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 $wgUser or $wgTitle. Keep them away!
  38  *
  39  * settings:
  40  *  $wgUseDynamicDates*, $wgInterwikiMagic*,
  41  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  42  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  43  *  $wgMaxArticleSize*
  44  *
  45  *  * only within ParserOptions
  46  * </pre>
  47  *
  48  * @ingroup Parser
  49  */
  50 class Parser {
  51         /**
  52          * Update this version number when the ParserOutput format
  53          * changes in an incompatible way, so the parser cache
  54          * can automatically discard old data.
  55          */
  56         const VERSION = '1.6.5';
  57
  58         /**
  59          * Update this version number when the output of serialiseHalfParsedText()
  60          * changes in an incompatible way
  61          */
  62         const HALF_PARSED_VERSION = 2;
  63
  64         # Flags for Parser::setFunctionHook
  65         # Also available as global constants from Defines.php
  66         const SFH_NO_HASH = 1;
  67         const SFH_OBJECT_ARGS = 2;
  68
  69         # Constants needed for external link processing
  70         # Everything except bracket, space, or control characters
  71         # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
  72         # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
  73         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
  74         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
  75                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
  76
  77         # State constants for the definition list colon extraction
  78         const COLON_STATE_TEXT = 0;
  79         const COLON_STATE_TAG = 1;
  80         const COLON_STATE_TAGSTART = 2;
  81         const COLON_STATE_CLOSETAG = 3;
  82         const COLON_STATE_TAGSLASH = 4;
  83         const COLON_STATE_COMMENT = 5;
  84         const COLON_STATE_COMMENTDASH = 6;
  85         const COLON_STATE_COMMENTDASHDASH = 7;
  86
  87         # Flags for preprocessToDom
  88         const PTD_FOR_INCLUSION = 1;
  89
  90         # Allowed values for $this->mOutputType
  91         # Parameter to startExternalParse().
  92         const OT_HTML = 1; # like parse()
  93         const OT_WIKI = 2; # like preSaveTransform()
  94         const OT_PREPROCESS = 3; # like preprocess()
  95         const OT_MSG = 3;
  96         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  97
  98         # Marker Suffix needs to be accessible staticly.
  99         const MARKER_SUFFIX = "-QINU\x7f";
 100
 101         # Persistent:
 102         var $mTagHooks = array();
 103         var $mTransparentTagHooks = array();
 104         var $mFunctionHooks = array();
 105         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
 106         var $mFunctionTagHooks = array();
 107         var $mStripList  = array();
 108         var $mDefaultStripList  = array();
 109         var $mVarCache = array();
 110         var $mImageParams = array();
 111         var $mImageParamsMagicArray = array();
 112         var $mMarkerIndex = 0;
 113         var $mFirstCall = true;
 114
 115         # Initialised by initialiseVariables()
 116
 117         /**
 118          * @var MagicWordArray
 119          */
 120         var $mVariables;
 121
 122         /**
 123          * @var MagicWordArray
 124          */
 125         var $mSubstWords;
 126         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 127
 128         # Cleared with clearState():
 129         /**
 130          * @var ParserOutput
 131          */
 132         var $mOutput;
 133         var $mAutonumber, $mDTopen;
 134
 135         /**
 136          * @var StripState
 137          */
 138         var $mStripState;
 139
 140         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 141         /**
 142          * @var LinkHolderArray
 143          */
 144         var $mLinkHolders;
 145
 146         var $mLinkID;
 147         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 148         var $mTplExpandCache; # empty-frame expansion cache
 149         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 150         var $mExpensiveFunctionCount; # number of expensive parser function calls
 151
 152         /**
 153          * @var User
 154          */
 155         var $mUser; # User object; only used when doing pre-save transform
 156
 157         # Temporary
 158         # These are variables reset at least once per parse regardless of $clearState
 159
 160         /**
 161          * @var ParserOptions
 162          */
 163         var $mOptions;
 164
 165         /**
 166          * @var Title
 167          */
 168         var $mTitle;        # Title context, used for self-link rendering and similar things
 169         var $mOutputType;   # Output type, one of the OT_xxx constants
 170         var $ot;            # Shortcut alias, see setOutputType()
 171         var $mRevisionObject; # The revision object of the specified revision ID
 172         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 173         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 174         var $mRevisionUser; # User to display in {{REVISIONUSER}} tag
 175         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 176
 177         /**
 178          * @var string
 179          */
 180         var $mUniqPrefix;
 181
 182         /**
 183          * Constructor
 184          */
 185         public function __construct( $conf = array() ) {
 186                 $this->mConf = $conf;
 187                 $this->mUrlProtocols = wfUrlProtocols();
 188                 $this->mExtLinkBracketedRegex = '/\[((' . wfUrlProtocols() . ')'.
 189                         self::EXT_LINK_URL_CLASS.'+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
 190                 if ( isset( $conf['preprocessorClass'] ) ) {
 191                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 192                 } elseif ( defined( 'MW_COMPILED' ) ) {
 193                         # Preprocessor_Hash is much faster than Preprocessor_DOM in compiled mode
 194                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 195                 } elseif ( extension_loaded( 'domxml' ) ) {
 196                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 197                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 198                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 199                 } elseif ( extension_loaded( 'dom' ) ) {
 200                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 201                 } else {
 202                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 203                 }
 204                 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
 205         }
 206
 207         /**
 208          * Reduce memory usage to reduce the impact of circular references
 209          */
 210         function __destruct() {
 211                 if ( isset( $this->mLinkHolders ) ) {
 212                         unset( $this->mLinkHolders );
 213                 }
 214                 foreach ( $this as $name => $value ) {
 215                         unset( $this->$name );
 216                 }
 217         }
 218
 219         /**
 220          * Do various kinds of initialisation on the first call of the parser
 221          */
 222         function firstCallInit() {
 223                 if ( !$this->mFirstCall ) {
 224                         return;
 225                 }
 226                 $this->mFirstCall = false;
 227
 228                 wfProfileIn( __METHOD__ );
 229
 230                 CoreParserFunctions::register( $this );
 231                 CoreTagHooks::register( $this );
 232                 $this->initialiseVariables();
 233
 234                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 235                 wfProfileOut( __METHOD__ );
 236         }
 237
 238         /**
 239          * Clear Parser state
 240          *
 241          * @private
 242          */
 243         function clearState() {
 244                 wfProfileIn( __METHOD__ );
 245                 if ( $this->mFirstCall ) {
 246                         $this->firstCallInit();
 247                 }
 248                 $this->mOutput = new ParserOutput;
 249                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 250                 $this->mAutonumber = 0;
 251                 $this->mLastSection = '';
 252                 $this->mDTopen = false;
 253                 $this->mIncludeCount = array();
 254                 $this->mArgStack = false;
 255                 $this->mInPre = false;
 256                 $this->mLinkHolders = new LinkHolderArray( $this );
 257                 $this->mLinkID = 0;
 258                 $this->mRevisionObject = $this->mRevisionTimestamp =
 259                         $this->mRevisionId = $this->mRevisionUser = null;
 260                 $this->mVarCache = array();
 261                 $this->mUser = null;
 262
 263                 /**
 264                  * Prefix for temporary replacement strings for the multipass parser.
 265                  * \x07 should never appear in input as it's disallowed in XML.
 266                  * Using it at the front also gives us a little extra robustness
 267                  * since it shouldn't match when butted up against identifier-like
 268                  * string constructs.
 269                  *
 270                  * Must not consist of all title characters, or else it will change
 271                  * the behaviour of <nowiki> in a link.
 272                  */
 273                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 274                 # Changed to \x7f to allow XML double-parsing -- TS
 275                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 276                 $this->mStripState = new StripState( $this->mUniqPrefix );
 277
 278
 279                 # Clear these on every parse, bug 4549
 280                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 281
 282                 $this->mShowToc = true;
 283                 $this->mForceTocPosition = false;
 284                 $this->mIncludeSizes = array(
 285                         'post-expand' => 0,
 286                         'arg' => 0,
 287                 );
 288                 $this->mPPNodeCount = 0;
 289                 $this->mDefaultSort = false;
 290                 $this->mHeadings = array();
 291                 $this->mDoubleUnderscores = array();
 292                 $this->mExpensiveFunctionCount = 0;
 293
 294                 # Fix cloning
 295                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 296                         $this->mPreprocessor = null;
 297                 }
 298
 299                 wfRunHooks( 'ParserClearState', array( &$this ) );
 300                 wfProfileOut( __METHOD__ );
 301         }
 302
 303         /**
 304          * Convert wikitext to HTML
 305          * Do not call this function recursively.
 306          *
 307          * @param $text String: text we want to parse
 308          * @param $title Title object
 309          * @param $options ParserOptions
 310          * @param $linestart boolean
 311          * @param $clearState boolean
 312          * @param $revid Int: number to pass in {{REVISIONID}}
 313          * @return ParserOutput a ParserOutput
 314          */
 315         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 316                 /**
 317                  * First pass--just handle <nowiki> sections, pass the rest off
 318                  * to internalParse() which does all the real work.
 319                  */
 320
 321                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 322                 $fname = __METHOD__.'-' . wfGetCaller();
 323                 wfProfileIn( __METHOD__ );
 324                 wfProfileIn( $fname );
 325
 326                 $this->startParse( $title, $options, self::OT_HTML, $clearState );
 327
 328                 $oldRevisionId = $this->mRevisionId;
 329                 $oldRevisionObject = $this->mRevisionObject;
 330                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 331                 $oldRevisionUser = $this->mRevisionUser;
 332                 if ( $revid !== null ) {
 333                         $this->mRevisionId = $revid;
 334                         $this->mRevisionObject = null;
 335                         $this->mRevisionTimestamp = null;
 336                         $this->mRevisionUser = null;
 337                 }
 338
 339                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 340                 # No more strip!
 341                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 342                 $text = $this->internalParse( $text );
 343
 344                 $text = $this->mStripState->unstripGeneral( $text );
 345
 346                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 347                 $fixtags = array(
 348                         # french spaces, last one Guillemet-left
 349                         # only if there is something before the space
 350                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 351                         # french spaces, Guillemet-right
 352                         '/(\\302\\253) /' => '\\1&#160;',
 353                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 354                 );
 355                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 356
 357                 $text = $this->doBlockLevels( $text, $linestart );
 358
 359                 $this->replaceLinkHolders( $text );
 360
 361                 /**
 362                  * The page doesn't get language converted if
 363                  * a) It's disabled
 364                  * b) Content isn't converted
 365                  * c) It's a conversion table
 366                  */
 367                 if ( !( $wgDisableLangConversion
 368                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 369                                 || $this->mTitle->isConversionTable() ) ) {
 370
 371                         # The position of the convert() call should not be changed. it
 372                         # assumes that the links are all replaced and the only thing left
 373                         # is the <nowiki> mark.
 374
 375                         $text = $wgContLang->convert( $text );
 376                 }
 377
 378                 /**
 379                  * A converted title will be provided in the output object if title and
 380                  * content conversion are enabled, the article text does not contain
 381                  * a conversion-suppressing double-underscore tag, and no
 382                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 383                  * automatic link conversion.
 384                  */
 385                 if ( !( $wgDisableLangConversion
 386                                 || $wgDisableTitleConversion
 387                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 388                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 389                                 || $this->mOutput->getDisplayTitle() !== false ) )
 390                 {
 391                         $convruletitle = $wgContLang->getConvRuleTitle();
 392                         if ( $convruletitle ) {
 393                                 $this->mOutput->setTitleText( $convruletitle );
 394                         } else {
 395                                 $titleText = $wgContLang->convertTitle( $title );
 396                                 $this->mOutput->setTitleText( $titleText );
 397                         }
 398                 }
 399
 400                 $text = $this->mStripState->unstripNoWiki( $text );
 401
 402                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 403
 404                 $text = $this->replaceTransparentTags( $text );
 405                 $text = $this->mStripState->unstripGeneral( $text );
 406
 407                 $text = Sanitizer::normalizeCharReferences( $text );
 408
 409                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 410                         $text = MWTidy::tidy( $text );
 411                 } else {
 412                         # attempt to sanitize at least some nesting problems
 413                         # (bug #2702 and quite a few others)
 414                         $tidyregs = array(
 415                                 # ''Something [http://www.cool.com cool''] -->
 416                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 417                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 418                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 419                                 # fix up an anchor inside another anchor, only
 420                                 # at least for a single single nested link (bug 3695)
 421                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 422                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 423                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 424                                 # contains a div, so fix it up here; replace
 425                                 # div with escaped text
 426                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 427                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 428                                 # remove empty italic or bold tag pairs, some
 429                                 # introduced by rules above
 430                                 '/<([bi])><\/\\1>/' => '',
 431                         );
 432
 433                         $text = preg_replace(
 434                                 array_keys( $tidyregs ),
 435                                 array_values( $tidyregs ),
 436                                 $text );
 437                 }
 438                 global $wgExpensiveParserFunctionLimit;
 439                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 440                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 441                 }
 442
 443                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 444
 445                 # Information on include size limits, for the benefit of users who try to skirt them
 446                 if ( $this->mOptions->getEnableLimitReport() ) {
 447                         $max = $this->mOptions->getMaxIncludeSize();
 448                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 449                         $limitReport =
 450                                 "NewPP limit report\n" .
 451                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 452                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 453                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 454                                 $PFreport;
 455                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 456                         $text .= "\n<!-- \n$limitReport-->\n";
 457                 }
 458                 $this->mOutput->setText( $text );
 459
 460                 $this->mRevisionId = $oldRevisionId;
 461                 $this->mRevisionObject = $oldRevisionObject;
 462                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 463                 $this->mRevisionUser = $oldRevisionUser;
 464                 wfProfileOut( $fname );
 465                 wfProfileOut( __METHOD__ );
 466
 467                 return $this->mOutput;
 468         }
 469
 470         /**
 471          * Recursive parser entry point that can be called from an extension tag
 472          * hook.
 473          *
 474          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 475          *
 476          * @param $text String: text extension wants to have parsed
 477          * @param $frame PPFrame: The frame to use for expanding any template variables
 478          *
 479          * @return string
 480          */
 481         function recursiveTagParse( $text, $frame=false ) {
 482                 wfProfileIn( __METHOD__ );
 483                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 484                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 485                 $text = $this->internalParse( $text, false, $frame );
 486                 wfProfileOut( __METHOD__ );
 487                 return $text;
 488         }
 489
 490         /**
 491          * Expand templates and variables in the text, producing valid, static wikitext.
 492          * Also removes comments.
 493          */
 494         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 495                 wfProfileIn( __METHOD__ );
 496                 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
 497                 if ( $revid !== null ) {
 498                         $this->mRevisionId = $revid;
 499                 }
 500                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 501                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 502                 $text = $this->replaceVariables( $text );
 503                 $text = $this->mStripState->unstripBoth( $text );
 504                 wfProfileOut( __METHOD__ );
 505                 return $text;
 506         }
 507
 508         /**
 509          * Process the wikitext for the ?preload= feature. (bug 5210)
 510          *
 511          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 512          * comments, templates, arguments, tags hooks and parser functions are untouched.
 513          */
 514         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 515                 # Parser (re)initialisation
 516                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 517
 518                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 519                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 520                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 521                 $text = $this->mStripState->unstripBoth( $text );
 522                 return $text;
 523         }
 524
 525         /**
 526          * Get a random string
 527          *
 528          * @return string
 529          */
 530         static public function getRandomString() {
 531                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 532         }
 533
 534         /**
 535          * Set the current user.
 536          * Should only be used when doing pre-save transform.
 537          *
 538          * @param $user Mixed: User object or null (to reset)
 539          */
 540         function setUser( $user ) {
 541                 $this->mUser = $user;
 542         }
 543
 544         /**
 545          * Accessor for mUniqPrefix.
 546          *
 547          * @return String
 548          */
 549         public function uniqPrefix() {
 550                 if ( !isset( $this->mUniqPrefix ) ) {
 551                         # @todo FIXME: This is probably *horribly wrong*
 552                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 553                         # if this is called for a parser cache hit, the parser may not
 554                         # have ever been initialized in the first place.
 555                         # Not really sure what the heck is supposed to be going on here.
 556                         return '';
 557                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 558                 }
 559                 return $this->mUniqPrefix;
 560         }
 561
 562         /**
 563          * Set the context title
 564          *
 565          * @param $t Title
 566          */
 567         function setTitle( $t ) {
 568                 if ( !$t || $t instanceof FakeTitle ) {
 569                         $t = Title::newFromText( 'NO TITLE' );
 570                 }
 571
 572                 if ( strval( $t->getFragment() ) !== '' ) {
 573                         # Strip the fragment to avoid various odd effects
 574                         $this->mTitle = clone $t;
 575                         $this->mTitle->setFragment( '' );
 576                 } else {
 577                         $this->mTitle = $t;
 578                 }
 579         }
 580
 581         /**
 582          * Accessor for the Title object
 583          *
 584          * @return Title object
 585          */
 586         function getTitle() {
 587                 return $this->mTitle;
 588         }
 589
 590         /**
 591          * Accessor/mutator for the Title object
 592          *
 593          * @param $x New Title object or null to just get the current one
 594          * @return Title object
 595          */
 596         function Title( $x = null ) {
 597                 return wfSetVar( $this->mTitle, $x );
 598         }
 599
 600         /**
 601          * Set the output type
 602          *
 603          * @param $ot Integer: new value
 604          */
 605         function setOutputType( $ot ) {
 606                 $this->mOutputType = $ot;
 607                 # Shortcut alias
 608                 $this->ot = array(
 609                         'html' => $ot == self::OT_HTML,
 610                         'wiki' => $ot == self::OT_WIKI,
 611                         'pre' => $ot == self::OT_PREPROCESS,
 612                         'plain' => $ot == self::OT_PLAIN,
 613                 );
 614         }
 615
 616         /**
 617          * Accessor/mutator for the output type
 618          *
 619          * @param $x New value or null to just get the current one
 620          * @return Integer
 621          */
 622         function OutputType( $x = null ) {
 623                 return wfSetVar( $this->mOutputType, $x );
 624         }
 625
 626         /**
 627          * Get the ParserOutput object
 628          *
 629          * @return ParserOutput object
 630          */
 631         function getOutput() {
 632                 return $this->mOutput;
 633         }
 634
 635         /**
 636          * Get the ParserOptions object
 637          *
 638          * @return ParserOptions object
 639          */
 640         function getOptions() {
 641                 return $this->mOptions;
 642         }
 643
 644         /**
 645          * Accessor/mutator for the ParserOptions object
 646          *
 647          * @param $x New value or null to just get the current one
 648          * @return Current ParserOptions object
 649          */
 650         function Options( $x = null ) {
 651                 return wfSetVar( $this->mOptions, $x );
 652         }
 653
 654         /**
 655          * @return int
 656          */
 657         function nextLinkID() {
 658                 return $this->mLinkID++;
 659         }
 660
 661         /**
 662          * @param $id int
 663          */
 664         function setLinkID( $id ) {
 665                 $this->mLinkID = $id;
 666         }
 667
 668         /**
 669          * @return Language
 670          */
 671         function getFunctionLang() {
 672                 global $wgLang;
 673
 674                 $target = $this->mOptions->getTargetLanguage();
 675                 if ( $target !== null ) {
 676                         return $target;
 677                 } else {
 678                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $this->mTitle->getPageLanguage();
 679                 }
 680         }
 681
 682         /**
 683          * Get a User object either from $this->mUser, if set, or from the
 684          * ParserOptions object otherwise
 685          *
 686          * @return User object
 687          */
 688         function getUser() {
 689                 if ( !is_null( $this->mUser ) ) {
 690                         return $this->mUser;
 691                 }
 692                 return $this->mOptions->getUser();
 693         }
 694
 695         /**
 696          * Get a preprocessor object
 697          *
 698          * @return Preprocessor instance
 699          */
 700         function getPreprocessor() {
 701                 if ( !isset( $this->mPreprocessor ) ) {
 702                         $class = $this->mPreprocessorClass;
 703                         $this->mPreprocessor = new $class( $this );
 704                 }
 705                 return $this->mPreprocessor;
 706         }
 707
 708         /**
 709          * Replaces all occurrences of HTML-style comments and the given tags
 710          * in the text with a random marker and returns the next text. The output
 711          * parameter $matches will be an associative array filled with data in
 712          * the form:
 713          *   'UNIQ-xxxxx' => array(
 714          *     'element',
 715          *     'tag content',
 716          *     array( 'param' => 'x' ),
 717          *     '<element param="x">tag content</element>' ) )
 718          *
 719          * @param $elements array list of element names. Comments are always extracted.
 720          * @param $text string Source text string.
 721          * @param $matches array Out parameter, Array: extracted tags
 722          * @param $uniq_prefix string
 723          * @return String: stripped text
 724          */
 725         public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 726                 static $n = 1;
 727                 $stripped = '';
 728                 $matches = array();
 729
 730                 $taglist = implode( '|', $elements );
 731                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 732
 733                 while ( $text != '' ) {
 734                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 735                         $stripped .= $p[0];
 736                         if ( count( $p ) < 5 ) {
 737                                 break;
 738                         }
 739                         if ( count( $p ) > 5 ) {
 740                                 # comment
 741                                 $element    = $p[4];
 742                                 $attributes = '';
 743                                 $close      = '';
 744                                 $inside     = $p[5];
 745                         } else {
 746                                 # tag
 747                                 $element    = $p[1];
 748                                 $attributes = $p[2];
 749                                 $close      = $p[3];
 750                                 $inside     = $p[4];
 751                         }
 752
 753                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 754                         $stripped .= $marker;
 755
 756                         if ( $close === '/>' ) {
 757                                 # Empty element tag, <tag />
 758                                 $content = null;
 759                                 $text = $inside;
 760                                 $tail = null;
 761                         } else {
 762                                 if ( $element === '!--' ) {
 763                                         $end = '/(-->)/';
 764                                 } else {
 765                                         $end = "/(<\\/$element\\s*>)/i";
 766                                 }
 767                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 768                                 $content = $q[0];
 769                                 if ( count( $q ) < 3 ) {
 770                                         # No end tag -- let it run out to the end of the text.
