ;
1770                                 wfProfileOut( __METHOD__."-misc" );
1771                                 continue;
1772                         }
1773
1774                         # Make subpage if necessary
1775                         if ( $useSubpages ) {
1776                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1777                         } else {
1778                                 $link = $m[1];
1779                         }
1780
1781                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1782                         if ( !$noforce ) {
1783                                 # Strip off leading ':'
1784                                 $link = substr( $link, 1 );
1785                         }
1786
1787                         wfProfileOut( __METHOD__."-misc" );
1788                         wfProfileIn( __METHOD__."-title" );
1789                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1790                         if ( $nt === null ) {
1791                                 $s .= $prefix . '[[' . $line;
1792                                 wfProfileOut( __METHOD__."-title" );
1793                                 continue;
1794                         }
1795
1796                         $ns = $nt->getNamespace();
1797                         $iw = $nt->getInterWiki();
1798                         wfProfileOut( __METHOD__."-title" );
1799
1800                         if ( $might_be_img ) { # if this is actually an invalid link
1801                                 wfProfileIn( __METHOD__."-might_be_img" );
1802                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1803                                         $found = false;
1804                                         while ( true ) {
1805                                                 # look at the next 'line' to see if we can close it there
1806                                                 $a->next();
1807                                                 $next_line = $a->current();
1808                                                 if ( $next_line === false || $next_line === null ) {
1809                                                         break;
1810                                                 }
1811                                                 $m = explode( ']]', $next_line, 3 );
1812                                                 if ( count( $m ) == 3 ) {
1813                                                         # the first ]] closes the inner link, the second the image
1814                                                         $found = true;
1815                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1816                                                         $trail = $m[2];
1817                                                         break;
1818                                                 } elseif ( count( $m ) == 2 ) {
1819                                                         # if there's exactly one ]] that's fine, we'll keep looking
1820                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1821                                                 } else {
1822                                                         # if $next_line is invalid too, we need look no further
1823                                                         $text .= '[[' . $next_line;
1824                                                         break;
1825                                                 }
1826                                         }
1827                                         if ( !$found ) {
1828                                                 # we couldn't find the end of this imageLink, so output it raw
1829                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1830                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1831                                                 $s .= "{$prefix}[[$link|$text";
1832                                                 # note: no $trail, because without an end, there *is* no trail
1833                                                 wfProfileOut( __METHOD__."-might_be_img" );
1834                                                 continue;
1835                                         }
1836                                 } else { # it's not an image, so output it raw
1837                                         $s .= "{$prefix}[[$link|$text";
1838                                         # note: no $trail, because without an end, there *is* no trail
1839                                         wfProfileOut( __METHOD__."-might_be_img" );
1840                                         continue;
1841                                 }
1842                                 wfProfileOut( __METHOD__."-might_be_img" );
1843                         }
1844
1845                         $wasblank = ( $text  == '' );
1846                         if ( $wasblank ) {
1847                                 $text = $link;
1848                         } else {
1849                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1850                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1851                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1852                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1853                                 $text = $this->doQuotes( $text );
1854                         }
1855
1856                         # Link not escaped by : , create the various objects
1857                         if ( $noforce ) {
1858
1859                                 # Interwikis
1860                                 wfProfileIn( __METHOD__."-interwiki" );
1861                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1862                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1863                                         $s = rtrim( $s . $prefix );
1864                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1865                                         wfProfileOut( __METHOD__."-interwiki" );
1866                                         continue;
1867                                 }
1868                                 wfProfileOut( __METHOD__."-interwiki" );
1869
1870                                 if ( $ns == NS_FILE ) {
1871                                         wfProfileIn( __METHOD__."-image" );
1872                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1873                                                 if ( $wasblank ) {
1874                                                         # if no parameters were passed, $text
1875                                                         # becomes something like "File:Foo.png",
1876                                                         # which we don't want to pass on to the
1877                                                         # image generator
1878                                                         $text = '';
1879                                                 } else {
1880                                                         # recursively parse links inside the image caption
1881                                                         # actually, this will parse them in any other parameters, too,
1882                                                         # but it might be hard to fix that, and it doesn't matter ATM
1883                                                         $text = $this->replaceExternalLinks( $text );
1884                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1885                                                 }
1886                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1887                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1888                                         } else {
1889                                                 $s .= $prefix . $trail;
1890                                         }
1891                                         $this->mOutput->addImage( $nt->getDBkey() );
1892                                         wfProfileOut( __METHOD__."-image" );
1893                                         continue;
1894
1895                                 }
1896
1897                                 if ( $ns == NS_CATEGORY ) {
1898                                         wfProfileIn( __METHOD__."-category" );
1899                                         $s = rtrim( $s . "\n" ); # bug 87
1900
1901                                         if ( $wasblank ) {
1902                                                 $sortkey = $this->getDefaultSort();
1903                                         } else {
1904                                                 $sortkey = $text;
1905                                         }
1906                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1907                                         $sortkey = str_replace( "\n", '', $sortkey );
1908                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1909                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1910
1911                                         /**
1912                                          * Strip the whitespace Category links produce, see bug 87
1913                                          * @todo We might want to use trim($tmp, "\n") here.
1914                                          */
1915                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1916
1917                                         wfProfileOut( __METHOD__."-category" );
1918                                         continue;
1919                                 }
1920                         }
1921
1922                         # Self-link checking
1923                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1924                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1925                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1926                                         continue;
1927                                 }
1928                         }
1929
1930                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1931                         # FIXME: Should do batch file existence checks, see comment below
1932                         if ( $ns == NS_MEDIA ) {
1933                                 wfProfileIn( __METHOD__."-media" );
1934                                 # Give extensions a chance to select the file revision for us
1935                                 $skip = $time = false;
1936                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1937                                 if ( $skip ) {
1938                                         $link = $sk->link( $nt );
1939                                 } else {
1940                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1941                                 }
1942                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1943                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1944                                 $this->mOutput->addImage( $nt->getDBkey() );
1945                                 wfProfileOut( __METHOD__."-media" );
1946                                 continue;
1947                         }
1948
1949                         wfProfileIn( __METHOD__."-always_known" );
1950                         # Some titles, such as valid special pages or files in foreign repos, should
1951                         # be shown as bluelinks even though they're not included in the page table
1952                         #
1953                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1954                         # batch file existence checks for NS_FILE and NS_MEDIA
1955                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1956                                 $this->mOutput->addLink( $nt );
1957                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1958                         } else {
1959                                 # Links will be added to the output link list after checking
1960                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1961                         }
1962                         wfProfileOut( __METHOD__."-always_known" );
1963                 }
1964                 wfProfileOut( __METHOD__ );
1965                 return $holders;
1966         }
1967
1968         /**
1969          * Make a link placeholder. The text returned can be later resolved to a real link with
1970          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1971          * parsing of interwiki links, and secondly to allow all existence checks and
1972          * article length checks (for stub links) to be bundled into a single query.
1973          *
1974          * @deprecated
1975          */
1976         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1977                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1978         }
1979
1980         /**
1981          * Render a forced-blue link inline; protect against double expansion of
1982          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1983          * Since this little disaster has to split off the trail text to avoid
1984          * breaking URLs in the following text without breaking trails on the
1985          * wiki links, it's been made into a horrible function.
1986          *
1987          * @param $nt Title
1988          * @param $text String
1989          * @param $query String
1990          * @param $trail String
1991          * @param $prefix String
1992          * @return String: HTML-wikitext mix oh yuck
1993          */
1994         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1995                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1996                 $sk = $this->mOptions->getSkin( $this->mTitle );
1997                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1998                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1999                 return $this->armorLinks( $link ) . $trail;
2000         }
2001
2002         /**
2003          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2004          * going to go through further parsing steps before inline URL expansion.
2005          *
2006          * Not needed quite as much as it used to be since free links are a bit
2007          * more sensible these days. But bracketed links are still an issue.
2008          *
2009          * @param $text String: more-or-less HTML
2010          * @return String: less-or-more HTML with NOPARSE bits
2011          */
2012         function armorLinks( $text ) {
2013                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2014                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2015         }
2016
2017         /**
2018          * Return true if subpage links should be expanded on this page.
2019          * @return Boolean
2020          */
2021         function areSubpagesAllowed() {
2022                 # Some namespaces don't allow subpages
2023                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2024         }
2025
2026         /**
2027          * Handle link to subpage if necessary
2028          *
2029          * @param $target String: the source of the link
2030          * @param &$text String: the link text, modified as necessary
2031          * @return string the full name of the link
2032          * @private
2033          */
2034         function maybeDoSubpageLink( $target, &$text ) {
2035                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2036         }
2037
2038         /**#@+
2039          * Used by doBlockLevels()
2040          * @private
2041          */
2042         function closeParagraph() {
2043                 $result = '';
2044                 if ( $this->mLastSection != '' ) {
2045                         $result = '</' . $this->mLastSection  . ">\n";
2046                 }
2047                 $this->mInPre = false;
2048                 $this->mLastSection = '';
2049                 return $result;
2050         }
2051
2052         /**
2053          * getCommon() returns the length of the longest common substring
2054          * of both arguments, starting at the beginning of both.
2055          * @private
2056          */
2057         function getCommon( $st1, $st2 ) {
2058                 $fl = strlen( $st1 );
2059                 $shorter = strlen( $st2 );
2060                 if ( $fl < $shorter ) {
2061                         $shorter = $fl;
2062                 }
2063
2064                 for ( $i = 0; $i < $shorter; ++$i ) {
2065                         if ( $st1{$i} != $st2{$i} ) {
2066                                 break;
2067                         }
2068                 }
2069                 return $i;
2070         }
2071
2072         /**
2073          * These next three functions open, continue, and close the list
2074          * element appropriate to the prefix character passed into them.
2075          * @private
2076          */
2077         function openList( $char ) {
2078                 $result = $this->closeParagraph();
2079
2080                 if ( '*' === $char ) {
2081                         $result .= '<ul><li>';
2082                 } elseif ( '#' === $char ) {
2083                         $result .= '<ol><li>';
2084                 } elseif ( ':' === $char ) {
2085                         $result .= '<dl><dd>';
2086                 } elseif ( ';' === $char ) {
2087                         $result .= '<dl><dt>';
2088                         $this->mDTopen = true;
2089                 } else {
2090                         $result = '<!-- ERR 1 -->';
2091                 }
2092
2093                 return $result;
2094         }
2095
2096         /**
2097          * TODO: document
2098          * @param $char String
2099          * @private
2100          */
2101         function nextItem( $char ) {
2102                 if ( '*' === $char || '#' === $char ) {
2103                         return '</li><li>';
2104                 } elseif ( ':' === $char || ';' === $char ) {
2105                         $close = '</dd>';
2106                         if ( $this->mDTopen ) {
2107                                 $close = '</dt>';
2108                         }
2109                         if ( ';' === $char ) {
2110                                 $this->mDTopen = true;
2111                                 return $close . '<dt>';
2112                         } else {
2113                                 $this->mDTopen = false;
2114                                 return $close . '<dd>';
2115                         }
2116                 }
2117                 return '<!-- ERR 2 -->';
2118         }
2119
2120         /**
2121          * TODO: document
2122          * @param $char String
2123          * @private
2124          */
2125         function closeList( $char ) {
2126                 if ( '*' === $char ) {
2127                         $text = '</li></ul>';
2128                 } elseif ( '#' === $char ) {
2129                         $text = '</li></ol>';
2130                 } elseif ( ':' === $char ) {
2131                         if ( $this->mDTopen ) {
2132                                 $this->mDTopen = false;
2133                                 $text = '</dt></dl>';
2134                         } else {
2135                                 $text = '</dd></dl>';
2136                         }
2137                 } else {
2138                         return '<!-- ERR 3 -->';
2139                 }
2140                 return $text."\n";
2141         }
2142         /**#@-*/
2143
2144         /**
2145          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2146          *
2147          * @param $text String
2148          * @param $linestart Boolean: whether or not this is at the start of a line.
2149          * @private
2150          * @return string the lists rendered as HTML
2151          */
2152         function doBlockLevels( $text, $linestart ) {
2153                 wfProfileIn( __METHOD__ );
2154
2155                 # Parsing through the text line by line.  The main thing
2156                 # happening here is handling of block-level elements p, pre,
2157                 # and making lists from lines starting with * # : etc.
2158                 #
2159                 $textLines = StringUtils::explode( "\n", $text );
2160
2161                 $lastPrefix = $output = '';
2162                 $this->mDTopen = $inBlockElem = false;
2163                 $prefixLength = 0;
2164                 $paragraphStack = false;
2165
2166                 foreach ( $textLines as $oLine ) {
2167                         # Fix up $linestart
2168                         if ( !$linestart ) {
2169                                 $output .= $oLine;
2170                                 $linestart = true;
2171                                 continue;
2172                         }
2173                         # * = ul
2174                         # # = ol
2175                         # ; = dt
2176                         # : = dd
2177
2178                         $lastPrefixLength = strlen( $lastPrefix );
2179                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2180                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2181                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2182                         if ( !$this->mInPre ) {
2183                                 # Multiple prefixes may abut each other for nested lists.
2184                                 $prefixLength = strspn( $oLine, '*#:;' );
2185                                 $prefix = substr( $oLine, 0, $prefixLength );
2186
2187                                 # eh?
2188                                 # ; and : are both from definition-lists, so they're equivalent
2189                                 #  for the purposes of determining whether or not we need to open/close
2190                                 #  elements.
2191                                 $prefix2 = str_replace( ';', ':', $prefix );
2192                                 $t = substr( $oLine, $prefixLength );
2193                                 $this->mInPre = (bool)$preOpenMatch;
2194                         } else {
2195                                 # Don't interpret any other prefixes in preformatted text
2196                                 $prefixLength = 0;
2197                                 $prefix = $prefix2 = '';
2198                                 $t = $oLine;
2199                         }
2200
2201                         # List generation
2202                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2203                                 # Same as the last item, so no need to deal with nesting or opening stuff
2204                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2205                                 $paragraphStack = false;
2206
2207                                 if ( substr( $prefix, -1 ) === ';') {
2208                                         # The one nasty exception: definition lists work like this:
2209                                         # ; title : definition text
2210                                         # So we check for : in the remainder text to split up the
2211                                         # title and definition, without b0rking links.
2212                                         $term = $t2 = '';
2213                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2214                                                 $t = $t2;
2215                                                 $output .= $term . $this->nextItem( ':' );
2216                                         }
2217                                 }
2218                         } elseif ( $prefixLength || $lastPrefixLength ) {
2219                                 # We need to open or close prefixes, or both.
2220
2221                                 # Either open or close a level...
2222                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2223                                 $paragraphStack = false;
2224
2225                                 # Close all the prefixes which aren't shared.