 771                                         $tail = '';
 772                                         $text = '';
 773                                 } else {
 774                                         $tail = $q[1];
 775                                         $text = $q[2];
 776                                 }
 777                         }
 778
 779                         $matches[$marker] = array( $element,
 780                                 $content,
 781                                 Sanitizer::decodeTagAttributes( $attributes ),
 782                                 "<$element$attributes$close$content$tail" );
 783                 }
 784                 return $stripped;
 785         }
 786
 787         /**
 788          * Get a list of strippable XML-like elements
 789          *
 790          * @return array
 791          */
 792         function getStripList() {
 793                 return $this->mStripList;
 794         }
 795
 796         /**
 797          * Add an item to the strip state
 798          * Returns the unique tag which must be inserted into the stripped text
 799          * The tag will be replaced with the original text in unstrip()
 800          */
 801         function insertStripItem( $text ) {
 802                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 803                 $this->mMarkerIndex++;
 804                 $this->mStripState->addGeneral( $rnd, $text );
 805                 return $rnd;
 806         }
 807
 808         /**
 809          * parse the wiki syntax used to render tables
 810          *
 811          * @private
 812          */
 813         function doTableStuff( $text ) {
 814                 wfProfileIn( __METHOD__ );
 815
 816                 $lines = StringUtils::explode( "\n", $text );
 817                 $out = '';
 818                 $output =& $out;
 819
 820                 foreach ( $lines as $outLine ) {
 821                         $line = trim( $outLine );
 822
 823                         # empty line, go to next line,
 824                         # but only append \n if outside of table
 825                         if ( $line === '') {
 826                                 $output .= $outLine . "\n";
 827                                 continue;
 828                         }
 829                         $firstChars = $line[0];
 830                         if ( strlen( $line ) > 1 ) {
 831                                 $firstChars .= in_array( $line[1], array( '}', '+', '-' ) ) ? $line[1] : '';
 832                         }
 833                         $matches = array();
 834
 835                         if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line , $matches ) ) {
 836                                 $tables[] = array();
 837                                 $table =& $this->last( $tables );
 838                                 $table[0] = array(); // first row
 839                                 $currentRow =& $table[0];
 840                                 $table['indent'] = strlen( $matches[1] );
 841
 842                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 843                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 844
 845                                 if ( $attributes !== '' ) {
 846                                         $table['attributes'] = $attributes;
 847                                 }
 848                         } elseif ( !isset( $tables[0] ) ) {
 849                                 // we're outside the table
 850
 851                                 $out .= $outLine . "\n";
 852                         } elseif ( $firstChars === '|}' ) {
 853                                 // trim the |} code from the line
 854                                 $line = substr ( $line , 2 );
 855
 856                                 // Shorthand for last row
 857                                 $lastRow =& $this->last( $table );
 858
 859                                 // a thead at the end becomes a tfoot, unless there is only one row
 860                                 // Do this before deleting empty last lines to allow headers at the bottom of tables
 861                                 if ( isset( $lastRow['type'] ) && $lastRow['type'] == 'thead' && isset( $table[1] ) ) {
 862                                         $lastRow['type'] = 'tfoot';
 863                                         for ( $i = 0; isset( $lastRow[$i] ); $i++ ) {
 864                                                 $lastRow[$i]['type'] = 'th';
 865                                         }
 866                                 }
 867
 868                                 // Delete empty last lines
 869                                 if ( empty( $lastRow ) ) {
 870                                         $lastRow = NULL;
 871                                 }
 872                                 $o = '';
 873                                 $curtable = array_pop( $tables );
 874
 875                                 #Add a line-ending before the table, but only if there isn't one already
 876                                 if ( substr( $out, -1 ) !== "\n" ) {
 877                                         $o .= "\n";
 878                                 }
 879                                 $o .= $this->generateTableHTML( $curtable ) . $line . "\n";
 880
 881                                 if ( count( $tables ) > 0 ) {
 882                                         $table =& $this->last( $tables );
 883                                         $currentRow =& $this->last( $table );
 884                                         $currentElement =& $this->last( $currentRow );
 885
 886                                         $output =& $currentElement['content'];
 887                                 } else {
 888                                         $output =& $out;
 889                                 }
 890
 891                                 $output .= $o;
 892
 893                         } elseif ( $firstChars === '|-' ) {
 894                                 // start a new row element
 895                                 // but only when we haven't started one already
 896                                 if ( count( $currentRow ) != 0 ) {
 897                                         $table[] = array();
 898                                         $currentRow =& $this->last( $table );
 899                                 }
 900                                 // Get the attributes, there's nothing else useful in $line now
 901                                 $line = substr ( $line , 2 );
 902                                 $attributes = $this->mStripState->unstripBoth( $line );
 903                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 904                                 if ( $attributes !== '' ) {
 905                                         $currentRow['attributes'] = $attributes;
 906                                 }
 907
 908                         } elseif ( $firstChars  === '|+' ) {
 909                                 // a table caption, but only proceed if there isn't one already
 910                                 if ( !isset ( $table['caption'] ) ) {
 911                                         $line = substr ( $line , 2 );
 912
 913                                         $c = $this->getCellAttr( $line , 'caption' );
 914                                         $table['caption'] = array();
 915                                         $table['caption']['content'] = $c[0];
 916                                         if ( isset( $c[1] ) ) $table['caption']['attributes'] = $c[1];
 917                                         unset( $c );
 918                                         $output =& $table['caption']['content'];
 919                                 }
 920                         } elseif ( $firstChars === '|' || $firstChars === '!' || $firstChars === '!+' ) {
 921                                 // Which kind of cells are we dealing with
 922                                 $currentTag = 'td';
 923                                 $line = substr ( $line , 1 );
 924
 925                                 if ( $firstChars === '!'  || $firstChars === '!+' ) {
 926                                         $line = str_replace ( '!!' , '||' , $line );
 927                                         $currentTag = 'th';
 928                                 }
 929
 930                                 // Split up multiple cells on the same line.
 931                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 932                                 $line = ''; // save memory
 933
 934                                 // decide whether thead to tbody
 935                                 if ( !array_key_exists( 'type', $currentRow ) ) {
 936                                         $currentRow['type'] = ( $firstChars === '!' ) ? 'thead' : 'tbody' ;
 937                                 } elseif ( $firstChars === '|' ) {
 938                                         $currentRow['type'] = 'tbody';
 939                                 }
 940
 941                                 // Loop through each table cell
 942                                 foreach ( $cells as $cell ) {
 943                                         // a new cell
 944                                         $currentRow[] = array();
 945                                         $currentElement =& $this->last( $currentRow );
 946
 947                                         $currentElement['type'] = $currentTag;
 948
 949                                         $c = $this->getCellAttr( $cell , $currentTag );
 950                                         $currentElement['content'] = $c[0];
 951                                         if ( isset( $c[1] ) ) $currentElement['attributes'] = $c[1];
 952                                         unset( $c );
 953                                 }
 954                                 $output =& $currentElement['content'];
 955
 956                         } else {
 957                                 $output .= "\n$outLine";
 958                         }
 959                 }
 960
 961                 # Remove trailing line-ending (b/c)
 962                 if ( substr( $out, -1 ) === "\n" ) {
 963                         $out = substr( $out, 0, -1 );
 964                 }
 965
 966                 # Close any unclosed tables
 967                 if ( isset( $tables ) && count( $tables ) > 0 ) {
 968                         for ( $i = 0; $i < count( $tables ); $i++ ) {
 969                                 $curtable = array_pop( $tables );
 970                                 $curtable = $this->generateTableHTML( $curtable );
 971                                 #Add a line-ending before the table, but only if there isn't one already
 972                                 if ( substr( $out, -1 ) !== "\n"  && $curtable !== "" ) {
 973                                         $out .= "\n";
 974                                 }
 975                                 $out .= $curtable;
 976                         }
 977                 }
 978
 979                 wfProfileOut( __METHOD__ );
 980
 981                 return $out;
 982         }
 983
 984         /**
 985          * Helper function for doTableStuff() separating the contents of cells from
 986          * attributes. Particularly useful as there's a possible bug and this action
 987          * is repeated twice.
 988          *
 989          * @private
 990          * @param $cell
 991          * @param $tagName
 992          * @return array
 993          */
 994         function getCellAttr ( $cell, $tagName ) {
 995                 $attributes = null;
 996
 997                 $cell = trim ( $cell );
 998
 999                 // A cell could contain both parameters and data
1000                 $cellData = explode ( '|' , $cell , 2 );
1001
1002                 // Bug 553: Note that a '|' inside an invalid link should not
1003                 // be mistaken as delimiting cell parameters
1004                 if ( strpos( $cellData[0], '[[' ) !== false ) {
1005                         $content = trim ( $cell );
1006                 }
1007                 elseif ( count ( $cellData ) == 1 ) {
1008                         $content = trim ( $cellData[0] );
1009                 } else {
1010                         $attributes = $this->mStripState->unstripBoth( $cellData[0] );
1011                         $attributes = Sanitizer::fixTagAttributes( $attributes , $tagName );
1012
1013                         $content = trim ( $cellData[1] );
1014                 }
1015                 return array( $content, $attributes );
1016         }
1017
1018
1019         /**
1020          * Helper function for doTableStuff(). This converts the structured array into html.
1021          *
1022          * @private
1023          */
1024         function generateTableHTML( &$table ) {
1025                 $return = "";
1026                 $return .= str_repeat( '<dl><dd>' , $table['indent'] );
1027                 $return .= '<table';
1028                 $return .= isset( $table['attributes'] ) ? $table['attributes'] : '';
1029                 $return .= '>';
1030                 unset( $table['attributes'] );
1031
1032                 if ( isset( $table['caption'] ) ) {
1033                         $return .= "\n<caption";
1034                         $return .= isset( $table['caption']['attributes'] ) ? $table['caption']['attributes'] : '';
1035                         $return .= '>';
1036                         $return .= $table['caption']['content'];
1037                         $return .= "\n</caption>";
1038                 }
1039                 $lastSection = '';
1040                 $empty = true;
1041                 $simple = true;
1042
1043                 // If we only have tbodies, mark table as simple
1044                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1045                         if ( !count( $table[$i] ) ) continue;
1046                         if ( !isset( $table[$i]['type'] ) ) {
1047                                 $table[$i]['type'] = 'tbody';
1048                         }
1049                         if ( !$lastSection ) {
1050                                 $lastSection = $table[$i]['type'];
1051                         } elseif ( $lastSection != $table[$i]['type'] ) {
1052                                 $simple = false;
1053                         }
1054                 }
1055                 $lastSection = '';
1056                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1057                         if ( !count( $table[$i] ) ) continue;
1058                         $empty = false; // check for empty tables
1059
1060                         if ( $table[$i]['type'] != $lastSection && !$simple ) {
1061                                 $return .= "\n<" . $table[$i]['type'] . '>';
1062                         }
1063
1064                         $return .= "\n<tr";
1065                         $return .= isset( $table[$i]['attributes'] ) ? $table[$i]['attributes'] : '';
1066                         $return .= '>';
1067                         for ( $j = 0; isset( $table[$i][$j] ); $j++ ) {
1068                                 if ( !isset( $table[$i][$j]['type'] ) ) $table[$i][$j]['type'] = 'td';
1069                                 $return .= "\n<" . $table[$i][$j]['type'];
1070                                 $return .= isset( $table[$i][$j]['attributes'] ) ? $table[$i][$j]['attributes'] : '';
1071                                 $return .= '>';
1072
1073                                 $return .= $table[$i][$j]['content'];
1074                                 if ( $table[$i][$j]['content'] != '' )
1075                                         $return .= "\n";
1076
1077                                 $return .= '</' . $table[$i][$j]['type'] . '>';
1078                                 unset( $table[$i][$j] );
1079                         }
1080                         $return .= "\n</tr>";
1081
1082                         if ( ( !isset( $table[$i + 1] ) && !$simple ) || ( isset( $table[$i + 1] ) && isset( $table[$i + 1]['type'] ) && $table[$i]['type'] != $table[$i + 1]['type'] ) ) {
1083                                 $return .= '</' . $table[$i]['type'] . '>';
1084                         }
1085                         $lastSection = $table[$i]['type'];
1086                         unset( $table[$i] );
1087                 }
1088                 if ( $empty ) {
1089                         if ( isset( $table['caption'] ) ) {
1090                                 $return .= "\n<tr><td></td></tr>";
1091                         } else {
1092                                 return '';
1093                         }
1094                 }
1095                 $return .= "\n</table>";
1096                 $return .= str_repeat( '</dd></dl>' , $table['indent'] );
1097
1098                 return $return;
1099         }
1100
1101         /**
1102          * like end() but only works on the numeric array index and php's internal pointers
1103          * returns a reference to the last element of an array much like "\$arr[-1]" in perl
1104          * ignores associative elements and will create a 0 key will a NULL value if there were
1105          * no numric elements and an array itself if not previously defined.
1106          *
1107          * @private
1108          */
1109         function &last ( &$arr ) {
1110                 for ( $i = count( $arr ); ( !isset( $arr[$i] ) && $i > 0 ); $i-- ) {  }
1111                 return $arr[$i];
1112         }
1113
1114         /**
1115          * Helper function for parse() that transforms wiki markup into
1116          * HTML. Only called for $mOutputType == self::OT_HTML.
1117          *
1118          * @private
1119          */
1120         function internalParse( $text, $isMain = true, $frame=false ) {
1121                 wfProfileIn( __METHOD__ );
1122
1123                 $origText = $text;
1124
1125                 # Hook to suspend the parser in this state
1126                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1127                         wfProfileOut( __METHOD__ );
1128                         return $text ;
1129                 }
1130
1131                 # if $frame is provided, then use $frame for replacing any variables
1132                 if ( $frame ) {
1133                         # use frame depth to infer how include/noinclude tags should be handled
1134                         # depth=0 means this is the top-level document; otherwise it's an included document
1135                         if ( !$frame->depth ) {
1136                                 $flag = 0;
1137                         } else {
1138                                 $flag = Parser::PTD_FOR_INCLUSION;
1139                         }
1140                         $dom = $this->preprocessToDom( $text, $flag );
1141                         $text = $frame->expand( $dom );
1142                 } else {
1143                         # if $frame is not provided, then use old-style replaceVariables
1144                         $text = $this->replaceVariables( $text );
1145                 }
1146
1147                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1148                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1149
1150                 # Tables need to come after variable replacement for things to work
1151                 # properly; putting them before other transformations should keep
1152                 # exciting things like link expansions from showing up in surprising
1153                 # places.
1154                 $text = $this->doTableStuff( $text );
1155
1156                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1157
1158                 $text = $this->doDoubleUnderscore( $text );
1159
1160                 $text = $this->doHeadings( $text );
1161                 if ( $this->mOptions->getUseDynamicDates() ) {
1162                         $df = DateFormatter::getInstance();
1163                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1164                 }
1165                 $text = $this->replaceInternalLinks( $text );
1166                 $text = $this->doAllQuotes( $text );
1167                 $text = $this->replaceExternalLinks( $text );
1168
1169                 # replaceInternalLinks may sometimes leave behind
1170                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1171                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1172
1173                 $text = $this->doMagicLinks( $text );
1174                 $text = $this->formatHeadings( $text, $origText, $isMain );
1175
1176                 wfProfileOut( __METHOD__ );
1177                 return $text;
1178         }
1179
1180         /**
1181          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1182          * magic external links.
1183          *
1184          * DML
1185          * @private
1186          */
1187         function doMagicLinks( $text ) {
1188                 wfProfileIn( __METHOD__ );
1189                 $prots = $this->mUrlProtocols;
1190                 $urlChar = self::EXT_LINK_URL_CLASS;
1191                 $text = preg_replace_callback(
1192                         '!(?:                           # Start cases
1193                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1194                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1195                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1196                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1197                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1198                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1199                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1200                                         [0-9Xx]                 # check digit
1201                                         \b)
1202                         )!xu', array( &$this, 'magicLinkCallback' ), $text );
1203                 wfProfileOut( __METHOD__ );
1204                 return $text;
1205         }
1206
1207         /**
1208          * @throws MWException
1209          * @param $m array
1210          * @return HTML|string
1211          */
1212         function magicLinkCallback( $m ) {
1213                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1214                         # Skip anchor
1215                         return $m[0];
1216                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1217                         # Skip HTML element
1218                         return $m[0];
1219                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1220                         # Free external link
1221                         return $this->makeFreeExternalLink( $m[0] );
1222                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1223                         # RFC or PMID
1224                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1225                                 $keyword = 'RFC';
1226                                 $urlmsg = 'rfcurl';
1227                                 $CssClass = 'mw-magiclink-rfc';
1228                                 $id = $m[4];
1229                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1230                                 $keyword = 'PMID';
1231                                 $urlmsg = 'pubmedurl';
1232                                 $CssClass = 'mw-magiclink-pmid';
1233                                 $id = $m[4];
1234                         } else {
1235                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1236                                         substr( $m[0], 0, 20 ) . '"' );
1237                         }
1238                         $url = wfMsgForContent( $urlmsg, $id );
1239                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1240                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1241                         # ISBN
1242                         $isbn = $m[5];
1243                         $num = strtr( $isbn, array(
1244                                 '-' => '',
1245                                 ' ' => '',
1246                                 'x' => 'X',
1247                         ));
1248                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1249                         return'<a href="' .
1250                                 $titleObj->escapeLocalUrl() .
1251                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1252                 } else {
1253                         return $m[0];
1254                 }
1255         }
1256
1257         /**
1258          * Make a free external link, given a user-supplied URL
1259          * @return HTML
1260          * @private
1261          */
1262         function makeFreeExternalLink( $url ) {
1263                 global $wgContLang;
1264                 wfProfileIn( __METHOD__ );
1265
1266                 $trail = '';
1267
1268                 # The characters '<' and '>' (which were escaped by
1269                 # removeHTMLtags()) should not be included in
1270                 # URLs, per RFC 2396.
1271                 $m2 = array();
1272                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1273                         $trail = substr( $url, $m2[0][1] ) . $trail;
1274                         $url = substr( $url, 0, $m2[0][1] );
1275                 }
1276
1277                 # Move trailing punctuation to $trail
1278                 $sep = ',;\.:!?';
1279                 # If there is no left bracket, then consider right brackets fair game too
1280                 if ( strpos( $url, '(' ) === false ) {
1281                         $sep .= ')';
1282                 }
1283
1284                 $numSepChars = strspn( strrev( $url ), $sep );
1285                 if ( $numSepChars ) {
1286                         $trail = substr( $url, -$numSepChars ) . $trail;
1287                         $url = substr( $url, 0, -$numSepChars );
1288                 }
1289
1290                 $url = Sanitizer::cleanUrl( $url );
1291
1292                 # Is this an external image?
1293                 $text = $this->maybeMakeExternalImage( $url );
1294                 if ( $text === false ) {
1295                         # Not an image, make a link
1296                         $text = Linker::makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1297                                 $this->getExternalLinkAttribs( $url ) );
1298                         # Register it in the output object...
1299                         # Replace unnecessary URL escape codes with their equivalent characters
1300                         $pasteurized = self::replaceUnusualEscapes( $url );
1301                         $this->mOutput->addExternalLink( $pasteurized );
1302                 }
1303                 wfProfileOut( __METHOD__ );
1304                 return $text . $trail;
1305         }
1306
1307
1308         /**
1309          * Parse headers and return html
1310          *
1311          * @private
1312          */
1313         function doHeadings( $text ) {
1314                 wfProfileIn( __METHOD__ );
1315                 for ( $i = 6; $i >= 1; --$i ) {
1316                         $h = str_repeat( '=', $i );
1317                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1318                           "<h$i>\\1</h$i>", $text );
1319                 }
1320                 wfProfileOut( __METHOD__ );
1321                 return $text;
1322         }
1323
1324         /**
1325          * Replace single quotes with HTML markup
1326          * @private
1327          * @return string the altered text
1328          */
1329         function doAllQuotes( $text ) {
1330                 wfProfileIn( __METHOD__ );
1331                 $outtext = '';
1332                 $lines = StringUtils::explode( "\n", $text );
1333                 foreach ( $lines as $line ) {
1334                         $outtext .= $this->doQuotes( $line ) . "\n";
1335                 }
1336                 $outtext = substr( $outtext, 0,-1 );
1337                 wfProfileOut( __METHOD__ );
1338                 return $outtext;
1339         }
1340
1341         /**
1342          * Helper function for doAllQuotes()
1343          */
1344         public function doQuotes( $text ) {
1345                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1346                 if ( count( $arr ) == 1 ) {
1347                         return $text;
1348                 } else {
1349                         # First, do some preliminary work. This may shift some apostrophes from
1350                         # being mark-up to being text. It also counts the number of occurrences
1351                         # of bold and italics mark-ups.
1352                         $numbold = 0;
1353                         $numitalics = 0;
1354                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1355                                 if ( ( $i % 2 ) == 1 ) {
1356                                         # If there are ever four apostrophes, assume the first is supposed to
1357                                         # be text, and the remaining three constitute mark-up for bold text.
1358                                         if ( strlen( $arr[$i] ) == 4 ) {
1359                                                 $arr[$i-1] .= "'";
1360                                                 $arr[$i] = "'''";
1361                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1362                                                 # If there are more than 5 apostrophes in a row, assume they're all
1363                                                 # text except for the last 5.
1364                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1365                                                 $arr[$i] = "'''''";
1366                                         }
1367                                         # Count the number of occurrences of bold and italics mark-ups.
1368                                         # We are not counting sequences of five apostrophes.
1369                                         if ( strlen( $arr[$i] ) == 2 ) {
1370                                                 $numitalics++;
1371                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1372                                                 $numbold++;
1373                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1374                                                 $numitalics++;
1375                                                 $numbold++;
1376                                         }
1377                                 }
1378                         }
1379
1380                         # If there is an odd number of both bold and italics, it is likely
1381                         # that one of the bold ones was meant to be an apostrophe followed
1382                         # by italics. Which one we cannot know for certain, but it is more
1383                         # likely to be one that has a single-letter word before it.
1384                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1385                                 $i = 0;
1386                                 $firstsingleletterword = -1;
1387                                 $firstmultiletterword = -1;
1388                                 $firstspace = -1;
1389                                 foreach ( $arr as $r ) {
1390                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1391                                                 $x1 = substr( $arr[$i-1], -1 );
1392                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1393                                                 if ( $x1 === ' ' ) {
1394                                                         if ( $firstspace == -1 ) {
1395                                                                 $firstspace = $i;
1396                                                         }
1397                                                 } elseif ( $x2 === ' ') {
1398                                                         if ( $firstsingleletterword == -1 ) {
1399                                                                 $firstsingleletterword = $i;
1400                                                         }
1401                                                 } else {
1402                                                         if ( $firstmultiletterword == -1 ) {
1403                                                                 $firstmultiletterword = $i;
1404                                                         }
1405                                                 }
1406                                         }
1407                                         $i++;
1408                                 }
1409
1410                                 # If there is a single-letter word, use it!
1411                                 if ( $firstsingleletterword > -1 ) {
1412                                         $arr[$firstsingleletterword] = "''";
1413                                         $arr[$firstsingleletterword-1] .= "'";
1414                                 } elseif ( $firstmultiletterword > -1 ) {
1415                                         # If not, but there's a multi-letter word, use that one.
1416                                         $arr[$firstmultiletterword] = "''";
1417                                         $arr[$firstmultiletterword-1] .= "'";
1418                                 } elseif ( $firstspace > -1 ) {
1419                                         # ... otherwise use the first one that has neither.
1420                                         # (notice that it is possible for all three to be -1 if, for example,
1421                                         # there is only one pentuple-apostrophe in the line)
1422                                         $arr[$firstspace] = "''";
1423                                         $arr[$firstspace-1] .= "'";
1424                                 }
1425                         }
1426
1427                         # Now let's actually convert our apostrophic mush to HTML!
1428                         $output = '';
1429                         $buffer = '';
1430                         $state = '';
1431                         $i = 0;
1432                         foreach ( $arr as $r ) {
1433                                 if ( ( $i % 2 ) == 0 ) {
1434                                         if ( $state === 'both' ) {
1435                                                 $buffer .= $r;
1436                                         } else {
1437                                                 $output .= $r;
1438                                         }
1439                                 } else {
1440                                         if ( strlen( $r ) == 2 ) {
1441                                                 if ( $state === 'i' ) {
1442                                                         $output .= '</i>'; $state = '';
1443                                                 } elseif ( $state === 'bi' ) {
1444                                                         $output .= '</i>'; $state = 'b';
1445                                                 } elseif ( $state === 'ib' ) {
1446                                                         $output .= '</b></i><b>'; $state = 'b';
1447                                                 } elseif ( $state === 'both' ) {
1448                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1449                                                 } else { # $state can be 'b' or ''
1450                                                         $output .= '<i>'; $state .= 'i';
1451                                                 }
1452                                         } elseif ( strlen( $r ) == 3 ) {
1453                                                 if ( $state === 'b' ) {
1454                                                         $output .= '</b>'; $state = '';
1455                                                 } elseif ( $state === 'bi' ) {
1456                                                         $output .= '</i></b><i>'; $state = 'i';
1457                                                 } elseif ( $state === 'ib' ) {
1458                                                         $output .= '</b>'; $state = 'i';
1459                                                 } elseif ( $state === 'both' ) {
1460                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1461                                                 } else { # $state can be 'i' or ''
1462                                                         $output .= '<b>'; $state .= 'b';
1463                                                 }
1464                                         } elseif ( strlen( $r ) == 5 ) {
1465                                                 if ( $state === 'b' ) {
1466                                                         $output .= '</b><i>'; $state = 'i';
1467                                                 } elseif ( $state === 'i' ) {
1468                                                         $output .= '</i><b>'; $state = 'b';
1469                                                 } elseif ( $state === 'bi' ) {
1470                                                         $output .= '</i></b>'; $state = '';
1471                                                 } elseif ( $state === 'ib' ) {
1472                                                         $output .= '</b></i>'; $state = '';
1473                                                 } elseif ( $state === 'both' ) {
1474                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1475                                                 } else { # ($state == '')
1476                                                         $buffer = ''; $state = 'both';
1477                                                 }
1478                                         }
1479                                 }
1480                                 $i++;
1481                         }
1482                         # Now close all remaining tags.  Notice that the order is important.
1483                         if ( $state === 'b' || $state === 'ib' ) {
1484                                 $output .= '</b>';
1485                         }
1486                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1487                                 $output .= '</i>';
1488                         }
1489                         if ( $state === 'bi' ) {
1490                                 $output .= '</b>';
1491                         }
1492                         # There might be lonely ''''', so make sure we have a buffer
1493                         if ( $state === 'both' && $buffer ) {
1494                                 $output .= '<b><i>'.$buffer.'</i></b>';
1495                         }
1496                         return $output;
1497                 }
1498         }
1499
1500         /**
1501          * Replace external links (REL)
1502          *
1503          * Note: this is all very hackish and the order of execution matters a lot.
1504          * Make sure to run maintenance/parserTests.php if you change this code.
1505          *
1506          * @private
1507          */
1508         function replaceExternalLinks( $text ) {
1509                 global $wgContLang;
1510                 wfProfileIn( __METHOD__ );
1511
1512                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1513                 $s = array_shift( $bits );
1514
1515                 $i = 0;
1516                 while ( $i<count( $bits ) ) {
1517                         $url = $bits[$i++];
1518                         $protocol = $bits[$i++];
1519                         $text = $bits[$i++];
1520                         $trail = $bits[$i++];
1521
1522                         # The characters '<' and '>' (which were escaped by
1523                         # removeHTMLtags()) should not be included in
1524                         # URLs, per RFC 2396.
1525                         $m2 = array();
1526                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1527                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1528                                 $url = substr( $url, 0, $m2[0][1] );
1529                         }
1530
1531                         # If the link text is an image URL, replace it with an <img> tag
1532                         # This happened by accident in the original parser, but some people used it extensively
1533                         $img = $this->maybeMakeExternalImage( $text );
1534                         if ( $img !== false ) {
1535                                 $text = $img;
1536                         }
1537
1538                         $dtrail = '';
1539
1540                         # Set linktype for CSS - if URL==text, link is essentially free
1541                         $linktype = ( $text === $url ) ? 'free' : 'text';
1542
1543                         # No link text, e.g. [http://domain.tld/some.link]
1544                         if ( $text == '' ) {
1545                                 # Autonumber
1546                                 $langObj = $this->getFunctionLang();
1547                                 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1548                                 $linktype = 'autonumber';
1549                         } else {
1550                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1551                                 # Check for trail
1552                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1553                         }
1554
1555                         $text = $wgContLang->markNoConversion( $text );
1556
1557                         $url = Sanitizer::cleanUrl( $url );
1558
1559                         # Use the encoded URL
1560                         # This means that users can paste URLs directly into the text
1561                         # Funny characters like ö aren't valid in URLs anyway
1562                         # This was changed in August 2004
1563                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1564                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1565
1566                         # Register link in the output object.
1567                         # Replace unnecessary URL escape codes with the referenced character
1568                         # This prevents spammers from hiding links from the filters
1569                         $pasteurized = self::replaceUnusualEscapes( $url );
1570                         $this->mOutput->addExternalLink( $pasteurized );
1571                 }
1572
1573                 wfProfileOut( __METHOD__ );
1574                 return $s;
1575         }
1576
1577         /**
1578          * Get an associative array of additional HTML attributes appropriate for a
1579          * particular external link.  This currently may include rel => nofollow
1580          * (depending on configuration, namespace, and the URL's domain) and/or a
1581          * target attribute (depending on configuration).
1582          *
1583          * @param $url String: optional URL, to extract the domain from for rel =>
1584          *   nofollow if appropriate
1585          * @return Array: associative array of HTML attributes
1586          */
1587         function getExternalLinkAttribs( $url = false ) {
1588                 $attribs = array();
1589                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1590                 $ns = $this->mTitle->getNamespace();
1591                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1592                         $attribs['rel'] = 'nofollow';
1593
1594                         global $wgNoFollowDomainExceptions;
1595                         if ( $wgNoFollowDomainExceptions ) {
1596                                 $bits = wfParseUrl( $url );
1597                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1598                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1599                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1600                                                         unset( $attribs['rel'] );
1601                                                         break;
1602                                                 }
1603                                         }
1604                                 }
1605                         }
1606                 }
1607                 if ( $this->mOptions->getExternalLinkTarget() ) {
1608                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1609                 }
1610                 return $attribs;
1611         }
1612
1613         /**
1614          * Replace unusual URL escape codes with their equivalent characters
1615          *
1616          * @param $url String
1617          * @return String
1618          *
1619          * @todo  This can merge genuinely required bits in the path or query string,
1620          *        breaking legit URLs. A proper fix would treat the various parts of
1621          *        the URL differently; as a workaround, just use the output for
1622          *        statistical records, not for actual linking/output.
1623          */
1624         static function replaceUnusualEscapes( $url ) {
1625                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1626                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1627         }
1628
1629         /**
1630          * Callback function used in replaceUnusualEscapes().
1631          * Replaces unusual URL escape codes with their equivalent character
1632          */
1633         private static function replaceUnusualEscapesCallback( $matches ) {
1634                 $char = urldecode( $matches[0] );
1635                 $ord = ord( $char );
1636                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1637                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1638                         # No, shouldn't be escaped
1639                         return $char;
1640                 } else {
1641                         # Yes, leave it escaped
1642                         return $matches[0];
1643                 }
1644         }
1645
1646         /**
1647          * make an image if it's allowed, either through the global
1648          * option, through the exception, or through the on-wiki whitelist
1649          * @private
1650          */
1651         function maybeMakeExternalImage( $url ) {
1652                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1653                 $imagesexception = !empty( $imagesfrom );
1654                 $text = false;
1655                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1656                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1657                         $imagematch = false;
1658                         foreach ( $imagesfrom as $match ) {
1659                                 if ( strpos( $url, $match ) === 0 ) {
1660                                         $imagematch = true;
1661                                         break;
1662                                 }
1663                         }
1664                 } elseif ( $imagesexception ) {
1665                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1666                 } else {
1667                         $imagematch = false;
1668                 }
1669                 if ( $this->mOptions->getAllowExternalImages()
1670                          || ( $imagesexception && $imagematch ) ) {
1671                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1672                                 # Image found
1673                                 $text = Linker::makeExternalImage( $url );
1674                         }
1675                 }
1676                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1677                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1678                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1679                         foreach ( $whitelist as $entry ) {
1680                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1681                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1682                                         continue;
1683                                 }
1684                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1685                                         # Image matches a whitelist entry
1686                                         $text = Linker::makeExternalImage( $url );
1687                                         break;
1688                                 }
1689                         }
1690                 }
1691                 return $text;
1692         }
1693
1694         /**
1695          * Process [[ ]] wikilinks
1696          * @return String: processed text
1697          *
1698          * @private
1699          */
1700         function replaceInternalLinks( $s ) {
1701                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1702                 return $s;
1703         }
1704
1705         /**
1706          * Process [[ ]] wikilinks (RIL)
1707          * @return LinkHolderArray
1708          *
1709          * @private
1710          */
1711         function replaceInternalLinks2( &$s ) {
1712                 global $wgContLang;
1713
1714                 wfProfileIn( __METHOD__ );
1715
1716                 wfProfileIn( __METHOD__.'-setup' );
1717                 static $tc = FALSE, $e1, $e1_img;
1718                 # the % is needed to support urlencoded titles as well
1719                 if ( !$tc ) {
1720                         $tc = Title::legalChars() . '#%';
1721                         # Match a link having the form [[namespace:link|alternate]]trail
1722                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1723                         # Match cases where there is no "]]", which might still be images
1724                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1725                 }
1726
1727                 $holders = new LinkHolderArray( $this );
1728
1729                 # split the entire text string on occurences of [[
1730                 $a = StringUtils::explode( '[[', ' ' . $s );
1731                 # get the first element (all text up to first [[), and remove the space we added
1732                 $s = $a->current();
1733                 $a->next();
1734                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1735                 $s = substr( $s, 1 );
1736
1737                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1738                 $e2 = null;
1739                 if ( $useLinkPrefixExtension ) {
1740                         # Match the end of a line for a word that's not followed by whitespace,
1741                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1742                         $e2 = wfMsgForContent( 'linkprefix' );
1743                 }
1744
1745                 if ( is_null( $this->mTitle ) ) {
1746                         wfProfileOut( __METHOD__.'-setup' );
1747                         wfProfileOut( __METHOD__ );
1748                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1749                 }
1750                 $nottalk = !$this->mTitle->isTalkPage();
1751
1752                 if ( $useLinkPrefixExtension ) {
1753                         $m = array();
1754                         if ( preg_match( $e2, $s, $m ) ) {
1755                                 $first_prefix = $m[2];
1756                         } else {
1757                                 $first_prefix = false;
1758                         }
1759                 } else {
1760                         $prefix = '';
1761                 }
1762
1763                 if ( $wgContLang->hasVariants() ) {
1764                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1765                 } else {
1766                         $selflink = array( $this->mTitle->getPrefixedText() );
1767                 }
1768                 $useSubpages = $this->areSubpagesAllowed();
1769                 wfProfileOut( __METHOD__.'-setup' );
1770
1771                 # Loop for each link
1772                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1773                         # Check for excessive memory usage
1774                         if ( $holders->isBig() ) {
1775                                 # Too big
1776                                 # Do the existence check, replace the link holders and clear the array
1777                                 $holders->replace( $s );
1778                                 $holders->clear();
1779                         }
1780
1781                         if ( $useLinkPrefixExtension ) {
1782                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1783                                 if ( preg_match( $e2, $s, $m ) ) {
1784                                         $prefix = $m[2];
1785                                         $s = $m[1];
1786                                 } else {
1787                                         $prefix='';
1788                                 }
1789                                 # first link
1790                                 if ( $first_prefix ) {
1791                                         $prefix = $first_prefix;
1792                                         $first_prefix = false;
1793                                 }
1794                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1795                         }
1796
1797                         $might_be_img = false;
1798
1799                         wfProfileIn( __METHOD__."-e1" );
1800                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1801                                 $text = $m[2];
1802                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1803                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1804                                 # the real problem is with the $e1 regex
1805                                 # See bug 1300.
1806                                 #
1807                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1808                                 # and no image is in sight. See bug 2095.
1809                                 #
1810                                 if ( $text !== '' &&
1811                                         substr( $m[3], 0, 1 ) === ']' &&
1812                                         strpos( $text, '[' ) !== false
1813                                 )
1814                                 {
1815                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1816                                         $m[3] = substr( $m[3], 1 );
1817                                 }
1818                                 # fix up urlencoded title texts
1819                                 if ( strpos( $m[1], '%' ) !== false ) {
1820                                         # Should anchors '#' also be rejected?
1821                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1822                                 }
1823                                 $trail = $m[3];
1824                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1825                                 $might_be_img = true;
1826                                 $text = $m[2];
1827                                 if ( strpos( $m[1], '%' ) !== false ) {
1828                                         $m[1] = rawurldecode( $m[1] );
1829                                 }
1830                                 $trail = "";
1831                         } else { # Invalid form; output directly
1832                                 $s .= $prefix . '[[' . $line ;
1833                                 wfProfileOut( __METHOD__."-e1" );
1834                                 continue;
1835                         }
1836                         wfProfileOut( __METHOD__."-e1" );
1837                         wfProfileIn( __METHOD__."-misc" );
1838
1839                         # Don't allow internal links to pages containing
1840                         # PROTO: where PROTO is a valid URL protocol; these
1841                         # should be external links.
1842                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1843                                 $s .= $prefix . '[[' . $line ;
1844                                 wfProfileOut( __METHOD__."-misc" );
1845                                 continue;
1846                         }
1847
1848                         # Make subpage if necessary
1849                         if ( $useSubpages ) {
1850                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1851                         } else {
1852                                 $link = $m[1];
1853                         }
1854
1855                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1856                         if ( !$noforce ) {
1857                                 # Strip off leading ':'
1858                                 $link = substr( $link, 1 );
1859                         }
1860
1861                         wfProfileOut( __METHOD__."-misc" );
1862                         wfProfileIn( __METHOD__."-title" );
1863                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1864                         if ( $nt === null ) {
1865                                 $s .= $prefix . '[[' . $line;
1866                                 wfProfileOut( __METHOD__."-title" );
1867                                 continue;
1868                         }
1869
1870                         $ns = $nt->getNamespace();
1871                         $iw = $nt->getInterWiki();
1872                         wfProfileOut( __METHOD__."-title" );
1873
1874                         if ( $might_be_img ) { # if this is actually an invalid link
1875                                 wfProfileIn( __METHOD__."-might_be_img" );
1876                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1877                                         $found = false;
1878                                         while ( true ) {
1879                                                 # look at the next 'line' to see if we can close it there
1880                                                 $a->next();
1881                                                 $next_line = $a->current();
1882                                                 if ( $next_line === false || $next_line === null ) {
1883                                                         break;
1884                                                 }
1885                                                 $m = explode( ']]', $next_line, 3 );
1886                                                 if ( count( $m ) == 3 ) {
1887                                                         # the first ]] closes the inner link, the second the image
1888                                                         $found = true;
1889                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1890                                                         $trail = $m[2];
1891                                                         break;
1892                                                 } elseif ( count( $m ) == 2 ) {
1893                                                         # if there's exactly one ]] that's fine, we'll keep looking
1894                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1895                                                 } else {
1896                                                         # if $next_line is invalid too, we need look no further
1897                                                         $text .= '[[' . $next_line;
1898                                                         break;
1899                                                 }
1900                                         }
1901                                         if ( !$found ) {
1902                                                 # we couldn't find the end of this imageLink, so output it raw
1903                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1904                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1905                                                 $s .= "{$prefix}[[$link|$text";
1906                                                 # note: no $trail, because without an end, there *is* no trail
1907                                                 wfProfileOut( __METHOD__."-might_be_img" );
1908                                                 continue;
1909                                         }
1910                                 } else { # it's not an image, so output it raw
1911                                         $s .= "{$prefix}[[$link|$text";
1912                                         # note: no $trail, because without an end, there *is* no trail
1913                                         wfProfileOut( __METHOD__."-might_be_img" );
1914                                         continue;
1915                                 }
1916                                 wfProfileOut( __METHOD__."-might_be_img" );
1917                         }
1918
1919                         $wasblank = ( $text  == '' );
1920                         if ( $wasblank ) {
1921                                 $text = $link;
1922                         } else {
1923                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1924                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1925                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1926                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1927                                 $text = $this->doQuotes( $text );
1928                         }
1929
1930                         # Link not escaped by : , create the various objects
1931                         if ( $noforce ) {
1932
1933                                 # Interwikis
1934                                 wfProfileIn( __METHOD__."-interwiki" );
1935                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1936                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1937                                         $s = rtrim( $s . $prefix );
1938                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1939                                         wfProfileOut( __METHOD__."-interwiki" );
1940                                         continue;
1941                                 }
1942                                 wfProfileOut( __METHOD__."-interwiki" );
1943
1944                                 if ( $ns == NS_FILE ) {
1945                                         wfProfileIn( __METHOD__."-image" );
1946                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1947                                                 if ( $wasblank ) {
1948                                                         # if no parameters were passed, $text
1949                                                         # becomes something like "File:Foo.png",
1950                                                         # which we don't want to pass on to the
1951                                                         # image generator
1952                                                         $text = '';
1953                                                 } else {
1954                                                         # recursively parse links inside the image caption
1955                                                         # actually, this will parse them in any other parameters, too,
1956                                                         # but it might be hard to fix that, and it doesn't matter ATM
1957                                                         $text = $this->replaceExternalLinks( $text );
1958                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1959                                                 }
1960                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1961                                                 $s .= $prefix . $this->armorLinks(
1962                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
1963                                         } else {
1964                                                 $s .= $prefix . $trail;
1965                                         }
1966                                         wfProfileOut( __METHOD__."-image" );
1967                                         continue;
1968                                 }
1969
1970                                 if ( $ns == NS_CATEGORY ) {
1971                                         wfProfileIn( __METHOD__."-category" );
1972                                         $s = rtrim( $s . "\n" ); # bug 87
1973
1974                                         if ( $wasblank ) {
1975                                                 $sortkey = $this->getDefaultSort();
1976                                         } else {
1977                                                 $sortkey = $text;
1978                                         }
1979                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1980                                         $sortkey = str_replace( "\n", '', $sortkey );
1981                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1982                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1983
1984                                         /**
1985                                          * Strip the whitespace Category links produce, see bug 87
1986                                          * @todo We might want to use trim($tmp, "\n") here.
1987                                          */
1988                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1989
1990                                         wfProfileOut( __METHOD__."-category" );
1991                                         continue;
1992                                 }
1993                         }
1994
1995                         # Self-link checking
1996                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1997                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1998                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
1999                                         continue;
2000                                 }
2001                         }
2002
2003                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
2004                         # @todo FIXME: Should do batch file existence checks, see comment below
2005                         if ( $ns == NS_MEDIA ) {
2006                                 wfProfileIn( __METHOD__."-media" );
2007                                 # Give extensions a chance to select the file revision for us
2008                                 $time = $sha1 = $descQuery = false;
2009                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2010                                         array( $this, $nt, &$time, &$sha1, &$descQuery ) );
2011                                 # Fetch and register the file (file title may be different via hooks)
2012                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $time, $sha1 );
2013                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2014                                 $s .= $prefix . $this->armorLinks(
2015                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2016                                 wfProfileOut( __METHOD__."-media" );
2017                                 continue;
2018                         }
2019
2020                         wfProfileIn( __METHOD__."-always_known" );
2021                         # Some titles, such as valid special pages or files in foreign repos, should
2022                         # be shown as bluelinks even though they're not included in the page table
2023                         #
2024                         # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2025                         # batch file existence checks for NS_FILE and NS_MEDIA
2026                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
2027                                 $this->mOutput->addLink( $nt );
2028                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2029                         } else {
2030                                 # Links will be added to the output link list after checking
2031                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2032                         }
2033                         wfProfileOut( __METHOD__."-always_known" );
2034                 }
2035                 wfProfileOut( __METHOD__ );
2036                 return $holders;
2037         }
2038
2039         /**
2040          * Render a forced-blue link inline; protect against double expansion of
2041          * URLs if we're in a mode that prepends full URL prefixes to internal links.
2042          * Since this little disaster has to split off the trail text to avoid
2043          * breaking URLs in the following text without breaking trails on the
2044          * wiki links, it's been made into a horrible function.
2045          *
2046          * @param $nt Title
2047          * @param $text String
2048          * @param $query Array or String
2049          * @param $trail String
2050          * @param $prefix String
2051          * @return String: HTML-wikitext mix oh yuck
2052          */
2053         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2054                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2055
2056                 if ( is_string( $query ) ) {
2057                         $query = wfCgiToArray( $query );
2058                 }
2059                 if ( $text == '' ) {
2060                         $text = htmlspecialchars( $nt->getPrefixedText() );
2061                 }
2062
2063                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2064
2065                 return $this->armorLinks( $link ) . $trail;
2066         }
2067
2068         /**
2069          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2070          * going to go through further parsing steps before inline URL expansion.
2071          *
2072          * Not needed quite as much as it used to be since free links are a bit
2073          * more sensible these days. But bracketed links are still an issue.
2074          *
2075          * @param $text String: more-or-less HTML
2076          * @return String: less-or-more HTML with NOPARSE bits
2077          */
2078         function armorLinks( $text ) {
2079                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2080                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2081         }
2082
2083         /**
2084          * Return true if subpage links should be expanded on this page.
2085          * @return Boolean
2086          */
2087         function areSubpagesAllowed() {
2088                 # Some namespaces don't allow subpages
2089                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2090         }
2091
2092         /**
2093          * Handle link to subpage if necessary
2094          *
2095          * @param $target String: the source of the link
2096          * @param &$text String: the link text, modified as necessary
2097          * @return string the full name of the link
2098          * @private
2099          */
2100         function maybeDoSubpageLink( $target, &$text ) {
2101                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2102         }
2103
2104         /**#@+
2105          * Used by doBlockLevels()
2106          * @private
2107          *
2108          * @return string
2109          */
2110         function closeParagraph() {
2111                 $result = '';
2112                 if ( $this->mLastSection != '' ) {
2113                         $result = '</' . $this->mLastSection  . ">\n";
2114                 }
2115                 $this->mInPre = false;
2116                 $this->mLastSection = '';
2117                 return $result;
2118         }
2119
2120         /**
2121          * getCommon() returns the length of the longest common substring
2122          * of both arguments, starting at the beginning of both.