2226                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2227                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2228                                         --$lastPrefixLength;
2229                                 }
2230
2231                                 # Continue the current prefix if appropriate.
2232                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2233                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2234                                 }
2235
2236                                 # Open prefixes where appropriate.
2237                                 while ( $prefixLength > $commonPrefixLength ) {
2238                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2239                                         $output .= $this->openList( $char );
2240
2241                                         if ( ';' === $char ) {
2242                                                 # FIXME: This is dupe of code above
2243                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2244                                                         $t = $t2;
2245                                                         $output .= $term . $this->nextItem( ':' );
2246                                                 }
2247                                         }
2248                                         ++$commonPrefixLength;
2249                                 }
2250                                 $lastPrefix = $prefix2;
2251                         }
2252
2253                         # If we have no prefixes, go to paragraph mode.
2254                         if ( 0 == $prefixLength ) {
2255                                 wfProfileIn( __METHOD__."-paragraph" );
2256                                 # No prefix (not in list)--go to paragraph mode
2257                                 # XXX: use a stack for nestable elements like span, table and div
2258                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2259                                 $closematch = preg_match(
2260                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2261                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2262                                 if ( $openmatch or $closematch ) {
2263
2264                                         $paragraphStack = false;
2265                                         # TODO bug 5718: paragraph closed
2266                                         $output .= $this->closeParagraph();
2267                                         if ( $preOpenMatch and !$preCloseMatch ) {
2268                                                 $this->mInPre = true;
2269                                         }
2270                                         $inBlockElem = !$closematch;
2271                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2272                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2273                                                 # pre
2274                                                 if ( $this->mLastSection !== 'pre' ) {
2275                                                         $paragraphStack = false;
2276                                                         $output .= $this->closeParagraph().'<pre>';
2277                                                         $this->mLastSection = 'pre';
2278                                                 }
2279                                                 $t = substr( $t, 1 );
2280                                         } else {
2281                                                 # paragraph
2282                                                 if ( trim( $t ) === '' ) {
2283                                                         if ( $paragraphStack ) {
2284                                                                 $output .= $paragraphStack.'<br />';
2285                                                                 $paragraphStack = false;
2286                                                                 $this->mLastSection = 'p';
2287                                                         } else {
2288                                                                 if ( $this->mLastSection !== 'p' ) {
2289                                                                         $output .= $this->closeParagraph();
2290                                                                         $this->mLastSection = '';
2291                                                                         $paragraphStack = '<p>';
2292                                                                 } else {
2293                                                                         $paragraphStack = '</p><p>';
2294                                                                 }
2295                                                         }
2296                                                 } else {
2297                                                         if ( $paragraphStack ) {
2298                                                                 $output .= $paragraphStack;
2299                                                                 $paragraphStack = false;
2300                                                                 $this->mLastSection = 'p';
2301                                                         } elseif ( $this->mLastSection !== 'p' ) {
2302                                                                 $output .= $this->closeParagraph().'<p>';
2303                                                                 $this->mLastSection = 'p';
2304                                                         }
2305                                                 }
2306                                         }
2307                                 }
2308                                 wfProfileOut( __METHOD__."-paragraph" );
2309                         }
2310                         # somewhere above we forget to get out of pre block (bug 785)
2311                         if ( $preCloseMatch && $this->mInPre ) {
2312                                 $this->mInPre = false;
2313                         }
2314                         if ( $paragraphStack === false ) {
2315                                 $output .= $t."\n";
2316                         }
2317                 }
2318                 while ( $prefixLength ) {
2319                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2320                         --$prefixLength;
2321                 }
2322                 if ( $this->mLastSection != '' ) {
2323                         $output .= '</' . $this->mLastSection . '>';
2324                         $this->mLastSection = '';
2325                 }
2326
2327                 wfProfileOut( __METHOD__ );
2328                 return $output;
2329         }
2330
2331         /**
2332          * Split up a string on ':', ignoring any occurences inside tags
2333          * to prevent illegal overlapping.
2334          *
2335          * @param $str String: the string to split
2336          * @param &$before String: set to everything before the ':'
2337          * @param &$after String: set to everything after the ':'
2338          * return String: the position of the ':', or false if none found
2339          */
2340         function findColonNoLinks( $str, &$before, &$after ) {
2341                 wfProfileIn( __METHOD__ );
2342
2343                 $pos = strpos( $str, ':' );
2344                 if ( $pos === false ) {
2345                         # Nothing to find!
2346                         wfProfileOut( __METHOD__ );
2347                         return false;
2348                 }
2349
2350                 $lt = strpos( $str, '<' );
2351                 if ( $lt === false || $lt > $pos ) {
2352                         # Easy; no tag nesting to worry about
2353                         $before = substr( $str, 0, $pos );
2354                         $after = substr( $str, $pos+1 );
2355                         wfProfileOut( __METHOD__ );
2356                         return $pos;
2357                 }
2358
2359                 # Ugly state machine to walk through avoiding tags.
2360                 $state = self::COLON_STATE_TEXT;
2361                 $stack = 0;
2362                 $len = strlen( $str );
2363                 for( $i = 0; $i < $len; $i++ ) {
2364                         $c = $str{$i};
2365
2366                         switch( $state ) {
2367                         # (Using the number is a performance hack for common cases)
2368                         case 0: # self::COLON_STATE_TEXT:
2369                                 switch( $c ) {
2370                                 case "<":
2371                                         # Could be either a <start> tag or an </end> tag
2372                                         $state = self::COLON_STATE_TAGSTART;
2373                                         break;
2374                                 case ":":
2375                                         if ( $stack == 0 ) {
2376                                                 # We found it!
2377                                                 $before = substr( $str, 0, $i );
2378                                                 $after = substr( $str, $i + 1 );
2379                                                 wfProfileOut( __METHOD__ );
2380                                                 return $i;
2381                                         }
2382                                         # Embedded in a tag; don't break it.
2383                                         break;
2384                                 default:
2385                                         # Skip ahead looking for something interesting
2386                                         $colon = strpos( $str, ':', $i );
2387                                         if ( $colon === false ) {
2388                                                 # Nothing else interesting
2389                                                 wfProfileOut( __METHOD__ );
2390                                                 return false;
2391                                         }
2392                                         $lt = strpos( $str, '<', $i );
2393                                         if ( $stack === 0 ) {
2394                                                 if ( $lt === false || $colon < $lt ) {
2395                                                         # We found it!
2396                                                         $before = substr( $str, 0, $colon );
2397                                                         $after = substr( $str, $colon + 1 );
2398                                                         wfProfileOut( __METHOD__ );
2399                                                         return $i;
2400                                                 }
2401                                         }
2402                                         if ( $lt === false ) {
2403                                                 # Nothing else interesting to find; abort!
2404                                                 # We're nested, but there's no close tags left. Abort!
2405                                                 break 2;
2406                                         }
2407                                         # Skip ahead to next tag start
2408                                         $i = $lt;
2409                                         $state = self::COLON_STATE_TAGSTART;
2410                                 }
2411                                 break;
2412                         case 1: # self::COLON_STATE_TAG:
2413                                 # In a <tag>
2414                                 switch( $c ) {
2415                                 case ">":
2416                                         $stack++;
2417                                         $state = self::COLON_STATE_TEXT;
2418                                         break;
2419                                 case "/":
2420                                         # Slash may be followed by >?
2421                                         $state = self::COLON_STATE_TAGSLASH;
2422                                         break;
2423                                 default:
2424                                         # ignore
2425                                 }
2426                                 break;
2427                         case 2: # self::COLON_STATE_TAGSTART:
2428                                 switch( $c ) {
2429                                 case "/":
2430                                         $state = self::COLON_STATE_CLOSETAG;
2431                                         break;
2432                                 case "!":
2433                                         $state = self::COLON_STATE_COMMENT;
2434                                         break;
2435                                 case ">":
2436                                         # Illegal early close? This shouldn't happen D:
2437                                         $state = self::COLON_STATE_TEXT;
2438                                         break;
2439                                 default:
2440                                         $state = self::COLON_STATE_TAG;
2441                                 }
2442                                 break;
2443                         case 3: # self::COLON_STATE_CLOSETAG:
2444                                 # In a </tag>
2445                                 if ( $c === ">" ) {
2446                                         $stack--;
2447                                         if ( $stack < 0 ) {
2448                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2449                                                 wfProfileOut( __METHOD__ );
2450                                                 return false;
2451                                         }
2452                                         $state = self::COLON_STATE_TEXT;
2453                                 }
2454                                 break;
2455                         case self::COLON_STATE_TAGSLASH:
2456                                 if ( $c === ">" ) {
2457                                         # Yes, a self-closed tag <blah/>
2458                                         $state = self::COLON_STATE_TEXT;
2459                                 } else {
2460                                         # Probably we're jumping the gun, and this is an attribute
2461                                         $state = self::COLON_STATE_TAG;
2462                                 }
2463                                 break;
2464                         case 5: # self::COLON_STATE_COMMENT:
2465                                 if ( $c === "-" ) {
2466                                         $state = self::COLON_STATE_COMMENTDASH;
2467                                 }
2468                                 break;
2469                         case self::COLON_STATE_COMMENTDASH:
2470                                 if ( $c === "-" ) {
2471                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2472                                 } else {
2473                                         $state = self::COLON_STATE_COMMENT;
2474                                 }
2475                                 break;
2476                         case self::COLON_STATE_COMMENTDASHDASH:
2477                                 if ( $c === ">" ) {
2478                                         $state = self::COLON_STATE_TEXT;
2479                                 } else {
2480                                         $state = self::COLON_STATE_COMMENT;
2481                                 }
2482                                 break;
2483                         default:
2484                                 throw new MWException( "State machine error in " . __METHOD__ );
2485                         }
2486                 }
2487                 if ( $stack > 0 ) {
2488                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2489                         return false;
2490                 }
2491                 wfProfileOut( __METHOD__ );
2492                 return false;
2493         }
2494
2495         /**
2496          * Return value of a magic variable (like PAGENAME)
2497          *
2498          * @private
2499          */
2500         function getVariableValue( $index, $frame=false ) {
2501                 global $wgContLang, $wgSitename, $wgServer;
2502                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2503
2504                 /**
2505                  * Some of these require message or data lookups and can be
2506                  * expensive to check many times.
2507                  */
2508                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2509                         if ( isset( $this->mVarCache[$index] ) ) {
2510                                 return $this->mVarCache[$index];
2511                         }
2512                 }
2513
2514                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2515                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2516
2517                 # Use the time zone
2518                 global $wgLocaltimezone;
2519                 if ( isset( $wgLocaltimezone ) ) {
2520                         $oldtz = date_default_timezone_get();
2521                         date_default_timezone_set( $wgLocaltimezone );
2522                 }
2523
2524                 $localTimestamp = date( 'YmdHis', $ts );
2525                 $localMonth = date( 'm', $ts );
2526                 $localMonth1 = date( 'n', $ts );
2527                 $localMonthName = date( 'n', $ts );
2528                 $localDay = date( 'j', $ts );
2529                 $localDay2 = date( 'd', $ts );
2530                 $localDayOfWeek = date( 'w', $ts );
2531                 $localWeek = date( 'W', $ts );
2532                 $localYear = date( 'Y', $ts );
2533                 $localHour = date( 'H', $ts );
2534                 if ( isset( $wgLocaltimezone ) ) {
2535                         date_default_timezone_set( $oldtz );
2536                 }
2537
2538                 switch ( $index ) {
2539                         case 'currentmonth':
2540                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2541                                 break;
2542                         case 'currentmonth1':
2543                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2544                                 break;
2545                         case 'currentmonthname':
2546                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2547                                 break;
2548                         case 'currentmonthnamegen':
2549                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2550                                 break;
2551                         case 'currentmonthabbrev':
2552                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2553                                 break;
2554                         case 'currentday':
2555                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2556                                 break;
2557                         case 'currentday2':
2558                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2559                                 break;
2560                         case 'localmonth':
2561                                 $value = $wgContLang->formatNum( $localMonth );
2562                                 break;
2563                         case 'localmonth1':
2564                                 $value = $wgContLang->formatNum( $localMonth1 );
2565                                 break;
2566                         case 'localmonthname':
2567                                 $value = $wgContLang->getMonthName( $localMonthName );
2568                                 break;
2569                         case 'localmonthnamegen':
2570                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2571                                 break;
2572                         case 'localmonthabbrev':
2573                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2574                                 break;
2575                         case 'localday':
2576                                 $value = $wgContLang->formatNum( $localDay );
2577                                 break;
2578                         case 'localday2':
2579                                 $value = $wgContLang->formatNum( $localDay2 );
2580                                 break;
2581                         case 'pagename':
2582                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2583                                 break;
2584                         case 'pagenamee':
2585                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2586                                 break;
2587                         case 'fullpagename':
2588                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2589                                 break;
2590                         case 'fullpagenamee':
2591                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2592                                 break;
2593                         case 'subpagename':
2594                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2595                                 break;
2596                         case 'subpagenamee':
2597                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2598                                 break;
2599                         case 'basepagename':
2600                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2601                                 break;
2602                         case 'basepagenamee':
2603                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2604                                 break;
2605                         case 'talkpagename':
2606                                 if ( $this->mTitle->canTalk() ) {
2607                                         $talkPage = $this->mTitle->getTalkPage();
2608                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2609                                 } else {
2610                                         $value = '';
2611                                 }
2612                                 break;
2613                         case 'talkpagenamee':
2614                                 if ( $this->mTitle->canTalk() ) {
2615                                         $talkPage = $this->mTitle->getTalkPage();
2616                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2617                                 } else {
2618                                         $value = '';
2619                                 }
2620                                 break;
2621                         case 'subjectpagename':
2622                                 $subjPage = $this->mTitle->getSubjectPage();
2623                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2624                                 break;
2625                         case 'subjectpagenamee':
2626                                 $subjPage = $this->mTitle->getSubjectPage();
2627                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2628                                 break;
2629                         case 'revisionid':
2630                                 # Let the edit saving system know we should parse the page
2631                                 # *after* a revision ID has been assigned.
2632                                 $this->mOutput->setFlag( 'vary-revision' );
2633                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2634                                 $value = $this->mRevisionId;
2635                                 break;
2636                         case 'revisionday':
2637                                 # Let the edit saving system know we should parse the page
2638                                 # *after* a revision ID has been assigned. This is for null edits.
2639                                 $this->mOutput->setFlag( 'vary-revision' );
2640                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2641                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2642                                 break;
2643                         case 'revisionday2':
2644                                 # Let the edit saving system know we should parse the page
2645                                 # *after* a revision ID has been assigned. This is for null edits.