2123          * @private
2124          */
2125         function getCommon( $st1, $st2 ) {
2126                 $fl = strlen( $st1 );
2127                 $shorter = strlen( $st2 );
2128                 if ( $fl < $shorter ) {
2129                         $shorter = $fl;
2130                 }
2131
2132                 for ( $i = 0; $i < $shorter; ++$i ) {
2133                         if ( $st1[$i] != $st2[$i] ) {
2134                                 break;
2135                         }
2136                 }
2137                 return $i;
2138         }
2139
2140         /**
2141          * These next three functions open, continue, and close the list
2142          * element appropriate to the prefix character passed into them.
2143          * @private
2144          *
2145          * @return string
2146          */
2147         function openList( $char ) {
2148                 $result = $this->closeParagraph();
2149
2150                 if ( '*' === $char ) {
2151                         $result .= '<ul><li>';
2152                 } elseif ( '#' === $char ) {
2153                         $result .= '<ol><li>';
2154                 } elseif ( ':' === $char ) {
2155                         $result .= '<dl><dd>';
2156                 } elseif ( ';' === $char ) {
2157                         $result .= '<dl><dt>';
2158                         $this->mDTopen = true;
2159                 } else {
2160                         $result = '<!-- ERR 1 -->';
2161                 }
2162
2163                 return $result;
2164         }
2165
2166         /**
2167          * TODO: document
2168          * @param $char String
2169          * @private
2170          *
2171          * @return string
2172          */
2173         function nextItem( $char ) {
2174                 if ( '*' === $char || '#' === $char ) {
2175                         return '</li><li>';
2176                 } elseif ( ':' === $char || ';' === $char ) {
2177                         $close = '</dd>';
2178                         if ( $this->mDTopen ) {
2179                                 $close = '</dt>';
2180                         }
2181                         if ( ';' === $char ) {
2182                                 $this->mDTopen = true;
2183                                 return $close . '<dt>';
2184                         } else {
2185                                 $this->mDTopen = false;
2186                                 return $close . '<dd>';
2187                         }
2188                 }
2189                 return '<!-- ERR 2 -->';
2190         }
2191
2192         /**
2193          * TODO: document
2194          * @param $char String
2195          * @private
2196          *
2197          * @return string
2198          */
2199         function closeList( $char ) {
2200                 if ( '*' === $char ) {
2201                         $text = '</li></ul>';
2202                 } elseif ( '#' === $char ) {
2203                         $text = '</li></ol>';
2204                 } elseif ( ':' === $char ) {
2205                         if ( $this->mDTopen ) {
2206                                 $this->mDTopen = false;
2207                                 $text = '</dt></dl>';
2208                         } else {
2209                                 $text = '</dd></dl>';
2210                         }
2211                 } else {
2212                         return '<!-- ERR 3 -->';
2213                 }
2214                 return $text."\n";
2215         }
2216         /**#@-*/
2217
2218         /**
2219          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2220          *
2221          * @param $text String
2222          * @param $linestart Boolean: whether or not this is at the start of a line.
2223          * @private
2224          * @return string the lists rendered as HTML
2225          */
2226         function doBlockLevels( $text, $linestart ) {
2227                 wfProfileIn( __METHOD__ );
2228
2229                 # Parsing through the text line by line.  The main thing
2230                 # happening here is handling of block-level elements p, pre,
2231                 # and making lists from lines starting with * # : etc.
2232                 #
2233                 $textLines = StringUtils::explode( "\n", $text );
2234
2235                 $lastPrefix = $output = '';
2236                 $this->mDTopen = $inBlockElem = false;
2237                 $prefixLength = 0;
2238                 $paragraphStack = false;
2239
2240                 foreach ( $textLines as $oLine ) {
2241                         # Fix up $linestart
2242                         if ( !$linestart ) {
2243                                 $output .= $oLine;
2244                                 $linestart = true;
2245                                 continue;
2246                         }
2247                         # * = ul
2248                         # # = ol
2249                         # ; = dt
2250                         # : = dd
2251
2252                         $lastPrefixLength = strlen( $lastPrefix );
2253                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2254                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2255                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2256                         if ( !$this->mInPre ) {
2257                                 # Multiple prefixes may abut each other for nested lists.
2258                                 $prefixLength = strspn( $oLine, '*#:;' );
2259                                 $prefix = substr( $oLine, 0, $prefixLength );
2260
2261                                 # eh?
2262                                 # ; and : are both from definition-lists, so they're equivalent
2263                                 #  for the purposes of determining whether or not we need to open/close
2264                                 #  elements.
2265                                 $prefix2 = str_replace( ';', ':', $prefix );
2266                                 $t = substr( $oLine, $prefixLength );
2267                                 $this->mInPre = (bool)$preOpenMatch;
2268                         } else {
2269                                 # Don't interpret any other prefixes in preformatted text
2270                                 $prefixLength = 0;
2271                                 $prefix = $prefix2 = '';
2272                                 $t = $oLine;
2273                         }
2274
2275                         # List generation
2276                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2277                                 # Same as the last item, so no need to deal with nesting or opening stuff
2278                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2279                                 $paragraphStack = false;
2280
2281                                 if ( substr( $prefix, -1 ) === ';') {
2282                                         # The one nasty exception: definition lists work like this:
2283                                         # ; title : definition text
2284                                         # So we check for : in the remainder text to split up the
2285                                         # title and definition, without b0rking links.
2286                                         $term = $t2 = '';
2287                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2288                                                 $t = $t2;
2289                                                 $output .= $term . $this->nextItem( ':' );
2290                                         }
2291                                 }
2292                         } elseif ( $prefixLength || $lastPrefixLength ) {
2293                                 # We need to open or close prefixes, or both.
2294
2295                                 # Either open or close a level...
2296                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2297                                 $paragraphStack = false;
2298
2299                                 # Close all the prefixes which aren't shared.
2300                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2301                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2302                                         --$lastPrefixLength;
2303                                 }
2304
2305                                 # Continue the current prefix if appropriate.
2306                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2307                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2308                                 }
2309
2310                                 # Open prefixes where appropriate.
2311                                 while ( $prefixLength > $commonPrefixLength ) {
2312                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2313                                         $output .= $this->openList( $char );
2314
2315                                         if ( ';' === $char ) {
2316                                                 # @todo FIXME: This is dupe of code above
2317                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2318                                                         $t = $t2;
2319                                                         $output .= $term . $this->nextItem( ':' );
2320                                                 }
2321                                         }
2322                                         ++$commonPrefixLength;
2323                                 }
2324                                 $lastPrefix = $prefix2;
2325                         }
2326
2327                         # If we have no prefixes, go to paragraph mode.
2328                         if ( 0 == $prefixLength ) {
2329                                 wfProfileIn( __METHOD__."-paragraph" );
2330                                 # No prefix (not in list)--go to paragraph mode
2331                                 # XXX: use a stack for nestable elements like span, table and div
2332                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2333                                 $closematch = preg_match(
2334                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2335                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2336                                 if ( $openmatch or $closematch ) {
2337                                         $paragraphStack = false;
2338                                         # TODO bug 5718: paragraph closed
2339                                         $output .= $this->closeParagraph();
2340                                         if ( $preOpenMatch and !$preCloseMatch ) {
2341                                                 $this->mInPre = true;
2342                                         }
2343                                         $inBlockElem = !$closematch;
2344                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2345                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2346                                                 # pre
2347                                                 if ( $this->mLastSection !== 'pre' ) {
2348                                                         $paragraphStack = false;
2349                                                         $output .= $this->closeParagraph().'<pre>';
2350                                                         $this->mLastSection = 'pre';
2351                                                 }
2352                                                 $t = substr( $t, 1 );
2353                                         } else {
2354                                                 # paragraph
2355                                                 if ( trim( $t ) === '' ) {
2356                                                         if ( $paragraphStack ) {
2357                                                                 $output .= $paragraphStack.'<br />';
2358                                                                 $paragraphStack = false;
2359                                                                 $this->mLastSection = 'p';
2360                                                         } else {
2361                                                                 if ( $this->mLastSection !== 'p' ) {
2362                                                                         $output .= $this->closeParagraph();
2363                                                                         $this->mLastSection = '';
2364                                                                         $paragraphStack = '<p>';
2365                                                                 } else {
2366                                                                         $paragraphStack = '</p><p>';
2367                                                                 }
2368                                                         }
2369                                                 } else {
2370                                                         if ( $paragraphStack ) {
2371                                                                 $output .= $paragraphStack;
2372                                                                 $paragraphStack = false;
2373                                                                 $this->mLastSection = 'p';
2374                                                         } elseif ( $this->mLastSection !== 'p' ) {
2375                                                                 $output .= $this->closeParagraph().'<p>';
2376                                                                 $this->mLastSection = 'p';
2377                                                         }
2378                                                 }
2379                                         }
2380                                 }
2381                                 wfProfileOut( __METHOD__."-paragraph" );
2382                         }
2383                         # somewhere above we forget to get out of pre block (bug 785)
2384                         if ( $preCloseMatch && $this->mInPre ) {
2385                                 $this->mInPre = false;
2386                         }
2387                         if ( $paragraphStack === false ) {
2388                                 $output .= $t."\n";
2389                         }
2390                 }
2391                 while ( $prefixLength ) {
2392                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2393                         --$prefixLength;
2394                 }
2395                 if ( $this->mLastSection != '' ) {
2396                         $output .= '</' . $this->mLastSection . '>';
2397                         $this->mLastSection = '';
2398                 }
2399
2400                 wfProfileOut( __METHOD__ );
2401                 return $output;
2402         }
2403
2404         /**
2405          * Split up a string on ':', ignoring any occurences inside tags
2406          * to prevent illegal overlapping.
2407          *
2408          * @param $str String: the string to split
2409          * @param &$before String: set to everything before the ':'
2410          * @param &$after String: set to everything after the ':'
2411          * return String: the position of the ':', or false if none found
2412          */
2413         function findColonNoLinks( $str, &$before, &$after ) {
2414                 wfProfileIn( __METHOD__ );
2415
2416                 $pos = strpos( $str, ':' );
2417                 if ( $pos === false ) {
2418                         # Nothing to find!
2419                         wfProfileOut( __METHOD__ );
2420                         return false;
2421                 }
2422
2423                 $lt = strpos( $str, '<' );
2424                 if ( $lt === false || $lt > $pos ) {
2425                         # Easy; no tag nesting to worry about
2426                         $before = substr( $str, 0, $pos );
2427                         $after = substr( $str, $pos+1 );
2428                         wfProfileOut( __METHOD__ );
2429                         return $pos;
2430                 }
2431
2432                 # Ugly state machine to walk through avoiding tags.
2433                 $state = self::COLON_STATE_TEXT;
2434                 $stack = 0;
2435                 $len = strlen( $str );
2436                 for( $i = 0; $i < $len; $i++ ) {
2437                         $c = $str[$i];
2438
2439                         switch( $state ) {
2440                         # (Using the number is a performance hack for common cases)
2441                         case 0: # self::COLON_STATE_TEXT:
2442                                 switch( $c ) {
2443                                 case "<":
2444                                         # Could be either a <start> tag or an </end> tag
2445                                         $state = self::COLON_STATE_TAGSTART;
2446                                         break;
2447                                 case ":":
2448                                         if ( $stack == 0 ) {
2449                                                 # We found it!
2450                                                 $before = substr( $str, 0, $i );
2451                                                 $after = substr( $str, $i + 1 );
2452                                                 wfProfileOut( __METHOD__ );
2453                                                 return $i;
2454                                         }
2455                                         # Embedded in a tag; don't break it.
2456                                         break;
2457                                 default:
2458                                         # Skip ahead looking for something interesting
2459                                         $colon = strpos( $str, ':', $i );
2460                                         if ( $colon === false ) {
2461                                                 # Nothing else interesting
2462                                                 wfProfileOut( __METHOD__ );
2463                                                 return false;
2464                                         }
2465                                         $lt = strpos( $str, '<', $i );
2466                                         if ( $stack === 0 ) {
2467                                                 if ( $lt === false || $colon < $lt ) {
2468                                                         # We found it!
2469                                                         $before = substr( $str, 0, $colon );
2470                                                         $after = substr( $str, $colon + 1 );
2471                                                         wfProfileOut( __METHOD__ );
2472                                                         return $i;
2473                                                 }
2474                                         }
2475                                         if ( $lt === false ) {
2476                                                 # Nothing else interesting to find; abort!
2477                                                 # We're nested, but there's no close tags left. Abort!
2478                                                 break 2;
2479                                         }
2480                                         # Skip ahead to next tag start
2481                                         $i = $lt;
2482                                         $state = self::COLON_STATE_TAGSTART;
2483                                 }
2484                                 break;
2485                         case 1: # self::COLON_STATE_TAG:
2486                                 # In a <tag>
2487                                 switch( $c ) {
2488                                 case ">":
2489                                         $stack++;
2490                                         $state = self::COLON_STATE_TEXT;
2491                                         break;
2492                                 case "/":
2493                                         # Slash may be followed by >?
2494                                         $state = self::COLON_STATE_TAGSLASH;
2495                                         break;
2496                                 default:
2497                                         # ignore
2498                                 }
2499                                 break;
2500                         case 2: # self::COLON_STATE_TAGSTART:
2501                                 switch( $c ) {
2502                                 case "/":
2503                                         $state = self::COLON_STATE_CLOSETAG;
2504                                         break;
2505                                 case "!":
2506                                         $state = self::COLON_STATE_COMMENT;
2507                                         break;
2508                                 case ">":
2509                                         # Illegal early close? This shouldn't happen D:
2510                                         $state = self::COLON_STATE_TEXT;
2511                                         break;
2512                                 default:
2513                                         $state = self::COLON_STATE_TAG;
2514                                 }
2515                                 break;
2516                         case 3: # self::COLON_STATE_CLOSETAG:
2517                                 # In a </tag>
2518                                 if ( $c === ">" ) {
2519                                         $stack--;
2520                                         if ( $stack < 0 ) {
2521                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2522                                                 wfProfileOut( __METHOD__ );
2523                                                 return false;
2524                                         }
2525                                         $state = self::COLON_STATE_TEXT;
2526                                 }
2527                                 break;
2528                         case self::COLON_STATE_TAGSLASH:
2529                                 if ( $c === ">" ) {
2530                                         # Yes, a self-closed tag <blah/>
2531                                         $state = self::COLON_STATE_TEXT;
2532                                 } else {
2533                                         # Probably we're jumping the gun, and this is an attribute
2534                                         $state = self::COLON_STATE_TAG;
2535                                 }
2536                                 break;
2537                         case 5: # self::COLON_STATE_COMMENT:
2538                                 if ( $c === "-" ) {
2539                                         $state = self::COLON_STATE_COMMENTDASH;
2540                                 }
2541                                 break;
2542                         case self::COLON_STATE_COMMENTDASH:
2543                                 if ( $c === "-" ) {
2544                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2545                                 } else {
2546                                         $state = self::COLON_STATE_COMMENT;
2547                                 }
2548                                 break;
2549                         case self::COLON_STATE_COMMENTDASHDASH:
2550                                 if ( $c === ">" ) {
2551                                         $state = self::COLON_STATE_TEXT;
2552                                 } else {
2553                                         $state = self::COLON_STATE_COMMENT;
2554                                 }
2555                                 break;
2556                         default:
2557                                 throw new MWException( "State machine error in " . __METHOD__ );
2558                         }
2559                 }
2560                 if ( $stack > 0 ) {
2561                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2562                         wfProfileOut( __METHOD__ );
2563                         return false;
2564                 }
2565                 wfProfileOut( __METHOD__ );
2566                 return false;
2567         }
2568
2569         /**
2570          * Return value of a magic variable (like PAGENAME)
2571          *
2572          * @private
2573          *
2574          * @param $index integer
2575          * @param $frame PPFrame
2576          */
2577         function getVariableValue( $index, $frame=false ) {
2578                 global $wgContLang, $wgSitename, $wgServer;
2579                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2580
2581                 /**
2582                  * Some of these require message or data lookups and can be
2583                  * expensive to check many times.
2584                  */
2585                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2586                         if ( isset( $this->mVarCache[$index] ) ) {
2587                                 return $this->mVarCache[$index];
2588                         }
2589                 }
2590
2591                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2592                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2593
2594                 # Use the time zone
2595                 global $wgLocaltimezone;
2596                 if ( isset( $wgLocaltimezone ) ) {
2597                         $oldtz = date_default_timezone_get();
2598                         date_default_timezone_set( $wgLocaltimezone );
2599                 }
2600
2601                 $localTimestamp = date( 'YmdHis', $ts );
2602                 $localMonth = date( 'm', $ts );
2603                 $localMonth1 = date( 'n', $ts );
2604                 $localMonthName = date( 'n', $ts );
2605                 $localDay = date( 'j', $ts );
2606                 $localDay2 = date( 'd', $ts );
2607                 $localDayOfWeek = date( 'w', $ts );
2608                 $localWeek = date( 'W', $ts );
2609                 $localYear = date( 'Y', $ts );
2610                 $localHour = date( 'H', $ts );
2611                 if ( isset( $wgLocaltimezone ) ) {
2612                         date_default_timezone_set( $oldtz );
2613                 }
2614
2615                 switch ( $index ) {
2616                         case 'currentmonth':
2617                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2618                                 break;
2619                         case 'currentmonth1':
2620                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2621                                 break;
2622                         case 'currentmonthname':
2623                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2624                                 break;
2625                         case 'currentmonthnamegen':
2626                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2627                                 break;
2628                         case 'currentmonthabbrev':
2629                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2630                                 break;
2631                         case 'currentday':
2632                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2633                                 break;
2634                         case 'currentday2':
2635                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2636                                 break;
2637                         case 'localmonth':
2638                                 $value = $wgContLang->formatNum( $localMonth );
2639                                 break;
2640                         case 'localmonth1':
2641                                 $value = $wgContLang->formatNum( $localMonth1 );
2642                                 break;
2643                         case 'localmonthname':
2644                                 $value = $wgContLang->getMonthName( $localMonthName );
2645                                 break;
2646                         case 'localmonthnamegen':
2647                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2648                                 break;
2649                         case 'localmonthabbrev':
2650                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2651                                 break;
2652                         case 'localday':
2653                                 $value = $wgContLang->formatNum( $localDay );
2654                                 break;
2655                         case 'localday2':
2656                                 $value = $wgContLang->formatNum( $localDay2 );
2657                                 break;
2658                         case 'pagename':
2659                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2660                                 break;
2661                         case 'pagenamee':
2662                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2663                                 break;
2664                         case 'fullpagename':
2665                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2666                                 break;
2667                         case 'fullpagenamee':
2668                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2669                                 break;
2670                         case 'subpagename':
2671                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2672                                 break;
2673                         case 'subpagenamee':
2674                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2675                                 break;
2676                         case 'basepagename':
2677                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2678                                 break;
2679                         case 'basepagenamee':
2680                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2681                                 break;
2682                         case 'talkpagename':
2683                                 if ( $this->mTitle->canTalk() ) {
2684                                         $talkPage = $this->mTitle->getTalkPage();
2685                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2686                                 } else {
2687                                         $value = '';
2688                                 }
2689                                 break;
2690                         case 'talkpagenamee':
2691                                 if ( $this->mTitle->canTalk() ) {
2692                                         $talkPage = $this->mTitle->getTalkPage();
2693                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2694                                 } else {
2695                                         $value = '';
2696                                 }
2697                                 break;
2698                         case 'subjectpagename':
2699                                 $subjPage = $this->mTitle->getSubjectPage();
2700                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2701                                 break;
2702                         case 'subjectpagenamee':
2703                                 $subjPage = $this->mTitle->getSubjectPage();
2704                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2705                                 break;
2706                         case 'revisionid':
2707                                 # Let the edit saving system know we should parse the page
2708                                 # *after* a revision ID has been assigned.
2709                                 $this->mOutput->setFlag( 'vary-revision' );
2710                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2711                                 $value = $this->mRevisionId;
2712                                 break;
2713                         case 'revisionday':
2714                                 # Let the edit saving system know we should parse the page
2715                                 # *after* a revision ID has been assigned. This is for null edits.
2716                                 $this->mOutput->setFlag( 'vary-revision' );
2717                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2718                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2719                                 break;
2720                         case 'revisionday2':
2721                                 # Let the edit saving system know we should parse the page
2722                                 # *after* a revision ID has been assigned. This is for null edits.
2723                                 $this->mOutput->setFlag( 'vary-revision' );
2724                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2725                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2726                                 break;
2727                         case 'revisionmonth':
2728                                 # Let the edit saving system know we should parse the page
2729                                 # *after* a revision ID has been assigned. This is for null edits.
2730                                 $this->mOutput->setFlag( 'vary-revision' );
2731                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2732                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2733                                 break;
2734                         case 'revisionmonth1':
2735                                 # Let the edit saving system know we should parse the page
2736                                 # *after* a revision ID has been assigned. This is for null edits.
2737                                 $this->mOutput->setFlag( 'vary-revision' );
2738                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2739                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2740                                 break;
2741                         case 'revisionyear':
2742                                 # Let the edit saving system know we should parse the page
2743                                 # *after* a revision ID has been assigned. This is for null edits.
2744                                 $this->mOutput->setFlag( 'vary-revision' );
2745                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2746                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2747                                 break;
2748                         case 'revisiontimestamp':
2749                                 # Let the edit saving system know we should parse the page
2750                                 # *after* a revision ID has been assigned. This is for null edits.
2751                                 $this->mOutput->setFlag( 'vary-revision' );
2752                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2753                                 $value = $this->getRevisionTimestamp();
2754                                 break;
2755                         case 'revisionuser':
2756                                 # Let the edit saving system know we should parse the page
2757                                 # *after* a revision ID has been assigned. This is for null edits.
2758                                 $this->mOutput->setFlag( 'vary-revision' );
2759                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2760                                 $value = $this->getRevisionUser();
2761                                 break;
2762                         case 'namespace':
2763                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2764                                 break;
2765                         case 'namespacee':
2766                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2767                                 break;
2768                         case 'talkspace':
2769                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2770                                 break;
2771                         case 'talkspacee':
2772                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2773                                 break;
2774                         case 'subjectspace':
2775                                 $value = $this->mTitle->getSubjectNsText();
2776                                 break;
2777                         case 'subjectspacee':
2778                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2779                                 break;
2780                         case 'currentdayname':
2781                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2782                                 break;
2783                         case 'currentyear':
2784                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2785                                 break;
2786                         case 'currenttime':
2787                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2788                                 break;
2789                         case 'currenthour':
2790                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2791                                 break;
2792                         case 'currentweek':
2793                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2794                                 # int to remove the padding
2795                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2796                                 break;
2797                         case 'currentdow':
2798                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2799                                 break;
2800                         case 'localdayname':
2801                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2802                                 break;
2803                         case 'localyear':
2804                                 $value = $wgContLang->formatNum( $localYear, true );
2805                                 break;
2806                         case 'localtime':
2807                                 $value = $wgContLang->time( $localTimestamp, false, false );
2808                                 break;
2809                         case 'localhour':
2810                                 $value = $wgContLang->formatNum( $localHour, true );
2811                                 break;
2812                         case 'localweek':
2813                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2814                                 # int to remove the padding
2815                                 $value = $wgContLang->formatNum( (int)$localWeek );
2816                                 break;
2817                         case 'localdow':
2818                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2819                                 break;
2820                         case 'numberofarticles':
2821                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2822                                 break;
2823                         case 'numberoffiles':
2824                                 $value = $wgContLang->formatNum( SiteStats::images() );
2825                                 break;
2826                         case 'numberofusers':
2827                                 $value = $wgContLang->formatNum( SiteStats::users() );
2828                                 break;
2829                         case 'numberofactiveusers':
2830                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2831                                 break;
2832                         case 'numberofpages':
2833                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2834                                 break;
2835                         case 'numberofadmins':
2836                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2837                                 break;
2838                         case 'numberofedits':
2839                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2840                                 break;
2841                         case 'numberofviews':
2842                                 $value = $wgContLang->formatNum( SiteStats::views() );
2843                                 break;
2844                         case 'currenttimestamp':
2845                                 $value = wfTimestamp( TS_MW, $ts );
2846                                 break;
2847                         case 'localtimestamp':
2848                                 $value = $localTimestamp;
2849                                 break;
2850                         case 'currentversion':
2851                                 $value = SpecialVersion::getVersion();
2852                                 break;
2853                         case 'articlepath':
2854                                 return $wgArticlePath;
2855                         case 'sitename':
2856                                 return $wgSitename;
2857                         case 'server':
2858                                 return $wgServer;
2859                         case 'servername':
2860                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2861                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2862                                 wfRestoreWarnings();
2863                                 return $serverName ? $serverName : $wgServer;
2864                         case 'scriptpath':
2865                                 return $wgScriptPath;
2866                         case 'stylepath':
2867                                 return $wgStylePath;
2868                         case 'directionmark':
2869                                 return $wgContLang->getDirMark();
2870                         case 'contentlanguage':
2871                                 global $wgLanguageCode;
2872                                 return $wgLanguageCode;
2873                         default:
2874                                 $ret = null;
2875                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2876                                         return $ret;
2877                                 } else {
2878                                         return null;
2879                                 }
2880                 }
2881
2882                 if ( $index )
2883                         $this->mVarCache[$index] = $value;
2884
2885                 return $value;
2886         }
2887
2888         /**
2889          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2890          *
2891          * @private
2892          */
2893         function initialiseVariables() {
2894                 wfProfileIn( __METHOD__ );
2895                 $variableIDs = MagicWord::getVariableIDs();
2896                 $substIDs = MagicWord::getSubstIDs();
2897
2898                 $this->mVariables = new MagicWordArray( $variableIDs );
2899                 $this->mSubstWords = new MagicWordArray( $substIDs );
2900                 wfProfileOut( __METHOD__ );
2901         }
2902
2903         /**
2904          * Preprocess some wikitext and return the document tree.
2905          * This is the ghost of replace_variables().
2906          *
2907          * @param $text String: The text to parse
2908          * @param $flags Integer: bitwise combination of:
2909          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2910          *                                     included. Default is to assume a direct page view.
2911          *
2912          * The generated DOM tree must depend only on the input text and the flags.
2913          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2914          *
2915          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2916          * change in the DOM tree for a given text, must be passed through the section identifier
2917          * in the section edit link and thus back to extractSections().
2918          *
2919          * The output of this function is currently only cached in process memory, but a persistent
2920          * cache may be implemented at a later date which takes further advantage of these strict
2921          * dependency requirements.
2922          *
2923          * @private
2924          *
2925          * @return PPNode
2926          */
2927         function preprocessToDom( $text, $flags = 0 ) {
2928                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2929                 return $dom;
2930         }
2931
2932         /**
2933          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2934          *
2935          * @return array
2936          */
2937         public static function splitWhitespace( $s ) {
2938                 $ltrimmed = ltrim( $s );
2939                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2940                 $trimmed = rtrim( $ltrimmed );
2941                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2942                 if ( $diff > 0 ) {
2943                         $w2 = substr( $ltrimmed, -$diff );
2944                 } else {
2945                         $w2 = '';
2946                 }
2947                 return array( $w1, $trimmed, $w2 );
2948         }
2949
2950         /**
2951          * Replace magic variables, templates, and template arguments
2952          * with the appropriate text. Templates are substituted recursively,
2953          * taking care to avoid infinite loops.
2954          *
2955          * Note that the substitution depends on value of $mOutputType:
2956          *  self::OT_WIKI: only {{subst:}} templates
2957          *  self::OT_PREPROCESS: templates but not extension tags
2958          *  self::OT_HTML: all templates and extension tags
2959          *
2960          * @param $text String: the text to transform
2961          * @param $frame PPFrame Object describing the arguments passed to the template.
2962          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2963          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2964          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2965          * @private
2966          *
2967          * @return string
2968          */
2969         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2970                 # Is there any text? Also, Prevent too big inclusions!
2971                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2972                         return $text;
2973                 }
2974                 wfProfileIn( __METHOD__ );
2975
2976                 if ( $frame === false ) {
2977                         $frame = $this->getPreprocessor()->newFrame();
2978                 } elseif ( !( $frame instanceof PPFrame ) ) {
2979                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2980                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2981                 }
2982
2983                 $dom = $this->preprocessToDom( $text );
2984                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2985                 $text = $frame->expand( $dom, $flags );
2986
2987                 wfProfileOut( __METHOD__ );
2988                 return $text;
2989         }
2990
2991         /**
2992          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2993          *
2994          * @return array
2995          */
2996         static function createAssocArgs( $args ) {
2997                 $assocArgs = array();
2998                 $index = 1;
2999                 foreach ( $args as $arg ) {
3000                         $eqpos = strpos( $arg, '=' );
3001                         if ( $eqpos === false ) {
3002                                 $assocArgs[$index++] = $arg;
3003                         } else {
3004                                 $name = trim( substr( $arg, 0, $eqpos ) );
3005                                 $value = trim( substr( $arg, $eqpos+1 ) );
3006                                 if ( $value === false ) {
3007                                         $value = '';
3008                                 }
3009                                 if ( $name !== false ) {
3010                                         $assocArgs[$name] = $value;
3011                                 }
3012                         }
3013                 }
3014
3015                 return $assocArgs;
3016         }
3017
3018         /**
3019          * Warn the user when a parser limitation is reached
3020          * Will warn at most once the user per limitation type
3021          *
3022          * @param $limitationType String: should be one of:
3023          *   'expensive-parserfunction' (corresponding messages:
3024          *       'expensive-parserfunction-warning',
3025          *       'expensive-parserfunction-category')
3026          *   'post-expand-template-argument' (corresponding messages:
3027          *       'post-expand-template-argument-warning',
3028          *       'post-expand-template-argument-category')
3029          *   'post-expand-template-inclusion' (corresponding messages:
3030          *       'post-expand-template-inclusion-warning',
3031          *       'post-expand-template-inclusion-category')
3032          * @param $current Current value
3033          * @param $max Maximum allowed, when an explicit limit has been
3034          *       exceeded, provide the values (optional)
3035          */
3036         function limitationWarn( $limitationType, $current=null, $max=null) {
3037                 # does no harm if $current and $max are present but are unnecessary for the message
3038                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3039                 $this->mOutput->addWarning( $warning );
3040                 $this->addTrackingCategory( "$limitationType-category" );
3041         }
3042
3043         /**
3044          * Return the text of a template, after recursively
3045          * replacing any variables or templates within the template.
3046          *
3047          * @param $piece Array: the parts of the template
3048          *  $piece['title']: the title, i.e. the part before the |
3049          *  $piece['parts']: the parameter array
3050          *  $piece['lineStart']: whether the brace was at the start of a line
3051          * @param $frame PPFrame The current frame, contains template arguments
3052          * @return String: the text of the template
3053          * @private
3054          */
3055         function braceSubstitution( $piece, $frame ) {
3056                 global $wgContLang, $wgNonincludableNamespaces;
3057                 wfProfileIn( __METHOD__ );
3058                 wfProfileIn( __METHOD__.'-setup' );
3059
3060                 # Flags
3061                 $found = false;             # $text has been filled
3062                 $nowiki = false;            # wiki markup in $text should be escaped
3063                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3064                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3065                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3066                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3067
3068                 # Title object, where $text came from
3069                 $title = null;
3070
3071                 # $part1 is the bit before the first |, and must contain only title characters.
3072                 # Various prefixes will be stripped from it later.
3073                 $titleWithSpaces = $frame->expand( $piece['title'] );
3074                 $part1 = trim( $titleWithSpaces );
3075                 $titleText = false;
3076
3077                 # Original title text preserved for various purposes
3078                 $originalTitle = $part1;
3079
3080                 # $args is a list of argument nodes, starting from index 0, not including $part1
3081                 # @todo FIXME: If piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3082                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3083                 wfProfileOut( __METHOD__.'-setup' );
3084                 wfProfileIn( __METHOD__."-title-$originalTitle" );
3085
3086                 # SUBST
3087                 wfProfileIn( __METHOD__.'-modifiers' );
3088                 if ( !$found ) {
3089
3090                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3091
3092                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3093                         # Decide whether to expand template or keep wikitext as-is.
3094                         if ( $this->ot['wiki'] ) {
3095                                 if ( $substMatch === false ) {
3096                                         $literal = true;  # literal when in PST with no prefix
3097                                 } else {
3098                                         $literal = false; # expand when in PST with subst: or safesubst:
3099                                 }
3100                         } else {
3101                                 if ( $substMatch == 'subst' ) {
3102                                         $literal = true;  # literal when not in PST with plain subst:
3103                                 } else {
3104                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3105                                 }
3106                         }
3107                         if ( $literal ) {
3108                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3109                                 $isLocalObj = true;
3110                                 $found = true;
3111                         }
3112                 }
3113
3114                 # Variables
3115                 if ( !$found && $args->getLength() == 0 ) {
3116                         $id = $this->mVariables->matchStartToEnd( $part1 );
3117                         if ( $id !== false ) {
3118                                 $text = $this->getVariableValue( $id, $frame );
3119                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3120                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3121                                 }
3122                                 $found = true;
3123                         }
3124                 }
3125
3126                 # MSG, MSGNW and RAW
3127                 if ( !$found ) {
3128                         # Check for MSGNW:
3129                         $mwMsgnw = MagicWord::get( 'msgnw' );
3130                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3131                                 $nowiki = true;
3132                         } else {
3133                                 # Remove obsolete MSG:
3134                                 $mwMsg = MagicWord::get( 'msg' );
3135                                 $mwMsg->matchStartAndRemove( $part1 );
3136                         }
3137
3138                         # Check for RAW:
3139                         $mwRaw = MagicWord::get( 'raw' );
3140                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3141                                 $forceRawInterwiki = true;
3142                         }
3143                 }
3144                 wfProfileOut( __METHOD__.'-modifiers' );
3145
3146                 # Parser functions
3147                 if ( !$found ) {
3148                         wfProfileIn( __METHOD__ . '-pfunc' );
3149
3150                         $colonPos = strpos( $part1, ':' );
3151                         if ( $colonPos !== false ) {
3152                                 # Case sensitive functions
3153                                 $function = substr( $part1, 0, $colonPos );
3154                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3155                                         $function = $this->mFunctionSynonyms[1][$function];
3156                                 } else {
3157                                         # Case insensitive functions
3158                                         $function = $wgContLang->lc( $function );
3159                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3160                                                 $function = $this->mFunctionSynonyms[0][$function];
3161                                         } else {
3162                                                 $function = false;
3163                                         }
3164                                 }
3165                                 if ( $function ) {
3166                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3167                                         $initialArgs = array( &$this );
3168                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3169                                         if ( $flags & SFH_OBJECT_ARGS ) {
3170                                                 # Add a frame parameter, and pass the arguments as an array
3171                                                 $allArgs = $initialArgs;
3172                                                 $allArgs[] = $frame;
3173                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3174                                                         $funcArgs[] = $args->item( $i );
3175                                                 }
3176                                                 $allArgs[] = $funcArgs;
3177                                         } else {
3178                                                 # Convert arguments to plain text
3179                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3180                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3181                                                 }
3182                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3183                                         }
3184
3185                                         # Workaround for PHP bug 35229 and similar
3186                                         if ( !is_callable( $callback ) ) {
3187                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3188                                                 wfProfileOut( __METHOD__ );
3189                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3190                                         }
3191                                         $result = call_user_func_array( $callback, $allArgs );
3192                                         $found = true;
3193                                         $noparse = true;
3194                                         $preprocessFlags = 0;
3195
3196                                         if ( is_array( $result ) ) {
3197                                                 if ( isset( $result[0] ) ) {
3198                                                         $text = $result[0];
3199                                                         unset( $result[0] );
3200                                                 }
3201
3202                                                 # Extract flags into the local scope
3203                                                 # This allows callers to set flags such as nowiki, found, etc.
3204                                                 extract( $result );
3205                                         } else {
3206                                                 $text = $result;
3207                                         }
3208                                         if ( !$noparse ) {
3209                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3210                                                 $isChildObj = true;
3211                                         }
3212                                 }
3213                         }
3214                         wfProfileOut( __METHOD__ . '-pfunc' );
3215                 }
3216
3217                 # Finish mangling title and then check for loops.
3218                 # Set $title to a Title object and $titleText to the PDBK
3219                 if ( !$found ) {
3220                         $ns = NS_TEMPLATE;
3221                         # Split the title into page and subpage
3222                         $subpage = '';
3223                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3224                         if ( $subpage !== '' ) {
3225                                 $ns = $this->mTitle->getNamespace();
3226                         }
3227                         $title = Title::newFromText( $part1, $ns );
3228                         if ( $title ) {
3229                                 $titleText = $title->getPrefixedText();
3230                                 # Check for language variants if the template is not found
3231                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3232                                         $wgContLang->findVariantLink( $part1, $title, true );
3233                                 }
3234                                 # Do recursion depth check
3235                                 $limit = $this->mOptions->getMaxTemplateDepth();
3236                                 if ( $frame->depth >= $limit ) {
3237                                         $found = true;
3238                                         $text = '<span class="error">'
3239                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3240                                                 . '</span>';
3241                                 }
3242                         }
3243                 }
3244
3245                 # Load from database
3246                 if ( !$found && $title ) {
3247                         wfProfileIn( __METHOD__ . '-loadtpl' );
3248                         if ( !$title->isExternal() ) {
3249                                 if ( $title->getNamespace() == NS_SPECIAL
3250                                         && $this->mOptions->getAllowSpecialInclusion()
3251                                         && $this->ot['html'] )
3252                                 {
3253                                         $text = SpecialPageFactory::capturePath( $title );
3254                                         if ( is_string( $text ) ) {
3255                                                 $found = true;
3256                                                 $isHTML = true;
3257                                                 $this->disableCache();
3258                                         }
3259                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3260                                         $found = false; # access denied
3261                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3262                                 } else {
3263                                         list( $text, $title ) = $this->getTemplateDom( $title );
3264                                         if ( $text !== false ) {
3265                                                 $found = true;
3266                                                 $isChildObj = true;
3267                                         }
3268                                 }
3269
3270                                 # If the title is valid but undisplayable, make a link to it
3271                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3272                                         $text = "[[:$titleText]]";
3273                                         $found = true;
3274                                 }
3275                         } elseif ( $title->isTrans() ) {
3276                                 # Interwiki transclusion
3277                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3278                                         $text = $this->interwikiTransclude( $title, 'render' );
3279                                         $isHTML = true;
3280                                 } else {
3281                                         $text = $this->interwikiTransclude( $title, 'raw' );
3282                                         # Preprocess it like a template
3283                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3284                                         $isChildObj = true;
3285                                 }
3286                                 $found = true;
3287                         }
3288
3289                         # Do infinite loop check
3290                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3291                         if ( !$frame->loopCheck( $title ) ) {
3292                                 $found = true;
3293                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3294                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3295                         }
3296                         wfProfileOut( __METHOD__ . '-loadtpl' );
3297                 }
3298
3299                 # If we haven't found text to substitute by now, we're done
3300                 # Recover the source wikitext and return it
3301                 if ( !$found ) {
3302                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3303                         wfProfileOut( __METHOD__."-title-$originalTitle" );
3304                         wfProfileOut( __METHOD__ );
3305                         return array( 'object' => $text );
3306                 }
3307
3308                 # Expand DOM-style return values in a child frame
3309                 if ( $isChildObj ) {
3310                         # Clean up argument array
3311                         $newFrame = $frame->newChild( $args, $title );
3312
3313                         if ( $nowiki ) {
3314                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3315                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3316                                 # Expansion is eligible for the empty-frame cache
3317                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3318                                         $text = $this->mTplExpandCache[$titleText];
3319                                 } else {
3320                                         $text = $newFrame->expand( $text );
3321                                         $this->mTplExpandCache[$titleText] = $text;
3322                                 }
3323                         } else {
3324                                 # Uncached expansion
3325                                 $text = $newFrame->expand( $text );
3326                         }
3327                 }
3328                 if ( $isLocalObj && $nowiki ) {
3329                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3330                         $isLocalObj = false;
3331                 }
3332
3333                 # Replace raw HTML by a placeholder
3334                 # Add a blank line preceding, to prevent it from mucking up
3335                 # immediately preceding headings
3336                 if ( $isHTML ) {
3337                         $text = "\n\n" . $this->insertStripItem( $text );
3338                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3339                         # Escape nowiki-style return values
3340                         $text = wfEscapeWikiText( $text );
3341                 } elseif ( is_string( $text )
3342                         && !$piece['lineStart']
3343                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3344                 {
3345                         # Bug 529: if the template begins with a table or block-level
3346                         # element, it should be treated as beginning a new line.
3347                         # This behaviour is somewhat controversial.
3348                         $text = "\n" . $text;
3349                 }
3350
3351                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3352                         # Error, oversize inclusion
3353                         if ( $titleText !== false ) {
3354                                 # Make a working, properly escaped link if possible (bug 23588)
3355                                 $text = "[[:$titleText]]";
3356                         } else {
3357                                 # This will probably not be a working link, but at least it may
3358                                 # provide some hint of where the problem is
3359                                 preg_replace( '/^:/', '', $originalTitle );
3360                                 $text = "[[:$originalTitle]]";
3361                         }
3362                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3363                         $this->limitationWarn( 'post-expand-template-inclusion' );
3364                 }
3365
3366                 if ( $isLocalObj ) {
3367                         $ret = array( 'object' => $text );
3368                 } else {
3369                         $ret = array( 'text' => $text );
3370                 }
3371
3372                 wfProfileOut( __METHOD__."-title-$originalTitle" );
3373                 wfProfileOut( __METHOD__ );
3374                 return $ret;
3375         }
3376
3377         /**
3378          * Get the semi-parsed DOM representation of a template with a given title,
3379          * and its redirect destination title. Cached.
3380          *
3381          * @return array
3382          */
3383         function getTemplateDom( $title ) {
3384                 $cacheTitle = $title;
3385                 $titleText = $title->getPrefixedDBkey();
3386
3387                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3388                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3389                         $title = Title::makeTitle( $ns, $dbk );
3390                         $titleText = $title->getPrefixedDBkey();
3391                 }
3392                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3393                         return array( $this->mTplDomCache[$titleText], $title );
3394                 }
3395
3396                 # Cache miss, go to the database
3397                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3398
3399                 if ( $text === false ) {
3400                         $this->mTplDomCache[$titleText] = false;
3401                         return array( false, $title );
3402                 }
3403
3404                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3405                 $this->mTplDomCache[ $titleText ] = $dom;
3406
3407                 if ( !$title->equals( $cacheTitle ) ) {
3408                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3409                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3410                 }
3411
3412                 return array( $dom, $title );
3413         }
3414
3415         /**
3416          * Fetch the unparsed text of a template and register a reference to it.
3417          * @param Title $title
3418          * @return Array ( string or false, Title )
3419          */
3420         function fetchTemplateAndTitle( $title ) {
3421                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3422                 $stuff = call_user_func( $templateCb, $title, $this );
3423                 $text = $stuff['text'];
3424                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3425                 if ( isset( $stuff['deps'] ) ) {
3426                         foreach ( $stuff['deps'] as $dep ) {
3427                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3428                         }
3429                 }
3430                 return array( $text, $finalTitle );
3431         }
3432
3433         /**
3434          * Fetch the unparsed text of a template and register a reference to it.
3435          * @param Title $title
3436          * @return mixed string or false
3437          */
3438         function fetchTemplate( $title ) {
3439                 $rv = $this->fetchTemplateAndTitle( $title );
3440                 return $rv[0];
3441         }
3442
3443         /**
3444          * Static function to get a template
3445          * Can be overridden via ParserOptions::setTemplateCallback().
3446          *
3447          * @return array
3448          */
3449         static function statelessFetchTemplate( $title, $parser = false ) {
3450                 $text = $skip = false;
3451                 $finalTitle = $title;
3452                 $deps = array();
3453
3454                 # Loop to fetch the article, with up to 1 redirect
3455                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3456                         # Give extensions a chance to select the revision instead
3457                         $id = false; # Assume current
3458                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3459                                 array( $parser, $title, &$skip, &$id ) );
3460
3461                         if ( $skip ) {
3462                                 $text = false;
3463                                 $deps[] = array(
3464                                         'title'         => $title,
3465                                         'page_id'       => $title->getArticleID(),
3466                                         'rev_id'        => null
3467                                 );
3468                                 break;
3469                         }
3470                         # Get the revision
3471                         $rev = $id
3472                                 ? Revision::newFromId( $id )
3473                                 : Revision::newFromTitle( $title );
3474                         $rev_id = $rev ? $rev->getId() : 0;
3475                         # If there is no current revision, there is no page
3476                         if ( $id === false && !$rev ) {
3477                                 $linkCache = LinkCache::singleton();
3478                                 $linkCache->addBadLinkObj( $title );
3479                         }
3480
3481                         $deps[] = array(
3482                                 'title'         => $title,
3483                                 'page_id'       => $title->getArticleID(),
3484                                 'rev_id'        => $rev_id );
3485                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3486                                 # We fetched a rev from a different title; register it too...