2646                                 $this->mOutput->setFlag( 'vary-revision' );
2647                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2648                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2649                                 break;
2650                         case 'revisionmonth':
2651                                 # Let the edit saving system know we should parse the page
2652                                 # *after* a revision ID has been assigned. This is for null edits.
2653                                 $this->mOutput->setFlag( 'vary-revision' );
2654                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2655                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2656                                 break;
2657                         case 'revisionmonth1':
2658                                 # Let the edit saving system know we should parse the page
2659                                 # *after* a revision ID has been assigned. This is for null edits.
2660                                 $this->mOutput->setFlag( 'vary-revision' );
2661                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2662                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2663                                 break;
2664                         case 'revisionyear':
2665                                 # Let the edit saving system know we should parse the page
2666                                 # *after* a revision ID has been assigned. This is for null edits.
2667                                 $this->mOutput->setFlag( 'vary-revision' );
2668                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2669                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2670                                 break;
2671                         case 'revisiontimestamp':
2672                                 # Let the edit saving system know we should parse the page
2673                                 # *after* a revision ID has been assigned. This is for null edits.
2674                                 $this->mOutput->setFlag( 'vary-revision' );
2675                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2676                                 $value = $this->getRevisionTimestamp();
2677                                 break;
2678                         case 'revisionuser':
2679                                 # Let the edit saving system know we should parse the page
2680                                 # *after* a revision ID has been assigned. This is for null edits.
2681                                 $this->mOutput->setFlag( 'vary-revision' );
2682                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2683                                 $value = $this->getRevisionUser();
2684                                 break;
2685                         case 'namespace':
2686                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2687                                 break;
2688                         case 'namespacee':
2689                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2690                                 break;
2691                         case 'talkspace':
2692                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2693                                 break;
2694                         case 'talkspacee':
2695                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2696                                 break;
2697                         case 'subjectspace':
2698                                 $value = $this->mTitle->getSubjectNsText();
2699                                 break;
2700                         case 'subjectspacee':
2701                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2702                                 break;
2703                         case 'currentdayname':
2704                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2705                                 break;
2706                         case 'currentyear':
2707                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2708                                 break;
2709                         case 'currenttime':
2710                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2711                                 break;
2712                         case 'currenthour':
2713                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2714                                 break;
2715                         case 'currentweek':
2716                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2717                                 # int to remove the padding
2718                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2719                                 break;
2720                         case 'currentdow':
2721                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2722                                 break;
2723                         case 'localdayname':
2724                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2725                                 break;
2726                         case 'localyear':
2727                                 $value = $wgContLang->formatNum( $localYear, true );
2728                                 break;
2729                         case 'localtime':
2730                                 $value = $wgContLang->time( $localTimestamp, false, false );
2731                                 break;
2732                         case 'localhour':
2733                                 $value = $wgContLang->formatNum( $localHour, true );
2734                                 break;
2735                         case 'localweek':
2736                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2737                                 # int to remove the padding
2738                                 $value = $wgContLang->formatNum( (int)$localWeek );
2739                                 break;
2740                         case 'localdow':
2741                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2742                                 break;
2743                         case 'numberofarticles':
2744                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2745                                 break;
2746                         case 'numberoffiles':
2747                                 $value = $wgContLang->formatNum( SiteStats::images() );
2748                                 break;
2749                         case 'numberofusers':
2750                                 $value = $wgContLang->formatNum( SiteStats::users() );
2751                                 break;
2752                         case 'numberofactiveusers':
2753                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2754                                 break;
2755                         case 'numberofpages':
2756                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2757                                 break;
2758                         case 'numberofadmins':
2759                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2760                                 break;
2761                         case 'numberofedits':
2762                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2763                                 break;
2764                         case 'numberofviews':
2765                                 $value = $wgContLang->formatNum( SiteStats::views() );
2766                                 break;
2767                         case 'currenttimestamp':
2768                                 $value = wfTimestamp( TS_MW, $ts );
2769                                 break;
2770                         case 'localtimestamp':
2771                                 $value = $localTimestamp;
2772                                 break;
2773                         case 'currentversion':
2774                                 $value = SpecialVersion::getVersion();
2775                                 break;
2776                         case 'articlepath':
2777                                 return $wgArticlePath;
2778                         case 'sitename':
2779                                 return $wgSitename;
2780                         case 'server':
2781                                 return $wgServer;
2782                         case 'servername':
2783                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2784                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2785                                 wfRestoreWarnings();
2786                                 return $serverName ? $serverName : $wgServer;
2787                         case 'scriptpath':
2788                                 return $wgScriptPath;
2789                         case 'stylepath':
2790                                 return $wgStylePath;
2791                         case 'directionmark':
2792                                 return $wgContLang->getDirMark();
2793                         case 'contentlanguage':
2794                                 global $wgLanguageCode;
2795                                 return $wgLanguageCode;
2796                         default:
2797                                 $ret = null;
2798                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2799                                         return $ret;
2800                                 } else {
2801                                         return null;
2802                                 }
2803                 }
2804
2805                 if ( $index )
2806                         $this->mVarCache[$index] = $value;
2807
2808                 return $value;
2809         }
2810
2811         /**
2812          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2813          *
2814          * @private
2815          */
2816         function initialiseVariables() {
2817                 wfProfileIn( __METHOD__ );
2818                 $variableIDs = MagicWord::getVariableIDs();
2819                 $substIDs = MagicWord::getSubstIDs();
2820
2821                 $this->mVariables = new MagicWordArray( $variableIDs );
2822                 $this->mSubstWords = new MagicWordArray( $substIDs );
2823                 wfProfileOut( __METHOD__ );
2824         }
2825
2826         /**
2827          * Preprocess some wikitext and return the document tree.
2828          * This is the ghost of replace_variables().
2829          *
2830          * @param $text String: The text to parse
2831          * @param $flags Integer: bitwise combination of:
2832          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2833          *                                     included. Default is to assume a direct page view.
2834          *
2835          * The generated DOM tree must depend only on the input text and the flags.
2836          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2837          *
2838          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2839          * change in the DOM tree for a given text, must be passed through the section identifier
2840          * in the section edit link and thus back to extractSections().
2841          *
2842          * The output of this function is currently only cached in process memory, but a persistent
2843          * cache may be implemented at a later date which takes further advantage of these strict
2844          * dependency requirements.
2845          *
2846          * @private
2847          */
2848         function preprocessToDom( $text, $flags = 0 ) {
2849                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2850                 return $dom;
2851         }
2852
2853         /**
2854          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2855          */
2856         public static function splitWhitespace( $s ) {
2857                 $ltrimmed = ltrim( $s );
2858                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2859                 $trimmed = rtrim( $ltrimmed );
2860                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2861                 if ( $diff > 0 ) {
2862                         $w2 = substr( $ltrimmed, -$diff );
2863                 } else {
2864                         $w2 = '';
2865                 }
2866                 return array( $w1, $trimmed, $w2 );
2867         }
2868
2869         /**
2870          * Replace magic variables, templates, and template arguments
2871          * with the appropriate text. Templates are substituted recursively,
2872          * taking care to avoid infinite loops.
2873          *
2874          * Note that the substitution depends on value of $mOutputType:
2875          *  self::OT_WIKI: only {{subst:}} templates
2876          *  self::OT_PREPROCESS: templates but not extension tags
2877          *  self::OT_HTML: all templates and extension tags
2878          *
2879          * @param $text String: the text to transform
2880          * @param $frame PPFrame Object describing the arguments passed to the template.
2881          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2882          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2883          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2884          * @private
2885          */
2886         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2887                 # Is there any text? Also, Prevent too big inclusions!
2888                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2889                         return $text;
2890                 }
2891                 wfProfileIn( __METHOD__ );
2892
2893                 if ( $frame === false ) {
2894                         $frame = $this->getPreprocessor()->newFrame();
2895                 } elseif ( !( $frame instanceof PPFrame ) ) {
2896                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2897                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2898                 }
2899
2900                 $dom = $this->preprocessToDom( $text );
2901                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2902                 $text = $frame->expand( $dom, $flags );
2903
2904                 wfProfileOut( __METHOD__ );
2905                 return $text;
2906         }
2907
2908         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2909         static function createAssocArgs( $args ) {
2910                 $assocArgs = array();
2911                 $index = 1;
2912                 foreach ( $args as $arg ) {
2913                         $eqpos = strpos( $arg, '=' );
2914                         if ( $eqpos === false ) {
2915                                 $assocArgs[$index++] = $arg;
2916                         } else {
2917                                 $name = trim( substr( $arg, 0, $eqpos ) );
2918                                 $value = trim( substr( $arg, $eqpos+1 ) );
2919                                 if ( $value === false ) {
2920                                         $value = '';
2921                                 }
2922                                 if ( $name !== false ) {
2923                                         $assocArgs[$name] = $value;
2924                                 }
2925                         }
2926                 }
2927
2928                 return $assocArgs;
2929         }
2930
2931         /**
2932          * Warn the user when a parser limitation is reached
2933          * Will warn at most once the user per limitation type
2934          *
2935          * @param $limitationType String: should be one of:
2936          *   'expensive-parserfunction' (corresponding messages:
2937          *       'expensive-parserfunction-warning',
2938          *       'expensive-parserfunction-category')
2939          *   'post-expand-template-argument' (corresponding messages:
2940          *       'post-expand-template-argument-warning',
2941          *       'post-expand-template-argument-category')
2942          *   'post-expand-template-inclusion' (corresponding messages:
2943          *       'post-expand-template-inclusion-warning',
2944          *       'post-expand-template-inclusion-category')
2945          * @param $current Current value
2946          * @param $max Maximum allowed, when an explicit limit has been
2947          *       exceeded, provide the values (optional)
2948          */
2949         function limitationWarn( $limitationType, $current=null, $max=null) {
2950                 # does no harm if $current and $max are present but are unnecessary for the message
2951                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2952                 $this->mOutput->addWarning( $warning );
2953                 $this->addTrackingCategory( "$limitationType-category" );
2954         }
2955
2956         /**
2957          * Return the text of a template, after recursively
2958          * replacing any variables or templates within the template.
2959          *
2960          * @param $piece Array: the parts of the template
2961          *  $piece['title']: the title, i.e. the part before the |
2962          *  $piece['parts']: the parameter array
2963          *  $piece['lineStart']: whether the brace was at the start of a line
2964          * @param $frame PPFrame The current frame, contains template arguments
2965          * @return String: the text of the template
2966          * @private
2967          */
2968         function braceSubstitution( $piece, $frame ) {
2969                 global $wgContLang, $wgNonincludableNamespaces;
2970                 wfProfileIn( __METHOD__ );
2971                 wfProfileIn( __METHOD__.'-setup' );
2972
2973                 # Flags
2974                 $found = false;             # $text has been filled
2975                 $nowiki = false;            # wiki markup in $text should be escaped
2976                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2977                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2978                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2979                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2980
2981                 # Title object, where $text came from
2982                 $title = null;
2983
2984                 # $part1 is the bit before the first |, and must contain only title characters.
2985                 # Various prefixes will be stripped from it later.
2986                 $titleWithSpaces = $frame->expand( $piece['title'] );
2987                 $part1 = trim( $titleWithSpaces );
2988                 $titleText = false;
2989
2990                 # Original title text preserved for various purposes
2991                 $originalTitle = $part1;
2992
2993                 # $args is a list of argument nodes, starting from index 0, not including $part1
2994                 # *** FIXME if piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
2995                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2996                 wfProfileOut( __METHOD__.'-setup' );
2997
2998                 # SUBST
2999                 wfProfileIn( __METHOD__.'-modifiers' );
3000                 if ( !$found ) {
3001
3002                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3003
3004                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3005                         # Decide whether to expand template or keep wikitext as-is.
3006                         if ( $this->ot['wiki'] ) {
3007                                 if ( $substMatch === false ) {
3008                                         $literal = true;  # literal when in PST with no prefix
3009                                 } else {
3010                                         $literal = false; # expand when in PST with subst: or safesubst:
3011                                 }
3012                         } else {
3013                                 if ( $substMatch == 'subst' ) {
3014                                         $literal = true;  # literal when not in PST with plain subst:
3015                                 } else {
3016                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3017                                 }
3018                         }
3019                         if ( $literal ) {
3020                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3021                                 $isLocalObj = true;
3022                                 $found = true;
3023                         }
3024                 }
3025
3026                 # Variables
3027                 if ( !$found && $args->getLength() == 0 ) {
3028                         $id = $this->mVariables->matchStartToEnd( $part1 );
3029                         if ( $id !== false ) {
3030                                 $text = $this->getVariableValue( $id, $frame );
3031                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3032                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3033                                 }
3034                                 $found = true;
3035                         }
3036                 }
3037
3038                 # MSG, MSGNW and RAW
3039                 if ( !$found ) {
3040                         # Check for MSGNW:
3041                         $mwMsgnw = MagicWord::get( 'msgnw' );
3042                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3043                                 $nowiki = true;
3044                         } else {
3045                                 # Remove obsolete MSG:
3046                                 $mwMsg = MagicWord::get( 'msg' );
3047                                 $mwMsg->matchStartAndRemove( $part1 );
3048                         }
3049
3050                         # Check for RAW:
3051                         $mwRaw = MagicWord::get( 'raw' );
3052                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3053                                 $forceRawInterwiki = true;
3054                         }
3055                 }
3056                 wfProfileOut( __METHOD__.'-modifiers' );
3057
3058                 # Parser functions
3059                 if ( !$found ) {
3060                         wfProfileIn( __METHOD__ . '-pfunc' );
3061
3062                         $colonPos = strpos( $part1, ':' );
3063                         if ( $colonPos !== false ) {
3064                                 # Case sensitive functions
3065                                 $function = substr( $part1, 0, $colonPos );
3066                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3067                                         $function = $this->mFunctionSynonyms[1][$function];
3068                                 } else {
3069                                         # Case insensitive functions
3070                                         $function = $wgContLang->lc( $function );
3071                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3072                                                 $function = $this->mFunctionSynonyms[0][$function];
3073                                         } else {
3074                                                 $function = false;
3075                                         }
3076                                 }
3077                                 if ( $function ) {
3078                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3079                                         $initialArgs = array( &$this );
3080                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3081                                         if ( $flags & SFH_OBJECT_ARGS ) {
3082                                                 # Add a frame parameter, and pass the arguments as an array
3083                                                 $allArgs = $initialArgs;
3084                                                 $allArgs[] = $frame;
3085                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3086                                                         $funcArgs[] = $args->item( $i );
3087                                                 }
3088                                                 $allArgs[] = $funcArgs;
3089                                         } else {
3090                                                 # Convert arguments to plain text
3091                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3092                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3093                                                 }
3094                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3095                                         }
3096
3097                                         # Workaround for PHP bug 35229 and similar
3098                                         if ( !is_callable( $callback ) ) {
3099                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3100                                                 wfProfileOut( __METHOD__ );
3101                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3102                                         }
3103                                         $result = call_user_func_array( $callback, $allArgs );
3104                                         $found = true;
3105                                         $noparse = true;
3106                                         $preprocessFlags = 0;
3107
3108                                         if ( is_array( $result ) ) {
3109                                                 if ( isset( $result[0] ) ) {
3110                                                         $text = $result[0];
3111                                                         unset( $result[0] );
3112                                                 }
3113
3114                                                 # Extract flags into the local scope
3115                                                 # This allows callers to set flags such as nowiki, found, etc.