3487                                 $deps[] = array(
3488                                         'title'         => $rev->getTitle(),
3489                                         'page_id'       => $rev->getPage(),
3490                                         'rev_id'        => $rev_id );
3491                         }
3492
3493                         if ( $rev ) {
3494                                 $text = $rev->getText();
3495                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3496                                 global $wgContLang;
3497                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3498                                 if ( !$message->exists() ) {
3499                                         $text = false;
3500                                         break;
3501                                 }
3502                                 $text = $message->plain();
3503                         } else {
3504                                 break;
3505                         }
3506                         if ( $text === false ) {
3507                                 break;
3508                         }
3509                         # Redirect?
3510                         $finalTitle = $title;
3511                         $title = Title::newFromRedirect( $text );
3512                 }
3513                 return array(
3514                         'text' => $text,
3515                         'finalTitle' => $finalTitle,
3516                         'deps' => $deps );
3517         }
3518
3519         /**
3520          * Fetch a file and its title and register a reference to it.
3521          * @param Title $title
3522          * @param string $time MW timestamp
3523          * @param string $sha1 base 36 SHA-1
3524          * @return mixed File or false
3525          */
3526         function fetchFile( $title, $time = false, $sha1 = false ) {
3527                 $res = $this->fetchFileAndTitle( $title, $time, $sha1 );
3528                 return $res[0];
3529         }
3530
3531         /**
3532          * Fetch a file and its title and register a reference to it.
3533          * @param Title $title
3534          * @param string $time MW timestamp
3535          * @param string $sha1 base 36 SHA-1
3536          * @return Array ( File or false, Title of file )
3537          */
3538         function fetchFileAndTitle( $title, $time = false, $sha1 = false ) {
3539                 if ( $time === '0' ) {
3540                         $file = false; // broken thumbnail forced by hook
3541                 } elseif ( $sha1 ) { // get by (sha1,timestamp)
3542                         $file = RepoGroup::singleton()->findFileFromKey( $sha1, array( 'time' => $time ) );
3543                 } else { // get by (name,timestamp)
3544                         $file = wfFindFile( $title, array( 'time' => $time ) );
3545                 }
3546                 $time = $file ? $file->getTimestamp() : false;
3547                 $sha1 = $file ? $file->getSha1() : false;
3548                 # Register the file as a dependency...
3549                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3550                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3551                         # Update fetched file title
3552                         $title = $file->getTitle();
3553                         if ( is_null( $file->getRedirectedTitle() ) ) {
3554                                 # This file was not a redirect, but the title does not match.
3555                                 # Register under the new name because otherwise the link will
3556                                 # get lost.
3557                                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3558                         }
3559                 }
3560                 return array( $file, $title );
3561         }
3562
3563         /**
3564          * Transclude an interwiki link.
3565          *
3566          * @param $title Title
3567          * @param $action
3568          *
3569          * @return string
3570          */
3571         function interwikiTransclude( $title, $action ) {
3572                 global $wgEnableScaryTranscluding;
3573
3574                 if ( !$wgEnableScaryTranscluding ) {
3575                         return wfMsgForContent('scarytranscludedisabled');
3576                 }
3577
3578                 $url = $title->getFullUrl( "action=$action" );
3579
3580                 if ( strlen( $url ) > 255 ) {
3581                         return wfMsgForContent( 'scarytranscludetoolong' );
3582                 }
3583                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3584         }
3585
3586         /**
3587          * @param $url string
3588          * @return Mixed|String
3589          */
3590         function fetchScaryTemplateMaybeFromCache( $url ) {
3591                 global $wgTranscludeCacheExpiry;
3592                 $dbr = wfGetDB( DB_SLAVE );
3593                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3594                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3595                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3596                 if ( $obj ) {
3597                         return $obj->tc_contents;
3598                 }
3599
3600                 $text = Http::get( $url );
3601                 if ( !$text ) {
3602                         return wfMsgForContent( 'scarytranscludefailed', $url );
3603                 }
3604
3605                 $dbw = wfGetDB( DB_MASTER );
3606                 $dbw->replace( 'transcache', array('tc_url'), array(
3607                         'tc_url' => $url,
3608                         'tc_time' => $dbw->timestamp( time() ),
3609                         'tc_contents' => $text)
3610                 );
3611                 return $text;
3612         }
3613
3614         /**
3615          * Triple brace replacement -- used for template arguments
3616          * @private
3617          *
3618          * @param $peice array
3619          * @param $frame PPFrame
3620          *
3621          * @return array
3622          */
3623         function argSubstitution( $piece, $frame ) {
3624                 wfProfileIn( __METHOD__ );
3625
3626                 $error = false;
3627                 $parts = $piece['parts'];
3628                 $nameWithSpaces = $frame->expand( $piece['title'] );
3629                 $argName = trim( $nameWithSpaces );
3630                 $object = false;
3631                 $text = $frame->getArgument( $argName );
3632                 if (  $text === false && $parts->getLength() > 0
3633                   && (
3634                         $this->ot['html']
3635                         || $this->ot['pre']
3636                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3637                   )
3638                 ) {
3639                         # No match in frame, use the supplied default
3640                         $object = $parts->item( 0 )->getChildren();
3641                 }
3642                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3643                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3644                         $this->limitationWarn( 'post-expand-template-argument' );
3645                 }
3646
3647                 if ( $text === false && $object === false ) {
3648                         # No match anywhere
3649                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3650                 }
3651                 if ( $error !== false ) {
3652                         $text .= $error;
3653                 }
3654                 if ( $object !== false ) {
3655                         $ret = array( 'object' => $object );
3656                 } else {
3657                         $ret = array( 'text' => $text );
3658                 }
3659
3660                 wfProfileOut( __METHOD__ );
3661                 return $ret;
3662         }
3663
3664         /**
3665          * Return the text to be used for a given extension tag.
3666          * This is the ghost of strip().
3667          *
3668          * @param $params Associative array of parameters:
3669          *     name       PPNode for the tag name
3670          *     attr       PPNode for unparsed text where tag attributes are thought to be
3671          *     attributes Optional associative array of parsed attributes
3672          *     inner      Contents of extension element
3673          *     noClose    Original text did not have a close tag
3674          * @param $frame PPFrame
3675          *
3676          * @return string
3677          */
3678         function extensionSubstitution( $params, $frame ) {
3679                 $name = $frame->expand( $params['name'] );
3680                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3681                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3682                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3683
3684                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3685                         ( $this->ot['html'] || $this->ot['pre'] );
3686                 if ( $isFunctionTag ) {
3687                         $markerType = 'none';
3688                 } else {
3689                         $markerType = 'general';
3690                 }
3691                 if ( $this->ot['html'] || $isFunctionTag ) {
3692                         $name = strtolower( $name );
3693                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3694                         if ( isset( $params['attributes'] ) ) {
3695                                 $attributes = $attributes + $params['attributes'];
3696                         }
3697
3698                         if ( isset( $this->mTagHooks[$name] ) ) {
3699                                 # Workaround for PHP bug 35229 and similar
3700                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3701                                         throw new MWException( "Tag hook for $name is not callable\n" );
3702                                 }
3703                                 $output = call_user_func_array( $this->mTagHooks[$name],
3704                                         array( $content, $attributes, $this, $frame ) );
3705                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3706                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3707                                 if ( !is_callable( $callback ) ) {
3708                                         throw new MWException( "Tag hook for $name is not callable\n" );
3709                                 }
3710
3711                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3712                         } else {
3713                                 $output = '<span class="error">Invalid tag extension name: ' .
3714                                         htmlspecialchars( $name ) . '</span>';
3715                         }
3716
3717                         if ( is_array( $output ) ) {
3718                                 # Extract flags to local scope (to override $markerType)
3719                                 $flags = $output;
3720                                 $output = $flags[0];
3721                                 unset( $flags[0] );
3722                                 extract( $flags );
3723                         }
3724                 } else {
3725                         if ( is_null( $attrText ) ) {
3726                                 $attrText = '';
3727                         }
3728                         if ( isset( $params['attributes'] ) ) {
3729                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3730                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3731                                                 htmlspecialchars( $attrValue ) . '"';
3732                                 }
3733                         }
3734                         if ( $content === null ) {
3735                                 $output = "<$name$attrText/>";
3736                         } else {
3737                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3738                                 $output = "<$name$attrText>$content$close";
3739                         }
3740                 }
3741
3742                 if ( $markerType === 'none' ) {
3743                         return $output;
3744                 } elseif ( $markerType === 'nowiki' ) {
3745                         $this->mStripState->addNoWiki( $marker, $output );
3746                 } elseif ( $markerType === 'general' ) {
3747                         $this->mStripState->addGeneral( $marker, $output );
3748                 } else {
3749                         throw new MWException( __METHOD__.': invalid marker type' );
3750                 }
3751                 return $marker;
3752         }
3753
3754         /**
3755          * Increment an include size counter
3756          *
3757          * @param $type String: the type of expansion
3758          * @param $size Integer: the size of the text
3759          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3760          */
3761         function incrementIncludeSize( $type, $size ) {
3762                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3763                         return false;
3764                 } else {
3765                         $this->mIncludeSizes[$type] += $size;
3766                         return true;
3767                 }
3768         }
3769
3770         /**
3771          * Increment the expensive function count
3772          *
3773          * @return Boolean: false if the limit has been exceeded
3774          */
3775         function incrementExpensiveFunctionCount() {
3776                 global $wgExpensiveParserFunctionLimit;
3777                 $this->mExpensiveFunctionCount++;
3778                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3779                         return true;
3780                 }
3781                 return false;
3782         }
3783
3784         /**
3785          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3786          * Fills $this->mDoubleUnderscores, returns the modified text
3787          */
3788         function doDoubleUnderscore( $text ) {
3789                 wfProfileIn( __METHOD__ );
3790
3791                 # The position of __TOC__ needs to be recorded
3792                 $mw = MagicWord::get( 'toc' );
3793                 if ( $mw->match( $text ) ) {
3794                         $this->mShowToc = true;
3795                         $this->mForceTocPosition = true;
3796
3797                         # Set a placeholder. At the end we'll fill it in with the TOC.
3798                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3799
3800                         # Only keep the first one.
3801                         $text = $mw->replace( '', $text );
3802                 }
3803
3804                 # Now match and remove the rest of them
3805                 $mwa = MagicWord::getDoubleUnderscoreArray();
3806                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3807
3808                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3809                         $this->mOutput->mNoGallery = true;
3810                 }
3811                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3812                         $this->mShowToc = false;
3813                 }
3814                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3815                         $this->addTrackingCategory( 'hidden-category-category' );
3816                 }
3817                 # (bug 8068) Allow control over whether robots index a page.
3818                 #
3819                 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here!  This
3820                 # is not desirable, the last one on the page should win.
3821                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3822                         $this->mOutput->setIndexPolicy( 'noindex' );
3823                         $this->addTrackingCategory( 'noindex-category' );
3824                 }
3825                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3826                         $this->mOutput->setIndexPolicy( 'index' );
3827                         $this->addTrackingCategory( 'index-category' );
3828                 }
3829
3830                 # Cache all double underscores in the database
3831                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3832                         $this->mOutput->setProperty( $key, '' );
3833                 }
3834
3835                 wfProfileOut( __METHOD__ );
3836                 return $text;
3837         }
3838
3839         /**
3840          * Add a tracking category, getting the title from a system message,
3841          * or print a debug message if the title is invalid.
3842          *
3843          * @param $msg String: message key
3844          * @return Boolean: whether the addition was successful
3845          */
3846         protected function addTrackingCategory( $msg ) {
3847                 $cat = wfMsgForContent( $msg );
3848
3849                 # Allow tracking categories to be disabled by setting them to "-"
3850                 if ( $cat === '-' ) {
3851                         return false;
3852                 }
3853
3854                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3855                 if ( $containerCategory ) {
3856                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3857                         return true;
3858                 } else {
3859                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3860                         return false;
3861                 }
3862         }
3863
3864         /**
3865          * This function accomplishes several tasks:
3866          * 1) Auto-number headings if that option is enabled
3867          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3868          * 3) Add a Table of contents on the top for users who have enabled the option
3869          * 4) Auto-anchor headings
3870          *
3871          * It loops through all headlines, collects the necessary data, then splits up the
3872          * string and re-inserts the newly formatted headlines.
3873          *
3874          * @param $text String
3875          * @param $origText String: original, untouched wikitext
3876          * @param $isMain Boolean
3877          * @private
3878          */
3879         function formatHeadings( $text, $origText, $isMain=true ) {
3880                 global $wgMaxTocLevel, $wgHtml5, $wgExperimentalHtmlIds;
3881
3882                 # Inhibit editsection links if requested in the page
3883                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3884                         $showEditLink = 0;
3885                 } else {
3886                         $showEditLink = $this->mOptions->getEditSection();
3887                 }
3888                 if ( $showEditLink ) {
3889                         $this->mOutput->setEditSectionTokens( true );
3890                 }
3891
3892                 # Get all headlines for numbering them and adding funky stuff like [edit]
3893                 # links - this is for later, but we need the number of headlines right now
3894                 $matches = array();
3895                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3896
3897                 # if there are fewer than 4 headlines in the article, do not show TOC
3898                 # unless it's been explicitly enabled.
3899                 $enoughToc = $this->mShowToc &&
3900                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3901
3902                 # Allow user to stipulate that a page should have a "new section"
3903                 # link added via __NEWSECTIONLINK__
3904                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3905                         $this->mOutput->setNewSection( true );
3906                 }
3907
3908                 # Allow user to remove the "new section"
3909                 # link via __NONEWSECTIONLINK__
3910                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3911                         $this->mOutput->hideNewSection( true );
3912                 }
3913
3914                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3915                 # override above conditions and always show TOC above first header
3916                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3917                         $this->mShowToc = true;
3918                         $enoughToc = true;
3919                 }
3920
3921                 # headline counter
3922                 $headlineCount = 0;
3923                 $numVisible = 0;
3924
3925                 # Ugh .. the TOC should have neat indentation levels which can be
3926                 # passed to the skin functions. These are determined here
3927                 $toc = '';
3928                 $full = '';
3929                 $head = array();
3930                 $sublevelCount = array();
3931                 $levelCount = array();
3932                 $level = 0;
3933                 $prevlevel = 0;
3934                 $toclevel = 0;
3935                 $prevtoclevel = 0;
3936                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3937                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3938                 $oldType = $this->mOutputType;
3939                 $this->setOutputType( self::OT_WIKI );
3940                 $frame = $this->getPreprocessor()->newFrame();
3941                 $root = $this->preprocessToDom( $origText );
3942                 $node = $root->getFirstChild();
3943                 $byteOffset = 0;
3944                 $tocraw = array();
3945                 $refers = array();
3946
3947                 foreach ( $matches[3] as $headline ) {
3948                         $isTemplate = false;
3949                         $titleText = false;
3950                         $sectionIndex = false;
3951                         $numbering = '';
3952                         $markerMatches = array();
3953                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3954                                 $serial = $markerMatches[1];
3955                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3956                                 $isTemplate = ( $titleText != $baseTitleText );
3957                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3958                         }
3959
3960                         if ( $toclevel ) {
3961                                 $prevlevel = $level;
3962                         }
3963                         $level = $matches[1][$headlineCount];
3964
3965                         if ( $level > $prevlevel ) {
3966                                 # Increase TOC level
3967                                 $toclevel++;
3968                                 $sublevelCount[$toclevel] = 0;
3969                                 if ( $toclevel<$wgMaxTocLevel ) {
3970                                         $prevtoclevel = $toclevel;
3971                                         $toc .= Linker::tocIndent();
3972                                         $numVisible++;
3973                                 }
3974                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3975                                 # Decrease TOC level, find level to jump to
3976
3977                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3978                                         if ( $levelCount[$i] == $level ) {
3979                                                 # Found last matching level
3980                                                 $toclevel = $i;
3981                                                 break;
3982                                         } elseif ( $levelCount[$i] < $level ) {
3983                                                 # Found first matching level below current level
3984                                                 $toclevel = $i + 1;
3985                                                 break;
3986                                         }
3987                                 }
3988                                 if ( $i == 0 ) {
3989                                         $toclevel = 1;
3990                                 }
3991                                 if ( $toclevel<$wgMaxTocLevel ) {
3992                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3993                                                 # Unindent only if the previous toc level was shown :p
3994                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
3995                                                 $prevtoclevel = $toclevel;
3996                                         } else {
3997                                                 $toc .= Linker::tocLineEnd();
3998                                         }
3999                                 }
4000                         } else {
4001                                 # No change in level, end TOC line
4002                                 if ( $toclevel<$wgMaxTocLevel ) {
4003                                         $toc .= Linker::tocLineEnd();
4004                                 }
4005                         }
4006
4007                         $levelCount[$toclevel] = $level;
4008
4009                         # count number of headlines for each level
4010                         @$sublevelCount[$toclevel]++;
4011                         $dot = 0;
4012                         for( $i = 1; $i <= $toclevel; $i++ ) {
4013                                 if ( !empty( $sublevelCount[$i] ) ) {
4014                                         if ( $dot ) {
4015                                                 $numbering .= '.';
4016                                         }
4017                                         $numbering .= $this->getFunctionLang()->formatNum( $sublevelCount[$i] );
4018                                         $dot = 1;
4019                                 }
4020                         }
4021
4022                         # The safe header is a version of the header text safe to use for links
4023                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4024                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
4025
4026                         # Remove link placeholders by the link text.
4027                         #     <!--LINK number-->
4028                         # turns into
4029                         #     link text with suffix
4030                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
4031
4032                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
4033                         $tocline = preg_replace(
4034                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
4035                                 array( '',                          '<$1>' ),
4036                                 $safeHeadline
4037                         );
4038                         $tocline = trim( $tocline );
4039
4040                         # For the anchor, strip out HTML-y stuff period
4041                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4042                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4043
4044                         # Save headline for section edit hint before it's escaped
4045                         $headlineHint = $safeHeadline;
4046
4047                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4048                                 # For reverse compatibility, provide an id that's
4049                                 # HTML4-compatible, like we used to.
4050                                 #
4051                                 # It may be worth noting, academically, that it's possible for
4052                                 # the legacy anchor to conflict with a non-legacy headline
4053                                 # anchor on the page.  In this case likely the "correct" thing
4054                                 # would be to either drop the legacy anchors or make sure
4055                                 # they're numbered first.  However, this would require people
4056                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4057                                 # manually, so let's not bother worrying about it.
4058                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4059                                         array( 'noninitial', 'legacy' ) );
4060                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4061
4062                                 if ( $legacyHeadline == $safeHeadline ) {
4063                                         # No reason to have both (in fact, we can't)
4064                                         $legacyHeadline = false;
4065                                 }
4066                         } else {
4067                                 $legacyHeadline = false;
4068                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4069                                         'noninitial' );
4070                         }
4071
4072                         # HTML names must be case-insensitively unique (bug 10721).
4073                         # This does not apply to Unicode characters per
4074                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4075                         # @todo FIXME: We may be changing them depending on the current locale.
4076                         $arrayKey = strtolower( $safeHeadline );
4077                         if ( $legacyHeadline === false ) {
4078                                 $legacyArrayKey = false;
4079                         } else {
4080                                 $legacyArrayKey = strtolower( $legacyHeadline );
4081                         }
4082
4083                         # count how many in assoc. array so we can track dupes in anchors
4084                         if ( isset( $refers[$arrayKey] ) ) {
4085                                 $refers[$arrayKey]++;
4086                         } else {
4087                                 $refers[$arrayKey] = 1;
4088                         }
4089                         if ( isset( $refers[$legacyArrayKey] ) ) {
4090                                 $refers[$legacyArrayKey]++;
4091                         } else {
4092                                 $refers[$legacyArrayKey] = 1;
4093                         }
4094
4095                         # Don't number the heading if it is the only one (looks silly)
4096                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4097                                 # the two are different if the line contains a link
4098                                 $headline = $numbering . ' ' . $headline;
4099                         }
4100
4101                         # Create the anchor for linking from the TOC to the section
4102                         $anchor = $safeHeadline;
4103                         $legacyAnchor = $legacyHeadline;
4104                         if ( $refers[$arrayKey] > 1 ) {
4105                                 $anchor .= '_' . $refers[$arrayKey];
4106                         }
4107                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4108                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4109                         }
4110                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4111                                 $toc .= Linker::tocLine( $anchor, $tocline,
4112                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4113                         }
4114
4115                         # Add the section to the section tree
4116                         # Find the DOM node for this header
4117                         while ( $node && !$isTemplate ) {
4118                                 if ( $node->getName() === 'h' ) {
4119                                         $bits = $node->splitHeading();
4120                                         if ( $bits['i'] == $sectionIndex ) {
4121                                                 break;
4122                                         }
4123                                 }
4124                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4125                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4126                                 $node = $node->getNextSibling();
4127                         }
4128                         $tocraw[] = array(
4129                                 'toclevel' => $toclevel,
4130                                 'level' => $level,
4131                                 'line' => $tocline,
4132                                 'number' => $numbering,
4133                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4134                                 'fromtitle' => $titleText,
4135                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4136                                 'anchor' => $anchor,
4137                         );
4138
4139                         # give headline the correct <h#> tag
4140                         if ( $showEditLink && $sectionIndex !== false ) {
4141                                 // Output edit section links as markers with styles that can be customized by skins
4142                                 if ( $isTemplate ) {
4143                                         # Put a T flag in the section identifier, to indicate to extractSections()
4144                                         # that sections inside <includeonly> should be counted.
4145                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4146                                 } else {
4147                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4148                                 }
4149                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4150                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4151                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4152                                 // so we don't have to worry about a user trying to input one of these markers directly.
4153                                 // We use a page and section attribute to stop the language converter from converting these important bits
4154                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4155                                 // be able to convert that piece of data.
4156                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4157                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4158                                 if ( isset($editlinkArgs[2]) ) {
4159                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4160                                 } else {
4161                                         $editlink .= '/>';
4162                                 }
4163                         } else {
4164                                 $editlink = '';
4165                         }
4166                         $head[$headlineCount] = Linker::makeHeadline( $level,
4167                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4168                                 $editlink, $legacyAnchor );
4169
4170                         $headlineCount++;
4171                 }
4172
4173                 $this->setOutputType( $oldType );
4174
4175                 # Never ever show TOC if no headers
4176                 if ( $numVisible < 1 ) {
4177                         $enoughToc = false;
4178                 }
4179
4180                 if ( $enoughToc ) {
4181                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4182                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4183                         }
4184                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLang() );
4185                         $this->mOutput->setTOCHTML( $toc );
4186                 }
4187
4188                 if ( $isMain ) {
4189                         $this->mOutput->setSections( $tocraw );
4190                 }
4191
4192                 # split up and insert constructed headlines
4193                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4194                 $i = 0;
4195
4196                 // build an array of document sections
4197                 $sections = array();
4198                 foreach ( $blocks as $block ) {
4199                         // $head is zero-based, sections aren't.