3116                                                 extract( $result );
3117                                         } else {
3118                                                 $text = $result;
3119                                         }
3120                                         if ( !$noparse ) {
3121                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3122                                                 $isChildObj = true;
3123                                         }
3124                                 }
3125                         }
3126                         wfProfileOut( __METHOD__ . '-pfunc' );
3127                 }
3128
3129                 # Finish mangling title and then check for loops.
3130                 # Set $title to a Title object and $titleText to the PDBK
3131                 if ( !$found ) {
3132                         $ns = NS_TEMPLATE;
3133                         # Split the title into page and subpage
3134                         $subpage = '';
3135                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3136                         if ( $subpage !== '' ) {
3137                                 $ns = $this->mTitle->getNamespace();
3138                         }
3139                         $title = Title::newFromText( $part1, $ns );
3140                         if ( $title ) {
3141                                 $titleText = $title->getPrefixedText();
3142                                 # Check for language variants if the template is not found
3143                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3144                                         $wgContLang->findVariantLink( $part1, $title, true );
3145                                 }
3146                                 # Do recursion depth check
3147                                 $limit = $this->mOptions->getMaxTemplateDepth();
3148                                 if ( $frame->depth >= $limit ) {
3149                                         $found = true;
3150                                         $text = '<span class="error">'
3151                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3152                                                 . '</span>';
3153                                 }
3154                         }
3155                 }
3156
3157                 # Load from database
3158                 if ( !$found && $title ) {
3159                         wfProfileIn( __METHOD__ . '-loadtpl' );
3160                         if ( !$title->isExternal() ) {
3161                                 if ( $title->getNamespace() == NS_SPECIAL
3162                                         && $this->mOptions->getAllowSpecialInclusion()
3163                                         && $this->ot['html'] )
3164                                 {
3165                                         $text = SpecialPage::capturePath( $title );
3166                                         if ( is_string( $text ) ) {
3167                                                 $found = true;
3168                                                 $isHTML = true;
3169                                                 $this->disableCache();
3170                                         }
3171                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3172                                         $found = false; # access denied
3173                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3174                                 } else {
3175                                         list( $text, $title ) = $this->getTemplateDom( $title );
3176                                         if ( $text !== false ) {
3177                                                 $found = true;
3178                                                 $isChildObj = true;
3179                                         }
3180                                 }
3181
3182                                 # If the title is valid but undisplayable, make a link to it
3183                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3184                                         $text = "[[:$titleText]]";
3185                                         $found = true;
3186                                 }
3187                         } elseif ( $title->isTrans() ) {
3188                                 # Interwiki transclusion
3189                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3190                                         $text = $this->interwikiTransclude( $title, 'render' );
3191                                         $isHTML = true;
3192                                 } else {
3193                                         $text = $this->interwikiTransclude( $title, 'raw' );
3194                                         # Preprocess it like a template
3195                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3196                                         $isChildObj = true;
3197                                 }
3198                                 $found = true;
3199                         }
3200
3201                         # Do infinite loop check
3202                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3203                         if ( !$frame->loopCheck( $title ) ) {
3204                                 $found = true;
3205                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3206                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3207                         }
3208                         wfProfileOut( __METHOD__ . '-loadtpl' );
3209                 }
3210
3211                 # If we haven't found text to substitute by now, we're done
3212                 # Recover the source wikitext and return it
3213                 if ( !$found ) {
3214                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3215                         wfProfileOut( __METHOD__ );
3216                         return array( 'object' => $text );
3217                 }
3218
3219                 # Expand DOM-style return values in a child frame
3220                 if ( $isChildObj ) {
3221                         # Clean up argument array
3222                         $newFrame = $frame->newChild( $args, $title );
3223
3224                         if ( $nowiki ) {
3225                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3226                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3227                                 # Expansion is eligible for the empty-frame cache
3228                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3229                                         $text = $this->mTplExpandCache[$titleText];
3230                                 } else {
3231                                         $text = $newFrame->expand( $text );
3232                                         $this->mTplExpandCache[$titleText] = $text;
3233                                 }
3234                         } else {
3235                                 # Uncached expansion
3236                                 $text = $newFrame->expand( $text );
3237                         }
3238                 }
3239                 if ( $isLocalObj && $nowiki ) {
3240                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3241                         $isLocalObj = false;
3242                 }
3243
3244                 # Replace raw HTML by a placeholder
3245                 # Add a blank line preceding, to prevent it from mucking up
3246                 # immediately preceding headings
3247                 if ( $isHTML ) {
3248                         $text = "\n\n" . $this->insertStripItem( $text );
3249                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3250                         # Escape nowiki-style return values
3251                         $text = wfEscapeWikiText( $text );
3252                 } elseif ( is_string( $text )
3253                         && !$piece['lineStart']
3254                         && preg_match( '/^{\\|/', $text ) )
3255                 {
3256                         # Bug 529: if the template begins with a table, it should be treated as
3257                         # beginning a new line.  This previously handled other block-level elements
3258                         # such as #, :, etc, but these have many false-positives (bug 12974).
3259                         $text = "\n" . $text;
3260                 }
3261
3262                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3263                         # Error, oversize inclusion
3264                         if ( $titleText !== false ) {
3265                                 # Make a working, properly escaped link if possible (bug 23588)
3266                                 $text = "[[:$titleText]]";
3267                         } else {
3268                                 # This will probably not be a working link, but at least it may
3269                                 # provide some hint of where the problem is
3270                                 preg_replace( '/^:/', '', $originalTitle );
3271                                 $text = "[[:$originalTitle]]";
3272                         }
3273                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3274                         $this->limitationWarn( 'post-expand-template-inclusion' );
3275                 }
3276
3277                 if ( $isLocalObj ) {
3278                         $ret = array( 'object' => $text );
3279                 } else {
3280                         $ret = array( 'text' => $text );
3281                 }
3282
3283                 wfProfileOut( __METHOD__ );
3284                 return $ret;
3285         }
3286
3287         /**
3288          * Get the semi-parsed DOM representation of a template with a given title,
3289          * and its redirect destination title. Cached.
3290          */
3291         function getTemplateDom( $title ) {
3292                 $cacheTitle = $title;
3293                 $titleText = $title->getPrefixedDBkey();
3294
3295                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3296                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3297                         $title = Title::makeTitle( $ns, $dbk );
3298                         $titleText = $title->getPrefixedDBkey();
3299                 }
3300                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3301                         return array( $this->mTplDomCache[$titleText], $title );
3302                 }
3303
3304                 # Cache miss, go to the database
3305                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3306
3307                 if ( $text === false ) {
3308                         $this->mTplDomCache[$titleText] = false;
3309                         return array( false, $title );
3310                 }
3311
3312                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3313                 $this->mTplDomCache[ $titleText ] = $dom;
3314
3315                 if ( !$title->equals( $cacheTitle ) ) {
3316                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3317                                 array( $title->getNamespace(), $title->getDBkey() );
3318                 }
3319
3320                 return array( $dom, $title );
3321         }
3322
3323         /**
3324          * Fetch the unparsed text of a template and register a reference to it.
3325          */
3326         function fetchTemplateAndTitle( $title ) {
3327                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3328                 $stuff = call_user_func( $templateCb, $title, $this );
3329                 $text = $stuff['text'];
3330                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3331                 if ( isset( $stuff['deps'] ) ) {
3332                         foreach ( $stuff['deps'] as $dep ) {
3333                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3334                         }
3335                 }
3336                 return array( $text, $finalTitle );
3337         }
3338
3339         function fetchTemplate( $title ) {
3340                 $rv = $this->fetchTemplateAndTitle( $title );
3341                 return $rv[0];
3342         }
3343
3344         /**
3345          * Static function to get a template
3346          * Can be overridden via ParserOptions::setTemplateCallback().
3347          */
3348         static function statelessFetchTemplate( $title, $parser=false ) {
3349                 $text = $skip = false;
3350                 $finalTitle = $title;
3351                 $deps = array();
3352
3353                 # Loop to fetch the article, with up to 1 redirect
3354                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3355                         # Give extensions a chance to select the revision instead
3356                         $id = false; # Assume current
3357                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3358
3359                         if ( $skip ) {
3360                                 $text = false;
3361                                 $deps[] = array(
3362                                         'title' => $title,
3363                                         'page_id' => $title->getArticleID(),
3364                                         'rev_id' => null );
3365                                 break;
3366                         }
3367                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3368                         $rev_id = $rev ? $rev->getId() : 0;
3369                         # If there is no current revision, there is no page
3370                         if ( $id === false && !$rev ) {
3371                                 $linkCache = LinkCache::singleton();
3372                                 $linkCache->addBadLinkObj( $title );
3373                         }
3374
3375                         $deps[] = array(
3376                                 'title' => $title,
3377                                 'page_id' => $title->getArticleID(),
3378                                 'rev_id' => $rev_id );
3379
3380                         if ( $rev ) {
3381                                 $text = $rev->getText();
3382                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3383                                 global $wgContLang;
3384                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3385                                 if ( !$message->exists() ) {
3386                                         $text = false;
3387                                         break;
3388                                 }
3389                                 $text = $message->plain();
3390                         } else {
3391                                 break;
3392                         }
3393                         if ( $text === false ) {
3394                                 break;
3395                         }
3396                         # Redirect?
3397                         $finalTitle = $title;
3398                         $title = Title::newFromRedirect( $text );
3399                 }
3400                 return array(
3401                         'text' => $text,
3402                         'finalTitle' => $finalTitle,
3403                         'deps' => $deps );
3404         }
3405
3406         /**
3407          * Transclude an interwiki link.
3408          */
3409         function interwikiTransclude( $title, $action ) {
3410                 global $wgEnableScaryTranscluding;
3411
3412                 if ( !$wgEnableScaryTranscluding ) {
3413                         return wfMsgForContent('scarytranscludedisabled');
3414                 }
3415
3416                 $url = $title->getFullUrl( "action=$action" );
3417
3418                 if ( strlen( $url ) > 255 ) {
3419                         return wfMsgForContent( 'scarytranscludetoolong' );
3420                 }
3421                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3422         }
3423
3424         function fetchScaryTemplateMaybeFromCache( $url ) {
3425                 global $wgTranscludeCacheExpiry;
3426                 $dbr = wfGetDB( DB_SLAVE );
3427                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3428                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3429                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3430                 if ( $obj ) {
3431                         return $obj->tc_contents;
3432                 }
3433
3434                 $text = Http::get( $url );
3435                 if ( !$text ) {
3436                         return wfMsgForContent( 'scarytranscludefailed', $url );
3437                 }
3438
3439                 $dbw = wfGetDB( DB_MASTER );
3440                 $dbw->replace( 'transcache', array('tc_url'), array(
3441                         'tc_url' => $url,
3442                         'tc_time' => $dbw->timestamp( time() ),
3443                         'tc_contents' => $text)
3444                 );
3445                 return $text;
3446         }
3447
3448
3449         /**
3450          * Triple brace replacement -- used for template arguments
3451          * @private
3452          */
3453         function argSubstitution( $piece, $frame ) {
3454                 wfProfileIn( __METHOD__ );
3455
3456                 $error = false;
3457                 $parts = $piece['parts'];
3458                 $nameWithSpaces = $frame->expand( $piece['title'] );
3459                 $argName = trim( $nameWithSpaces );
3460                 $object = false;
3461                 $text = $frame->getArgument( $argName );
3462                 if (  $text === false && $parts->getLength() > 0
3463                   && (
3464                         $this->ot['html']
3465                         || $this->ot['pre']
3466                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3467                   )
3468                 ) {
3469                         # No match in frame, use the supplied default
3470                         $object = $parts->item( 0 )->getChildren();
3471                 }
3472                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3473                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3474                         $this->limitationWarn( 'post-expand-template-argument' );
3475                 }
3476
3477                 if ( $text === false && $object === false ) {
3478                         # No match anywhere
3479                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3480                 }
3481                 if ( $error !== false ) {
3482                         $text .= $error;
3483                 }
3484                 if ( $object !== false ) {
3485                         $ret = array( 'object' => $object );
3486                 } else {
3487                         $ret = array( 'text' => $text );
3488                 }
3489
3490                 wfProfileOut( __METHOD__ );
3491                 return $ret;
3492         }
3493
3494         /**
3495          * Return the text to be used for a given extension tag.
3496          * This is the ghost of strip().
3497          *
3498          * @param $params Associative array of parameters:
3499          *     name       PPNode for the tag name
3500          *     attr       PPNode for unparsed text where tag attributes are thought to be
3501          *     attributes Optional associative array of parsed attributes
3502          *     inner      Contents of extension element
3503          *     noClose    Original text did not have a close tag
3504          * @param $frame PPFrame
3505          */
3506         function extensionSubstitution( $params, $frame ) {
3507                 $name = $frame->expand( $params['name'] );
3508                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3509                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3510                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3511
3512                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3513                         ( $this->ot['html'] || $this->ot['pre'] );
3514                 if ( $isFunctionTag ) {
3515                         $markerType = 'none';
3516                 } else {
3517                         $markerType = 'general';
3518                 }
3519                 if ( $this->ot['html'] || $isFunctionTag ) {
3520                         $name = strtolower( $name );
3521                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3522                         if ( isset( $params['attributes'] ) ) {
3523                                 $attributes = $attributes + $params['attributes'];
3524                         }
3525
3526                         if ( isset( $this->mTagHooks[$name] ) ) {
3527                                 # Workaround for PHP bug 35229 and similar
3528                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3529                                         throw new MWException( "Tag hook for $name is not callable\n" );
3530                                 }
3531                                 $output = call_user_func_array( $this->mTagHooks[$name],
3532                                         array( $content, $attributes, $this, $frame ) );
3533                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3534                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3535                                 if ( !is_callable( $callback ) ) {
3536                                         throw new MWException( "Tag hook for $name is not callable\n" );
3537                                 }
3538
3539                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3540                         } else {
3541                                 $output = '<span class="error">Invalid tag extension name: ' .