4200                         if ( empty( $head[$i - 1] ) ) {
4201                                 $sections[$i] = $block;
4202                         } else {
4203                                 $sections[$i] = $head[$i - 1] . $block;
4204                         }
4205
4206                         /**
4207                          * Send a hook, one per section.
4208                          * The idea here is to be able to make section-level DIVs, but to do so in a
4209                          * lower-impact, more correct way than r50769
4210                          *
4211                          * $this : caller
4212                          * $section : the section number
4213                          * &$sectionContent : ref to the content of the section
4214                          * $showEditLinks : boolean describing whether this section has an edit link
4215                          */
4216                         wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4217
4218                         $i++;
4219                 }
4220
4221                 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4222                         // append the TOC at the beginning
4223                         // Top anchor now in skin
4224                         $sections[0] = $sections[0] . $toc . "\n";
4225                 }
4226
4227                 $full .= join( '', $sections );
4228
4229                 if ( $this->mForceTocPosition ) {
4230                         return str_replace( '<!--MWTOC-->', $toc, $full );
4231                 } else {
4232                         return $full;
4233                 }
4234         }
4235
4236         /**
4237          * Transform wiki markup when saving a page by doing \r\n -> \n
4238          * conversion, substitting signatures, {{subst:}} templates, etc.
4239          *
4240          * @param $text String: the text to transform
4241          * @param $title Title: the Title object for the current article
4242          * @param $user User: the User object describing the current user
4243          * @param $options ParserOptions: parsing options
4244          * @param $clearState Boolean: whether to clear the parser state first
4245          * @return String: the altered wiki markup
4246          */
4247         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4248                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4249                 $this->setUser( $user );
4250
4251                 $pairs = array(
4252                         "\r\n" => "\n",
4253                 );
4254                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4255                 if( $options->getPreSaveTransform() ) {
4256                         $text = $this->pstPass2( $text, $user );
4257                 }
4258                 $text = $this->mStripState->unstripBoth( $text );
4259
4260                 $this->setUser( null ); #Reset
4261
4262                 return $text;
4263         }
4264
4265         /**
4266          * Pre-save transform helper function
4267          * @private
4268          */
4269         function pstPass2( $text, $user ) {
4270                 global $wgContLang, $wgLocaltimezone;
4271
4272                 # Note: This is the timestamp saved as hardcoded wikitext to
4273                 # the database, we use $wgContLang here in order to give
4274                 # everyone the same signature and use the default one rather
4275                 # than the one selected in each user's preferences.
4276                 # (see also bug 12815)
4277                 $ts = $this->mOptions->getTimestamp();
4278                 if ( isset( $wgLocaltimezone ) ) {
4279                         $tz = $wgLocaltimezone;
4280                 } else {
4281                         $tz = date_default_timezone_get();
4282                 }
4283
4284                 $unixts = wfTimestamp( TS_UNIX, $ts );
4285                 $oldtz = date_default_timezone_get();
4286                 date_default_timezone_set( $tz );
4287                 $ts = date( 'YmdHis', $unixts );
4288                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4289
4290                 # Allow translation of timezones through wiki. date() can return
4291                 # whatever crap the system uses, localised or not, so we cannot
4292                 # ship premade translations.
4293                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4294                 $msg = wfMessage( $key )->inContentLanguage();
4295                 if ( $msg->exists() ) {
4296                         $tzMsg = $msg->text();
4297                 }
4298
4299                 date_default_timezone_set( $oldtz );
4300
4301                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4302
4303                 # Variable replacement
4304                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4305                 $text = $this->replaceVariables( $text );
4306
4307                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4308                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4309
4310                 # Signatures
4311                 $sigText = $this->getUserSig( $user );
4312                 $text = strtr( $text, array(
4313                         '~~~~~' => $d,
4314                         '~~~~' => "$sigText $d",
4315                         '~~~' => $sigText
4316                 ) );
4317
4318                 # Context links: [[|name]] and [[name (context)|]]
4319                 global $wgLegalTitleChars;
4320                 $tc = "[$wgLegalTitleChars]";
4321                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4322
4323                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)(\s?\\($tc+\\))\\|]]/";          # [[ns:page (context)|]]
4324                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(\s?（$tc+）)\\|]]/";          # [[ns:page（context）|]]
4325                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)(\s?\\($tc+\\)|)(, $tc+|)\\|]]/";        # [[ns:page (context), context|]]
4326                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4327
4328                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4329                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4330                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4331                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4332
4333                 $t = $this->mTitle->getText();
4334                 $m = array();
4335                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4336                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4337                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4338                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4339                 } else {
4340                         # if there's no context, don't bother duplicating the title
4341                         $text = preg_replace( $p2, '[[\\1]]', $text );
4342                 }
4343
4344                 # Trim trailing whitespace
4345                 $text = rtrim( $text );
4346
4347                 return $text;
4348         }
4349
4350         /**
4351          * Fetch the user's signature text, if any, and normalize to
4352          * validated, ready-to-insert wikitext.
4353          * If you have pre-fetched the nickname or the fancySig option, you can
4354          * specify them here to save a database query.
4355          * Do not reuse this parser instance after calling getUserSig(),
4356          * as it may have changed if it's the $wgParser.
4357          *
4358          * @param $user User
4359          * @param $nickname String: nickname to use or false to use user's default nickname
4360          * @param $fancySig Boolean: whether the nicknname is the complete signature
4361          *                  or null to use default value
4362          * @return string
4363          */
4364         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4365                 global $wgMaxSigChars;
4366
4367                 $username = $user->getName();
4368
4369                 # If not given, retrieve from the user object.
4370                 if ( $nickname === false )
4371                         $nickname = $user->getOption( 'nickname' );
4372
4373                 if ( is_null( $fancySig ) ) {
4374                         $fancySig = $user->getBoolOption( 'fancysig' );
4375                 }
4376
4377                 $nickname = $nickname == null ? $username : $nickname;
4378
4379                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4380                         $nickname = $username;
4381                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4382                 } elseif ( $fancySig !== false ) {
4383                         # Sig. might contain markup; validate this
4384                         if ( $this->validateSig( $nickname ) !== false ) {
4385                                 # Validated; clean up (if needed) and return it
4386                                 return $this->cleanSig( $nickname, true );
4387                         } else {
4388                                 # Failed to validate; fall back to the default
4389                                 $nickname = $username;
4390                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4391                         }
4392                 }
4393
4394                 # Make sure nickname doesnt get a sig in a sig
4395                 $nickname = $this->cleanSigInSig( $nickname );
4396
4397                 # If we're still here, make it a link to the user page
4398                 $userText = wfEscapeWikiText( $username );
4399                 $nickText = wfEscapeWikiText( $nickname );
4400                 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4401
4402                 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()->title( $this->getTitle() )->text();
4403         }
4404
4405         /**
4406          * Check that the user's signature contains no bad XML
4407          *
4408          * @param $text String
4409          * @return mixed An expanded string, or false if invalid.
4410          */
4411         function validateSig( $text ) {
4412                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4413         }
4414
4415         /**
4416          * Clean up signature text
4417          *
4418          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4419          * 2) Substitute all transclusions
4420          *
4421          * @param $text String
4422          * @param $parsing Whether we're cleaning (preferences save) or parsing
4423          * @return String: signature text
4424          */
4425         function cleanSig( $text, $parsing = false ) {
4426                 if ( !$parsing ) {
4427                         global $wgTitle;
4428                         $this->mOptions = new ParserOptions;
4429                         $this->clearState();
4430                         $this->setTitle( $wgTitle );
4431                         $this->setOutputType = self::OT_PREPROCESS;
4432                 }
4433
4434                 # Option to disable this feature
4435                 if ( !$this->mOptions->getCleanSignatures() ) {
4436                         return $text;
4437                 }
4438
4439                 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4440                 #  => Move this logic to braceSubstitution()
4441                 $substWord = MagicWord::get( 'subst' );
4442                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4443                 $substText = '{{' . $substWord->getSynonym( 0 );
4444
4445                 $text = preg_replace( $substRegex, $substText, $text );
4446                 $text = $this->cleanSigInSig( $text );
4447                 $dom = $this->preprocessToDom( $text );
4448                 $frame = $this->getPreprocessor()->newFrame();
4449                 $text = $frame->expand( $dom );
4450
4451                 if ( !$parsing ) {
4452                         $text = $this->mStripState->unstripBoth( $text );
4453                 }
4454
4455                 return $text;
4456         }
4457
4458         /**
4459          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4460          *
4461          * @param $text String
4462          * @return String: signature text with /~{3,5}/ removed
4463          */
4464         function cleanSigInSig( $text ) {
4465                 $text = preg_replace( '/~{3,5}/', '', $text );
4466                 return $text;
4467         }
4468
4469         /**
4470          * Set up some variables which are usually set up in parse()
4471          * so that an external function can call some class members with confidence
4472          */
4473         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4474                 $this->startParse( $title, $options, $outputType, $clearState );
4475         }
4476
4477         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4478                 $this->setTitle( $title );
4479                 $this->mOptions = $options;
4480                 $this->setOutputType( $outputType );
4481                 if ( $clearState ) {
4482                         $this->clearState();
4483                 }
4484         }
4485
4486         /**
4487          * Wrapper for preprocess()
4488          *
4489          * @param $text String: the text to preprocess
4490          * @param $options ParserOptions: options
4491          * @param $title Title object or null to use $wgTitle
4492          * @return String
4493          */
4494         public function transformMsg( $text, $options, $title = null ) {
4495                 static $executing = false;
4496
4497                 # Guard against infinite recursion
4498                 if ( $executing ) {
4499                         return $text;
4500                 }
4501                 $executing = true;
4502
4503                 wfProfileIn( __METHOD__ );
4504                 if ( !$title ) {
4505                         global $wgTitle;
4506                         $title = $wgTitle;
4507                 }
4508                 if ( !$title ) {
4509                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4510                         # Create a ghost title in such case
4511                         $title = Title::newFromText( 'Dwimmerlaik' );
4512                 }
4513                 $text = $this->preprocess( $text, $title, $options );
4514
4515                 $executing = false;
4516                 wfProfileOut( __METHOD__ );
4517                 return $text;
4518         }
4519
4520         /**
4521          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4522          * The callback should have the following form:
4523          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4524          *
4525          * Transform and return $text. Use $parser for any required context, e.g. use
4526          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4527          *
4528          * Hooks may return extended information by returning an array, of which the
4529          * first numbered element (index 0) must be the return string, and all other
4530          * entries are extracted into local variables within an internal function
4531          * in the Parser class.
4532          *
4533          * This interface (introduced r61913) appears to be undocumented, but
4534          * 'markerName' is used by some core tag hooks to override which strip
4535          * array their results are placed in. **Use great caution if attempting
4536          * this interface, as it is not documented and injudicious use could smash
4537          * private variables.**
4538          *
4539          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4540          * @param $callback Mixed: the callback function (and object) to use for the tag
4541          * @return The old value of the mTagHooks array associated with the hook
4542          */
4543         public function setHook( $tag, $callback ) {
4544                 $tag = strtolower( $tag );
4545                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4546                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4547                 $this->mTagHooks[$tag] = $callback;
4548                 if ( !in_array( $tag, $this->mStripList ) ) {
4549                         $this->mStripList[] = $tag;
4550                 }
4551
4552                 return $oldVal;
4553         }
4554
4555         /**
4556          * As setHook(), but letting the contents be parsed.
4557          *
4558          * Transparent tag hooks are like regular XML-style tag hooks, except they
4559          * operate late in the transformation sequence, on HTML instead of wikitext.
4560          *
4561          * This is probably obsoleted by things dealing with parser frames?
4562          * The only extension currently using it is geoserver.
4563          *
4564          * @since 1.10
4565          * @todo better document or deprecate this
4566          *
4567          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4568          * @param $callback Mixed: the callback function (and object) to use for the tag
4569          * @return The old value of the mTagHooks array associated with the hook
4570          */
4571         function setTransparentTagHook( $tag, $callback ) {
4572                 $tag = strtolower( $tag );
4573                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4574                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4575                 $this->mTransparentTagHooks[$tag] = $callback;
4576
4577                 return $oldVal;
4578         }
4579
4580         /**
4581          * Remove all tag hooks
4582          */
4583         function clearTagHooks() {
4584                 $this->mTagHooks = array();
4585                 $this->mStripList = $this->mDefaultStripList;
4586         }
4587
4588         /**
4589          * Remove a specific tag hook. Should not be called on $wgParser.
4590          * Does not change the strip list.
4591          *
4592          * @param string $tag
4593          * @return void
4594          */
4595         function clearTagHook( $tag ) {
4596                 if ( isset( $this->mTagHooks[$tag] ) ) {
4597                         unset( $this->mTagHooks[$tag] );
4598                 }
4599         }
4600
4601         /**
4602          * Create a function, e.g. {{sum:1|2|3}}
4603          * The callback function should have the form:
4604          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4605          *
4606          * Or with SFH_OBJECT_ARGS:
4607          *    function myParserFunction( $parser, $frame, $args ) { ... }
4608          *
4609          * The callback may either return the text result of the function, or an array with the text
4610          * in element 0, and a number of flags in the other elements. The names of the flags are
4611          * specified in the keys. Valid flags are:
4612          *   found                     The text returned is valid, stop processing the template. This
4613          *                             is on by default.
4614          *   nowiki                    Wiki markup in the return value should be escaped
4615          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4616          *
4617          * @param $id String: The magic word ID
4618          * @param $callback Mixed: the callback function (and object) to use
4619          * @param $flags Integer: a combination of the following flags:
4620          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4621          *
4622          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4623          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4624          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4625          *     the arguments, and to control the way they are expanded.
4626          *
4627          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4628          *     arguments, for instance:
4629          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4630          *
4631          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4632          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4633          *     working if/when this is changed.
4634          *
4635          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4636          *     expansion.
4637          *
4638          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4639          *     about the methods available in PPFrame and PPNode.
4640          *
4641          * @return The old callback function for this name, if any
4642          */
4643         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4644                 global $wgContLang;
4645
4646                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4647                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4648
4649                 # Add to function cache
4650                 $mw = MagicWord::get( $id );
4651                 if ( !$mw )
4652                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4653
4654                 $synonyms = $mw->getSynonyms();
4655                 $sensitive = intval( $mw->isCaseSensitive() );
4656
4657                 foreach ( $synonyms as $syn ) {
4658                         # Case
4659                         if ( !$sensitive ) {
4660                                 $syn = $wgContLang->lc( $syn );
4661                         }
4662                         # Add leading hash
4663                         if ( !( $flags & SFH_NO_HASH ) ) {
4664                                 $syn = '#' . $syn;
4665                         }
4666                         # Remove trailing colon
4667                         if ( substr( $syn, -1, 1 ) === ':' ) {
4668                                 $syn = substr( $syn, 0, -1 );
4669                         }
4670                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4671                 }
4672                 return $oldVal;
4673         }
4674
4675         /**
4676          * Get all registered function hook identifiers
4677          *
4678          * @return Array
4679          */
4680         function getFunctionHooks() {
4681                 return array_keys( $this->mFunctionHooks );
4682         }
4683
4684         /**
4685          * Create a tag function, e.g. <test>some stuff</test>.
4686          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4687          * Unlike parser functions, their content is not preprocessed.
4688          */
4689         function setFunctionTagHook( $tag, $callback, $flags ) {
4690                 $tag = strtolower( $tag );
4691                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4692                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4693                         $this->mFunctionTagHooks[$tag] : null;
4694                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4695
4696                 if ( !in_array( $tag, $this->mStripList ) ) {
4697                         $this->mStripList[] = $tag;
4698                 }
4699
4700                 return $old;
4701         }
4702
4703         /**
4704          * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
4705          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4706          * Placeholders created in Skin::makeLinkObj()
4707          * Returns an array of link CSS classes, indexed by PDBK.
4708          */
4709         function replaceLinkHolders( &$text, $options = 0 ) {
4710                 return $this->mLinkHolders->replace( $text );
4711         }
4712
4713         /**
4714          * Replace <!--LINK--> link placeholders with plain text of links
4715          * (not HTML-formatted).
4716          *
4717          * @param $text String
4718          * @return String
4719          */
4720         function replaceLinkHoldersText( $text ) {
4721                 return $this->mLinkHolders->replaceText( $text );
4722         }
4723
4724         /**
4725          * Renders an image gallery from a text with one line per image.
4726          * text labels may be given by using |-style alternative text. E.g.
4727          *   Image:one.jpg|The number "1"
4728          *   Image:tree.jpg|A tree
4729          * given as text will return the HTML of a gallery with two images,
4730          * labeled 'The number "1"' and
4731          * 'A tree'.
4732          *
4733          * @param string $text
4734          * @param array $param
4735          * @return string HTML
4736          */
4737         function renderImageGallery( $text, $params ) {
4738                 $ig = new ImageGallery();
4739                 $ig->setContextTitle( $this->mTitle );
4740                 $ig->setShowBytes( false );
4741                 $ig->setShowFilename( false );
4742                 $ig->setParser( $this );
4743                 $ig->setHideBadImages();
4744                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4745
4746                 if ( isset( $params['showfilename'] ) ) {
4747                         $ig->setShowFilename( true );
4748                 } else {
4749                         $ig->setShowFilename( false );
4750                 }
4751                 if ( isset( $params['caption'] ) ) {
4752                         $caption = $params['caption'];
4753                         $caption = htmlspecialchars( $caption );
4754                         $caption = $this->replaceInternalLinks( $caption );
4755                         $ig->setCaptionHtml( $caption );
4756                 }
4757                 if ( isset( $params['perrow'] ) ) {
4758                         $ig->setPerRow( $params['perrow'] );
4759                 }
4760                 if ( isset( $params['widths'] ) ) {
4761                         $ig->setWidths( $params['widths'] );
4762                 }
4763                 if ( isset( $params['heights'] ) ) {
4764                         $ig->setHeights( $params['heights'] );
4765                 }
4766
4767                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4768
4769                 $lines = StringUtils::explode( "\n", $text );
4770                 foreach ( $lines as $line ) {
4771                         # match lines like these:
4772                         # Image:someimage.jpg|This is some image
4773                         $matches = array();
4774                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4775                         # Skip empty lines
4776                         if ( count( $matches ) == 0 ) {
4777                                 continue;
4778                         }
4779
4780                         if ( strpos( $matches[0], '%' ) !== false ) {
4781                                 $matches[1] = rawurldecode( $matches[1] );
4782                         }
4783                         $title = Title::newFromText( $matches[1], NS_FILE );
4784                         if ( is_null( $title ) ) {
4785                                 # Bogus title. Ignore these so we don't bomb out later.
4786                                 continue;
4787                         }
4788
4789                         $label = '';
4790                         $alt = '';
4791                         if ( isset( $matches[3] ) ) {
4792                                 // look for an |alt= definition while trying not to break existing
4793                                 // captions with multiple pipes (|) in it, until a more sensible grammar
4794                                 // is defined for images in galleries
4795
4796                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4797                                 $altmatches = StringUtils::explode('|', $matches[3]);
4798                                 $magicWordAlt = MagicWord::get( 'img_alt' );
4799
4800                                 foreach ( $altmatches as $altmatch ) {
4801                                         $match = $magicWordAlt->matchVariableStartToEnd( $altmatch );
4802                                         if ( $match ) {
4803                                                 $alt = $this->stripAltText( $match, false );
4804                                         }
4805                                         else {
4806                                                 // concatenate all other pipes
4807                                                 $label .= '|' . $altmatch;
4808                                         }
4809                                 }
4810                                 // remove the first pipe
4811                                 $label = substr( $label, 1 );
4812                         }
4813
4814                         $ig->add( $title, $label, $alt );
4815                 }
4816                 return $ig->toHTML();
4817         }
4818
4819         function getImageParams( $handler ) {
4820                 if ( $handler ) {
4821                         $handlerClass = get_class( $handler );
4822                 } else {
4823                         $handlerClass = '';
4824                 }
4825                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4826                         # Initialise static lists
4827                         static $internalParamNames = array(
4828                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4829                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4830                                         'bottom', 'text-bottom' ),
4831                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4832                                         'upright', 'border', 'link', 'alt' ),
4833                         );
4834                         static $internalParamMap;
4835                         if ( !$internalParamMap ) {
4836                                 $internalParamMap = array();
4837                                 foreach ( $internalParamNames as $type => $names ) {
4838                                         foreach ( $names as $name ) {
4839                                                 $magicName = str_replace( '-', '_', "img_$name" );
4840                                                 $internalParamMap[$magicName] = array( $type, $name );
4841                                         }
4842                                 }
4843                         }
4844
4845                         # Add handler params
4846                         $paramMap = $internalParamMap;
4847                         if ( $handler ) {
4848                                 $handlerParamMap = $handler->getParamMap();
4849                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4850                                         $paramMap[$magic] = array( 'handler', $paramName );
4851                                 }
4852                         }
4853                         $this->mImageParams[$handlerClass] = $paramMap;
4854                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4855                 }
4856                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4857         }
4858
4859         /**
4860          * Parse image options text and use it to make an image
4861          *
4862          * @param $title Title
4863          * @param $options String
4864          * @param $holders LinkHolderArray
4865          * @return string HTML
4866          */
4867         function makeImage( $title, $options, $holders = false ) {
4868                 # Check if the options text is of the form "options|alt text"
4869                 # Options are:
4870                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4871                 #  * left       no resizing, just left align. label is used for alt= only
4872                 #  * right      same, but right aligned
4873                 #  * none       same, but not aligned
4874                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4875                 #  * center     center the image
4876                 #  * frame      Keep original image size, no magnify-button.