3542                                         htmlspecialchars( $name ) . '</span>';
3543                         }
3544
3545                         if ( is_array( $output ) ) {
3546                                 # Extract flags to local scope (to override $markerType)
3547                                 $flags = $output;
3548                                 $output = $flags[0];
3549                                 unset( $flags[0] );
3550                                 extract( $flags );
3551                         }
3552                 } else {
3553                         if ( is_null( $attrText ) ) {
3554                                 $attrText = '';
3555                         }
3556                         if ( isset( $params['attributes'] ) ) {
3557                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3558                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3559                                                 htmlspecialchars( $attrValue ) . '"';
3560                                 }
3561                         }
3562                         if ( $content === null ) {
3563                                 $output = "<$name$attrText/>";
3564                         } else {
3565                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3566                                 $output = "<$name$attrText>$content$close";
3567                         }
3568                 }
3569
3570                 if ( $markerType === 'none' ) {
3571                         return $output;
3572                 } elseif ( $markerType === 'nowiki' ) {
3573                         $this->mStripState->nowiki->setPair( $marker, $output );
3574                 } elseif ( $markerType === 'general' ) {
3575                         $this->mStripState->general->setPair( $marker, $output );
3576                 } else {
3577                         throw new MWException( __METHOD__.': invalid marker type' );
3578                 }
3579                 return $marker;
3580         }
3581
3582         /**
3583          * Increment an include size counter
3584          *
3585          * @param $type String: the type of expansion
3586          * @param $size Integer: the size of the text
3587          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3588          */
3589         function incrementIncludeSize( $type, $size ) {
3590                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3591                         return false;
3592                 } else {
3593                         $this->mIncludeSizes[$type] += $size;
3594                         return true;
3595                 }
3596         }
3597
3598         /**
3599          * Increment the expensive function count
3600          *
3601          * @return Boolean: false if the limit has been exceeded
3602          */
3603         function incrementExpensiveFunctionCount() {
3604                 global $wgExpensiveParserFunctionLimit;
3605                 $this->mExpensiveFunctionCount++;
3606                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3607                         return true;
3608                 }
3609                 return false;
3610         }
3611
3612         /**
3613          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3614          * Fills $this->mDoubleUnderscores, returns the modified text
3615          */
3616         function doDoubleUnderscore( $text ) {
3617                 wfProfileIn( __METHOD__ );
3618
3619                 # The position of __TOC__ needs to be recorded
3620                 $mw = MagicWord::get( 'toc' );
3621                 if ( $mw->match( $text ) ) {
3622                         $this->mShowToc = true;
3623                         $this->mForceTocPosition = true;
3624
3625                         # Set a placeholder. At the end we'll fill it in with the TOC.
3626                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3627
3628                         # Only keep the first one.
3629                         $text = $mw->replace( '', $text );
3630                 }
3631
3632                 # Now match and remove the rest of them
3633                 $mwa = MagicWord::getDoubleUnderscoreArray();
3634                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3635
3636                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3637                         $this->mOutput->mNoGallery = true;
3638                 }
3639                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3640                         $this->mShowToc = false;
3641                 }
3642                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3643                         $this->addTrackingCategory( 'hidden-category-category' );
3644                 }
3645                 # (bug 8068) Allow control over whether robots index a page.
3646                 #
3647                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3648                 # is not desirable, the last one on the page should win.
3649                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3650                         $this->mOutput->setIndexPolicy( 'noindex' );
3651                         $this->addTrackingCategory( 'noindex-category' );
3652                 }
3653                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3654                         $this->mOutput->setIndexPolicy( 'index' );
3655                         $this->addTrackingCategory( 'index-category' );
3656                 }
3657
3658                 # Cache all double underscores in the database
3659                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3660                         $this->mOutput->setProperty( $key, '' );
3661                 }
3662
3663                 wfProfileOut( __METHOD__ );
3664                 return $text;
3665         }
3666
3667         /**
3668          * Add a tracking category, getting the title from a system message,
3669          * or print a debug message if the title is invalid.
3670          *
3671          * @param $msg String: message key
3672          * @return Boolean: whether the addition was successful
3673          */
3674         protected function addTrackingCategory( $msg ) {
3675                 $cat = wfMsgForContent( $msg );
3676
3677                 # Allow tracking categories to be disabled by setting them to "-"
3678                 if ( $cat === '-' ) {
3679                         return false;
3680                 }
3681
3682                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3683                 if ( $containerCategory ) {
3684                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3685                         return true;
3686                 } else {
3687                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3688                         return false;
3689                 }
3690         }
3691
3692         /**
3693          * This function accomplishes several tasks:
3694          * 1) Auto-number headings if that option is enabled
3695          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3696          * 3) Add a Table of contents on the top for users who have enabled the option
3697          * 4) Auto-anchor headings
3698          *
3699          * It loops through all headlines, collects the necessary data, then splits up the
3700          * string and re-inserts the newly formatted headlines.
3701          *
3702          * @param $text String
3703          * @param $origText String: original, untouched wikitext
3704          * @param $isMain Boolean
3705          * @private
3706          */
3707         function formatHeadings( $text, $origText, $isMain=true ) {
3708                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3709
3710                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3711
3712                 # Inhibit editsection links if requested in the page
3713                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3714                         $showEditLink = 0;
3715                 } else {
3716                         $showEditLink = $this->mOptions->getEditSection();
3717                 }
3718                 if ( $showEditLink ) {
3719                         $this->mOutput->setEditSectionTokens( true );
3720                 }
3721
3722                 # Get all headlines for numbering them and adding funky stuff like [edit]
3723                 # links - this is for later, but we need the number of headlines right now
3724                 $matches = array();
3725                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3726
3727                 # if there are fewer than 4 headlines in the article, do not show TOC
3728                 # unless it's been explicitly enabled.
3729                 $enoughToc = $this->mShowToc &&
3730                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3731
3732                 # Allow user to stipulate that a page should have a "new section"
3733                 # link added via __NEWSECTIONLINK__
3734                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3735                         $this->mOutput->setNewSection( true );
3736                 }
3737
3738                 # Allow user to remove the "new section"
3739                 # link via __NONEWSECTIONLINK__
3740                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3741                         $this->mOutput->hideNewSection( true );
3742                 }
3743
3744                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3745                 # override above conditions and always show TOC above first header
3746                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3747                         $this->mShowToc = true;
3748                         $enoughToc = true;
3749                 }
3750
3751                 # We need this to perform operations on the HTML
3752                 $sk = $this->mOptions->getSkin( $this->mTitle );
3753
3754                 # headline counter
3755                 $headlineCount = 0;
3756                 $numVisible = 0;
3757
3758                 # Ugh .. the TOC should have neat indentation levels which can be
3759                 # passed to the skin functions. These are determined here
3760                 $toc = '';
3761                 $full = '';
3762                 $head = array();
3763                 $sublevelCount = array();
3764                 $levelCount = array();
3765                 $level = 0;
3766                 $prevlevel = 0;
3767                 $toclevel = 0;
3768                 $prevtoclevel = 0;
3769                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3770                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3771                 $oldType = $this->mOutputType;
3772                 $this->setOutputType( self::OT_WIKI );
3773                 $frame = $this->getPreprocessor()->newFrame();
3774                 $root = $this->preprocessToDom( $origText );
3775                 $node = $root->getFirstChild();
3776                 $byteOffset = 0;
3777                 $tocraw = array();
3778                 $refers = array();
3779
3780                 foreach ( $matches[3] as $headline ) {
3781                         $isTemplate = false;
3782                         $titleText = false;
3783                         $sectionIndex = false;
3784                         $numbering = '';
3785                         $markerMatches = array();
3786                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3787                                 $serial = $markerMatches[1];
3788                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3789                                 $isTemplate = ( $titleText != $baseTitleText );
3790                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3791                         }
3792
3793                         if ( $toclevel ) {
3794                                 $prevlevel = $level;
3795                         }
3796                         $level = $matches[1][$headlineCount];
3797
3798                         if ( $level > $prevlevel ) {
3799                                 # Increase TOC level
3800                                 $toclevel++;
3801                                 $sublevelCount[$toclevel] = 0;
3802                                 if ( $toclevel<$wgMaxTocLevel ) {
3803                                         $prevtoclevel = $toclevel;
3804                                         $toc .= $sk->tocIndent();
3805                                         $numVisible++;
3806                                 }
3807                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3808                                 # Decrease TOC level, find level to jump to
3809
3810                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3811                                         if ( $levelCount[$i] == $level ) {
3812                                                 # Found last matching level
3813                                                 $toclevel = $i;
3814                                                 break;
3815                                         } elseif ( $levelCount[$i] < $level ) {
3816                                                 # Found first matching level below current level
3817                                                 $toclevel = $i + 1;
3818                                                 break;
3819                                         }
3820                                 }
3821                                 if ( $i == 0 ) {
3822                                         $toclevel = 1;
3823                                 }
3824                                 if ( $toclevel<$wgMaxTocLevel ) {
3825                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3826                                                 # Unindent only if the previous toc level was shown :p
3827                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3828                                                 $prevtoclevel = $toclevel;
3829                                         } else {
3830                                                 $toc .= $sk->tocLineEnd();
3831                                         }
3832                                 }
3833                         } else {
3834                                 # No change in level, end TOC line
3835                                 if ( $toclevel<$wgMaxTocLevel ) {
3836                                         $toc .= $sk->tocLineEnd();
3837                                 }
3838                         }
3839
3840                         $levelCount[$toclevel] = $level;
3841
3842                         # count number of headlines for each level
3843                         @$sublevelCount[$toclevel]++;
3844                         $dot = 0;
3845                         for( $i = 1; $i <= $toclevel; $i++ ) {
3846                                 if ( !empty( $sublevelCount[$i] ) ) {
3847                                         if ( $dot ) {
3848                                                 $numbering .= '.';
3849                                         }
3850                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3851                                         $dot = 1;
3852                                 }
3853                         }
3854
3855                         # The safe header is a version of the header text safe to use for links
3856                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3857                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3858
3859                         # Remove link placeholders by the link text.
3860                         #     <!--LINK number-->
3861                         # turns into
3862                         #     link text with suffix
3863                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3864
3865                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3866                         $tocline = preg_replace(
3867                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3868                                 array( '',                          '<$1>' ),
3869                                 $safeHeadline
3870                         );
3871                         $tocline = trim( $tocline );
3872
3873                         # For the anchor, strip out HTML-y stuff period
3874                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3875                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
3876
3877                         # Save headline for section edit hint before it's escaped
3878                         $headlineHint = $safeHeadline;
3879
3880                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3881                                 # For reverse compatibility, provide an id that's
3882                                 # HTML4-compatible, like we used to.
3883                                 #
3884                                 # It may be worth noting, academically, that it's possible for
3885                                 # the legacy anchor to conflict with a non-legacy headline
3886                                 # anchor on the page.  In this case likely the "correct" thing
3887                                 # would be to either drop the legacy anchors or make sure
3888                                 # they're numbered first.  However, this would require people
3889                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3890                                 # manually, so let's not bother worrying about it.
3891                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3892                                         array( 'noninitial', 'legacy' ) );
3893                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3894
3895                                 if ( $legacyHeadline == $safeHeadline ) {
3896                                         # No reason to have both (in fact, we can't)
3897                                         $legacyHeadline = false;
3898                                 }
3899                         } else {
3900                                 $legacyHeadline = false;
3901                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3902                                         'noninitial' );
3903                         }
3904
3905                         # HTML names must be case-insensitively unique (bug 10721).
3906                         # This does not apply to Unicode characters per
3907                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
3908                         # FIXME: We may be changing them depending on the current locale.
3909                         $arrayKey = strtolower( $safeHeadline );
3910                         if ( $legacyHeadline === false ) {
3911                                 $legacyArrayKey = false;
3912                         } else {
3913                                 $legacyArrayKey = strtolower( $legacyHeadline );
3914                         }
3915
3916                         # count how many in assoc. array so we can track dupes in anchors
3917                         if ( isset( $refers[$arrayKey] ) ) {
3918                                 $refers[$arrayKey]++;
3919                         } else {
3920                                 $refers[$arrayKey] = 1;
3921                         }
3922                         if ( isset( $refers[$legacyArrayKey] ) ) {
3923                                 $refers[$legacyArrayKey]++;
3924                         } else {
3925                                 $refers[$legacyArrayKey] = 1;
3926                         }
3927
3928                         # Don't number the heading if it is the only one (looks silly)
3929                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3930                                 # the two are different if the line contains a link
3931                                 $headline = $numbering . ' ' . $headline;
3932                         }
3933
3934                         # Create the anchor for linking from the TOC to the section
3935                         $anchor = $safeHeadline;
3936                         $legacyAnchor = $legacyHeadline;
3937                         if ( $refers[$arrayKey] > 1 ) {
3938                                 $anchor .= '_' . $refers[$arrayKey];
3939                         }
3940                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3941                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3942                         }
3943                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3944                                 $toc .= $sk->tocLine( $anchor, $tocline,
3945                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3946                         }
3947
3948                         # Add the section to the section tree
3949                         # Find the DOM node for this header
3950                         while ( $node && !$isTemplate ) {
3951                                 if ( $node->getName() === 'h' ) {
3952                                         $bits = $node->splitHeading();
3953                                         if ( $bits['i'] == $sectionIndex ) {
3954                                                 break;
3955                                         }
3956                                 }
3957                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3958                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3959                                 $node = $node->getNextSibling();
3960                         }
3961                         $tocraw[] = array(
3962                                 'toclevel' => $toclevel,
3963                                 'level' => $level,
3964                                 'line' => $tocline,
3965                                 'number' => $numbering,
3966                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3967                                 'fromtitle' => $titleText,
3968                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3969                                 'anchor' => $anchor,
3970                         );
3971
3972                         # give headline the correct <h#> tag
3973                         if ( $showEditLink && $sectionIndex !== false ) {
3974                                 // Output edit section links as markers with styles that can be customized by skins
3975                                 if ( $isTemplate ) {
3976                                         # Put a T flag in the section identifier, to indicate to extractSections()
3977                                         # that sections inside <includeonly> should be counted.
3978                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
3979                                 } else {
3980                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
3981                                 }
3982                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
3983                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
3984                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
3985                                 // so we don't have to worry about a user trying to input one of these markers directly.
3986                                 // We use a page and section attribute to stop the language converter from converting these important bits
3987                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
3988                                 // be able to convert that piece of data.
3989                                 $editlink = '<editsection page="' . htmlspecialchars($editlinkArgs[0]);
3990                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
3991                                 if ( isset($editlinkArgs[2]) ) {
3992                                         $editlink .= '>' . $editlinkArgs[2] . '</editsection>';
3993                                 } else {
3994                                         $editlink .= '/>';
3995                                 }
3996                         } else {
3997                                 $editlink = '';
3998                         }
3999                         $head[$headlineCount] = $sk->makeHeadline( $level,
4000                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4001                                 $editlink, $legacyAnchor );
4002
4003                         $headlineCount++;
4004                 }
4005
4006                 $this->setOutputType( $oldType );
4007
4008                 # Never ever show TOC if no headers
4009                 if ( $numVisible < 1 ) {
4010                         $enoughToc = false;
4011                 }
4012
4013                 if ( $enoughToc ) {
4014                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4015                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
4016                         }
4017                         $toc = $sk->tocList( $toc );
4018                         $this->mOutput->setTOCHTML( $toc );
4019                 }
4020
4021                 if ( $isMain ) {
4022                         $this->mOutput->setSections( $tocraw );
4023                 }
4024
4025                 # split up and insert constructed headlines
4026
4027                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4028                 $i = 0;
4029
4030                 foreach ( $blocks as $block ) {
4031                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4032                                 # This is the [edit] link that appears for the top block of text when
4033                                 # section editing is enabled
4034
4035                                 # Disabled because it broke block formatting
4036                                 # For example, a bullet point in the top line
4037                                 # $full .= $sk->editSectionLink(0);
4038                         }
4039                         $full .= $block;
4040                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4041                                 # Top anchor now in skin
4042                                 $full = $full.$toc;
4043                         }
4044
4045                         if ( !empty( $head[$i] ) ) {
4046                                 $full .= $head[$i];
4047                         }
4048                         $i++;
4049                 }
4050                 if ( $this->mForceTocPosition ) {
4051                         return str_replace( '<!--MWTOC-->', $toc, $full );
4052                 } else {
4053                         return $full;
4054                 }
4055         }
4056
4057         /**
4058          * Transform wiki markup when saving a page by doing \r\n -> \n
4059          * conversion, substitting signatures, {{subst:}} templates, etc.
4060          *
4061          * @param $text String: the text to transform
4062          * @param $title Title: the Title object for the current article
4063          * @param $user User: the User object describing the current user
4064          * @param $options ParserOptions: parsing options
4065          * @param $clearState Boolean: whether to clear the parser state first
4066          * @return String: the altered wiki markup
4067          */
4068         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4069                 $this->startExternalParse( $title, $options, self::OT_WIKI, $clearState );
4070                 $this->setUser( $user );
4071
4072                 $pairs = array(
4073                         "\r\n" => "\n",
4074                 );
4075                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4076                 if( $options->getPreSaveTransform() ) {
4077                         $text = $this->pstPass2( $text, $user );
4078                 }
4079                 $text = $this->mStripState->unstripBoth( $text );
4080
4081                 $this->setUser( null ); #Reset
4082
4083                 return $text;
4084         }
4085
4086         /**
4087          * Pre-save transform helper function
4088          * @private
4089          */
4090         function pstPass2( $text, $user ) {
4091                 global $wgContLang, $wgLocaltimezone;
4092
4093                 # Note: This is the timestamp saved as hardcoded wikitext to
4094                 # the database, we use $wgContLang here in order to give
4095                 # everyone the same signature and use the default one rather
4096                 # than the one selected in each user's preferences.
4097                 # (see also bug 12815)
4098                 $ts = $this->mOptions->getTimestamp();
4099                 if ( isset( $wgLocaltimezone ) ) {
4100                         $tz = $wgLocaltimezone;
4101                 } else {
4102                         $tz = date_default_timezone_get();
4103                 }
4104
4105                 $unixts = wfTimestamp( TS_UNIX, $ts );
4106                 $oldtz = date_default_timezone_get();
4107                 date_default_timezone_set( $tz );
4108                 $ts = date( 'YmdHis', $unixts );
4109                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4110
4111                 # Allow translation of timezones through wiki. date() can return
4112                 # whatever crap the system uses, localised or not, so we cannot
4113                 # ship premade translations.
4114                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4115                 $msg = wfMessage( $key )->inContentLanguage();
4116                 if ( $msg->exists() ) {
4117                         $tzMsg = $msg->text();
4118                 }
4119
4120                 date_default_timezone_set( $oldtz );
4121
4122                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4123
4124                 # Variable replacement
4125                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4126                 $text = $this->replaceVariables( $text );
4127
4128                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4129                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4130
4131                 # Signatures
4132                 $sigText = $this->getUserSig( $user );
4133                 $text = strtr( $text, array(
4134                         '~~~~~' => $d,
4135                         '~~~~' => "$sigText $d",
4136                         '~~~' => $sigText
4137                 ) );
4138
4139                 # Context links: [[|name]] and [[name (context)|]]
4140                 global $wgLegalTitleChars;
4141                 $tc = "[$wgLegalTitleChars]";
4142                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4143
4144                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4145                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4146                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4147                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4148
4149                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4150                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4151                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4152                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4153
4154                 $t = $this->mTitle->getText();
4155                 $m = array();
4156                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4157                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4158                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4159                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4160                 } else {
4161                         # if there's no context, don't bother duplicating the title
4162                         $text = preg_replace( $p2, '[[\\1]]', $text );
4163                 }
4164
4165                 # Trim trailing whitespace
4166                 $text = rtrim( $text );
4167
4168                 return $text;
4169         }
4170
4171         /**
4172          * Fetch the user's signature text, if any, and normalize to
4173          * validated, ready-to-insert wikitext.
4174          * If you have pre-fetched the nickname or the fancySig option, you can
4175          * specify them here to save a database query.
4176          * Do not reuse this parser instance after calling getUserSig(),
4177          * as it may have changed if it's the $wgParser.
4178          *
4179          * @param $user User
4180          * @param $nickname String: nickname to use or false to use user's default nickname
4181          * @param $fancySig Boolean: whether the nicknname is the complete signature
4182          *                  or null to use default value
4183          * @return string
4184          */
4185         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4186                 global $wgMaxSigChars;
4187
4188                 $username = $user->getName();
4189
4190                 # If not given, retrieve from the user object.
4191                 if ( $nickname === false )
4192                         $nickname = $user->getOption( 'nickname' );
4193
4194                 if ( is_null( $fancySig ) ) {
4195                         $fancySig = $user->getBoolOption( 'fancysig' );
4196                 }
4197
4198                 $nickname = $nickname == null ? $username : $nickname;
4199
4200                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4201                         $nickname = $username;
4202                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4203                 } elseif ( $fancySig !== false ) {
4204                         # Sig. might contain markup; validate this
4205                         if ( $this->validateSig( $nickname ) !== false ) {
4206                                 # Validated; clean up (if needed) and return it
4207                                 return $this->cleanSig( $nickname, true );
4208                         } else {
4209                                 # Failed to validate; fall back to the default
4210                                 $nickname = $username;
4211                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4212                         }
4213                 }
4214
4215                 # Make sure nickname doesnt get a sig in a sig
4216                 $nickname = $this->cleanSigInSig( $nickname );
4217
4218                 # If we're still here, make it a link to the user page
4219                 $userText = wfEscapeWikiText( $username );
4220                 $nickText = wfEscapeWikiText( $nickname );
4221                 if ( $user->isAnon() )  {
4222                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4223                 } else {
4224                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4225                 }
4226         }
4227
4228         /**
4229          * Check that the user's signature contains no bad XML
4230          *
4231          * @param $text String
4232          * @return mixed An expanded string, or false if invalid.
4233          */
4234         function validateSig( $text ) {
4235                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4236         }
4237
4238         /**
4239          * Clean up signature text
4240          *
4241          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4242          * 2) Substitute all transclusions
4243          *
4244          * @param $text String
4245          * @param $parsing Whether we're cleaning (preferences save) or parsing
4246          * @return String: signature text
4247          */
4248         function cleanSig( $text, $parsing = false ) {
4249                 if ( !$parsing ) {
4250                         global $wgTitle;
4251                         $this->mOptions = new ParserOptions;
4252                         $this->clearState();
4253                         $this->setTitle( $wgTitle );
4254                         $this->setOutputType = self::OT_PREPROCESS;
4255                 }
4256
4257                 # Option to disable this feature
4258                 if ( !$this->mOptions->getCleanSignatures() ) {
4259                         return $text;
4260                 }
4261
4262                 # FIXME: regex doesn't respect extension tags or nowiki
4263                 #  => Move this logic to braceSubstitution()
4264                 $substWord = MagicWord::get( 'subst' );
4265                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4266                 $substText = '{{' . $substWord->getSynonym( 0 );
4267
4268                 $text = preg_replace( $substRegex, $substText, $text );
4269                 $text = $this->cleanSigInSig( $text );
4270                 $dom = $this->preprocessToDom( $text );
4271                 $frame = $this->getPreprocessor()->newFrame();
4272                 $text = $frame->expand( $dom );
4273
4274                 if ( !$parsing ) {
4275                         $text = $this->mStripState->unstripBoth( $text );
4276                 }
4277
4278                 return $text;
4279         }
4280
4281         /**
4282          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4283          *
4284          * @param $text String
4285          * @return String: signature text with /~{3,5}/ removed
4286          */
4287         function cleanSigInSig( $text ) {
4288                 $text = preg_replace( '/~{3,5}/', '', $text );
4289                 return $text;
4290         }
4291
4292         /**
4293          * Set up some variables which are usually set up in parse()
4294          * so that an external function can call some class members with confidence
4295          */
4296         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4297                 $this->setTitle( $title );
4298                 $this->mOptions = $options;
4299                 $this->setOutputType( $outputType );
4300                 if ( $clearState ) {
4301                         $this->clearState();
4302                 }
4303         }
4304
4305         /**
4306          * Wrapper for preprocess()
4307          *
4308          * @param $text String: the text to preprocess
4309          * @param $options ParserOptions: options
4310          * @return String
4311          */
4312         public function transformMsg( $text, $options ) {
4313                 global $wgTitle;
4314                 static $executing = false;
4315
4316                 # Guard against infinite recursion
4317                 if ( $executing ) {
4318                         return $text;
4319                 }
4320                 $executing = true;
4321
4322                 wfProfileIn( __METHOD__ );
4323                 $text = $this->preprocess( $text, $wgTitle, $options );
4324
4325                 $executing = false;
4326                 wfProfileOut( __METHOD__ );
4327                 return $text;
4328         }
4329
4330         /**
4331          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4332          * The callback should have the following form:
4333          *    function myParserHook( $text, $params, $parser ) { ... }
4334          *
4335          * Transform and return $text. Use $parser for any required context, e.g. use
4336          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4337          *
4338          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4339          * @param $callback Mixed: the callback function (and object) to use for the tag
4340          * @return The old value of the mTagHooks array associated with the hook
4341          */
4342         public function setHook( $tag, $callback ) {
4343                 $tag = strtolower( $tag );
4344                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4345                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4346                 $this->mTagHooks[$tag] = $callback;
4347                 if ( !in_array( $tag, $this->mStripList ) ) {
4348                         $this->mStripList[] = $tag;
4349                 }
4350
4351                 return $oldVal;
4352         }
4353
4354         function setTransparentTagHook( $tag, $callback ) {
4355                 $tag = strtolower( $tag );
4356                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4357                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4358                 $this->mTransparentTagHooks[$tag] = $callback;
4359
4360                 return $oldVal;
4361         }
4362
4363         /**
4364          * Remove all tag hooks
4365          */
4366         function clearTagHooks() {
4367                 $this->mTagHooks = array();
4368                 $this->mStripList = $this->mDefaultStripList;
4369         }
4370
4371         /**
4372          * Create a function, e.g. {{sum:1|2|3}}
4373          * The callback function should have the form:
4374          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4375          *
4376          * Or with SFH_OBJECT_ARGS:
4377          *    function myParserFunction( $parser, $frame, $args ) { ... }
4378          *
4379          * The callback may either return the text result of the function, or an array with the text
4380          * in element 0, and a number of flags in the other elements. The names of the flags are
4381          * specified in the keys. Valid flags are:
4382          *   found                     The text returned is valid, stop processing the template. This
4383          *                             is on by default.
4384          *   nowiki                    Wiki markup in the return value should be escaped
4385          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4386          *
4387          * @param $id String: The magic word ID
4388          * @param $callback Mixed: the callback function (and object) to use
4389          * @param $flags Integer: a combination of the following flags:
4390          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4391          *
4392          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4393          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4394          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4395          *     the arguments, and to control the way they are expanded.
4396          *
4397          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4398          *     arguments, for instance:
4399          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4400          *
4401          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4402          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4403          *     working if/when this is changed.
4404          *
4405          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4406          *     expansion.
4407          *
4408          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4409          *     about the methods available in PPFrame and PPNode.
4410          *
4411          * @return The old callback function for this name, if any
4412          */
4413         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4414                 global $wgContLang;
4415
4416                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4417                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4418
4419                 # Add to function cache
4420                 $mw = MagicWord::get( $id );
4421                 if ( !$mw )
4422                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4423
4424                 $synonyms = $mw->getSynonyms();
4425                 $sensitive = intval( $mw->isCaseSensitive() );
4426
4427                 foreach ( $synonyms as $syn ) {
4428                         # Case
4429                         if ( !$sensitive ) {
4430                                 $syn = $wgContLang->lc( $syn );
4431                         }
4432                         # Add leading hash
4433                         if ( !( $flags & SFH_NO_HASH ) ) {
4434                                 $syn = '#' . $syn;
4435                         }
4436                         # Remove trailing colon
4437                         if ( substr( $syn, -1, 1 ) === ':' ) {
4438                                 $syn = substr( $syn, 0, -1 );
4439                         }
4440                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4441                 }
4442                 return $oldVal;
4443         }
4444
4445         /**
4446          * Get all registered function hook identifiers
4447          *
4448          * @return Array
4449          */
4450         function getFunctionHooks() {
4451                 return array_keys( $this->mFunctionHooks );
4452         }
4453
4454         /**
4455          * Create a tag function, e.g. <test>some stuff</test>.
4456          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4457          * Unlike parser functions, their content is not preprocessed.
4458          */
4459         function setFunctionTagHook( $tag, $callback, $flags ) {
4460                 $tag = strtolower( $tag );
4461                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4462                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4463                         $this->mFunctionTagHooks[$tag] : null;
4464                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4465
4466                 if ( !in_array( $tag, $this->mStripList ) ) {
4467                         $this->mStripList[] = $tag;
4468                 }
4469
4470                 return $old;
4471         }
4472
4473         /**
4474          * FIXME: update documentation. makeLinkObj() is deprecated.
4475          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4476          * Placeholders created in Skin::makeLinkObj()
4477          * Returns an array of link CSS classes, indexed by PDBK.
4478          */
4479         function replaceLinkHolders( &$text, $options = 0 ) {
4480                 return $this->mLinkHolders->replace( $text );
4481         }
4482
4483         /**
4484          * Replace <!--LINK--> link placeholders with plain text of links
4485          * (not HTML-formatted).