4877                 #  * framed     Same as "frame"
4878                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4879                 #  * upright    reduce width for upright images, rounded to full __0 px
4880                 #  * border     draw a 1px border around the image
4881                 #  * alt        Text for HTML alt attribute (defaults to empty)
4882                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4883                 # vertical-align values (no % or length right now):
4884                 #  * baseline
4885                 #  * sub
4886                 #  * super
4887                 #  * top
4888                 #  * text-top
4889                 #  * middle
4890                 #  * bottom
4891                 #  * text-bottom
4892
4893                 $parts = StringUtils::explode( "|", $options );
4894
4895                 # Give extensions a chance to select the file revision for us
4896                 $time = $sha1 = $descQuery = false;
4897                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
4898                         array( $this, $title, &$time, &$sha1, &$descQuery ) );
4899                 # Fetch and register the file (file title may be different via hooks)
4900                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $time, $sha1 );
4901
4902                 # Get parameter map
4903                 $handler = $file ? $file->getHandler() : false;
4904
4905                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4906
4907                 if ( !$file ) {
4908                         $this->addTrackingCategory( 'broken-file-category' );
4909                 }
4910
4911                 # Process the input parameters
4912                 $caption = '';
4913                 $params = array( 'frame' => array(), 'handler' => array(),
4914                         'horizAlign' => array(), 'vertAlign' => array() );
4915                 foreach ( $parts as $part ) {
4916                         $part = trim( $part );
4917                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4918                         $validated = false;
4919                         if ( isset( $paramMap[$magicName] ) ) {
4920                                 list( $type, $paramName ) = $paramMap[$magicName];
4921
4922                                 # Special case; width and height come in one variable together
4923                                 if ( $type === 'handler' && $paramName === 'width' ) {
4924                                         $m = array();
4925                                         # (bug 13500) In both cases (width/height and width only),
4926                                         # permit trailing "px" for backward compatibility.
4927                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4928                                                 $width = intval( $m[1] );
4929                                                 $height = intval( $m[2] );
4930                                                 if ( $handler->validateParam( 'width', $width ) ) {
4931                                                         $params[$type]['width'] = $width;
4932                                                         $validated = true;
4933                                                 }
4934                                                 if ( $handler->validateParam( 'height', $height ) ) {
4935                                                         $params[$type]['height'] = $height;
4936                                                         $validated = true;
4937                                                 }
4938                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4939                                                 $width = intval( $value );
4940                                                 if ( $handler->validateParam( 'width', $width ) ) {
4941                                                         $params[$type]['width'] = $width;
4942                                                         $validated = true;
4943                                                 }
4944                                         } # else no validation -- bug 13436
4945                                 } else {
4946                                         if ( $type === 'handler' ) {
4947                                                 # Validate handler parameter
4948                                                 $validated = $handler->validateParam( $paramName, $value );
4949                                         } else {
4950                                                 # Validate internal parameters
4951                                                 switch( $paramName ) {
4952                                                 case 'manualthumb':
4953                                                 case 'alt':
4954                                                         # @todo FIXME: Possibly check validity here for
4955                                                         # manualthumb? downstream behavior seems odd with
4956                                                         # missing manual thumbs.
4957                                                         $validated = true;
4958                                                         $value = $this->stripAltText( $value, $holders );
4959                                                         break;
4960                                                 case 'link':
4961                                                         $chars = self::EXT_LINK_URL_CLASS;
4962                                                         $prots = $this->mUrlProtocols;
4963                                                         if ( $value === '' ) {
4964                                                                 $paramName = 'no-link';
4965                                                                 $value = true;
4966                                                                 $validated = true;
4967                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4968                                                                 if ( preg_match( "/^($prots)$chars+$/u", $value, $m ) ) {
4969                                                                         $paramName = 'link-url';
4970                                                                         $this->mOutput->addExternalLink( $value );
4971                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4972                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4973                                                                         }
4974                                                                         $validated = true;
4975                                                                 }
4976                                                         } else {
4977                                                                 $linkTitle = Title::newFromText( $value );
4978                                                                 if ( $linkTitle ) {
4979                                                                         $paramName = 'link-title';
4980                                                                         $value = $linkTitle;
4981                                                                         $this->mOutput->addLink( $linkTitle );
4982                                                                         $validated = true;
4983                                                                 }
4984                                                         }
4985                                                         break;
4986                                                 default:
4987                                                         # Most other things appear to be empty or numeric...
4988                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4989                                                 }
4990                                         }
4991
4992                                         if ( $validated ) {
4993                                                 $params[$type][$paramName] = $value;
4994                                         }
4995                                 }
4996                         }
4997                         if ( !$validated ) {
4998                                 $caption = $part;
4999                         }
5000                 }
5001
5002                 # Process alignment parameters
5003                 if ( $params['horizAlign'] ) {
5004                         $params['frame']['align'] = key( $params['horizAlign'] );
5005                 }
5006                 if ( $params['vertAlign'] ) {
5007                         $params['frame']['valign'] = key( $params['vertAlign'] );
5008                 }
5009
5010                 $params['frame']['caption'] = $caption;
5011
5012                 # Will the image be presented in a frame, with the caption below?
5013                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5014                                                  isset( $params['frame']['framed'] ) ||
5015                                                  isset( $params['frame']['thumbnail'] ) ||
5016                                                  isset( $params['frame']['manualthumb'] );
5017
5018                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
5019                 # came to also set the caption, ordinary text after the image -- which
5020                 # makes no sense, because that just repeats the text multiple times in
5021                 # screen readers.  It *also* came to set the title attribute.
5022                 #
5023                 # Now that we have an alt attribute, we should not set the alt text to
5024                 # equal the caption: that's worse than useless, it just repeats the
5025                 # text.  This is the framed/thumbnail case.  If there's no caption, we
5026                 # use the unnamed parameter for alt text as well, just for the time be-
5027                 # ing, if the unnamed param is set and the alt param is not.
5028                 #
5029                 # For the future, we need to figure out if we want to tweak this more,
5030                 # e.g., introducing a title= parameter for the title; ignoring the un-
5031                 # named parameter entirely for images without a caption; adding an ex-
5032                 # plicit caption= parameter and preserving the old magic unnamed para-
5033                 # meter for BC; ...
5034                 if ( $imageIsFramed ) { # Framed image
5035                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5036                                 # No caption or alt text, add the filename as the alt text so
5037                                 # that screen readers at least get some description of the image
5038                                 $params['frame']['alt'] = $title->getText();
5039                         }
5040                         # Do not set $params['frame']['title'] because tooltips don't make sense
5041                         # for framed images
5042                 } else { # Inline image
5043                         if ( !isset( $params['frame']['alt'] ) ) {
5044                                 # No alt text, use the "caption" for the alt text
5045                                 if ( $caption !== '') {
5046                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5047                                 } else {
5048                                         # No caption, fall back to using the filename for the
5049                                         # alt text
5050                                         $params['frame']['alt'] = $title->getText();
5051                                 }
5052                         }
5053                         # Use the "caption" for the tooltip text
5054                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5055                 }
5056
5057                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
5058
5059                 # Linker does the rest
5060                 $ret = Linker::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
5061                         $time, $descQuery, $this->mOptions->getThumbSize() );
5062
5063                 # Give the handler a chance to modify the parser object
5064                 if ( $handler ) {
5065                         $handler->parserTransformHook( $this, $file );
5066                 }
5067
5068                 return $ret;
5069         }
5070
5071         /**
5072          * @param $caption
5073          * @param $holders LinkHolderArray
5074          * @return mixed|String
5075          */
5076         protected function stripAltText( $caption, $holders ) {
5077                 # Strip bad stuff out of the title (tooltip).  We can't just use
5078                 # replaceLinkHoldersText() here, because if this function is called
5079                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5080                 if ( $holders ) {
5081                         $tooltip = $holders->replaceText( $caption );
5082                 } else {
5083                         $tooltip = $this->replaceLinkHoldersText( $caption );
5084                 }
5085
5086                 # make sure there are no placeholders in thumbnail attributes
5087                 # that are later expanded to html- so expand them now and
5088                 # remove the tags
5089                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5090                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5091
5092                 return $tooltip;
5093         }
5094
5095         /**
5096          * Set a flag in the output object indicating that the content is dynamic and
5097          * shouldn't be cached.
5098          */
5099         function disableCache() {
5100                 wfDebug( "Parser output marked as uncacheable.\n" );
5101                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5102                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5103         }
5104
5105         /**
5106          * Callback from the Sanitizer for expanding items found in HTML attribute
5107          * values, so they can be safely tested and escaped.
5108          *
5109          * @param $text String
5110          * @param $frame PPFrame
5111          * @return String
5112          */
5113         function attributeStripCallback( &$text, $frame = false ) {
5114                 $text = $this->replaceVariables( $text, $frame );
5115                 $text = $this->mStripState->unstripBoth( $text );
5116                 return $text;
5117         }
5118
5119         /**
5120          * Accessor
5121          *
5122          * @return array
5123          */
5124         function getTags() {
5125                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
5126         }
5127
5128         /**
5129          * Replace transparent tags in $text with the values given by the callbacks.
5130          *
5131          * Transparent tag hooks are like regular XML-style tag hooks, except they
5132          * operate late in the transformation sequence, on HTML instead of wikitext.
5133          *
5134          * @param $text string
5135          *
5136          * @return string
5137          */
5138         function replaceTransparentTags( $text ) {
5139                 $matches = array();
5140                 $elements = array_keys( $this->mTransparentTagHooks );
5141                 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5142                 $replacements = array();
5143
5144                 foreach ( $matches as $marker => $data ) {
5145                         list( $element, $content, $params, $tag ) = $data;
5146                         $tagName = strtolower( $element );
5147                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5148                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5149                         } else {
5150                                 $output = $tag;
5151                         }
5152                         $replacements[$marker] = $output;
5153                 }
5154                 return strtr( $text, $replacements );
5155         }
5156
5157         /**
5158          * Break wikitext input into sections, and either pull or replace
5159          * some particular section's text.
5160          *
5161          * External callers should use the getSection and replaceSection methods.
5162          *
5163          * @param $text String: Page wikitext
5164          * @param $section String: a section identifier string of the form:
5165          *   <flag1> - <flag2> - ... - <section number>
5166          *
5167          * Currently the only recognised flag is "T", which means the target section number
5168          * was derived during a template inclusion parse, in other words this is a template
5169          * section edit link. If no flags are given, it was an ordinary section edit link.
5170          * This flag is required to avoid a section numbering mismatch when a section is
5171          * enclosed by <includeonly> (bug 6563).
5172          *
5173          * The section number 0 pulls the text before the first heading; other numbers will
5174          * pull the given section along with its lower-level subsections. If the section is
5175          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5176          *
5177          * @param $mode String: one of "get" or "replace"
5178          * @param $newText String: replacement text for section data.
5179          * @return String: for "get", the extracted section text.
5180          *                 for "replace", the whole page with the section replaced.
5181          *                 If the page is empty and section 0 is requested, $text (as '')
5182          *                  is returned
5183          */
5184         private function extractSections( $text, $section, $mode, $newText='' ) {
5185                 global $wgTitle; # not generally used but removes an ugly failure mode
5186                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5187                 $outText = '';
5188                 $frame = $this->getPreprocessor()->newFrame();
5189
5190                 # Process section extraction flags
5191                 $flags = 0;
5192                 $sectionParts = explode( '-', $section );
5193                 $sectionIndex = array_pop( $sectionParts );
5194                 foreach ( $sectionParts as $part ) {
5195                         if ( $part === 'T' ) {
5196                                 $flags |= self::PTD_FOR_INCLUSION;
5197                         }
5198                 }
5199                 # Preprocess the text
5200                 $root = $this->preprocessToDom( $text, $flags );
5201
5202                 # <h> nodes indicate section breaks
5203                 # They can only occur at the top level, so we can find them by iterating the root's children
5204                 $node = $root->getFirstChild();
5205
5206                 # Find the target section
5207                 if ( $sectionIndex == 0 ) {
5208                         # Section zero doesn't nest, level=big
5209                         $targetLevel = 1000;
5210                         if ( !$node ) {
5211                                 # The page definitely exists - we checked that earlier - so it must be blank: see bug #14005
5212                                 return $text;
5213                         }
5214                 } else {
5215                         while ( $node ) {
5216                                 if ( $node->getName() === 'h' ) {
5217                                         $bits = $node->splitHeading();
5218                                         if ( $bits['i'] == $sectionIndex ) {
5219                                                 $targetLevel = $bits['level'];
5220                                                 break;
5221                                         }
5222                                 }
5223                                 if ( $mode === 'replace' ) {
5224                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5225                                 }
5226                                 $node = $node->getNextSibling();
5227                         }
5228                 }
5229
5230                 if ( !$node ) {
5231                         # Not found
5232                         if ( $mode === 'get' ) {
5233                                 return $newText;
5234                         } else {
5235                                 return $text;
5236                         }
5237                 }
5238
5239                 # Find the end of the section, including nested sections
5240                 do {
5241                         if ( $node->getName() === 'h' ) {
5242                                 $bits = $node->splitHeading();
5243                                 $curLevel = $bits['level'];
5244                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5245                                         break;
5246                                 }
5247                         }
5248                         if ( $mode === 'get' ) {
5249                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5250                         }
5251                         $node = $node->getNextSibling();
5252                 } while ( $node );
5253
5254                 # Write out the remainder (in replace mode only)
5255                 if ( $mode === 'replace' ) {
5256                         # Output the replacement text
5257                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5258                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5259                         # Only add trailing whitespace if there is newText
5260                         if ( $newText != "" ) {
5261                                 $outText .= $newText . "\n\n";
5262                         }
5263
5264                         while ( $node ) {
5265                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5266                                 $node = $node->getNextSibling();
5267                         }
5268                 }
5269
5270                 if ( is_string( $outText ) ) {
5271                         # Re-insert stripped tags
5272                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5273                 }
5274
5275                 return $outText;
5276         }
5277
5278         /**
5279          * This function returns the text of a section, specified by a number ($section).
5280          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5281          * the first section before any such heading (section 0).
5282          *
5283          * If a section contains subsections, these are also returned.
5284          *
5285          * @param $text String: text to look in
5286          * @param $section String: section identifier
5287          * @param $deftext String: default to return if section is not found
5288          * @return string text of the requested section
5289          */
5290         public function getSection( $text, $section, $deftext='' ) {
5291                 return $this->extractSections( $text, $section, "get", $deftext );
5292         }
5293
5294         /**
5295          * This function returns $oldtext after the content of the section
5296          * specified by $section has been replaced with $text.
5297          *
5298          * @param $oldtext String: former text of the article
5299          * @param $section Numeric: section identifier
5300          * @param $text String: replacing text
5301          * @return String: modified text
5302          */
5303         public function replaceSection( $oldtext, $section, $text ) {
5304                 return $this->extractSections( $oldtext, $section, "replace", $text );
5305         }
5306
5307         /**
5308          * Get the ID of the revision we are parsing
5309          *
5310          * @return Mixed: integer or null
5311          */
5312         function getRevisionId() {
5313                 return $this->mRevisionId;
5314         }
5315
5316         /**
5317          * Get the revision object for $this->mRevisionId
5318          *
5319          * @return Revision|null either a Revision object or null
5320          */
5321         protected function getRevisionObject() {
5322                 if ( !is_null( $this->mRevisionObject ) ) {
5323                         return $this->mRevisionObject;
5324                 }
5325                 if ( is_null( $this->mRevisionId ) ) {
5326                         return null;
5327                 }
5328
5329                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5330                 return $this->mRevisionObject;
5331         }
5332
5333         /**
5334          * Get the timestamp associated with the current revision, adjusted for
5335          * the default server-local timestamp
5336          */
5337         function getRevisionTimestamp() {
5338                 if ( is_null( $this->mRevisionTimestamp ) ) {
5339                         wfProfileIn( __METHOD__ );
5340
5341                         $revObject = $this->getRevisionObject();
5342                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5343
5344                         if( $timestamp !== false ) {
5345                                 global $wgContLang;
5346
5347                                 # The cryptic '' timezone parameter tells to use the site-default
5348                                 # timezone offset instead of the user settings.
5349                                 #
5350                                 # Since this value will be saved into the parser cache, served
5351                                 # to other users, and potentially even used inside links and such,
5352                                 # it needs to be consistent for all visitors.
5353                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5354                         }
5355
5356                         wfProfileOut( __METHOD__ );
5357                 }
5358                 return $this->mRevisionTimestamp;
5359         }
5360
5361         /**
5362          * Get the name of the user that edited the last revision
5363          *
5364          * @return String: user name
5365          */
5366         function getRevisionUser() {
5367                 if( is_null( $this->mRevisionUser ) ) {
5368                         $revObject = $this->getRevisionObject();
5369
5370                         # if this template is subst: the revision id will be blank,
5371                         # so just use the current user's name
5372                         if( $revObject ) {
5373                                 $this->mRevisionUser = $revObject->getUserText();
5374                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5375                                 $this->mRevisionUser = $this->getUser()->getName();
5376                         }
5377                 }
5378                 return $this->mRevisionUser;
5379         }
5380
5381         /**
5382          * Mutator for $mDefaultSort
5383          *
5384          * @param $sort New value
5385          */
5386         public function setDefaultSort( $sort ) {
5387                 $this->mDefaultSort = $sort;
5388                 $this->mOutput->setProperty( 'defaultsort', $sort );
5389         }
5390
5391         /**
5392          * Accessor for $mDefaultSort
5393          * Will use the empty string if none is set.
5394          *
5395          * This value is treated as a prefix, so the
5396          * empty string is equivalent to sorting by
5397          * page name.
5398          *
5399          * @return string
5400          */
5401         public function getDefaultSort() {
5402                 if ( $this->mDefaultSort !== false ) {
5403                         return $this->mDefaultSort;
5404                 } else {
5405                         return '';
5406                 }
5407         }
5408
5409         /**
5410          * Accessor for $mDefaultSort
5411          * Unlike getDefaultSort(), will return false if none is set
5412          *
5413          * @return string or false
5414          */
5415         public function getCustomDefaultSort() {
5416                 return $this->mDefaultSort;
5417         }
5418
5419         /**
5420          * Try to guess the section anchor name based on a wikitext fragment
5421          * presumably extracted from a heading, for example "Header" from
5422          * "== Header ==".
5423          *
5424          * @param $text string
5425          *
5426          * @return string
5427          */
5428         public function guessSectionNameFromWikiText( $text ) {
5429                 # Strip out wikitext links(they break the anchor)
5430                 $text = $this->stripSectionName( $text );
5431                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5432                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5433         }
5434
5435         /**
5436          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5437          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5438          * as something other than UTF-8 (apparently?), resulting in breakage.
5439          *
5440          * @param $text String: The section name
5441          * @return string An anchor
5442          */
5443         public function guessLegacySectionNameFromWikiText( $text ) {
5444                 # Strip out wikitext links(they break the anchor)
5445                 $text = $this->stripSectionName( $text );
5446                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5447                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5448         }
5449
5450         /**
5451          * Strips a text string of wikitext for use in a section anchor
5452          *
5453          * Accepts a text string and then removes all wikitext from the
5454          * string and leaves only the resultant text (i.e. the result of
5455          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5456          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5457          * to create valid section anchors by mimicing the output of the
5458          * parser when headings are parsed.
5459          *
5460          * @param $text String: text string to be stripped of wikitext
5461          * for use in a Section anchor
5462          * @return Filtered text string
5463          */
5464         public function stripSectionName( $text ) {
5465                 # Strip internal link markup
5466                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5467                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5468
5469                 # Strip external link markup
5470                 # @todo FIXME: Not tolerant to blank link text
5471                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5472                 # on how many empty links there are on the page - need to figure that out.
5473                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5474
5475                 # Parse wikitext quotes (italics & bold)
5476                 $text = $this->doQuotes( $text );
5477
5478                 # Strip HTML tags
5479                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5480                 return $text;
5481         }
5482
5483         /**
5484          * strip/replaceVariables/unstrip for preprocessor regression testing
5485          *
5486          * @return string
5487          */
5488         function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5489                 $this->startParse( $title, $options, $outputType, true );
5490
5491                 $text = $this->replaceVariables( $text );
5492                 $text = $this->mStripState->unstripBoth( $text );
5493                 $text = Sanitizer::removeHTMLtags( $text );
5494                 return $text;
5495         }
5496
5497         function testPst( $text, Title $title, ParserOptions $options ) {
5498                 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5499         }
5500
5501         function testPreprocess( $text, Title $title, ParserOptions $options ) {
5502                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5503         }
5504
5505         /**
5506          * Call a callback function on all regions of the given text that are not
5507          * inside strip markers, and replace those regions with the return value
5508          * of the callback. For example, with input:
5509          *
5510          *  aaa<MARKER>bbb
5511          *
5512          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5513          * two strings will be replaced with the value returned by the callback in
5514          * each case.
5515          *
5516          * @return string
5517          */
5518         function markerSkipCallback( $s, $callback ) {
5519                 $i = 0;
5520                 $out = '';
5521                 while ( $i < strlen( $s ) ) {
5522                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5523                         if ( $markerStart === false ) {
5524                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5525                                 break;
5526                         } else {
5527                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5528                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5529                                 if ( $markerEnd === false ) {
5530                                         $out .= substr( $s, $markerStart );
5531                                         break;
5532                                 } else {
5533                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5534                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5535                                         $i = $markerEnd;
5536                                 }
5537                         }
5538                 }
5539                 return $out;
5540         }
5541
5542         /**
5543          * Save the parser state required to convert the given half-parsed text to
5544          * HTML. "Half-parsed" in this context means the output of
5545          * recursiveTagParse() or internalParse(). This output has strip markers
5546          * from replaceVariables (extensionSubstitution() etc.), and link
5547          * placeholders from replaceLinkHolders().
5548          *
5549          * Returns an array which can be serialized and stored persistently. This
5550          * array can later be loaded into another parser instance with
5551          * unserializeHalfParsedText(). The text can then be safely incorporated into
5552          * the return value of a parser hook.
5553          *
5554          * @return array
5555          */
5556         function serializeHalfParsedText( $text ) {
5557                 wfProfileIn( __METHOD__ );
5558                 $data = array(
5559                         'text' => $text,
5560                         'version' => self::HALF_PARSED_VERSION,
5561                         'stripState' => $this->mStripState->getSubState( $text ),
5562                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5563                 );
5564                 wfProfileOut( __METHOD__ );
5565                 return $data;
5566         }
5567
5568         /**
5569          * Load the parser state given in the $data array, which is assumed to
5570          * have been generated by serializeHalfParsedText(). The text contents is
5571          * extracted from the array, and its markers are transformed into markers
5572          * appropriate for the current Parser instance. This transformed text is
5573          * returned, and can be safely included in the return value of a parser
5574          * hook.
5575          *
5576          * If the $data array has been stored persistently, the caller should first
5577          * check whether it is still valid, by calling isValidHalfParsedText().
5578          *
5579          * @param $data Serialized data
5580          * @return String
5581          */
5582         function unserializeHalfParsedText( $data ) {
5583                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5584                         throw new MWException( __METHOD__.': invalid version' );
5585                 }
5586
5587                 # First, extract the strip state.
5588                 $texts = array( $data['text'] );
5589                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5590
5591                 # Now renumber links
5592                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5593
5594                 # Should be good to go.
5595                 return $texts[0];
5596         }
5597
5598         /**
5599          * Returns true if the given array, presumed to be generated by
5600          * serializeHalfParsedText(), is compatible with the current version of the
5601          * parser.
5602          *
5603          * @param $data Array
5604          *
5605          * @return bool
5606          */
5607         function isValidHalfParsedText( $data ) {
5608                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5609         }
5610 }