4486          *
4487          * @param $text String
4488          * @return String
4489          */
4490         function replaceLinkHoldersText( $text ) {
4491                 return $this->mLinkHolders->replaceText( $text );
4492         }
4493
4494         /**
4495          * Renders an image gallery from a text with one line per image.
4496          * text labels may be given by using |-style alternative text. E.g.
4497          *   Image:one.jpg|The number "1"
4498          *   Image:tree.jpg|A tree
4499          * given as text will return the HTML of a gallery with two images,
4500          * labeled 'The number "1"' and
4501          * 'A tree'.
4502          */
4503         function renderImageGallery( $text, $params ) {
4504                 $ig = new ImageGallery();
4505                 $ig->setContextTitle( $this->mTitle );
4506                 $ig->setShowBytes( false );
4507                 $ig->setShowFilename( false );
4508                 $ig->setParser( $this );
4509                 $ig->setHideBadImages();
4510                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4511                 $ig->useSkin( $this->mOptions->getSkin( $this->mTitle ) );
4512                 $ig->mRevisionId = $this->mRevisionId;
4513
4514                 if ( isset( $params['showfilename'] ) ) {
4515                         $ig->setShowFilename( true );
4516                 } else {
4517                         $ig->setShowFilename( false );
4518                 }
4519                 if ( isset( $params['caption'] ) ) {
4520                         $caption = $params['caption'];
4521                         $caption = htmlspecialchars( $caption );
4522                         $caption = $this->replaceInternalLinks( $caption );
4523                         $ig->setCaptionHtml( $caption );
4524                 }
4525                 if ( isset( $params['perrow'] ) ) {
4526                         $ig->setPerRow( $params['perrow'] );
4527                 }
4528                 if ( isset( $params['widths'] ) ) {
4529                         $ig->setWidths( $params['widths'] );
4530                 }
4531                 if ( isset( $params['heights'] ) ) {
4532                         $ig->setHeights( $params['heights'] );
4533                 }
4534
4535                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4536
4537                 $lines = StringUtils::explode( "\n", $text );
4538                 foreach ( $lines as $line ) {
4539                         # match lines like these:
4540                         # Image:someimage.jpg|This is some image
4541                         $matches = array();
4542                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4543                         # Skip empty lines
4544                         if ( count( $matches ) == 0 ) {
4545                                 continue;
4546                         }
4547
4548                         if ( strpos( $matches[0], '%' ) !== false ) {
4549                                 $matches[1] = rawurldecode( $matches[1] );
4550                         }
4551                         $tp = Title::newFromText( $matches[1], NS_FILE );
4552                         $nt =& $tp;
4553                         if ( is_null( $nt ) ) {
4554                                 # Bogus title. Ignore these so we don't bomb out later.
4555                                 continue;
4556                         }
4557                         if ( isset( $matches[3] ) ) {
4558                                 $label = $matches[3];
4559                         } else {
4560                                 $label = '';
4561                         }
4562
4563                         $html = $this->recursiveTagParse( trim( $label ) );
4564
4565                         $ig->add( $nt, $html );
4566
4567                         # Only add real images (bug #5586)
4568                         if ( $nt->getNamespace() == NS_FILE ) {
4569                                 $this->mOutput->addImage( $nt->getDBkey() );
4570                         }
4571                 }
4572                 return $ig->toHTML();
4573         }
4574
4575         function getImageParams( $handler ) {
4576                 if ( $handler ) {
4577                         $handlerClass = get_class( $handler );
4578                 } else {
4579                         $handlerClass = '';
4580                 }
4581                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4582                         # Initialise static lists
4583                         static $internalParamNames = array(
4584                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4585                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4586                                         'bottom', 'text-bottom' ),
4587                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4588                                         'upright', 'border', 'link', 'alt' ),
4589                         );
4590                         static $internalParamMap;
4591                         if ( !$internalParamMap ) {
4592                                 $internalParamMap = array();
4593                                 foreach ( $internalParamNames as $type => $names ) {
4594                                         foreach ( $names as $name ) {
4595                                                 $magicName = str_replace( '-', '_', "img_$name" );
4596                                                 $internalParamMap[$magicName] = array( $type, $name );
4597                                         }
4598                                 }
4599                         }
4600
4601                         # Add handler params
4602                         $paramMap = $internalParamMap;
4603                         if ( $handler ) {
4604                                 $handlerParamMap = $handler->getParamMap();
4605                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4606                                         $paramMap[$magic] = array( 'handler', $paramName );
4607                                 }
4608                         }
4609                         $this->mImageParams[$handlerClass] = $paramMap;
4610                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4611                 }
4612                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4613         }
4614
4615         /**
4616          * Parse image options text and use it to make an image
4617          *
4618          * @param $title Title
4619          * @param $options String
4620          * @param $holders LinkHolderArray
4621          */
4622         function makeImage( $title, $options, $holders = false ) {
4623                 # Check if the options text is of the form "options|alt text"
4624                 # Options are:
4625                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4626                 #  * left       no resizing, just left align. label is used for alt= only
4627                 #  * right      same, but right aligned
4628                 #  * none       same, but not aligned
4629                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4630                 #  * center     center the image
4631                 #  * frame      Keep original image size, no magnify-button.
4632                 #  * framed     Same as "frame"
4633                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4634                 #  * upright    reduce width for upright images, rounded to full __0 px
4635                 #  * border     draw a 1px border around the image
4636                 #  * alt        Text for HTML alt attribute (defaults to empty)
4637                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4638                 # vertical-align values (no % or length right now):
4639                 #  * baseline
4640                 #  * sub
4641                 #  * super
4642                 #  * top
4643                 #  * text-top
4644                 #  * middle
4645                 #  * bottom
4646                 #  * text-bottom
4647
4648                 $parts = StringUtils::explode( "|", $options );
4649                 $sk = $this->mOptions->getSkin( $this->mTitle );
4650
4651                 # Give extensions a chance to select the file revision for us
4652                 $skip = $time = $descQuery = false;
4653                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4654
4655                 if ( $skip ) {
4656                         return $sk->link( $title );
4657                 }
4658
4659                 # Get the file
4660                 $file = wfFindFile( $title, array( 'time' => $time ) );
4661                 # Get parameter map
4662                 $handler = $file ? $file->getHandler() : false;
4663
4664                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4665
4666                 # Process the input parameters
4667                 $caption = '';
4668                 $params = array( 'frame' => array(), 'handler' => array(),
4669                         'horizAlign' => array(), 'vertAlign' => array() );
4670                 foreach ( $parts as $part ) {
4671                         $part = trim( $part );
4672                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4673                         $validated = false;
4674                         if ( isset( $paramMap[$magicName] ) ) {
4675                                 list( $type, $paramName ) = $paramMap[$magicName];
4676
4677                                 # Special case; width and height come in one variable together
4678                                 if ( $type === 'handler' && $paramName === 'width' ) {
4679                                         $m = array();
4680                                         # (bug 13500) In both cases (width/height and width only),
4681                                         # permit trailing "px" for backward compatibility.
4682                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4683                                                 $width = intval( $m[1] );
4684                                                 $height = intval( $m[2] );
4685                                                 if ( $handler->validateParam( 'width', $width ) ) {
4686                                                         $params[$type]['width'] = $width;
4687                                                         $validated = true;
4688                                                 }
4689                                                 if ( $handler->validateParam( 'height', $height ) ) {
4690                                                         $params[$type]['height'] = $height;
4691                                                         $validated = true;
4692                                                 }
4693                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4694                                                 $width = intval( $value );
4695                                                 if ( $handler->validateParam( 'width', $width ) ) {
4696                                                         $params[$type]['width'] = $width;
4697                                                         $validated = true;
4698                                                 }
4699                                         } # else no validation -- bug 13436
4700                                 } else {
4701                                         if ( $type === 'handler' ) {
4702                                                 # Validate handler parameter
4703                                                 $validated = $handler->validateParam( $paramName, $value );
4704                                         } else {
4705                                                 # Validate internal parameters
4706                                                 switch( $paramName ) {
4707                                                 case 'manualthumb':
4708                                                 case 'alt':
4709                                                         # @todo Fixme: possibly check validity here for
4710                                                         # manualthumb? downstream behavior seems odd with
4711                                                         # missing manual thumbs.
4712                                                         $validated = true;
4713                                                         $value = $this->stripAltText( $value, $holders );
4714                                                         break;
4715                                                 case 'link':
4716                                                         $chars = self::EXT_LINK_URL_CLASS;
4717                                                         $prots = $this->mUrlProtocols;
4718                                                         if ( $value === '' ) {
4719                                                                 $paramName = 'no-link';
4720                                                                 $value = true;
4721                                                                 $validated = true;
4722                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4723                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4724                                                                         $paramName = 'link-url';
4725                                                                         $this->mOutput->addExternalLink( $value );
4726                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4727                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4728                                                                         }
4729                                                                         $validated = true;
4730                                                                 }
4731                                                         } else {
4732                                                                 $linkTitle = Title::newFromText( $value );
4733                                                                 if ( $linkTitle ) {
4734                                                                         $paramName = 'link-title';
4735                                                                         $value = $linkTitle;
4736                                                                         $this->mOutput->addLink( $linkTitle );
4737                                                                         $validated = true;
4738                                                                 }
4739                                                         }
4740                                                         break;
4741                                                 default:
4742                                                         # Most other things appear to be empty or numeric...
4743                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4744                                                 }
4745                                         }
4746
4747                                         if ( $validated ) {
4748                                                 $params[$type][$paramName] = $value;
4749                                         }
4750                                 }
4751                         }
4752                         if ( !$validated ) {
4753                                 $caption = $part;
4754                         }
4755                 }
4756
4757                 # Process alignment parameters
4758                 if ( $params['horizAlign'] ) {
4759                         $params['frame']['align'] = key( $params['horizAlign'] );
4760                 }
4761                 if ( $params['vertAlign'] ) {
4762                         $params['frame']['valign'] = key( $params['vertAlign'] );
4763                 }
4764
4765                 $params['frame']['caption'] = $caption;
4766
4767                 # Will the image be presented in a frame, with the caption below?
4768                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4769                                                  isset( $params['frame']['framed'] ) ||
4770                                                  isset( $params['frame']['thumbnail'] ) ||
4771                                                  isset( $params['frame']['manualthumb'] );
4772
4773                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4774                 # came to also set the caption, ordinary text after the image -- which
4775                 # makes no sense, because that just repeats the text multiple times in
4776                 # screen readers.  It *also* came to set the title attribute.
4777                 #
4778                 # Now that we have an alt attribute, we should not set the alt text to
4779                 # equal the caption: that's worse than useless, it just repeats the
4780                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4781                 # use the unnamed parameter for alt text as well, just for the time be-
4782                 # ing, if the unnamed param is set and the alt param is not.
4783                 #
4784                 # For the future, we need to figure out if we want to tweak this more,
4785                 # e.g., introducing a title= parameter for the title; ignoring the un-
4786                 # named parameter entirely for images without a caption; adding an ex-
4787                 # plicit caption= parameter and preserving the old magic unnamed para-
4788                 # meter for BC; ...
4789                 if ( $imageIsFramed ) { # Framed image
4790                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4791                                 # No caption or alt text, add the filename as the alt text so
4792                                 # that screen readers at least get some description of the image
4793                                 $params['frame']['alt'] = $title->getText();
4794                         }
4795                         # Do not set $params['frame']['title'] because tooltips don't make sense
4796                         # for framed images
4797                 } else { # Inline image
4798                         if ( !isset( $params['frame']['alt'] ) ) {
4799                                 # No alt text, use the "caption" for the alt text
4800                                 if ( $caption !== '') {
4801                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4802                                 } else {
4803                                         # No caption, fall back to using the filename for the
4804                                         # alt text
4805                                         $params['frame']['alt'] = $title->getText();
4806                                 }
4807                         }
4808                         # Use the "caption" for the tooltip text
4809                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4810                 }
4811
4812                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4813
4814                 # Linker does the rest
4815                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery, $this->mOptions->getThumbSize() );
4816
4817                 # Give the handler a chance to modify the parser object
4818                 if ( $handler ) {
4819                         $handler->parserTransformHook( $this, $file );
4820                 }
4821
4822                 return $ret;
4823         }
4824
4825         protected function stripAltText( $caption, $holders ) {
4826                 # Strip bad stuff out of the title (tooltip).  We can't just use
4827                 # replaceLinkHoldersText() here, because if this function is called
4828                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4829                 if ( $holders ) {
4830                         $tooltip = $holders->replaceText( $caption );
4831                 } else {
4832                         $tooltip = $this->replaceLinkHoldersText( $caption );
4833                 }
4834
4835                 # make sure there are no placeholders in thumbnail attributes
4836                 # that are later expanded to html- so expand them now and
4837                 # remove the tags
4838                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4839                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4840
4841                 return $tooltip;
4842         }
4843
4844         /**
4845          * Set a flag in the output object indicating that the content is dynamic and
4846          * shouldn't be cached.
4847          */
4848         function disableCache() {
4849                 wfDebug( "Parser output marked as uncacheable.\n" );
4850                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
4851                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
4852         }
4853
4854         /**
4855          * Callback from the Sanitizer for expanding items found in HTML attribute
4856          * values, so they can be safely tested and escaped.
4857          *
4858          * @param $text String
4859          * @param $frame PPFrame
4860          * @return String
4861          * @private
4862          */
4863         function attributeStripCallback( &$text, $frame = false ) {
4864                 $text = $this->replaceVariables( $text, $frame );
4865                 $text = $this->mStripState->unstripBoth( $text );
4866                 return $text;
4867         }
4868
4869         /**
4870          * Accessor
4871          */
4872         function getTags() {
4873                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4874         }
4875
4876         /**
4877          * Break wikitext input into sections, and either pull or replace
4878          * some particular section's text.
4879          *
4880          * External callers should use the getSection and replaceSection methods.
4881          *
4882          * @param $text String: Page wikitext
4883          * @param $section String: a section identifier string of the form:
4884          *   <flag1> - <flag2> - ... - <section number>
4885          *
4886          * Currently the only recognised flag is "T", which means the target section number
4887          * was derived during a template inclusion parse, in other words this is a template
4888          * section edit link. If no flags are given, it was an ordinary section edit link.
4889          * This flag is required to avoid a section numbering mismatch when a section is
4890          * enclosed by <includeonly> (bug 6563).
4891          *
4892          * The section number 0 pulls the text before the first heading; other numbers will
4893          * pull the given section along with its lower-level subsections. If the section is
4894          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4895          *
4896          * @param $mode String: one of "get" or "replace"
4897          * @param $newText String: replacement text for section data.
4898          * @return String: for "get", the extracted section text.
4899          *                 for "replace", the whole page with the section replaced.
4900          */
4901         private function extractSections( $text, $section, $mode, $newText='' ) {
4902                 global $wgTitle; # not generally used but removes an ugly failure mode
4903                 $this->startExternalParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
4904                 $outText = '';
4905                 $frame = $this->getPreprocessor()->newFrame();
4906
4907                 # Process section extraction flags
4908                 $flags = 0;
4909                 $sectionParts = explode( '-', $section );
4910                 $sectionIndex = array_pop( $sectionParts );
4911                 foreach ( $sectionParts as $part ) {
4912                         if ( $part === 'T' ) {
4913                                 $flags |= self::PTD_FOR_INCLUSION;
4914                         }
4915                 }
4916                 # Preprocess the text
4917                 $root = $this->preprocessToDom( $text, $flags );
4918
4919                 # <h> nodes indicate section breaks
4920                 # They can only occur at the top level, so we can find them by iterating the root's children
4921                 $node = $root->getFirstChild();
4922
4923                 # Find the target section
4924                 if ( $sectionIndex == 0 ) {
4925                         # Section zero doesn't nest, level=big
4926                         $targetLevel = 1000;
4927                 } else {
4928                         while ( $node ) {
4929                                 if ( $node->getName() === 'h' ) {
4930                                         $bits = $node->splitHeading();
4931                                         if ( $bits['i'] == $sectionIndex ) {
4932                                                 $targetLevel = $bits['level'];
4933                                                 break;
4934                                         }
4935                                 }
4936                                 if ( $mode === 'replace' ) {
4937                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4938                                 }
4939                                 $node = $node->getNextSibling();
4940                         }
4941                 }
4942
4943                 if ( !$node ) {
4944                         # Not found
4945                         if ( $mode === 'get' ) {
4946                                 return $newText;
4947                         } else {
4948                                 return $text;
4949                         }
4950                 }
4951
4952                 # Find the end of the section, including nested sections
4953                 do {
4954                         if ( $node->getName() === 'h' ) {
4955                                 $bits = $node->splitHeading();
4956                                 $curLevel = $bits['level'];
4957                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4958                                         break;
4959                                 }
4960                         }
4961                         if ( $mode === 'get' ) {
4962                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4963                         }
4964                         $node = $node->getNextSibling();
4965                 } while ( $node );
4966
4967                 # Write out the remainder (in replace mode only)
4968                 if ( $mode === 'replace' ) {
4969                         # Output the replacement text
4970                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4971                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4972                         # Only add trailing whitespace if there is newText
4973                         if ( $newText != "" ) {
4974                                 $outText .= $newText . "\n\n";
4975                         }
4976
4977                         while ( $node ) {
4978                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4979                                 $node = $node->getNextSibling();
4980                         }
4981                 }
4982
4983                 if ( is_string( $outText ) ) {
4984                         # Re-insert stripped tags
4985                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4986                 }
4987
4988                 return $outText;
4989         }
4990
4991         /**
4992          * This function returns the text of a section, specified by a number ($section).
4993          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4994          * the first section before any such heading (section 0).
4995          *
4996          * If a section contains subsections, these are also returned.
4997          *
4998          * @param $text String: text to look in
4999          * @param $section String: section identifier
5000          * @param $deftext String: default to return if section is not found
5001          * @return string text of the requested section
5002          */
5003         public function getSection( $text, $section, $deftext='' ) {
5004                 return $this->extractSections( $text, $section, "get", $deftext );
5005         }
5006
5007         /**
5008          * This function returns $oldtext after the content of the section
5009          * specified by $section has been replaced with $text.
5010          *
5011          * @param $text String: former text of the article
5012          * @param $section Numeric: section identifier
5013          * @param $text String: replacing text
5014          * #return String: modified text
5015          */
5016         public function replaceSection( $oldtext, $section, $text ) {
5017                 return $this->extractSections( $oldtext, $section, "replace", $text );
5018         }
5019
5020         /**
5021          * Get the ID of the revision we are parsing
5022          *
5023          * @return Mixed: integer or null
5024          */
5025         function getRevisionId() {
5026                 return $this->mRevisionId;
5027         }
5028
5029         /**
5030          * Get the revision object for $this->mRevisionId
5031          *
5032          * @return either a Revision object or null
5033          */
5034         protected function getRevisionObject() {
5035                 if ( !is_null( $this->mRevisionObject ) ) {
5036                         return $this->mRevisionObject;
5037                 }
5038                 if ( is_null( $this->mRevisionId ) ) {
5039                         return null;
5040                 }
5041
5042                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5043                 return $this->mRevisionObject;
5044         }
5045
5046         /**
5047          * Get the timestamp associated with the current revision, adjusted for
5048          * the default server-local timestamp
5049          */
5050         function getRevisionTimestamp() {
5051                 if ( is_null( $this->mRevisionTimestamp ) ) {
5052                         wfProfileIn( __METHOD__ );
5053
5054                         $revObject = $this->getRevisionObject();
5055                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5056
5057                         if( $timestamp !== false ) {
5058                                 global $wgContLang;
5059
5060                                 # The cryptic '' timezone parameter tells to use the site-default
5061                                 # timezone offset instead of the user settings.
5062                                 #
5063                                 # Since this value will be saved into the parser cache, served
5064                                 # to other users, and potentially even used inside links and such,
5065                                 # it needs to be consistent for all visitors.
5066                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5067                         }
5068
5069                         wfProfileOut( __METHOD__ );
5070                 }
5071                 return $this->mRevisionTimestamp;
5072         }
5073
5074         /**
5075          * Get the name of the user that edited the last revision
5076          *
5077          * @return String: user name
5078          */
5079         function getRevisionUser() {
5080                 if( is_null( $this->mRevisionUser ) ) {
5081                         $revObject = $this->getRevisionObject();
5082
5083                         # if this template is subst: the revision id will be blank,
5084                         # so just use the current user's name
5085                         if( $revObject ) {
5086                                 $this->mRevisionUser = $revObject->getUserText();
5087                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5088                                 $this->mRevisionUser = $this->getUser()->getName();
5089                         }
5090                 }
5091                 return $this->mRevisionUser;
5092         }
5093
5094         /**
5095          * Mutator for $mDefaultSort
5096          *
5097          * @param $sort New value
5098          */
5099         public function setDefaultSort( $sort ) {
5100                 $this->mDefaultSort = $sort;
5101                 $this->mOutput->setProperty( 'defaultsort', $sort );
5102         }
5103
5104         /**
5105          * Accessor for $mDefaultSort
5106          * Will use the title/prefixed title if none is set
5107          *
5108          * @return string
5109          */
5110         public function getDefaultSort() {
5111                 if ( $this->mDefaultSort !== false ) {
5112                         return $this->mDefaultSort;
5113                 } else {
5114                         return $this->mTitle->getCategorySortkey();
5115                 }
5116         }
5117
5118         /**
5119          * Accessor for $mDefaultSort
5120          * Unlike getDefaultSort(), will return false if none is set
5121          *
5122          * @return string or false
5123          */
5124         public function getCustomDefaultSort() {
5125                 return $this->mDefaultSort;
5126         }
5127
5128         /**
5129          * Try to guess the section anchor name based on a wikitext fragment
5130          * presumably extracted from a heading, for example "Header" from
5131          * "== Header ==".
5132          */
5133         public function guessSectionNameFromWikiText( $text ) {
5134                 # Strip out wikitext links(they break the anchor)
5135                 $text = $this->stripSectionName( $text );
5136                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5137                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5138         }
5139
5140         /**
5141          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5142          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5143          * as something other than UTF-8 (apparently?), resulting in breakage.
5144          *
5145          * @param $text String: The section name
5146          * @return string An anchor
5147          */
5148         public function guessLegacySectionNameFromWikiText( $text ) {
5149                 # Strip out wikitext links(they break the anchor)
5150                 $text = $this->stripSectionName( $text );
5151                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5152                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5153         }
5154
5155         /**
5156          * Strips a text string of wikitext for use in a section anchor
5157          *
5158          * Accepts a text string and then removes all wikitext from the
5159          * string and leaves only the resultant text (i.e. the result of
5160          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5161          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5162          * to create valid section anchors by mimicing the output of the
5163          * parser when headings are parsed.
5164          *
5165          * @param $text String: text string to be stripped of wikitext
5166          * for use in a Section anchor
5167          * @return Filtered text string
5168          */
5169         public function stripSectionName( $text ) {
5170                 # Strip internal link markup
5171                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5172                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5173
5174                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5175                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5176                 # on how many empty links there are on the page - need to figure that out.
5177                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5178
5179                 # Parse wikitext quotes (italics & bold)
5180                 $text = $this->doQuotes( $text );
5181
5182                 # Strip HTML tags
5183                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5184                 return $text;
5185         }
5186
5187         /**
5188          * strip/replaceVariables/unstrip for preprocessor regression testing
5189          */
5190         function testSrvus( $text, $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5191                 if ( !$title instanceof Title ) {
5192                         $title = Title::newFromText( $title );
5193                 }
5194                 $this->startExternalParse( $title, $options, $outputType, true );
5195
5196                 $text = $this->replaceVariables( $text );
5197                 $text = $this->mStripState->unstripBoth( $text );
5198                 $text = Sanitizer::removeHTMLtags( $text );
5199                 return $text;
5200         }
5201
5202         function testPst( $text, $title, $options ) {
5203                 global $wgUser;
5204                 if ( !$title instanceof Title ) {
5205                         $title = Title::newFromText( $title );
5206                 }
5207                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5208         }
5209
5210         function testPreprocess( $text, $title, $options ) {
5211                 if ( !$title instanceof Title ) {
5212                         $title = Title::newFromText( $title );
5213                 }
5214                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5215         }
5216
5217         function markerSkipCallback( $s, $callback ) {
5218                 $i = 0;
5219                 $out = '';
5220                 while ( $i < strlen( $s ) ) {
5221                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5222                         if ( $markerStart === false ) {
5223                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5224                                 break;
5225                         } else {
5226                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5227                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5228                                 if ( $markerEnd === false ) {
5229                                         $out .= substr( $s, $markerStart );
5230                                         break;
5231                                 } else {
5232                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5233                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5234                                         $i = $markerEnd;
5235                                 }
5236                         }
5237                 }
5238                 return $out;
5239         }
5240
5241         function serialiseHalfParsedText( $text ) {
5242                 $data = array();
5243                 $data['text'] = $text;
5244
5245                 # First, find all strip markers, and store their
5246                 #  data in an array.
5247                 $stripState = new StripState;
5248                 $pos = 0;
5249                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5250                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5251                 {
5252                         $end_pos += strlen( self::MARKER_SUFFIX );
5253                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5254
5255                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5256                                 $replaceArray = $stripState->general;
5257                                 $stripText = $this->mStripState->general->data[$marker];
5258                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5259                                 $replaceArray = $stripState->nowiki;
5260                                 $stripText = $this->mStripState->nowiki->data[$marker];
5261                         } else {
5262                                 throw new MWException( "Hanging strip marker: '$marker'." );
5263                         }
5264
5265                         $replaceArray->setPair( $marker, $stripText );
5266                         $pos = $end_pos;
5267                 }
5268                 $data['stripstate'] = $stripState;
5269
5270                 # Now, find all of our links, and store THEIR
5271                 #  data in an array! :)
5272                 $links = array( 'internal' => array(), 'interwiki' => array() );
5273                 $pos = 0;
5274
5275                 # Internal links
5276                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5277                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5278
5279                         $ns = trim( $ns );
5280                         if ( empty( $links['internal'][$ns] ) ) {
5281                                 $links['internal'][$ns] = array();
5282                         }
5283
5284                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5285                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5286                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5287                 }
5288
5289                 $pos = 0;
5290
5291                 # Interwiki links
5292                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5293                         $data = substr( $text, $start_pos );
5294                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5295                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5296                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5297                 }
5298
5299                 $data['linkholder'] = $links;
5300
5301                 return $data;
5302         }
5303
5304         /**
5305          * TODO: document
5306          * @param $data Array
5307          * @param $intPrefix String unique identifying prefix
5308          * @return String
5309          */
5310         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5311                 if ( !$intPrefix ) {
5312                         $intPrefix = self::getRandomString();
5313                 }
5314
5315                 # First, extract the strip state.
5316                 $stripState = $data['stripstate'];
5317                 $this->mStripState->general->merge( $stripState->general );
5318                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5319
5320                 # Now, extract the text, and renumber links
5321                 $text = $data['text'];
5322                 $links = $data['linkholder'];
5323
5324                 # Internal...
5325                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5326                         foreach ( $nsLinks as $key => $entry ) {
5327                                 $newKey = $intPrefix . '-' . $key;
5328                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5329
5330                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5331                         }
5332                 }
5333
5334                 # Interwiki...
5335                 foreach ( $links['interwiki'] as $key => $entry ) {
5336                         $newKey = "$intPrefix-$key";
5337                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5338
5339                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5340                 }
5341
5342                 # Should be good to go.
5343                 return $text;
5344         }
5345 }
5346
5347 /**
5348  * @todo document, briefly.
5349  * @ingroup Parser
5350  */
5351 class StripState {
5352         var $general, $nowiki;
5353
5354         function __construct() {
5355                 $this->general = new ReplacementArray;
5356                 $this->nowiki = new ReplacementArray;
5357         }
5358
5359         function unstripGeneral( $text ) {
5360                 wfProfileIn( __METHOD__ );
5361                 do {
5362                         $oldText = $text;
5363                         $text = $this->general->replace( $text );
5364                 } while ( $text !== $oldText );
5365                 wfProfileOut( __METHOD__ );
5366                 return $text;
5367         }
5368
5369         function unstripNoWiki( $text ) {
5370                 wfProfileIn( __METHOD__ );
5371                 do {
5372                         $oldText = $text;
5373                         $text = $this->nowiki->replace( $text );
5374                 } while ( $text !== $oldText );
5375                 wfProfileOut( __METHOD__ );
5376                 return $text;
5377         }
5378
5379         function unstripBoth( $text ) {
5380                 wfProfileIn( __METHOD__ );
5381                 do {
5382                         $oldText = $text;
5383                         $text = $this->general->replace( $text );
5384                         $text = $this->nowiki->replace( $text );
5385                 } while ( $text !== $oldText );
5386                 wfProfileOut( __METHOD__ );
5387                 return $text;
5388         }
5389 }
5390
5391 /**
5392  * @todo document, briefly.
5393  * @ingroup Parser
5394  */
5395 class OnlyIncludeReplacer {
5396         var $output = '';
5397
5398         function replace( $matches ) {
5399                 if ( substr( $matches[1], -1 ) === "\n" ) {
5400                         $this->output .= substr( $matches[1], 0, -1 );
5401                 } else {
5402                         $this->output .= $matches[1];
5403                 }
5404         }
5405 }